Effect of Test Dust on Performance Test for a Pleated Filter Cartridge

This study focuses on the test performance discrepancies for filter cartridges resulting from different kinds of test dust. A comprehensive experimental set-up including pulse-jet cleaning system for cleanable pleated filter cartridges was developed. Then, three identical cartridge samples were tested in this experimental set-up using different synthetic test dust, i.e., 325 mesh talcum powder, ASHRAE dust and ISO A2 dust respectively. Results showed that the type of test dust had significant impacts on dust holding capacity and cleaning effectiveness of pulse-jet cleaned cartridges. Cartridge sample loaded with ASHRAE dust has the smallest dust holding capacity and the worst cleaning effectiveness. However, considering its short experimental period, size distribution and actual simulation effects of atmospheric aerosol, ASHRAE dust is still recommended.


INTRODUCTION
Pleated filter cartridges have been widely applied to control particulate entrance and emission in many fields such as general ventilation dust collection, thermal power generation and marine power plant.Cleanablity is an important feature of pleated filter cartridge.After a certain service time, the filter cartridges can be cleaned to keep pressure drop within a reasonable range.Mechanically shaking, reverse airflow and pulse-jet are the common cleaning methods and currently pulse-jet cleaning is the most widely used.
The pulse-jet cleaning method was proposed first in 1950s, applied to filter bags.Design cleaning parameters and operation conditions would have a direct influence on the cleaning effectiveness and filtration performance.Through relevant tests, Humphries (1981) proved that the residual pressure drop and other residual parameters were strongly dependent on cloth structure of fabric filter bags.Humphries and Madden (1983) demonstrated that, the key factors in pulse-jet cleaning effectiveness were the overpressure and the duration attaining overpressure.Filter bags would have a better cleaning effectiveness with a higher overpressure and shorter cleaning time.Hindy (1986) investigated the effect of cleaning modes on the performance of fabric filter bags.Clean-on-time mode and clean-on-demand mode with different terminal pressure drop (1500 Pa and 2500 Pa) were respectively used in this study.The experimental results showed that the cleaning modes differed greatly in residual pressure drop and filtration cycle time.Lu and Tsai (1996, 1998, 1999) conducted numerical and experimental researches on pulse-jet cleaning process of fabric bag house.Results of these studies showed that tank volume, initial tank pressure, nozzle diameter, distance between nozzle and bag top, and pulse duration were major cleaning parameters influencing static pressure and pressure impulse in the bag, which contribute to filtration performance.It was also found that there was a critical value of the tank pressure and accordingly a critical effective residual resistance existed for the pulsejet cleaning system.Similar studies were also conducted by Theodore and Reynolds (1975), Rothwell (1986) and Simon et al. (2004), on the influence of these cleaning parameters in pulse-jet cleaning.In addition, Tanabe et al. (2011) tested three types of synthetic fabric filter (acrylic, polypropylene and polyester), using phosphate rock as test dust.It was observed that residual pressure drop of polypropylene filter was higher than other filters due to adhesion force, which made it hard to remove the dust cake.The cleaning method used is reverse airflow cleaning.
In respect of pulse-jet cleaning of pleated filter cartridges, Lo et al. (2010a, b) conducted a series of studies on pulsejet cleaning process of pleated filter cartridges.In the experimental study, six cartridges with different base media and geometrical dimensions were tested under clean-ontime mode and clean-on-demand mode.It was found that the pleat ratio had a great influence on cleaning performance of different cleaning mode.The tank pressure was critical for cleaning effectiveness, and conversely the pulse duration had a little impact, which agreed with the previous work of filter bags conducted by Lu and Tsai (1996, 1998, 1999).Then, a numerical model was established to observe transient flow behavior during pulse-jet cleaning through computational fluid dynamics (CFD) analysis, and the numerical results were validated by the experimental results.The numerical simulation result of surface static pressure distribution along filter cartridges showed that, the average static pressure on the surface were more closely correlated to cleaning efficiency than overpressure, and the top area of the filter cartridge was hard to clean because of the lower surface pressure generated by pulse-jet.Analogous conclusions were also drawn by Yan et al. (2013).They tested and recorded the instantaneous pressure drop and static pressure across pleated filter cartridges during pulsejet cleaning.The experimental results showed that, the static pressure on the filter surface decreased from the top area of the cartridges to the bottom area.
Different from other particulate air filters, the pleated filter cartridge is largely dependent on the filtration performance after cleaning.According to Mao et al. (2008), it is impossible to clear particles completely out of the fabric so that the initial properties would not reappear.Therefore, the test method including cleaning method and test loading dust, is vital to performance test result of pleated filter cartridges.However, the test methods and corresponding evaluation criteria based on general particulate air filters are not fully applicable to filter cartridges, and the related standards for the performance test of cleanable pleated filter cartridges are still in progress.According to VDI 3926, Testing of cleanable filter media -Standard test for the evaluation of cleanable filter media, aluminum oxide is used as the test loading dust.On the other hand, ISO A2 dust or ISO A4 dust is used based on ISO 5011:2014, Inlet air cleaning equipment for internal combustion engines and compressors -Performance testing.Limestone or titanium dioxide powder is proposed in ISO 29461-2, Air intake filter systems for rotary machinery -Test methods -Part 2: Cleanable (pulse jet) filter system (first draft).Different kinds of test dust are applied for various performance tests currently, however, this will lead to different test result even if the test filters are absolutely identical.There is no agreed test loading dust till now, therefore, it is necessary to conduct the research concentrating on test loading dust.
In retrospect, experimental studies of pleated filter cartridges and pulse-jet cleaning method have been extensively carried out, using different kinds of test dust.However, the comparison of different test loading dust was not included.The objective of this work is to investigate the test performance discrepancies for pleated filter cartridges induced by different test loading dust.

EXPERIMENTAL SET-UP
The comprehensive experimental set-up, shown in Fig. 1, is built up based on the test rig of EN 779:2012, Particulate air filters for general ventilation -Determination of the filtration performance.It consists of 3 sections: upstream duct section, sample installation duct section and downstream duct section.Measurements of filter pressure drop are taken with static taps between upstream and downstream measure points located in the duct wall.A dust hopper is installed on the bottom of sample installation section for collecting the loading dust divorced from filter surface during pulse-jet cleaning.Ash valves are also set up with the hopper to discharge waste loading dust after test procedure.Meanwhile, a final filter classified as F9 collects particles passing through the tested filter to calculate dust holding mass and also protects the orifice plate flow measuring device and fan behind.The variable frequency device (VFD) in the end controls air flow rate and keeps the duct in negative pressure condition, which is consistent with actual operating conditions of pleated filter cartridges.

Pulse-Jet Cleaning System
The schematic of pulse-jet cleaning system is shown in Fig. 2. According to the results of previous work, factors influencing the cleaning effectiveness include the performance of the pulse valves, the capacity of the tank, the types of nozzles, pulse interval, pulse width, pulse pressure etc.
In this study, pulse-jet valves with threaded ports (CA25T&CA35T of Goyen) are used.The pulse-jet valve is controlled by a valve control and also connected with compressed air tank and blowtube.When receiving signals from pulse controller, the integral piloted valve will open the pulse-jet valve and the compressed air passing through pulse-jet valve and blowtube resulting in secondary airflow for cleaning in the nozzle.An independent control system and another control unit integrated in main test program are designed for pulse-jet cleaning system.The independent control system is equipped along with valves, can be used to start the pulse-jet cleaning system separately without main program.While the control unit is the section for reverse pulse-jet in main test program, operated for comprehensive test in cooperation with other equipments.The pulse-jet valve can be opened on demand, on time or manually by both control systems.In addition, pulse width, pulse interval and other parameters can be adjusted.
Compressed air tank, providing constant airflow rate and pressure, is another important device of pulse-jet cleaning system.The capacity of tank is 36 L based on previous work and the design bearing pressure is 1 MPa.To regulate compressed air pressure accurately, the input pressure of tank can be adjusted by a high precision electronic pressure reducing valve.Besides, a HEPA filter is installed in compressed air supply system to protect pressure reducing valve and pulse-jet valves.
The type of nozzles is crucial to airflow uniformity inside filter cartridges.The nozzle composed of a supersonic induced nozzle (VN25 of Goyen) and a cone shape air diffuser (CC200 of Goyen) is installed at the end of blowtube.Distance from pleated filter cartridge to nozzle outlet is adjustable by changing blowtubes with different lengths, generally is 100 mm.

Other Experimental Apparatus
Polydisperse Aerosol DEHS (DiEthylHexylSebacate) is generated by a Laskin nozzle, which produces droplets in sufficient concentration in the size range of 0.2 µm to 3.0 µm.In the mixing chamber, the aerosol for efficiency testing is dispersed and mixed to create a uniform concentration.The aerosol generator (model GABRPY-1) with 8 Laskin nozzles is used in this set-up, which can generate maximum generation concentration of 10 7 P L -1 .The generation concentration can be regulated by switching nozzles.
A handheld optical particle counter (TSI 9306) is used to sample air and record particle concentration upstream and downstream of the filter consecutively.The measurement is done by means of two rigid sample lines of equal length and equivalent geometry.The experiments in this study only include pressure drop and dust loading test, accordingly the aerosol generator and optical particle counter for efficiency test will not be used.
The loading dust fed by the dust feeder is injected into the duct and supplied with airflow.Design dust feeding rate is 10 g min -1 with compressed air in the flow rate range of 80 to 200 L min -1 .

Qualification Test
In order to meet requirements of relevant standards, qualification tests are carried out according to those specified in EN779:2012 (shown in Table 1).

Pleated Filter Cartridge Sample
Three of the same pleated filter cartridges (i.e., filters A, B and C) were tested in this experiment.The detailed information of the tested filters and filter materials are shown in Tables 2 and 3 respectively.The rated air flow rate of all filter samples is 1200 m 3 h -1 and initial pressure drop is 153-165 Pa.

Test Dust
The synthetic loading dusts used in this study were 325 mesh talcum powder, ASHRAE (American Society of Heating Refrigerating and Air-Conditioning Engineers) dust and ISO A2 dust.Each loading dust was fed respectively Filter sample Nozzle   4).The 325 mesh talcum powder, widely used in performance test for bag house and other dust collector, is mainly composed of silica and magnesia.
The size of ambient atmospheric aerosols ranges from a few nanometers up to tens of microns, in which the proportion of particles smaller than 1 µm is about 99% by quantity and smaller than 10 µm is about 50-95% by volume.Furthermore, the statistical average residence time of particles in ambient air varies with particle diameter and chemical composition.The largest particles (sizes ≥ 2.5 µm) have the shortest lifetimes due to gravity settling.Particles in the 0.1 to 2.5 µm range have the longest lifetimes in the atmosphere from days to weeks.These particles are not removed effectively by diffusion or gravitational settling, and predominantly removed by precipitation (i.e., rain-out).Other particles smaller than 10 µm according to Table 4. Therefore the two types of dust provide an insufficient concentration at or below 0.5 µm to simulate ambient atmospheric aerosols adequately.By contrast, the ASHRAE dust consists partly of carbon powder at 0.1 to 1 µm, which makes the median diameter smaller than talcum powder and ISO A2 dust.Moreover, the particle size distribution of ASHRAE dust is close to bimodal distribution, which has the same distribution law with particles in ambient atmospheric aerosols.

Experimental Design
The experiments were carried out basically in accordance with EN779:2012 to investigate the test performance discrepancies for filter cartridges caused by different test dust.The test air flow rate was kept constant at 1200 m 3 h -1 .Each loading dust was fed at a concentration of 100 mg m -3 until 450 Pa pressure drop was reached and the dust holding capacity was measured.Clogging cartridges were cleaned through pulse-jet cleaning process.After cleaning, residual pressure drop and residual dust mass were recorded.Test filters reach the stable filtration period at fifth cycle, therefore eight filtration-cleaning cycles were chosen in this experiment.
There are two cleaning modes generally used for cleanable pleated filter cartridges, clean-on-demand and clean-on-time.During filtration phase, the pressure drop of tested filter increases with dust loading.In the mode of clean-on-demand, when the pressure drop increases to the pre-set termination pressure drop, pulse-jet cleaning system would be triggered and make the filter back to a lower resistance again.Similarly, in the clean-on-time mode, a pre-set time value of filtration phase function as the ending signal of dust feeding, as well as the beginning of pulse-jet cleaning.According to Lo et al. (2010), test results such as residual pressure drop and other parameters in the stable operation phase are almost constant in whatever clean-on-demand mode or clean-ontime mode.Considering unavoidable interruption during test and the time value of filtration phase is not constant, the clean-on-demand mode was used with clean-on-time mode being supplemented.
The termination pressure drop was set at 450 Pa in this study.Investigation result shows that, in practice, the termination pressure drop normally is about 400 Pa and will be updated after clogging according to real time resistance of filter cartridge.The 450 Pa termination resistance is closer to the actual applications.In addition, 450 Pa is also the specified termination resistance in EN779, so that it is reasonable to compare filter cartridges with those filters for general ventilation tested based on EN779.
The value of tank pressure and pulse width is chosen according to pre-test results.A pleated filter cartridge sample after several filtration-cleaning cycles was cleaned with three different levels of tank pressure (0.5, 0.6 and 0.7 MPa).The residual resistance at 0.6 MPa is significantly lower than that at 0.5 MPa, but only 1.4% higher than that at 0.7 MPa.A higher tank pressure results in better cleaning effectiveness but consumes more compressed air.For the filter sample, therefore, 0.6 MPa is selected as an appropriate tank pressure.On the other hand, the filter sample was cleaned with different pulse width (50, 100 and 200 ms) at different tank pressure.The test results did not show any obvious difference, therefore, 100 ms was taken as the standard pulse width based on the operation condition of pulse-jet valve.

Effect on Dust Holding Capacity
During filtration process, the pressure drop varies with dust loading due to dust collection and deposition.Fig. 3 shows the curves that pressure drop increases as a function of dust fed for clean filter.The growth of pressure drop is slow and similar at the beginning of dust loading for three clean samples.As dust fed get larger (> 500 g), pressure drop of all filters increases rapidly.The pressure drop of filter loaded with ASHRAE dust whose average particle sizes is relatively small rises most quickly.On the contrary, the pressure drop grows gently using 325 mesh talcum powder which consists of the largest particles.
This phenomenon can be illustrated by the theory of Dennis and Dirgo (1981).Dust cake was built up on the filter surface in filtration process due to dust loading.The experimental results showed that, pressure drop of fabric filter increases linearly with cumulative thickness of deposited dust on surface, and meanwhile the deposited dust thickness is approximately proportional to the areal density of loading dust on the fabric.In consequence, during the filtration stage in which the dust cake has been formed, the relationship between pressure drop and dust holding capacity can be expressed by Eq. (1).

K V M
where ∆M (kg) is added mass of loading dust on filter, V (m s -1 ) is filtration velocity, A (m 2 ) is filtration area, K 2 is the specific resistance of dust cake.
According to the theory of granular bed filtration, thickness of one layer of dust cake on the filter surface is approximately equal to diameter of particles in loading dust.In the airflow with uniform dust concentration, fine dust has a smaller thickness of one layer of dust cake, so that the number of particle layers grows more quickly and so does the total thickness.Therefore, the pressure drop increases sharply with accumulated dust cake.Fig. 4 shows the variation trend of K 2 along with dust loading for three filter samples.In accordance with the curves of pressure drop, K 2 of sample B increases fastest among the filter samples.The negative value for K2 in Fig. 4 occurs because of the negative value for ∆P.Actually, the inhomogeneous filter material and nonuniform distribution of pleats may cause filtration velocity uneven on surface, which makes local turbulence and also resistance.As dust deposited, the part with lower resistance and higher velocity will filter more dust and therefore grow faster in pressure drop, so that the airflow everywhere tends towards balanced.Accordingly, the pressure drop for three test filters decreases or remains more or less constant at the beginning of dust loading.As a result, filter cartridge loaded with ASHRAE dust has the smallest dust holding capacity with 840 g, compared with 3549 g for the sample loaded with 325 mesh talcum powder at 450 Pa.At the same air flow rate and dust feeding concentration, it takes the shortest time for ASHRAE dust to complete dust loading.
Besides particle diameter, different bulk density of loading dust may be the other cause for discrepancy in dust holding capacity.The bulk density of talcum powder is 590 to 710 kg m -3 (Xiang, 2013), compared with 900 kg m -3 for ISO A2 dust (ISO 12103-1) and 40 kg m -3 for carbon powder, although there is no significant difference among true density.The dust holding capacity per unit volume of filter cartridge can be calculated by dividing dust fed mass by bulk density.

Effect on Cleaning Effectiveness
Basically, two parameters, i.e., residual pressure drop and filtration cycle time imply the cleaning effectiveness of a pulse-jet-cleaned cartridge.
Residual pressure drop of all test filters have the similar trend and reach the stable filtration period at fifth pulse-jet cleaning.The highest average residual pressure drop is 227 Pa for sample B, while lowest is 172 Pa for sample A as showed in Fig. 5.In the stable operation phase, duration of the filtration cycle (i.e., duration of dust loading to 450 Pa) for the same filter is nearly constant.The longest average duration time is 18 hours (as shown in Fig. 6) for sample A.
It can also be found that change trend of residual pressure drop and filtration cycle duration time of pleated filter cartridge are similar for three test dusts, however, the change values are much different.In the clean-on-demand mode, the filtration cycle time is dependent on the residual pressure drop.Cartridge loaded with 325 mesh talcum powder has lowest average residual pressure drop and the longest average duration of filtration cycle.Therefore, the filter loaded with 325 mesh talcum powder has the best cleaning effectiveness, conversely the filter with ASHRAE dust has the worst.
To determine the suitable dust for performance test, the lower residual pressure drop and the shorter duration of filtration cycle are better.The lower residual pressure drop represents the better cleaning effectiveness, as well as the longer filtration cycle time.However, in reality, to shorten testing hours, the shorter duration of filtration cycle is more practical.Hence, ASHRAE dust is recommended because of its short filtration cycle duration.
The scanning electron microscope photographs of the filter media (as shown in Figs.7-9) help to explain the reason for discrepancies of cleaning effectiveness.
The internal details of clean filter material are presented in Fig. 7. Sub-microfibers covered on the filter material surface are in the fiber diameter range of 0.2 to 1 µm.
Fig. 8 shows the media fibers of sample A in the stable operation phase after several filtration-cleaning cycles.Particles of 325 mesh talcum powder are in the shape of flake, with smooth surface and irregular edges.After reverse pulse-jet cleaning, large particles were captured and accumulated on the filter surface, while particles around or below 1 µm stuck to fibers contributing to residual resistance greatly.Besides, there are few larger particles remaining at the junction of multiple fibers.Since the overlapping space is not part of airflow passage, these particles will not cause resistance to airflow obviously.
Likewise, the scanning electron microscope photograph of filter loaded with ASHRAE dust (Fig. 9) reveals that, there is a large quantity of small particles sticking to fibers and meanwhile some large particles on the edge of airflow passage.However, compared with the sub-microfibers, particles on fibers are much smaller, which means that the average diameter of particles adhering to fibers is much less than 1 µm.Based on the composition of ASHRAE dust, it can be deduced that the majority of these ultrafine particles are carbon powder.The dendritic structure on fibers developed by carbon powder particles is the reason why residual resistance of sample B is much higher than

CONCLUSIONS
This study developed a comprehensive experimental setup for pulse-jet cleaned pleated filter cartridges to test the performance during several filtration-cleaning cycles.Based on the previous studies and pre-test results, the test method and relevant set-point of experimental parameters such as tank pressure and pulse width were determined.The cleanon-demand mode with 450 Pa termination pressure drop was recommended.
Three same pleated filter cartridges were tested with different loading dust (i.e., 325 mesh talcum powder, ASHRAE dust and ISO A2 dust).It was found that the loading dust would affect dust holding capacity result as well as cleaning effectiveness of pleated filter cartridge samples.
At the termination resistance 450 Pa, filter cartridge loaded with ASHRAE dust has the smallest dust holding capacity while sample loaded with 325 mesh talcum powder has the largest.Particle diameter and bulk density of loading dust may be the cause for discrepancy in dust holding capacity.
In addition, cartridge loaded with 325 mesh talcum powder has lowest average residual pressure drop and the longest average duration of filtration cycle, which means that the filter sample has the best cleaning effectiveness.Conversely, ASHRAE dust is hardest to be removed due to carbon powder particles inside.
If possible, the dust to be disposed in practical collection is recommended as the test loading dust.Generally, when atmospheric dust or other dust with massive ultrafine particles is going to be collected, ASHRAE dust is better choice due to its size distribution and actual simulation effects.To determine the suitable dust for performance test, ASHRAE dust is recommended because of its short filtration cycle duration.

Fig. 3 .
Fig. 3. Pressure drop as a function of dust fed for clean filter.

Table 1 .
Summary of qualification test results.

Table 2 .
Characteristics of test filters.

Table 3 .
Characteristics of filter materials.

Table 4 .
Limited size distribution of ISO A2 dust.