Influence of Oil Content on Particle Loading Characteristics of a Two-Stage Filtration System

Abstract

Filter media may encounter aerosols mixed with solid and oil ingredients from various sources, such as industries, transportation, and households, in the air purification process, while the influence such oil content has on the loading performance of single-stage and two-stage filtration systems is under-reported. Thus, this study aims to evaluate oil fraction effects on the loading performance of single-stage and two-stage filtration systems. First, to reveal the oil–solid mixed particle deposition mechanisms, the filter media parameters, i.e., specific cake resistance $\mathsf{\epsilon }$and cake porosity ${\mathrm{K}}_2,$were tested, indicating that a slight amount of oil can increase the dust holding capacity (DHC) of filters by forming a more porous cake, while an excess of oil results in reduced DHC by forming impermeable liquid films on the solid skeleton. Further two-stage experimental results indicate that the effectiveness of a pre-stage filter can be significantly affected by the properties of incoming aerosol and main-stage filters. The utilization of a pre-stage filter unintentionally deteriorated the service lifetime of the main-stage filter when challenged with contaminants with certain oil particles. This counter-intuitive negative phenomenon is due to the special loading behaviors of oil–solid mixed particles. The existing pre-stage filters allow a higher fine oil particle fraction to reach the main-stage downstream, while the induced cake filtration scenario leading to a film clogging scenario adversely reduced the lifetime of the main-stage filter. The findings suggest that the feasibility of a pre-stage in the filtration system requires compressive evaluations according to the specific oil-coated contaminants.

1. Introduction

Air filtration is an effective way to protect human beings and avoid equipment dysfunction (e.g., fouling, erosion and corrosion) by removing particle contaminants in residential houses and industrial buildings [1,2,3,4]. The lifetime of air filtration systems can be reduced by the pressure drop of high-efficiency filters in many circumstances, such as operating in heavy industrial areas, leading to the need to replace filters frequently [5,6]. Such abnormal replacements of expensive high-efficiency filters can increase company costs due to filter renewal and waste post-processing. Alternatively, a two-stage filtration system (i.e., a low-efficiency pre-filter and a high-efficiency main filter) provides a potential solution to address such issues. Specifically, it has been confirmed that pre-stage filters with coarse fibers can protect high-efficiency filters effectively by removing large particles with the advantages of being cost-effective, cleanable, and reusable, with flexible installation [7]. Following up, a main-stage filter with high filtration efficiency can capture finer particles below 1 μm, a measure widely employed in inlet air filtration systems [8,9,10,11,12].

However, detailed filtration processing can be complicated due to various operating conditions. For instance, the physical characteristics of particles are always nonuniform, such as various solid and liquid states [13,14,15,16,17]. One common aerosol contaminant in the air, especially in industries, is sticky oil, which can be formed from unburned hydrocarbons emitted from exhausted stacks, oil vapor from lubricating systems, metal processing fluid, exhaust gas from vehicles, and offshore oil drilling platforms. The percentage of oil particles and solid particles can vary significantly under different conditions [18,19,20,21]. The loading behaviors of filters can be extremely different between loaded solid particles and liquid sticky oil particles, while most investigations have focused on the filtration performance of filters on solid particles rather than liquid oil particles. Usually, the loading process of filters for solid particles goes through three stages, e.g., depth filtration stage, transition filtration stage, and surface filtration stage. For depth and transition filtration regimes, particles are collected inside the filters, and the pressure drop increases non-linearly. For the surface loading stage, a particle cake layer is formed on the surface of the filters, and the pressure drop increases almost linearly [22,23,24]. Unlike solid particles, the loading behaviors of pure oil particles can be divided into four stages [25,26,27,28]. Firstly, when the deposited oil droplets spread on single fibers, the pressure drop can remain constant. Secondly, the oil particles are rearranged in the filter interior, leading to a rapid increase of pressure drop. Then, liquid oil bridges, films, and pools are formed among fibers which cause the pressure drop to increase exponentially. Finally, the pressure drop remains in another constant stage as liquid oil is drained out of the filters. Based on the differences in the filtration mechanisms between oily particles and pure solid particles, it is essential to fill the knowledge gap and gain a better understanding of the loading behaviors of filters for mixed particles with different oil proportions [29,30,31].

The procedures for loading tests of air filters for solid-only particles mainly follow current international standards (e.g., EN 779, ASHR 52.2, ISO 16890, ISO 5011, ISO 19713), while those for liquid-only particles are based on ISO 12500. Filter media evaluation in laboratories usually use either pure oil or pure solid particles, which cannot reflect behaviors under real industrial conditions. To the best of our knowledge, there are few existing studies concerning the different physical characteristics on the loading performance of filters. Penicot et al. [32] investigated the loading behavior of high-efficiency particle air (HEPA) filters when loaded with solid particles and Dioctyl phthalate (DOP) liquid particles and showed that the loading behavior of filters when loaded with oil and solid particles are different. Bredin et al. [33] found that the oil viscosity had a major impact on the pressure drop behavior of the filters, and a higher soot content in the oil resulted in a higher oil viscosity and an increased final pressure drop. T.K. Müller [34] investigated the effect of a thin oil film on dust holding capacity and filtration efficiency of nylon and stainless-steel fiber media and concluded that particle immersion in the oil film is responsible for both the delayed increase in fiber efficiency and dust holding capacity. A.K. Maddineni [35] examined the pressure drop and particle filtration behaviors of oil-treated and untreated filters and found that the oil-treated filters exhibited lower pressure drop and lower filtration efficiency. Ta-Chih Hsiao [36] revealed that the overall particle size distribution, oil volume percentage, filter material, surface tension, and viscosity of the coating oils have a big influence on the loading characteristics of filters. As aforementioned, due to the difference in physical properties, the loading behavior of filters when loaded with oil–solid mixed aerosols is definitely different from that of pure oil or pure solid particles. However, how do the coarse mode solid dust particles interacting with fine mode oil particles in the environment affect the loading characteristics of a two-stage filtration system? Until now, there have been few research studies on the effect of oil–solid mixed particles on the filtration performance of a two-stage filtration system.

To partially fill the knowledge gap, the aim of this study is to investigate the effectiveness of pre-stage filters on different main-stage filters and associate overall performance of the two-stage filtration system under different oil–solid mixed particles. The tests were conducted in a novel two-stage filtration system consisting of a pre-stage and a main stage, which was developed in our previous study [37,38]. The findings of this study can better reveal the mechanisms for how the pre-stage filter performs and how it can extend or shorten the lifetime of main-stage filters under different oil–solid aerosol mixed conditions, which may provide new insights for researchers and engineers in designing energy-efficient two-stage or multi-stage filtration systems.

2. Materials and Methods

In this study, a G4 filter with low filtration efficiency was selected as the pre-stage filter. Another three filters, i.e., an F7 cellulose filter, an F9 melt-blown filter, and an E11 PTFE membrane-coated filter with high filtration efficiency, were chosen as the main-stage filters. The filtration efficiency rating for the four filters (i.e., G4, F7, F9, and E11) was ranked according to the European Standard EN 1822-1-2009. A higher number means a higher filter filtration efficiency. For example, F7 with a 7 is higher than G4 with a 4. Thus, three sets of two-stage filtration configurations (e.g., G4 + F7, G4 + F9, and G4 + E11) were developed, and specific physical properties are shown in

Table 1. The properties of pre-stage and main-stage filter samples.

Filter Samples		Material	Permeability [L/(m2·s1)]	Average Weight [g/m2]	Average Fiber Diameter [μm]	Average Pore Diameter [μm]
Pre-stage filter	G4	PP	3100 ± 400	173 ± 25	25.7 ± 8	14 ± 8
Main-stage filter	F7	Cellulose	64 ± 10	110 ± 13	15.3 ± 8	7 ± 5
	F9	Melt-blown fabrics	97 ± 9	131 ± 9	8.1 ± 3	7 ± 2
	E11	PTFE membrane	85 ± 6	108 ± 7	0.165 ± 8	0.05 ± 0.02

. The mean pore size of filter samples was tested by a Capillary Flow Porometer 7.0 (CFP-1100-A, Porous Material Inc., Ithaca, NY, USA). The average pore sizes of the G4, F7, F9, and E11 filters are 14, 7, 7, and 0.05 μm, respectively. It should be mentioned that the F9 melt-blown filter has a relatively narrower pore size distribution (SD = 2.18 μm), while the F7 cellulose filter has a broader pore size distribution from 1 to 20 μm (SD = 4.89 μm). The filters differ in structure and distributions of pore, as shown in the scanning electron microscope (SEM) images (see Figure 1). In addition, the oil contact angle of the filters was measured using an optical contact angle measuring system (JC2000C1), and the results showed that the F7, F9, and E11 with different pore structures share similar oil wettability features (see Table 1 and Figure 2).

Two types of testing particles were used to mimic both the oil and solid modes in various conditions. A2 fine dusts (ISO 12103-1) represent the coarse solid particles which were dispersed by a homemade dust disperser. The concentration of A2 fine particles was controlled by adjusting the dust feeding rate at the ejection pump which was used to deagglomerate dusts. Dioctyl sebacate (DEHS) was chosen as the raw material to generate the fine oil particles, in which the viscosity and the surface tension of DEHS were 26.2 Pa·s and 3.03 × 10⁻² N/m, respectively. DEHS was atomized by an aerosol nebulizer (Model 3076, TSI Inc., Shoreview, MN, USA). The concentration of DEHS particles was adjusted via regulating the inlet flow from a mass flow controller. To simulate designated distributions of oil–dust mixed particles, corresponding doses of DEHS oil particles and A2 solid particles were mixed in this study accordingly. The particle size distribution of DEHS fine and A2 coarse dust particles can be found in Figure 3. To facilitate the calculations, the gained weight of DEHS particles was replaced by the equivalent weight of A2 particles, as shown in Equation (1), and the concentration of emitted particles is written as Equation (2), i.e.,

$$△{{\mathrm{m}}^{\prime }}_{\mathrm{DEHS}}=\frac{△{\mathrm{m}}_{\mathrm{DEHS}}}{{\mathsf{\rho }}_{\mathrm{DEHS}}}·{\mathsf{\rho }}_{\mathrm{A}2}$$

(1)

$$△{\mathrm{C}}_{\mathrm{DEHS}}=\frac{△{{\mathrm{m}}^{\prime }}_{\mathrm{DEHS}}}{{\mathrm{v}}_2·\mathrm{S}·\mathrm{t}}$$

(2)

where △m′_DEHS denotes the equivalent weight of gained DEHS weight, mg; △m_DEHS denotes the mass gain of filters, mg; C_DEHS is the adjusted mass concentration of DEHS, mg/m³; ${\mathsf{\rho }}_{\mathrm{DEHS}}$ and ${\mathsf{\rho }}_{\mathrm{A}2}$ are the density of the DEHS and A2 particles with 900 kg/m³ and 2650 kg/m³, respectively; t is the loading time, s; and S is the effective filtration area, m².

To obtain five designated oil–solid mixed particles, the corresponding five mixing strategies were adopted by specifying the mass ratios of A2 vs. DEHS with 0:100% (pure solid), 25:75%, 50:50%, 75:25%, and 100:0% (pure oil), respectively. Depending on the mass ratios, the mass concentrations of A2 solid particles were 1040, 1293, 1166, 543, and 0 mg/m³, separately. The mass concentrations of DEHS oil particles were 0, 434, 1175, 1529, and 1075 mg/m³, respectively.

Figure 4 shows the schematic diagram of the self-assembled two-stage filtration experimental setup. Both the single-stage and two-stage filtration experiments can be performed. The two-stage filtration setup consists of an upstream pre-stage and a downstream main-stage filter holder, a fine oil generator and a dust disperser, mass flow meters, pressure sensors, and associated connecters. The airflow velocities were designated as 65.2 cm/s and 2.1 cm/s for the pre-stage and the main-stage for all loading conditions, respectively. It is worth mentioning that the low-cost pre-stage filters (i.e., G4) are usually replaced with a higher frequency than high-cost main-stage filters. Thus, the pre-stage filter samples were replaced three times in each complete test before the main-stage filter reached the terminal pressure drop. The two-stage filtration testing process is depicted in Figure 5, and the detailed testing procedures were specified in our previous study [37].

3. Results and Discussion

3.1. Loading Performance of a Single Main-Stage Only

3.1.1. Transition Loading Behaviors of Oil-Coated Particles in a Single Main Stage

Figure 6a–c depict that the growth rate of the pressure drop in the filters (i.e., F7 cellulose, F9 melt-blown, and E11 PTFE membrane-coated filters) varies significantly in a single-stage configuration as the oil content percentage in the oil–solid mixed particles increases from 0% to 100% (i.e., 0%, 25%, 50%, 75%, and 100%), specifically. When loaded with pure solid particles, a particle cake layer was formed on the surface of the high-efficiency filters, and the pressure drop in the filters has a positively linear-like relationship with the accumulated particles. However, the changes of pressure drop in the single main-stage present a different phenomenon when loaded with pure oil particles. With accumulating liquid oil particles in the filters, it can be observed that the pressure drop increases slightly before the deposited volume reaches a certain value, then an exponential rapid growth occurred due to the liquid film forming on the filter surface. According to the loading curve profiles of those three types of filters (see Figure 6), the loading behaviors of particles on the filters can be divided into two distinct categories, i.e., surface loading and film clogging. When the percentage of oil contents in the oil–solid mixed particle is smaller than 50%, the loading curves of the three filters are similar to that of loading pure solid particles. However, when the percentage is larger than 50%, the loading profiles show similar trends to that of capturing pure oil particles. Moreover, the pressure drop increases with a slow growth rate, similar to adding a small amount of oil on the solid particles. While the pressure drop increases dramatically with excess oil particles captured on the filters, the growth rate of the pressure drop slowed again when only pure oil particles were loaded.

Figure 7 illustrates the DHC performances of the F7 cellulose, F9 melt-blown, and E11 PTFE membrane-coated filters in single-stage configuration when loaded with different oil–solid mixed particles. With the oil percentage in oil-coated particles increasing from 0% to 100% with a gap of 25%, the DHC of the F7 cellulose filter is 0.33, 1.42, 0.13, 0.29, and 2.47 mm³/cm², respectively; similarly, the DHC of the F9 melt-blown filter is 0.49, 2.16, 2.51, 0.32, and 0.80 mm³/cm², respectively; the DHC of the E11 filter is 0.43, 2.14, 0.06, 0.09, and 0.76 mm³/cm², respectively. Figure 6 shows that the DHCs of the three filters increase significantly at the initial stage, then drop dramatically, and return to increase at the end in all oil–soil mixed conditions. It is worth mentioning that a slight change of oil particle fraction in oil-coated particles from 0% to 25% can yield a significant increase of the DHC, presenting an increase of 4.30, 4.41, and 4.98 times for the F7 cellulose, F9 melt-blown, and E11 PTFE membrane-coated filters, respectively. F7 cellulose, F9 melt-blown, and E11 PTFE membrane-coated filters have a super high DHC when the oil–solid mixed particles are coated with 25% oil. Moreover, the F9 melt-blown and PTFE membrane-coated filters performed better than the F7 cellulose filter in DHC when experiencing the particles with oil-free and 25% oil because the particles penetrated much deeper into the non-uniform pores of the F7 cellulose filter, causing a rapid increase in pressure drop, which was also the case in our previous study [39]. At the same time, the particles moved rapidly onto the surface of the F9 melt-blown and E11 PTFE membrane-coated filters with a high efficiency. When the oil percentage was greater than 50% (e.g., 50%, 75%, and 100%), the E11 PTFE membrane-coated filter, having the smallest pores, yielded the lowest DHC, thus displaying poor performance. The F7 cellulose filter yielded the maximum DHC when coping with pure oil particles due to the difficulty in forming liquid film on the large-size pores.

3.1.2. Oil-Coated Particle Deposition Mechanism

A cake layer can be formed as particles are captured on the surface of high efficiency filters and contributes to the increase in the pressure drop. Different increase rates of pressure drop are due to the changes in microstructure interactions between the fiber matrix and the deposited particles and inter-particles. The cake porosity $\mathsf{\epsilon }$ and specific cake resistance ${\mathrm{K}}_2$ are two essential indicators to evaluate the microstructure of deposited particles, which can reveal the deposition mechanism of filters directly. to the two indicators reveal the underlying mechanisms of the peculiar changing trend of the DHC and pressure drop for filters under different oil–solid mixed particle conditions.

With the known mass and height of the cake, the cake porosity $\mathsf{\epsilon }$can be calculated by the following equation:

$$\mathsf{\epsilon }=1-\frac{\mathrm{M}}{\mathrm{A}·{\mathsf{\rho }}_{\mathrm{m}}·\mathrm{th}}$$

(3)

where M is the mass of deposited dust, kg; $\mathrm{th}$ in m is the cake thickness which can be measured by a video-enhanced microscope (VEM) to observe the loaded filters; A is the effective filtration area, m²; and ρ_m is the density of testing particle, kg/m³.

A typical model to describe the overall pressure drop (∆P) of a loaded filters can be expressed as a sum of pressure drop across the cleaning filters (∆P_f) and pressure drop across the cake (∆P_c) [40,41], e.g.,

$$\Delta \mathrm{P}=\Delta {\mathrm{P}}_{\mathrm{f}}+\Delta {\mathrm{P}}_{\mathrm{c}}$$

(4)

$$\Delta {\mathrm{P}}_{\mathrm{f}}={\mathrm{k}}_1\mathrm{V}$$

(5)

where k₁ depends on the physical characteristics of filters, such as the filter packing density, fiber diameter, thickness, etc.; and V is the filtration velocity, where the pressure drop of cleaning filters is linearly proportional to the filtration velocity.

The pressure drop through the particle cake, ∆P_c, is given by:

$$\Delta {\mathrm{P}}_{\mathrm{c}}={\mathrm{k}}_2\mathrm{V}\frac{\mathrm{M}}{\mathrm{A}}$$

(6)

Thus, combining Equations (4) and (6), ${\mathrm{K}}_2$ can be written as:

$${\mathrm{k}}_2=\frac{\left(\Delta {\mathrm{P}}_{\mathrm{p}}-\Delta {\mathrm{P}}_{\mathrm{f}}\right)}{{\mathrm{V}}_{\mathrm{f}}}\frac{\mathrm{A}}{\mathrm{M}}$$

(7)

where ${\mathrm{K}}_2$ is the specific dust cake resistance coefficient and depends primarily upon the particle diameter and cake porosity; M is the dust mass deposited on the filters; and A is the filtration area.

The specific cake resistance $\mathsf{\epsilon }$ and cake porosity ${\mathrm{K}}_2$ of clogged F9 melt-blown filters were calculated by Equations (3)–(7) and are shown in Figure 8 and Figure 9, while the corresponding values for the F7 cellulose and E11 PTFE membrane-coated filters were not available because the formed cake layers were too thin to measure $\mathrm{t}\mathrm{h}$ accurately in the specific solid–oil mixed cases. With the percentage of oil contents rising at 0%, 25%, 50%, and 75%, the calculated ${\mathrm{K}}_2$ is 1.83 × 10⁵, 1.22 × 10⁵, 0.60 × 10⁵, and 5.60 × 10⁵, and $\mathsf{\epsilon }$ is 0.71, 0.85, 0.94, and 0.60, respectively. With the oil content increase to 25%, the value of $\mathsf{\epsilon }$ increases and ${\mathrm{K}}_2$ decreases, indicating that the cake layer becomes fluffier. When loaded with excess oil particles, such as a 75% oil condition, the values of $\mathsf{\epsilon }$ drop and ${\mathrm{K}}_2$ increase dramatically, indicating the cake layer become tighter.

The particle deposition mechanisms for oil–solid mixed particles with an elevated oil content from 0 to 100% were investigated thoroughly (see Figure 10) by tracking the changes in pressure drop, DHC, $\mathsf{\epsilon },$ and ${\mathrm{K}}_2$of filters, respectively. Loading particles with different oil contents on the filters can be classified into two loading patterns, i.e., surface loading and film clogging. In the case of 0% oil-coated conditions, particle dendrite-like chain structures were formed on fiber. Following up, particles were accumulated, causing the long chains collapse and a dense layer of packed particles, which is due to increasing gravity and air force in the longitudinal direction, which can be called an “accumulation–compression–accumulation” dynamic process. The compacted cake layer contributes the resistance increase on the filters, which can further prevent airflow through the filtration system. In the case of particles with slight oil contents, such as 25%, for F7 cellulose and E11 membrane-coated filters, and 50% for F9 melt-blown filters, a slight amount of oil coated solid particles and the fiber surface, forming a liquid bridge between fibers and particles or inter-particles. The oil liquid bridge increased the viscous resistance between oil-coated solid particles; thus, such particles with high viscosity tended to stick on the long chain firmly instead of causing collapse, which prevented the compression of the cake layer. As a result, the high porosity can allow airflow through the filters; thus a lower resistance of loaded filters are realized. When the oil content percentage is higher than 50% for F9 melt-blown filters and 25% for F7 cellulose and E11 membrane-coated filters, the excess amount of oil spread out on the skeleton of A2 dust particles and filters is then broken up solid dendrites and forms a continuous liquid film on the solid skeleton of fibers and particles or inter-particles on the top surface of the filters. All the solid particles were immersed in the oil, and the airflow was blocked by the impermeable liquid film, leading to the pressure drop increasing exponentially. When loaded with pure oil particles compared to the higher oil content of oil–solid mixed conditions, the initiation of the liquid film was delayed as it lacks a solid skeleton where oil can spread. The oil film slowly spread in the olyophilic fiber media; thus, the pressure drop increased slower than that of mixtures with higher oil contents.

3.2. Loading Performance of Main-Stage Filter in a Two-Stage System

After understanding the special loading behaviors of filters for different oil–solid mixed particles in single-stage configurations, the two-stage configurations were investigated, with a G4 filter used as the pre-stage. As shown in Figure 11a, the DHC of a (G4 + F9) two-stage configuration also exhibits a trend of “first increase significantly then drop dramatically and finally increase” with increasing oil content, similar to the performances of the DHC of F9 filters in the single-stage system. However, a counter-intuitive phenomenon was found in the following analysis. The DHCs of the (G4 + F9) two-stage configuration were even lower than its single-stage counterpart at oil percentages of 25% and 50%, suggesting that instead of protecting the main-stage, the added G4 pre-stage filters lead to even shorter service lifetime of F9 main-stage filters. To quantify and simplify comparisons on the effectiveness of pre-stage filters on the DHC of the whole filtration system and the service lifetime of the main-stage filters in the two-stage filtration system, a self-defined parameter M** was introduced and determined by the following Equations (8)–(13). M** means to what extent the mass holding capacity of a filtration system and the service lifetime of main-stage filters is changed under the protection of a pre-stage filter. (More detailed definitions are given in our previous study [37].)

$${\mathrm{M}}^{**}=\frac{{\mathrm{M}}_{\mathrm{total}}}{{\mathrm{M}}_{\mathrm{ref}}}$$

(8)

$$\mathrm{M}=\frac{\mathrm{m}}{\mathrm{u}·\mathrm{A}}$$

(9)

$${\mathrm{M}}_{\mathrm{total}}={\left({\mathrm{M}}_{\mathrm{P}}\right)}_{1–3}+{\mathrm{M}}_{\mathrm{f}}$$

(10)

$${\left({\mathrm{m}}_{\mathrm{P}}\right)}_{1–3}={\mathrm{m}}_{1,1}+{\mathrm{m}}_{1,2}+{\mathrm{m}}_{1,3}$$

(11)

$${\left({\mathrm{M}}_{\mathrm{P}}\right)}_{1–3}={(\frac{{\mathrm{m}}_{\mathrm{P}}}{{\mathrm{u}}_1·\mathrm{A}})}_{1–3}$$

(12)

$${\mathrm{M}}_{\mathrm{f}}=(\frac{{\mathrm{m}}_2}{{\mathrm{u}}_2·\mathrm{A}})$$

(13)

where M** is the ratio of the total velocity-normalized areal loading of both pre-stage and main-stage in the entire two-stage test case, M_total, divided by the velocity-normalized areal loading of the main-stage in the entire reference test case, M_total; (m_p)_1–3 is the mass gain of the pre-filter in the first three steps; m_f is the mass gain of the main filter in the whole two-stage test case; m_ref is the mass gain of main-filter in the whole single-stage test case; u₁ and u₂ is the face velocity on the pre-stage and the main stage, respectively; and A is the effective filtration area.

Using Equations (8)–(13), M** was calculated, showing 1.59, 0.85, 0.48, 2.21, and 1.05 for 0%, 25%, 50%, 75%, and 100% oil-coated conditions in the (G4 + F9) two-stage filtration system (see Figure 11b), respectively. For pure solid or pure oil mode aerosols, M** was greater than 1.0, indicating that the additional G4 pre-stage filters worked with extending the DHC of the filtration system and extending the service lifetime of F9 main-stage filters, while when the filters were challenged with oil-coated particles, the results were counter-intuitive, i.e., in 25% and 50% oil-coated particle cases, M** was substantially lower than 1, 0.85, and 0.48, respectively, which means adding a pre-stage filter in front of the F9 melt-blown main-stage filter can lead to a reduction of the DHC for the whole filtration system, and a shortened service lifetime of the main-stage filter. For the particles with a 75% oil coating condition, the life extension for the main-stage filters can reach to the maximum, with M** registered as 2.21. In addition, when pure oil particles were loaded, the G4 pre-stage filter only prolongs the F9 service lifetime by 5%.

To further explore the counter-intuitive phenomenon, the pressure drop and the DHC for F9 melt-blown main-stage filters in both single-stage and two-stage configurations were compared for different oil-coated conditions (see Figure 12). It needs to be mentioned that the loading performances of the pre-stage filters in two-stage loading configurations were deleted to facilitate comparisons under some oil coating percentages, where the replacement of the pre-stage filter was marked by arrows. As shown in Figure 12a, when pure solid was loaded, the pressure drop for F9 melt-blown main-stage filters increased gradually in the two-stage filtration configuration, and the pre-stage filter worked effectively as it can catch large particles and significantly reduce the weight of loaded particles on the main-stage filter, contributing to a service lifetime extension of 1.5 times for main-stage filters. For the pure oil loading case, the pre-stage filters with large pore structure are not able to capture oil fine particles, as shown in the calculated M** close to one. When challenged with 25% and 50% oil-coated particles, as shown in Figure 12b,c an interesting phenomenon can be observed from the pressure drop evolution profile. The pressure drop of the F9 melt-blown main-stage filter increases faster in the two-stage configuration than that in the single-stage configuration. This phenomenon can be explained as the pre-stage filter selectively removing coarse solid particles more efficiently than fine oil particles, leading to more oil fine particles reaching the main-stage downstream than in the referred single-stage. Additionally, the portion of solid particles entering the main-stage is sufficient to form a skeleton to assist oil particles to spread out and form an oil film. Thus, the existence of the G4 pre-stage filter can destroy the fluffy dust cake structure in the main-stage filter which is formed in the presence of a small amount of oil proportion. This made the loading mode of the high efficiency main-stage filter transit from surface loading to oil film clogging. Therefore, adding a pre-stage filter in front of the F9 melt-blown main-stage filter can cause a reduction of the DHC of the entire filtration system, as well as a shorter service lifetime of main-stage filters. With the oil percentage increase from 50% to 75%, as shown in from Figure 12d, the pressure drop of the F9 melt-blown main-stage filter increases more slowly in two-stage configuration than that in the corresponding single-stage configuration. This is because the existing solid particles are sufficient to serve as a skeleton, permitting rapid formation of an oil film without the pre-stage filter. However, the existing pre-stage filter can capture a considerable mass of solid particles so that the amount of solid particles entering the main-stage are dramatically reduced. The initiation of liquid film is delayed by the pre-stage as it takes time to form solid skeleton where oil can spread. As a result, the growth of pressure drop can be slowed down with the assistance of the pre-stage filter. Additionally, it can be observed that the DHC in the filtration system increases significantly, and a longer service lifetime of F9 main-stage filters can be achieved. Nevertheless, when solely loading pure oil particles, the G4 pre-stage filters have nearly no effect (~4%) on the extension of the service lifetime of F9 main-stage filters.

Figure 13 shows that the three designated two-stage filtration systems (G4 + F7, G4 + F9, G4 + E11) have similar performances on the two critical parameters, i.e., the DHC and M**, although some local differences can still be observed. When loaded with 25% oil-coated particles, M** for the (G4 + F9) and (G4 + E11) is smaller than 1.0 (0.85 and 0.87, respectively), which means the added pre-stage filters can shorten the service lifetime of E11 PTFE membrane-coated filters. Although different from the F9 melt-blown and E11 PTFE membrane coated filter whose M** was less than 1, the M** of the F7 cellulose filter was close to 1 (1.03). This means the existence of pre-stage filters had nearly no effect on extending the service lifetime of F7 cellulose filters. Since the G4 pre-stage filters can only allow limited oil to reach the F7 cellulose main-stage filter, the service lifetime of the F7 cellulose main-stage filters have nearly no major changes. However, when loaded against mixtures with 50% oil, the pre-stage filter shortens the service life-time of F9 melt-blown filters by 0.48 times, the pre-stage filter has significant effects on extending the service lifetime of E11 and F7 filters, whose lifespans are extended by 1.99 and 1.81 times, respectively. When loaded with 75% oil-coated particles, the M** of E11 PTFE membrane-coated filters was 1.12, and the pre-stage filters helped extend the service lifetime of the E11 PTFE membrane-coated main-stage filters. However, the DHCs in 50% and 75% oil-coated particle conditions were still much lower than oil-free and 25% oil-coated scenarios, whether or not a pre-satge filter was employed. The specific structures of the PTFE coating on the E11 filter with small inter-fiber distance limits its applications in high oil-containing containment removal conditions. In addition, G4 pre-stage filters can only prolong the service lifetime of F7 and E11 main-stage filters by 3% and 5%, respectively, when pure oil particles re loaded, which is similar to the performance on the (G4+F9) two-stage configuration.

4. Conclusions

The filtration effectiveness of the pre-stage under a two-stage filtration system during dust loading has received special attention because of its significant influences on the life span and energy consumption of the filtration systems. The filtration performance of filters was predominantly affected by oil–solid mixtures with different oil proportions. Therefore, the impacts of different oil–solid mixtures on the filtration of single-stage and two-stage systems were thoroughly investigated in this work. Specifically, key conclusions can be summarized, e.g.,

• The fraction of oil has a significant impact on the performance of single-stage filter loading. The DHC of filters exhibits a trend of “first increase significantly then drop dramatically and finally increase” with the percentage of oil content increase. A slight amount of oil helps increase the filter DHC by forming more porous cakes. However, an excess amount of oil can reduce the DHC by forming an impermeable liquid film on the solid skeleton.
• The utilization of a pre-stage filter may not increase the lifetime of the main-stage filter. The effectiveness of pre-stage filters can be strongly affected by the fraction of oil in the oil–solid particles; while the induced cake filtration scenario towards the film-clogging scenario will adversely reduce the lifetime of the main-stage filter. The micro-structures of different filters introduce extra complications towards this process, which requires case-specific testing of suitable filters and operations.

In summary, the effectiveness of a pre-stage filter medium was massively affected by the physical properties of incoming aerosols and the properties of main-stage filters. The utilization of a pre-stage filter unintentionally deteriorated the service lifetime of the main-stage filter when challenged with contaminants of a mixture of certain oil particles. This counter-intuitive negative phenomenon was due to the specific loading behavior of oil–solid mixed particles. The existing pre-stage filters allowed higher oil fine particles fraction to reach the main-stage downstream, and the cake filtration scenario transformed towards the film-clogging scenario. Therefore, both the particle physical states and properties of filters should be paid close attention when designing a two-stage or a multi-stage filtration system. The results of this work can further help optimize the filtration system design to further reduce operation cost, air pollution and energy consumption.