Liquid-mediated particle capture by nonwoven filter media for automotive engine intake air filtration

Abstract

This article reports on development, characterization, and performance of liquid-treated nonwoven air filter media for automotive engine intake application. A polypropylene fiber-based needle-punched nonwoven fabric was prepared for treatment with four viscous liquids (glycerol, SAE 20W/50 engine oil, PEG 400, and deionized water) by liquid spraying technique. The filtration performance was evaluated in terms of initial and final gravimetric filtration efficiencies, fractional filtration efficiency, evolution of pressure drop, and dust holding capacity. The liquid-treated filter media registered higher gravimetric as well as fractional filtration efficiency and higher dust holding capacity as compared to the untreated ones. The initial and final gravimetric filtration efficiencies were found to be directly related to liquid add-on via a power law relationship. The liquid-treated filter media also exhibited higher fractional filtration efficiency than their untreated counterparts for all sizes of tested particles. Interestingly, the increase of fractional efficiency was more for smaller particles as compared to larger ones. This was explained quantitatively through single fiber efficiency due to adhesion. The viscosity of liquid was found to be a very crucial parameter as the dust deposition morphology was contingent to the flow of liquid onto the filter media. The stickiest liquid yielded highest filtration efficiencies, displayed slowest rise of pressure drop, and exhibited highest dust holding capacity.

Introduction

Air filtration plays an important role in deciding the performance of automotives in terms of engine life, fuel economy, emission control, and passenger safety.^1–3 The air filtration system in automotives includes engine intake air filtration. It uses a filter medium for separation of hazardous contaminants from airstreams. Clean air is required for the supply of oxygen to the internal combustion engines for fuel combustion. In engine intake air filtration, airborne dust particles of 5 μm–10 μm size are responsible for major engine wear. ⁴ If the dust particles enter into the engines, they cause significant engine wear, create inadequate air-to-fuel stoichiometric ratio, give rise to malfunctioning of air flow sensors, and emit pollutants from the engines as exhaust. This leads to poor engine life, remarkable performance loss, high oil consumption, increased operation cost, enhanced air pollution, and greater CO₂ emission.^5–8 The modern automotive engine intake air filters are expected to comply with filtration performance as laid down in the following relevant specifications under all operating conditions. The filtration efficiency should be greater than or equal to a minimum specified efficiency. The pressure drop should be less than or equal to the terminal or maximum acceptable pressure drop. The dust holding capacity should be very high for greater service life. As known, the filtration efficiency controls the amount of particulate matter trapped in the filter media, the pressure drop influences the fuel economy, and the dust holding capacity determines the life of filter media. A high-quality engine intake air filter medium would have these parameters optimized to deliver the desired level of performance. In keeping view with this, three types of filter media are mostly used nowadays—foam filter media, paper filter media, and nonwoven filter media. Each of these media has its own advantages and disadvantages with respect to the material characteristics, manufacturing processes, structure–property relationship, and operating conditions, which are known to affect their filtration performance. While the foam filter media offer low filtration efficiency, low pressure drop, and high dust holding capacity, the paper filter media offers moderate filtration efficiency, high pressure drop, and low dust holding capacity. On the other hand, the nonwoven filter media offers high filtration efficiency, moderate pressure drop, and moderate dust holding capacity. Of them, the nonwoven filter media are comparatively “young” in engine intake air filtration. ⁹ Nevertheless, they are expected to evolve significantly for engine intake air filtration within the next decade.

In general, the nonwoven filter media separate particles from airstreams by means of depth filtration and depth straining. They capture particles via Brownian diffusion, direct interception, inertial impaction, and gravitational settling mechanisms. Of them, the inertial impaction has been the most dominating mechanism of particle capture in engine intake air filtration. Nevertheless, the ever-increasing trend of engine downsizing necessitates a reduction in the space available for installation of the filter media, which has resulted in an increase of face velocity of the particles. ⁶ This causes an increase of kinetic energy of the particles to such as extent that the particles sometimes bounce back from the media surface, re-entrain into the airstreams, and finally penetrate through the filter media.^6,10–12 This also occurs with the dust-loaded filter media as the non-uniform air flow and the associated dust deposition sometimes cause an increase of local velocities within the pore space of the filter media.^13,14 This leads to an elevated risk of exposure of the engine components to dust particles. The problem of particle bounce, re-entrainment, and penetration was studied earlier by a few researchers. Poon and Liu ¹⁵ reported on particle bounce and re-entrainment in case of foam filter media. They observed penetration of particles of greater than 2 μm diameter through the filter media at a face velocity of 6.5 m/s. They also found that the higher was the porosity of the filter media the higher was the particle penetration due to particle bounce and re-entrainment. Maus and Umhauer ¹⁶ also observed particle bounce and re-entrainment while experimenting with filtration of biological and non-biological aerosols through fibrous filter media for automotive engine intake air filtration. They noticed that the particles of greater than 5 μm diameter could bounce from the surface of the filter media at a velocity of 1 m/s. A similar behavior was also noticed in case of paper filter media by Mueller et al. ¹⁷ They examined particle filtration behavior of paper filter media at different face velocities ranging from 0.1 m/s to 0.3 m/s and observed that the particle penetration through the filter media increased with the increase of face velocity. Maddineni et al. ¹⁸ also observed a similar behavior with paper filter media and explained this behavior in the light of theoretical and empirical models of air filtration. Through computer simulations, it was established that the higher particle velocity, larger particle diameter, and higher porosity were the key determinants for particle bounce and re-entrainment. ¹⁹

In order to mitigate particle bounce and to counter the associated particle penetration, the engine intake air filter media are treated with viscous liquids. It is thought that the particles upon impact onto the fiber surface adhered to the liquid droplets, as a result, the particles do not get a chance to bounce back. Mullins et al. ²⁰ made an attempt to soak fibrous filter media with water and observed inhibition of particle bounce by the liquid film formed onto the filter surface. Boskovik et al. ²¹ observed suppression of particle bounce by a fibrous filter medium treated with mineral oil. A similar remedial measure was suggested by Mueller et al. ¹⁷ in case of paper filter media. Hubbard et al. ²² used glycerol to soak a very thin fibrous filter media and assessed its filtration behavior at very high velocities. They noticed that the liquid treated filter media were able to mitigate particle bounce. Maddeneni et al. ¹⁸ also observed inhibition of particle bounce by oil treatment to paper filter media. They prepared oil-treated filter media by spraying hydraulic oil and observed higher filtration efficiency due to oil treatment. They also observed that the oil-treated paper filter media exhibited higher pressure drop and higher dust holding capacity than the untreated ones. While the investigations established that the inhibition of particle bounce could be achieved by liquid treatment, it is yet to understand the effect of liquid properties on the adhesion of particles and suppression of particle bounce. Nevertheless, there were many attempts made to understand the physico-chemical and interfacial processes in order to modulate the adhesion properties of fibrous substrates. These processes are well summarized in a review by Molina et al. ²³ However, the role of liquid treatment process on the filtration performance of the filter media remains unknown. Needless to say, high performance engine intake air filter media can be developed if an appropriate liquid is identified and a suitable process is set up. This article tries to establish the roles of liquid properties and liquid treatment process in determining the adhesion of particles to liquid-treated fibers and deciding the filtration performance of liquid-treated filter media for automotive engine intake air filtration application.

Materials and methods

Materials

A needle-punched nonwoven filter medium was manufactured by using regular polypropylene fibers of 51 mm staple length and 6 denier fineness and employing a fiber opening machine, a carding machine, a cross-lapping machine, and a needle-punching machine. The basis weight of the filter medium was determined by a weighing balance as per ASTM D6242-98 standard. The average basis weight was measured to be 300 g/m² with a coefficient of variation of 3.12%. The thickness of the filter medium was measured by a digital thickness tester with the conditions being in accordance with ASTM standard D5729-97. The measured average thickness of the medium was recorded to be 3.5 mm with a coefficient of variation of 3.5%. The nonwoven filter medium was treated with four different liquids namely deionized water, PEG 400, SAE 20W/50 engine oil, and glycerol. The key properties of the liquids are mentioned in Table 1. It can be noted that PEG 400 and glycerol had a minimum purity level of 99.9% and 99.7%, respectively.

Table 1.

Properties of liquids.

@ 25°C	Deionized water	PEG 400	SAE 20W/50 oil	Glycerol
Density (kg/m3)	996	1128	880	1258
Dynamic viscosity (Pa·s)	8.9×10−4	0.102	0.140	0.906
Surface tension (N/m)	0.072	0.042	0.032	0.063

Spraying of viscous liquids

The dry nonwoven filter medium was treated with the aforesaid liquids by employing an in-house fabricated spraying apparatus as shown in Figure 1. The liquid from an oil tank and the air from a compressor were supplied through different channels to the spray atomizer. The compressor was operated automatically for a specific duration by means of a solenoid valve and an electronic timer. The atomizer was a twin-fluid spray nozzle with 0.14 mm size. It generated fine droplets at a pressure drop of 400 kPa. The liquid droplets were sprayed onto the filter medium which was held by a mounting plate kept at a distance of 15 cm from the atomizer. The mounting plate was rotated by using a rotating circular disk to achieve a uniform deposition of liquid onto the entire filter medium. The entire operation was conducted automatically to avoid any human interference. The extra liquid, if any, was collected in a tray which was kept below the fixture. The amount of liquid in the filter medium was varied at four levels viz. 40 g/m², 80 g/m², 120 g/m², and 160 g/m². The liquid was deposited only on one face of the filter medium. It can be noted that the untreated filter medium was considered to have 0 g/m² liquid for comparison of experimental data.

Figure 1.

Experimental set-up for spraying liquids onto filter medium (➊ Storage tank, ➋ air compressor, ➌ Atomizer, ➍ Liquid spray, ➎ Filter medium, ➏ Mounting plate, ➐ Tray, ➑ Rotating disk, and ➒ Base bed).

Testing of filtration performance

The gravimetric filtration behavior of the filter media was recorded in accordance with ISO 5011 standard using an air filter test rig as shown in Figure 2. ISO 12103-1 A2 fine test dust was used as a test dust to challenge the test media. Note that this test dust complies with ISO 5011 standard for testing single-stage air cleaners for IC engines. Before use, the test dust was heated in a vessel up to a temperature over 100°C to eliminate any humidity as recommended in the standard. The temperature and relative humidity were measured at the beginning of the measurement and were maintained throughout the test. The test dust was dispersed by a powder dispersion generator with a brush at a pressure drop of 1 bar. To ensure a dust concentration of 1 g/m³ as specified in ISO 5011 standard, the discharge pressure was adjusted accordingly to attain the dust feed rate as per the desired volumetric flow of air. The location of the dust feeding nozzle with respect to the position of the test channel was maintained at a distance of one diameter of the inlet channel dimension to ensure uniform distribution of dust aerosol. Necessary measures were taken as per ISO 5011 standard to ensure that the amount of dust fed into the filter media was within the accuracy of 0.1 g. Also, the mass balance at the end of the test was maintained within a range of 0.98–1.02. An air flow rate of 0.5 m³/min was controlled by a blower placed at downstream and the flow rate was precisely measured by using a vortex shedding flow sensor. The dust penetrated through the test medium was collected on an absolute filter medium placed at the downstream of the test medium. The absolute filter medium, made up of fiberglass, had a minimum efficiency of 99.58% at 8.5 m³/min flow rate as per the ISO 5011 standard. As a standard protocol, the tests were carried out only after gas flow equilibrium was achieved and practically no loss of liquid took place during the test. Gravimetric measurements on the test filter medium and the absolute filter medium were carried out at certain intervals to estimate the dust holding capacity of the test filter medium and the filtration efficiency was determined from the following expression

$$E=\frac{\mathrm{\Delta }MU}{\mathrm{\Delta }MU+\mathrm{\Delta }MABS}$$

(1)

where E refers to filtration efficiency, ΔMU indicates the increase in mass of the test filter medium, and ΔMABS denotes the increase in mass of the absolute filter medium. The pressure drop across the test medium was measured by using static pressure taps placed at the inlet duct and the outlet duct, connected each to a pressure sensor by a ring line. The test was carried out by injecting test dust at a feed rate of 0.5 g/min. The terminating pressure drops for the initial filtration efficiency and the final filtration efficiency were maintained, respectively, at 200 Pa and 1000 Pa above the initial pressure drop. The experimental data were analyzed to examine the evolutions of filtration efficiency, pressure loss, and dust holding capacity during dust loading of the tested filter medium.

Figure 2.

Air filter test setup for gravimetric measurement of filtration performance.

Also, in this work, the fibrous filter media were tested for determination of fractional filtration efficiency by employing a filter test setup shown in Figure 3. The setup comprised of an aerosol generator for generation of submicrometer-sized particles, a pair of particle counters to measure the upstream and downstream concentrations of particles, a pressure transducer for measurement of pressure drop, and a flow transducer for measurement of velocity. Polystyrene latex was used in the atomizer to generate particles in the sizes of 0.3 μm, 0.5 μm, 1 μm, 3 μm, 5 μm, and 10 μm. The solvent in the atomized droplets were dried in the dryer with silica gel as a desiccant. Electrical charges present on the dried particles were conditioned in a neutralizer. The test setup was equipped with two laser-based particle counters in order to count the number of particles in the upstream and downstream of the test filter media. The filtration efficiency at each particle size was calculated from the measured particle concentrations at upstream and downstream of test filter media as follows

$$E\left(d_x\right)=1-\frac{C_{\text{down}}\left(d_x\right)}{C_{\text{up}}\left(d_x\right)}$$

(2)

where $E\left(d_x\right)$ refers to the filtration efficiency for a test particle size, $C_{\text{down}}\left(d_x\right)$ denotes the concentration of particles of a specific size at the downstream of filter media, $C_{\text{up}}\left(d_x\right)$ indicates the concentration of particles of same size at the upstream of filter media. The test face velocity was kept at 0.3 m/s. The pressure drop across the test filter media was measured by using a pressure transducer. A flow transducer was used to determine the velocity of air flowing through the test specimen. A HEPA filter medium was mounted perpendicular to the air flow to ensure no particle in the exhaust.

Figure 3.

Air filter test setup for measurement of fractional filtration efficiency.

Results and Discussion

Gravimetric filtration efficiency

The automotive vehicle producers specify gravimetric filtration efficiency as a key requirement for engine intake air filter media. This is because the mass of dust particles accumulated inside the engine is known to be an estimate of engine wear. As per the gravimetric method, the filter media are challenged by dust particles of different sizes and the filtration efficiency is calculated by the ratio of the mass of dust particles collected by the filter media to that fed to the filter media. Often, the gravimetric filtration efficiency is expressed in terms of initial and final filtration efficiencies. The initial filtration efficiency is determined once the pressure drop across the filter media is increased by 200 Pa, while the final filtration efficiency is in conformity with a terminating pressure drop which is indicated by an increase of 1000 Pa. The initial filtration efficiency is a very critical parameter as at this initial stage of filtration the filter media exhibit their lowest efficiency, which can have an adverse effect on engine wear. The final filtration efficiency is also important as it corresponds to the maximum allowable pressure drop at which the power requirement of the engines can vary within a tolerance of ± 5%.

Figure 4 plots the initial and final gravimetric filtration efficiencies of the dry (untreated) as well as liquid-treated filter media. It can be seen that all the filter media exhibited an increase of both initial and final filtration efficiencies upon spraying of different liquids. The rise in filtration efficiency due to liquid treatment was associated with adhesion of particulate matter upon impact on the liquid-treated fibers, thus inhibiting particle bounce and re-entrainment into airstreams. The higher was the amount of liquid applied to the filter media the higher was their filtration efficiencies. The following power-law expression was found to explain the relation between the liquid weight and the filtration efficiency

$$E_{F,1}=E_F\hspace{0.17em}\left(1+{w^a}_1\right)$$

(3)

Figure 4.

Plots of gravimetric filtration performance of dry and liquid-treated filter media: (a) initial filtration efficiency and (b) final filtration efficiency.

where $E_{\text{F,l}}$ denotes the filtration efficiency of the liquid-treated filter media, $E_{\text{F}}$ refers to the filtration efficiency of untreated filter media, $w_{\text{l}}$ indicates the liquid add-on, and a is an exponent. The numerical values of the exponent for different levels of liquid add-on are reported in Table 2. The power of the exponent for the initial filtration efficiency was found to be higher than that for the final filtration efficiency. Looking at the coefficient of determination (R²), it can be said that equation (3) yielded data quite close to the reality. The comparison between the experimental and calculated ratios of treated and untreated filtration efficiencies is depicted in Figure 5. It can be inferred that the contribution of the viscous liquids to the initial filtration efficiency was more than that to the final filtration efficiency. A tentative explanation for this behavior is given below. The initial filtration efficiency was measured after an amount of dust was accumulated onto the filter media. Nevertheless, this amount was remarkably small. Hence, the initial filtration efficiency seemed to be decided by the fiber characteristics, media structure, and liquid properties. Whereas the final filtration efficiency was determined after a significant amount of dust was deposited onto the filter media. (The amount of dust can be calculated from the dust holding capacity of the filter media.) Hence, the final filtration efficiency was mainly decided by the accumulated dust particles. Therefore, the contribution of the viscous liquids to the initial filtration efficiency was more than that to the final filtration efficiency. As the initial filtration efficiency is a very critical parameter as it is highly correlated to engine wear, it can be inferred that the viscous liquids would seem to play a very important role in reducing the engine wear. Further, the filter media treated with glycerol exhibited highest initial filtration efficiency, followed by the filter media treated with SAE 20W/50 engine oil, PEG 400, and lastly deionized water. The same trend was followed in case of final filtration efficiency. This might be ascribed to glycerol producing the largest mean droplet size amongst all liquids for the same add-on. Please follow the next paragraph where a discussion is made on the droplet size. As a result, glycerol was able to capture the greatest number of dust particles of same size by means of adhesion mechanism.

Table 2.

Characteristics of power-law model.

Liquid	Initial filtration efficiency		Final filtration efficiency
	Exponent a	R 2	Exponent a	R 2
Deionized water	0.50	0.97	0.35	0.97
PEG 400	0.55	0.98	0.41	0.96
SAE 20W/50	0.59	0.96	0.48	0.96
Glycerol	0.61	0.92	0.50	0.96

Figure 5.

Plots of comparison between experimental and calculated ratios of treated and untreated filtration efficiencies: (a) initial filtration efficiency and (b) final filtration efficiency.

It would be interesting to analyze how the liquid jets emerged from the atomizer was broken down into droplets. This might give an insight of the spray characteristics. For this, it is necessary to capture the relative magnitude of different forces acted on the liquid jets while in motion. It is also required to think about the relative balance of timescales characterizing the breakup or thinning dynamics of the liquid jets. Table 3 reports on a set of important dimensionless numbers for the four liquids.

Table 3.

Non-dimensional parameters of liquids.

Dimensionless numbers	Definition	Deionized water	PEG 400	SAE 20W/50 oil	Glycerol
Reynolds number	$\mathit{R}\mathit{e}=\frac{{\rho }_{\mathrm{l}}{V}_l{d}_n}{{\mu }_{\mathrm{l}}}$	44403.04	412.31	265.33	49.02
Froude number	$\mathit{F}\mathit{r}=\frac{V_l}{\sqrt{g{d}_n}}$	240.61	226.09	255.98	214.09
Weber number	$\mathit{W}\mathit{e}=\frac{V_l}{\sqrt{{\sigma }_l/\left({\rho }_{\mathrm{l}}{d}_n\right)}}$	124.72	163.30	187.08	133.33
Ohnesorge number	$Oh=\frac{{\mu }_{\mathrm{l}}}{\sqrt{{\sigma }_l{\rho }_{\mathrm{l}}{d}_n}}$	0	0.40	0.71	2.72

Here, $V_{\text{l}}$ is velocity of liquid, ${\text{ρ}}_{\text{l}}$ is density of liquid, ${\text{μ}}_{\text{l}}$ is dynamic viscosity of liquid, ${\text{σ}}_{\text{l}}$ is surface tension of liquid, $d_{\text{n}}$ is nozzle diameter, and g is acceleration due to gravity. The velocity was calculated from the following expression: $V_{\text{l}}=\sqrt{2P_{\text{l}}/{\text{ρ}}_{\text{l}}}$ , where $P_{\text{l}}$ is the pressure acting on the liquid. The liquid velocity was found to be highest with SAE 20W/50 engine oil (30.15 m/s), followed by deionized water (28.34 m/s), PEG 400 (26.63 m/s), and glycerol (25.22 m/s). As the pressure was kept constant, this difference was caused due to the difference in the density of the liquids. Looking at the Reynolds number, it can be understood that deionized water showed a much greater relative contribution of the inertial force, while the other three liquids exhibited a greater relative contribution of the viscous force. The highest contribution of the viscous force was found with glycerol, followed by SAE 20W/50 engine oil, PEG 400, and deionized water, respectively. The Froude number, pointing at the relative contribution of inertial to gravitational force, followed the same order as the velocity of the liquids. The Weber number, expressing the ratio of inertial to surface tensional force of the liquids, is also often used to predict the droplet size. The most common way to characterize the droplet size is the median diameter based on volume. Hsiang et al. ²⁴ proposed the following equation for determination of the volume-based median diameter $\left(D_{50}\right)$

$$D_{50}=\mathrm{\alpha }\hspace{0.17em}\mathrm{\beta }{\left(\frac{{\rho }_{\text{a}}}{{\rho }_{\text{l}}}\right)}^{-1/6}{\left(We\right)}^{-1/3}$$

(4)

where $\alpha \hspace{0.17em}$ is a characteristic length of the nozzle, $\beta$ is a dimensionless constant, ${\rho }_{\text{a}}$ is density of air $\left(1183.9\text{kg}/{\text{m}}^3@{25}^{\circ }\text{C}\right)$ . It can be noticed that the droplet size would be high if the density and surface tension of the liquid are high. The ratio of volume-based median diameters of deionized water, PEG 400, SAE 20W/50 engine oil, and glycerol was computed as 0.38:0.77:0.78:1.19. Clearly, glycerol formed the largest droplet amongst the four liquids. There is another important dimensionless number often referred to as Ohnesorge number in spraying technology. It represents a ratio of viscous force to the product of inertial and surface tensional forces. Interestingly, this number does not involve liquid velocity. A zero Ohnesorge number for deionized water was ascribed to an exceptionally low viscous force. This number was found to be the highest with glycerol, followed by SAE engine oil and PEG 400, respectively. A larger Ohnesorge number indicated a greater influence of viscosity. A regime map based on Reynolds number and Ohnesorge number would suggest that the liquids had sufficient energy to form droplets via wave-like breakup by air friction. This kind of breakup usually happens at a distance closer to nozzle exit and the droplets tend to be smaller. The Ohnesorge number can also be expressed by the ratio of two timescales namely, visco-capillary timescale $\left(t_{\text{visc}}\sim {\text{μ}}_{\text{l}}{d}_{\text{n}}/{\text{σ}}_{\text{l}}\right)$ characterizing the thinning dynamics of a viscous liquid jet and Rayleigh timescale $\left(t_{\text{R}}\sim \sqrt{{\text{ρ}}_{\text{l}}{d}_{\text{n}}^3/{\text{σ}}_{\text{l}}}\right)$ describing the breakup of an inviscid liquid jet. This provides us with the relative contribution of how large each of these timescales is for a liquid jet of diameter approximately $d_{\text{n}}$ , when the density, viscosity, and surface tension of the liquid are known. Of all the liquids, glycerol was supposed to exhibit the slowest thinning of liquid jets, followed by SAE 20W/50 oil, PEG 400, and deionized water, respectively, however, their breakup velocities $\left(V_{\text{R}}\sim d_{\text{n}}/t_{\text{R}}\right)$ were likely to be more or less same.

It would be exciting to know what happens to the particles once they impact on the liquid-treated fibers. Buck et al. ²⁵ postulated that the particles would bounce back after impacting on the fiber surface if the kinetic energy of the particles after impact is greater than the adhesive forces experienced by the particles. Otherwise, the particles would remain captured by the filter media. Upon impact the particles undergo four stages of collision. In Stage I, the particles penetrate through the liquid layer. In Stage II, the particles hit the filter surface. In Stage III, the particles bounce back and emerge from the liquid. In Stage IV, the particles form a bridge with the liquid layer. Here, the forces acting on the particles are the viscous dissipation force during penetration and emergence; the contact force during collision of particle to the surface; the capillary force during penetration, emergence, and bridging; and the gravitational force. While we do not simulate any numerical model in the current study, we use the following force balance equation to qualitatively understand the role of liquids in hindering the particle bounce. The force balance equation can be written as follows

$$m_{\text{p}}\frac{d{v}_{\text{p}}}{d{t}_{\text{p}}}={\displaystyle \sum F_{\text{p}}}$$

(5)

where $m_{\text{p}}$ is the mass of the particle, $v_{\text{p}}$ is the velocity of the particle, $t_{\text{p}}$ is the time taken by the particle, and $F_{\text{p}}$ is the net force acting on the particle. According to Buck et al. ²⁵ , the net force can be expressed as a sum of all individual force components acting on the particles. This is shown below

$$F_{\text{p}}=F_{\text{vis, p}}+F_{\text{con, p}}+F_{\text{cap, p}}+F_{\text{gra, p}}$$

(6)

where $F_{\text{vis, p}}$ stands for viscous dissipative force, $F_{\text{con, p}}$ denotes contact force, $F_{\text{cap, p}}$ indicates capillary force, and $F_{\text{gra, p}}$ refers to gravitational force. Of all the forces, it was reported that the viscous dissipative force and the capillary force are the most dominating for hindering the bounce of particles upon impact. The viscous dissipative force, $F_{\text{vis, p}}$ , which is acting during penetration and emergence, is mathematically given by equations (7) and (8). Equation (7) is applicable for the viscous force $F_{\text{vis, p, }\infty }$ when the distance between the particle and the filter surface, $d$ , is very high. On the contrary, equation (8) describes the viscous force $F_{\text{vis, p, filter}}$ when the distance is small.

$$F_{\text{vis, p, }\infty }=6\hspace{0.17em}\mathrm{\pi }\hspace{0.17em}{\mathrm{\mu }}_{\text{l}}\hspace{0.17em}{r}_{\text{p}}\sin \left(\mathrm{\psi }\right)v_{\text{p}}\left[1+\frac{R{e}_{\text{p}}}{4\left(1+\sqrt{R{e}_{\text{p}}}\right)}+\frac{R{e}_{\text{p}}}{60}\right]R{e}_{\text{p}}=\frac{2\hspace{0.17em}{r}_{\text{p}}\sin \left(\mathrm{\psi }\right)v_{\text{p }}{\mathrm{\rho }}_{\text{l}}}{{\mathrm{\mu }}_{\text{l}}}$$

(7)

where ${\mu }_{\text{l}}$ is the viscosity of the liquid, $r_{\text{p}}$ is the radius of the particle, $\psi$ is the half-filling angle, $v_{\text{p}}$ is the velocity of the particle, $R{e}_{\text{p}}$ is the Reynolds number, and ${\rho }_{\text{l}}$ is the density of the liquid. The viscous dissipative force in a region close to the filter surface is given by

$$F_{\text{vis, p, filter}}=\frac{6\hspace{0.17em}\pi \hspace{0.17em}{\mu }_{\text{l}}\hspace{0.17em}{r_{\text{p}}}^2{v}_{\text{p}}}{d}$$

(8)

where $d$ stands for the distance between the particle center from the filter surface and the thickness of the liquid layer. Since this equation tends to infinity for the distance approaching to zero, a critical distance of 0.5 μm was considered. ²⁵ Depending upon the magnitude of the individual viscous dissipative force, the one with greater magnitude will be taken. This is shown below.

$$F_{\text{vis, p}}=\{\begin{array}{l}{F}_{\text{vis, p, }\infty }\hspace{1em}\hspace{1em}\hspace{0.17em}\text{if}\text{}{F}_{\text{vis, p, }\infty }\hspace{1em}\ge F_{\text{vis, p, filter}}\\ F_{\text{vis, p, filter}}\hspace{1em}\hspace{1em}\text{if}{F}_{\text{vis},\hspace{0.17em}\text{p},\hspace{0.17em}\text{filter}}>F_{\text{vis, p, }\infty }\end{array}$$

(9)

The surface tension of the liquid increases the capillary forces which is acting on the particle during penetration, emergence, and bridging to hinder particle bounce. The capillary force $F_{\text{cap, p}}$ can be expressed as a sum of two components, shown in equation (10). The first component $F_{\text{cap, p,}\text{}\sigma }$ relates to the surface tension of the liquid layer and the second term $F_{\text{cap, p, P}}$ is associated with the capillary pressure.

$$F_{\text{cap, p}}=F_{\text{cap, p, }\mathrm{\sigma }}+F_{\text{cap, p, P}}=-2\hspace{0.17em}\mathrm{\pi }\hspace{0.17em}{\mathrm{\sigma }}_{\text{l}}\hspace{0.17em}{r}_{\text{c}}\sin \left(\mathrm{\psi }+\mathrm{\vartheta }\right)+2\hspace{0.17em}\mathrm{\pi }\hspace{0.17em}{\mathrm{\sigma }}_{\text{l}}\hspace{0.17em}{r}_{\text{c}}$$

(10)

where ${\mathrm{\sigma }}_{\text{l}}$ is the surface tension of the liquid, $\mathrm{\vartheta }$ is the contact angle, $r_{\text{c}}$ is the radius of that part of the particle which is in contact with the liquid. Both of equations (7) and (10) involve a great complexity for the use of this theory as the half-filing angle, contact angle, and the contact radius are not known. As particle collision is a dynamic process, the said angles and radius are continuously changing while forming the capillary bridge between the particle and the liquid. Once the capillary bridge is fully developed and it achieves a constant shape, the half-filling angle, the contact angle, and the contact radius become constants. As our objective was not to simulate and verify this model, we did not make any attempt to evaluate the said angles and radius dynamically. Rather, in order to have a qualitative understanding of the role of liquids in hindering particle bounce, we hereby considered a set of values for the aforesaid parameters to obtain graphs. The half-filling angle was taken as 45°, the contact radius was considered as equal to the particle radius (maximum value), the contact angles for glycerol, oil, PEG, and deionized water were taken as 92°, 67°, 48°, and 101°, respectively.^25–27 Of course, these values could be taken differently owing to the dynamic character of the particle collision process, but this would not change the overall conclusion drawn on the effects of liquid properties on the bounce of particles. The trends of the viscous dissipating force and the capillary force against different particle sizes for the four different liquids taken in this work are displayed in Figure 6. Clearly, the viscous dissipative force was much higher in magnitude than the capillary force. This infers that the viscosity plays a much more important role than the surface tension for inhibiting particle bounce. The viscous dissipative force was the highest in case of filter media treated with glycerol, followed by SAE 20W/50, PEG 400, and deionized water, respectively. The duration for which the viscous dissipative force acted on a particle while penetrating and emerging would increase with the increase of thickness of the liquid layer. Hence, the higher was the add-on of the liquid, the higher was the inhibition of particle bounce, hence higher filtration efficiency.

Figure 6.

Plots of viscous dissipating force and capillary force against particle sizes.

Fractional filtration efficiency

The measurement of gravimetric efficiencies provides a limited information on filter behavior, since it does not give any information on the size of particles penetrating the filter media. As the size of particles is known to strongly influence the engine wear, it is critical to know about the efficacy of the engine intake filter media for different particle sizes. In addition, the gravimetric measurement of the initial efficiency does not provide an accurate information on what is exactly happening at the beginning of the particle filtration process. This is so because the initial gravimetric efficiency does not correspond to clean filter media. The measurement of initial filtration efficiency at zero dust loading would reflect the initial stage of particle filtration more accurately, however, the measurement variability will be significantly high there. Therefore, the measurement of fractional filtration efficiency is often recommended to carry out at on filter media zero dust loading.

The filtration efficiency of an untreated and clean filter medium can be expressed as stated hereunder

$$E_{\text{F}}=1-\exp \left[-\frac{4{\text{η}}_{\text{f}}\hspace{0.17em}{\text{μ}}_{\text{F}}\hspace{0.17em}{Z}_{\text{F}}}{\text{π}\hspace{0.17em}{d}_{\text{f}}\left(1-{\text{μ}}_{\text{F}}\right)}\right]$$

(11)

where E_F stands for overall filtration efficiency, η_f refers to single fiber efficiency, d_f denotes fiber diameter, μ_F indicates packing density of filter media, and Z_F represents thickness of filter media. The spraying of viscous liquids ultimately results in higher fiber diameter and increase of packing density. The increase of fiber diameter upon liquid treatment was modeled by Agranovski ²⁸ as shown below

$$d_{\text{f,l}}=d_{\text{f}}\sqrt{\left(1+\frac{w_{\text{l}}}{w_{\text{F}}}\frac{{\text{ρ}}_{\text{f}}}{{\text{ρ}}_{\text{l}}}\right)}$$

(12)

where $d_{\text{f,l}}$ denotes liquid-treated fiber diameter, $w_{\text{l}}$ stands for weight of liquid per unit area of filter media, $w_{\text{F}}$ stands for basis weight of filter media, ${\text{ρ}}_{\text{f}}$ indicates density of fiber, and ${\text{ρ}}_{\text{l}}$ represents density of liquid. The packing density of the liquid-treated filter medium was expressed by Maddineni et al. ¹⁸ as follows

$${\text{μ}}_{\text{F,}\text{l}}=\frac{w_{\text{F}}+w_{\text{l}}}{Z_{\text{F}}\left(\frac{g_{\text{f}\hspace{0.17em}}{\text{ρ}}_{\text{f}}+g_{\text{l}\hspace{0.17em}}{\text{ρ}}_{\text{l}}}{g_{\text{f}}+g_{\text{l}}}\right)}$$

(13)

where ${\text{μ}}_{\text{F,l}}$ refers to packing density of liquid-treated filter media, $g_{\text{f}}$ indicates weight fraction of fibers in filter media, and $g_{\text{l}}$ represents weight fraction of liquid in filter media. Using the aforesaid two expressions, the filtration efficiency of liquid-treated filter medium can be expressed as under

$$E_{\text{F,}\text{l}}=1-\exp \left[-\frac{4{\text{η}}_{\text{f,}\text{l}}\hspace{0.17em}{\text{μ}}_{\text{F,}\text{l}}\hspace{0.17em}{Z}_{\text{F,}\text{l}}}{\text{π}\hspace{0.17em}{d}_{\text{f,}\text{l}}\left(1-{\text{μ}}_{\text{F,}\text{l}}\right)}\right]$$

(14)

where $Z_{\text{F,l}}$ refers to thickness of liquid-treated filter medium and $\text{η}{\hspace{0.17em}}_{\text{f,l}}$ stands for single fiber efficiency of liquid-treated fiber. The last term involves two parameters namely single fiber efficiency due to collision $\text{η}{\hspace{0.17em}}_{\text{f}}$ (diffusion, interception, impaction, and gravitational settling) and single fiber efficiency due to adhesion $\text{η}{\hspace{0.17em}}_{\text{f,l,a}}$ . The relationship can be expressed as follows

$$\text{η}{}_{\text{f,}\text{l}}\hspace{0.17em}=\text{η}{}_{\text{f}}\mathrm{\cdot }\text{η}{}_{\text{f,}\text{l,}\text{a}}$$

(15)

It is assumed that the thickness of the filter medium did not change upon liquid spraying. Dividing equation (14) by (11) and then by using equations (12) and (14), we obtain

$$\text{η}{}_{\text{f,}\text{l,}\text{a}}\hspace{0.17em}=\hspace{0.17em}\frac{\ln \left(1-E_{\text{F,}\text{l}}\right)}{\ln \left(1-E_{\text{F}}\right)}\frac{d_{\text{f,}\text{l}}}{d_{\text{f}}}\frac{{\text{μ}}_{\text{F}}}{{\text{μ}}_{\text{F,}\text{l}}}\frac{\left(1-{\text{μ}}_{\text{F,}\text{l}}\right)}{\left(1-{\text{μ}}_{\text{F}}\right)}$$

(16)

This expression can be used to calculate single fiber efficiency due to adhesion upon liquid treatment.

Figure 7 plots the fractional filtration efficiencies of the liquid-treated filter media for different particle sizes. It can be observed that the filtration efficiencies increased due to liquid treatment regardless of any particle size. The adhesion of particles onto the liquid-treated fibers was significant. Interestingly, the increase of filtration efficiency was more for smaller particles as compared to larger ones. It was expected that a higher number of smaller particles would be adhered to a given volume of liquid droplet as compared to large particles. This was ascribed due to the fact that a smaller particle occupies less space than a larger one. The liquid treatment was therefore found to be more effective for filtering out smaller particles than the larger particles from the airstreams. Out of the four viscous liquids chosen for chemical treatment in the current study, glycerol resulted in the highest filtration efficiency followed by SAE 20W/50 oil, PEG 400, and deionized water. Being most sticky among the four liquids, glycerol offered the highest level of viscous dissipative force to adhere the particles onto the filter media. On the other hand, SAE 20W/50 oil offered a higher deposition volume than PEG 400, though the former was slightly more viscous than the latter. Here, the deposition volume outplayed the liquid viscosity, resulting the engine oil in capturing more particles than the low molecular weight grade of polyethylene glycol. In case of deionized water, its faster evaporation might have led to decrease the adhesion of particles. Glycerol was found to be the best of the four liquids used in this study for enhancement of fractional filtration efficiency.

Figure 7.

Plot of fractional filtration efficiency of dry and liquid-treated filter media: (a) Deionized water; (b) PEG 400; (c) SAE 20W/50 oil; (d) Glycerol.

Figure 8 displays the trend of single fiber efficiency due to adhesion as a function of weight of liquid in the liquid-treated nonwoven filter media. This efficiency was calculated from equation (16). As shown, this was decided by the overall fractional filtration efficiencies, fiber diameters, and packing densities of the liquid-treated and untreated filter media. It can be noticed that the adhesion efficiency generally increased with the increase of liquid add-on. This explains higher fractional filtration efficiency of the liquid-treated filter media as compared to that of the untreated one. Further, it can be noticed that the adhesion efficiency was higher for the smaller particles as compared to the larger ones. This was associated with the higher specific surface of smaller particles as compared to larger ones. As a result of this, a greater number of smaller particles can be adsorbed by a given volume of liquid present on the surface of a fiber. Interestingly, this trend of adhesion efficiency was found to be opposite to that of overall fractional filtration efficiency. The adhesion efficiency was found to be highest with glycerol-treated fiber, followed by the fibers treated with SAE 50W/40 engine oil, PEG-400, and deionized water, respectively.

Figure 8.

Plot of single fiber efficiency due to adhesion as a function of weight of liquid: (a) Deionized water; (b) PEG 400; (c) SAE 20W/50 oil; (d) Glycerol.

Evolution of pressure drop

Pressure drop is a key determinant of cost-efficiency for engine intake air filter media. The higher is the pressure drop the higher is the energy required for the air to flow through the filter media. The air is required for the supply of oxygen for fuel combustion. Figure 9 displays the trend of pressure drop of clean filter media treated with various levels of different liquids. In a true sense, this pressure drop can be termed as initial pressure drop. As shown, the untreated filter media exhibited lower initial pressure drop than the liquid-treated filter media. This is regardless of the kind of liquid employed to treat the filter media. The reason for this behavior can be explained based on the structure of the filter media. It is well known that the structural parameters on which the initial pressure drop mainly depends are thickness and porosity. The lower is the thickness and the higher is the porosity, the lower is the initial pressure drop and vice-versa. While the thickness of the filter media remained unchanged upon liquid treatment, it was the reduction in porosity that resulted in increase of initial pressure drop. Upon liquid deposition, the porosity of the filter media decreased due to reduction in pore volume. ¹⁷ Glycerol and engine oil showed the highest increase of initial pressure drop. While the volume of deposited engine oil was higher than that of glycerol, due to difference in their densities, the deposited glycerol might have occupied mostly on the upstream face of the filter media without seeping through due to high viscosity. Therefore, the initial pressure drop across the glycerol treated filter media matched up with the oil treated filter media. The PEG treated filter media showed an intermediate increase in pressure drop due to low deposition volume and low viscosity. The water treated filter media lay lowest in the plot. It is to be emphasized that other factors like evaporation during test might have reduced the quantity of water contained in the treated media.

Figure 9.

Plot of initial pressure drop across dry and liquid-treated filter media.

In reality, the pressure drop across the filter media continuously increases with the accumulation of dust particles. This is allowed till a terminating pressure drop is reached and the filter media is called for replacement. The rise of pressure drop is important to examine as it helps us draw a crucial inference regarding the life of the filter media. Figure 10 displays the evolution of pressure drop of the nonwoven filter media treated with four different liquids at five various levels of add-on. The pressure drop across various filter media was recorded against incremental dust feed. The experiments were continued till the pressure drop across the filter media rose to a terminating pressure drop of 1000 Pa above the initial pressure drop. It can be noticed that the evolution of pressure drop changes with a variation in liquid add on. As the liquid add on was higher the rise of pressure drop was slower. Each pressure drop curve can be characterized as a combination of an initial slow rise followed by an abrupt and markedly linear increase. In the case of untreated filter media, the initial slow rise of pressure drop was associated with dust accumulation in the depth of the filter media and the abrupt rise related to dust cake formation on the surface of the filter media. ²⁹ As compared to the untreated filter media, the gradual rise of pressure drop as well as the abrupt increase of pressure drop across the liquid-treated filter media happened further slowly. The initial slow rise of pressure drop across the liquid-treated filter media was an indicative of dust absorption by the liquid droplets. As this stage did not significantly alter the pore volume of the filter media (see Figure 11), the rise of pressure drop was slow. Whether this stage was completed or not, that is, the liquid droplets were fully or partially packed with the particulate matter, the dendrites started to form and grow. The dendrite growth of the particles caused a huge obstruction to the flow of air, which resulted in abrupt increase of pressure drop. It can be inferred from Figure 10 that glycerol showed the slowest rise of pressure drop at all treatment levels among all liquids. It is interesting to note that the pressure drop curves were most distinct for glycerol. This points out to the fact that the incremental volume of glycerol had the highest effect over the evolution of pressure drop among all liquids.

Figure 10.

Plots of evolution of pressure drop in case of dry and treated filter media with different liquid add-on: (a) Deionized water; (b) PEG 400; (c) SAE 20W/50 engine oil; (d) Glycerol.

Figure 11.

Depiction of dust deposition morphologies on a dry fiber (top row) and on a liquid-treated fiber (bottom row).

The dust particles get accumulated in the filter media during the filtration process. The weight of dust particles accumulated until the terminal pressure-drop is reached is known as dust holding capacity of the filter media. The higher is the dust holding capacity the higher is the life of the filter media. Figure 12 displays the trends of dust holding capacity in liquid-treated filter media as a function of the weight of liquid added to the filter media. The liquid-treated filter media exhibited much higher dust holding capacity than the untreated filter media. The liquid droplets on the fibers absorbed the particulate matter without imparting any significant change in the structure of the filter media. Hence, the pressure drop during the dust absorption phase remained practically unchanged. As a result, the amount of dust loaded onto the filter media was higher than the untreated media till the terminal pressure drop was reached. The dust holding capacity of glycerol treated filter media was recorded to be the highest at all levels of liquid add-on. This was associated with the slowest rise of pressure drop across the glycerol treated filter media. This might be due to high deposition volume of glycerol in the upstream of the filter media which facilitated to absorb the particulate matter and delay the dendrite formation of dust particles. The water-treated filter media showed the least dust holding capacity. As stated earlier, the evaporation of water during the test might have led to very small amount of water left deposited onto the filter media to absorb the particulate matter. Of the remaining two liquid-treated filter media, the oil treated filter media exhibited a little higher dust holding capacity than the PEG treated filter media. The volume of deposition of oil was higher than that of PEG, hence the former absorbed more dust particles than the latter.

Figure 12.

Plot of dust holding capacity of dry and liquid-treated filter media.

In order to see if there was any difference in the nature of dust accumulation onto the filter media treated with different liquids, the light microscopic images of the dust-loaded filter media were taken. This is shown in Figure 13. It can be noticed that there is not much difference in the pattern of dust deposition onto the untreated and water-treated filter media, barring the fact that the dust looks more brown in latter. This was supposed to be due to the interaction between the water and dust particles. Water has low viscosity due to which it may have transported to the downstream part of the filter media, reducing its content on the upstream part. Furthermore, evaporation may also have caused reduction of water content in the media, as discussed in previous section, causing the deposited dust cakes to appear similar to that in untreated media. For the other liquid-treated media, the deposition appears to be gel-like. This is presumably due to absorption of dust into the deposited liquid. Nevertheless, dendrites did not appear to have started forming on the treated or untreated filter media.

Figure 13.

Microscopic images of dust loaded filter media (at the same treatment level): (a) Untreated; (b) Water treated; (c) PEG 400 treated; (d) Engine oil treated; (e) Glycerol treated.

Conclusion

Spraying of viscous liquids onto nonwoven filter media resulted in higher gravimetric as well as fractional filtration efficiencies. The initial and final filtration efficiencies were directly related to liquid add-on via a power law relationship. The contribution of viscous liquid to the initial filtration efficiency was more than that to the final filtration efficiency. The increase in fractional filtration efficiency due to liquid treatment was more for smaller particles as compared to larger ones. This was explained through single fiber efficiency due to adhesion. In liquid-treated filter media, the initial pressure drop was higher, but the rise of pressure drop during dust loading was slower than the untreated media. The liquid-treated filter media also displayed a remarkably higher dust holding capacity than their untreated counterparts. However, no discernible difference in the dust deposition pattern was observed in the filter media upon liquid treatment. The viscosity of liquid was found to be a very crucial parameter as the dust deposition morphology was contingent to the flow of liquid onto the filter media. Of all the liquid-treated filter media, the one treated with glycerol exhibited highest filtration efficiency, slowest rise of pressure drop, and highest dust holding capacity.