TECHNICAL ARTICLE
Testing of Glass Fiber Coalescing Filters G. Belforte, T. Raparelli, and A. Trivella Department of Mechanical and Aerospace Engineering, Politecnico di Torino, Torino, Italy
Keywords Efficiency Air–Oil Filters, Coalescent Filters, Filter Media, Fiber Glass, Fibrous Filters Correspondence A. Trivella, Department of Mechanical and Aerospace Engineering, Politecnico di Torino, Torino, Italy Email:
[email protected] Received: October 20, 2011; accepted: December 19, 2011 doi:10.1111/j.1747-1567.2012.00806.x
Abstract The paper presents an experimental study of coalescing filters for separating oil from compressed air in industrial systems. A test bench for measuring filter efficiency was set up which reproduces several typical operating conditions of pneumatic systems. Tests were carried out on commercial products, using filters of different sizes as well as several borosilicate cartridges of similar size and shape. Preliminary analysis of cartridges indicated significant differences in glass fiber dimensions and binder composition. Test results made it possible to compare performance achieved by the different configurations in terms of efficiency and pressure drop. Further measurements were repeated with different cartridge supply system geometries. One of the tested filter-cartridgesupply system configurations was then used to investigate system behavior while varying certain operating parameters individually: air velocity, oil concentration, filter supply pressure, and operating time. Results are presented in statistical form.
Introduction The compressed air in a pneumatic system is normally moist and contaminated, containing by-products of wear from moving parts, metal particles, and oxides, and, often, quantities of oil resulting from compressor lubrication or production needs. Coolers and driers are used downstream of the compressor to reduce condensation, while filters of varying grade are provided along the line to retain solid particles of different sizes. Removing oil is essential both in order to ensure that the air is safe to breathe in the area where it is released, and to guarantee the cleanliness required for many applications, for example, food processing, pharmaceuticals, metrology, or electronics industries. ISO 8573-1 classifies the cleanliness of air for industrial use into several purity classes, according to the concentrations and particle sizes of the solid and liquid contaminants contained in the air. In particular, permissible oil concentrations range from 5 to 0.01 mg/m3 ANR (volume of air second the ‘‘Atmosfere Normale de Reference’’ conditions) with purity classes from 4 to 0. Considering that oil concentration at the compressor outlet can vary from 0.2 to 2 g/m3 , depending on machine wear, Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
the importance of the task performed by purification filters is clear. Coalescing filters are normally used to separate oil from compressed air. These filters consist of a random mass of inorganic fibers embedded in a binder featuring a high porosity of around 90%. Various kinds of binder are used: phenolic, epoxy, acrylic. As the binder is chemically inert and nonabsorbent, it is capable of maintaining sufficient structural integrity over time.1 The volume actually occupied by the fibers as a ratio of the total volume, that is, the packing fraction, is small, which makes it possible to achieve low pressure drops as air crosses the filter. Liquid particles are captured by the fibers and slide toward the outer surface of the filter, forming larger and larger droplets as a result of coalescence until they fall by gravity into the filter’s collection chamber. Coalescing filters can also operate as particulate filters, taking advantage of the fibers’ ability to capture solid contaminants. Filtration efficiency, or the ratio of the concentration of liquid retained and the amount entering the filter, is determined by a number of mechanisms involved in contaminant capture by the fibers.2 – 4 These mechanisms are heavily dependent on the relative dimensions of the particles and fibers, packing fraction, and flow conditions. 1
Coalescing Filters
Particles under 0.3 μm in size move chaotically without following the direction of air flow (Brownian motion): the smaller the particle, the more random the motion, and the higher the probability that the particle will be captured (Brownian diffusion). Particles measuring between 0.3 and 1 μm follow the air flow path until they collide with the surrounding fibers (direct interception). Particles between 1 and 10 μm in size have sufficient momentum to move away from the direction of flow until they strike the surrounding fibers (inertial impaction). Larger particles fall first as an effect of their weight while crossing the filter, and during their downward motion can be captured by the fibers and thus separated (gravity settling). Filter efficiency is the sum of the efficiencies resulting from each individual collection mechanism. Particles near 0.3 μm in size are the most difficult to capture, as the Brownian diffusion and direct impaction mechanisms are not particularly efficient in this size range. For this reason, coalescing filter manufacturers normally measure efficiency with the DOP (dioctyl phthalate) method, which uses a monodispersed aerosol of DOP oil droplets whose dimensions are close to 0.3 μm. With the DOP test, oil concentration averages 0.1 g/m3 (ANR), pressure is slightly above the ambient pressure and the velocity of the air crossing the filter is very low, not over 1 m/s. With these parameters, the duration of the test is not sufficient to bring the filter element to saturation. In other cases, performance is expressed by the quality factor, an index of efficiency that also takes pressure drop upstream and downstream of the filter into account. A number of different theoretical models have been developed to estimate filter performance under specific operating conditions. For example, Liew et al.5 demonstrate that higher efficiency is linked to an increase in inertial impaction mechanisms due to filter filling, while Raynor and Leith6 also consider the effects of liquid drainage, reentrainment, and evaporation on filter filling. The characteristics of the filter (materials, fiber size, packing fraction, wettability) and of the aerosol (surface tension, viscosity, dimension) influence filter efficiency. Currently, the use of glass microfibers embedded in polymer binders makes it possible to achieve liquid–air separation efficiencies of 0.95 or more as measured with the DOP method. A number of studies are now being carried out in order to improve filter efficiency in relation to the various parameters involved. Thus, Contal et al. and Charvet et al.7,8 investigate the effects of air velocity, 2
G. Belforte, T. Raparelli, and A. Trivella
filter type, and aerosol concentration and characteristics on filter clogging. The effect of glass fiber wettability on the coalescence mechanism is investigated in Ref. 9 using several water-in-oil emulsions and different surface coatings on a glass rod. In Ref. 10, similar studies are conducted directly on fibrous filters treated with nanoparticle coatings; Mullins et al.11 compare the capillarity of fibrous filter media, which is linked to the oil’s surface tension, with that of several theoretical models in order to predict packing density, while Yang and Chang12 evaluate the wettability of anthracite and quartz sand filters by measuring liquid penetration in specimens. In Ref. 13, it is demonstrated that an increase in temperature and humidity can significantly improve the performance of glass fiber filters. The use of a combination of B and E glass fibers that eliminate the need for common acrylic and epoxy binders to stiffen the structure, thus improving efficiency, is illustrated in Ref. 14. Currently, nanotechnologies make it possible to develop new, high efficiency filter media.15 – 16 A survey of the present status of nanofibrous filtering research is presented in Ref. 17. For example, Hajra et al.13 and Shin et al.18 demonstrate how adding polymer nanofibers to the glass fiber media can increase efficiency, while Shin et al.19 show that there is an optimum quantity of nanofibers that balances efficiency increases and pressure drop. More recently, Patel and Chase20 investigated the effect on quality factor of woven drainage structures placed on the filter surface and air flow direction on limiting filter-pressure drop. In Ref. 21, the hydrophobic and oleophobic effects of PTFE (polytetrafluoroethylene) foam coating on borosilicate filters for high temperature applications were investigated. A finite element method model for calculating liquid flows at the inlet of porous filters is presented in Ref. 22. Studies have been largely experimental, addressing only the filter element using special-purpose instrumentation and standard methods. Specimens, which are normally circular in shape, are cut from the filter and crossed by a homogeneous flow of air mixed with oil. Oil particles measuring <1 μm are produced using appropriate oil aerosol generators and detected by means of systems such as photoionization detectors. The normal operating conditions of a pneumatic system are generally quite different from those used in experimental tests as regards flow conditions, oil type, and the circuits followed by compressed air. The relatively high air flow rate in the system results in rapid filter saturation, and the filter surface area must be sized so that flow velocity does not exceed Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
Coalescing Filters
G. Belforte, T. Raparelli, and A. Trivella
the limits beyond which the collection mechanisms are no longer effective. Flow conditions, which vary from zone to zone, depend on operating pressures and line geometry. The complicated flow paths and high air velocities promote mixing and coalescing of the oil particles, which differ widely in dimensions from zone to zone.23 In Refs. 24–26, test benches for measuring the efficiency of commercial filters in capturing water and oil particles are presented. Liquid is emitted by means of appropriate metering and atomizing equipment. The higher the flow rate, the more polydispersed the aerosol will be, while the largest droplets, which may measure several dozen microns, influence collection mechanisms in relation to fiber dimensions. Particles of different sizes, from the largest down to the smallest, can be captured progressively with filters of decreasing porosity, with fiber diameters that generally vary from 30 to 0.5 μm. Obviously, low-porosity filters can drain larger particles, but excessive packing fractions can increase resistance to drainage and pressure drop, limiting filter performance.20,27 In pneumatic filters, the filter element consists of a cylindrical cartridge in which air flows from the interior toward the exterior. Borosilicate cartridges are tubular in shape, with a thickness ranging from 0.5 and 2 mm. Some cartridges have a layered structure: the filter layer per se is surrounded internally and externally by more porous layers. The largest droplets are captured first, followed by the smaller ones, while the outermost layer fulfills a drainage function. The cartridge is installed in the filter canister, which is provided with ports for connection to the circuit. Droplets are collected in a bowl located beneath the cartridge. Liquid collection capacity is limited by the presence of the cartridge: to operate correctly, the collected oil must be discharged before the level of its free surface reaches the cartridge. The filter bowl is provided with manual or automatic means for periodically discharging the drained oil. Cartridges can be installed in the canister in different ways, while the geometry of the cartridge supply system depends on the type of installation used. Filters sized for different air flow rate requirements are available on the market. This paper investigates the performance of coalescing filters with glass fiber cartridges in terms of efficiency and pressure drop under the actual operating conditions encountered in a pneumatic system. The test bench set up for this purpose reproduces several typical pneumatic system operating conditions, making it possible to measure the efficiency of the canister–cartridge system with a polydispersed oil Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
aerosol. Tests were carried out on several glass fiber cartridges of differing composition, with and without drainage layers, and with two different canister sizes. One of the cartridges under test was used to investigate the system behavior while varying supply system geometry (considering three different methods of installing the cartridge in the filter) and operating parameters (average air velocity, supply pressure, oil concentration, and operating time). To take the variability in the morphological characteristics of different cartridges of the same type into account, the tests were repeated several times on two cartridges of each type, obtaining results in statistical form.
Filter Cartridges Four cylindrical glass fiber filter cartridges of similar size were tested. Cartridges are shown in Fig. 1(a). Cartridges A, B, C, and D have similar percentages by weight of glass fiber, but differ significantly in average fiber size and binder material. They have an outside diameter of 18–20 mm, length of 56–57 mm, and thickness of approximately 2.5 mm. Types A, B, and D feature an outer synthetic fiber drainage layer from 0.5 to 1 mm thick. The cartridge weights and dimensions shown in Table 1 are the average of measurements on three specimens. Because of the specimens’ surface irregularity, dimensional measurements have an uncertainty of 0.5 mm and the dimensions shown are rounded of to the nearest unit. Specimen weight was determined with an electronic balance with an accuracy of 10−5 kg. Several preliminary analyses were carried out on virgin specimens to evaluate their main physical properties. The filter structure was observed under the electron microscope, evaluating mean diameter of the fibers in the cartridges and the drainage layers. Results were obtained by considering a sufficient number of fibers viewed in each enlargement. The photos in Fig. 1(b) show the enlargements for each specimen. Thermogravimetric analysis (TGA) was used to measure the percentage of glass by weight in both the filtration areas and the drainage layers, verifying the presence of glass fiber only in the former. Fourier transform infrared (FT-IR) spectroscopy was then used to determine the type of resin used. Values for investigated parameters are shown in Table 2. Porosity and DOP efficiency values are provided by the manufacturer for each type. While weight glass content is approximately 70% for all cartridges, mean glass fiber diameters differ significantly, with a range of 3–15 μm. Fiber dimensions in the drainage layers 3
Coalescing Filters
G. Belforte, T. Raparelli, and A. Trivella
for any given type of binder the declared DOP efficiencies are significantly higher for the internal cartridges with smaller diameter of the glass fibers.
(a)
A
B
C
D
Cartridge and Filter Supply Systems
(b)
Figure 1 (a) Cartridges from left to right A, B, C, D; (b) Enlargements of cartridges: (1) spec. A, cartridge; (2) spec. A, outer layer; (3) spec. B, cartridge; (4) spec. B, outer layer; (5) spec. C, cartridge; (6) spec. D, cartridge; and (7) spec. D, outer layer.
are 6–10 times greater than in the cartridges. Resins are phenolic and epoxy, with both materials showing similar structural stiffness and appropriate mechanical strength for the application. For all specimens, the porosity declared by the manufacturer is 95%, while 4
Three different cartridge supply systems were considered, viz.: S1 , S2 , and S3 . These systems can be installed on two filters of different sizes, F1 and F2 . Figure 2(a) shows the cartridges and supply systems installed in the smaller filter, while the photograph in Fig. 2(b) shows each of the supply systems used. Lubricated air reaches the interior of cartridge 2 through inlet ports in distributor 4, which is threaded into canister 1. Oil droplets intercepted by the fibers coalesce together and move to the outer surface of the cartridge, where they fall by gravity into bowl 5. Sealing between the canister and the distributor is accomplished by O-ring 6. In solutions S1 and S2 , threaded shaft 7, which is positively connected to the distributor, makes is possible to seal the cartridge by tightening lock ring 8. As the shaft must be secured to the distributor, it was necessary to provide air passage holes; for solution S1 , the distributor is provided with a collector tube that directs flow downward before crossing the cartridge, while for solution S2 , this tube is not provided and the inlet ports are positioned at a different height than in the first solution. The effective cross-sectional passage area of the ports in distributor S1 is 43 mm2 , while that of distributor S2 is 47 mm2 . In solution S3 , the cartridge is sealed to distributor 4 and end plate 9, thus making the threaded shaft unnecessary. Sealing between end plate and cartridge is ensured by a silicone gasket. The effective cross-sectional passage area of the ports on distributor S3 is 63 mm2 . Figure 3(a) shows the different sizes of filter used. Both feature 3/8 air inlet and outlet ports. An adapter ring was constructed to ensure that the same distributors can also be used on the larger filter. In this way, the supply systems are displaced by approximately 8 mm below the air inlet port in the filter. The liquid collection capacity of the two filters is 14 and 38 cm3 . The input ports on supply system S1 , which is installed in F1 and F2 with a sealed threaded fitting, are perpendicular to incoming air flow. To determine the influence of the different orientation α of the ports in system S1 , while at the same time guaranteeing sealing with the distributor, a modified supply system (S1m ) was produced in which the distributor and the collector tube consist of two separate pieces (Fig. 3(b)). The distributor is threaded into the canister, while the collector, which houses Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
Coalescing Filters
G. Belforte, T. Raparelli, and A. Trivella
Table 1 Cartridges weights and dimensions
d
D2
Type A B C D
D1
L
(mm) 13 13 13 13
20 20 20 21
18 18 — 18
57 57 57 57
Cartridge weight (10−3 kg) Total
Outer layer
2.15 2.44 1.55 2.11
0.43 0.62 — 0.74
Table 2 Cartridges parameters
Type A B C D
Glass weight (measured) (%) 74 76 67 70
Mean fiber diameter (measured) (μm) int
ext
Resin
Declared porosity (%)
3.3 6.4 6.6 9.6
30.2 20.6 — 15.4
Phenolic Phenolic Epoxy Epoxy
95 95 95 95
Declared DOP efficiency 0.9997 0.9500 0.9997 0.9500
(a)
(b)
Figure 2 (a) Supply systems installed in filter F1 and (b) supply systems S1 , S2 , S3 .
Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
5
Coalescing Filters
G. Belforte, T. Raparelli, and A. Trivella
(a)
(b)
Figure 3 (a) Filters of different sizes: F1 , F2 and (b) modified supply system S1m (filter F1 ) for different angles α inlet ports-air flow.
the ports, is positioned at an angle and secured in such a way as to align a set of equally spaced holes and several threaded holes in the first element. Collector length and diameter are the same as in solution S1 , so that the effective cross-sectional passage area is also the same. System S1m was produced in two versions, one for each filter F1 and F2 , with thread diameter equal to that of the corresponding seat in the filter; consequently, it was not necessary to use the adapter ring on the larger filter. In both versions, collector inlet ports are at the same height as the flow entering the filters, so that the performance is compared relative to filter size only.
(a)
(b)
Test Bench and Parameters The test bench, which was based partially on previous work,15 – 17 is shown schematically in Fig. 4(a). The test bench consists of a filter-pressure reducer 1, lubricator 2, filter under test 3, variable resistance 4, and oil removal filter 5 on discharge. Components are connected by Rilsan tubing. To simulate a user circuit, a 5-m long, 6-mm ID coiled line 6 is installed between the lubricator and the filter under test. All other lines have an inside diameter of 10 mm. Air flow is measured by float type flow meter 7 installed upstream of the lubricator. Supply pressure Ps is measured immediately upstream of the flow meter with the pressure gage 8. Pressure P1 upstream of the filter and pressure drop P are measured by pressure gages 9 and 10. A photograph of the bench is shown in Fig. 4(b), where the flow regularizing tubes to which the pressure gages are connected can be seen at filter inlet and outlet. The amount of oil introduced is regulated by the lubricator’s internal metering device, while air velocity and supply pressure are controlled by means of the variable resistance and the pressure reducer. 6
Figure 4 (a) Test bench schematics and (b) test bench.
For all tests, a paraffin oil for compressors with the properties shown in Table 3 was used. The following parameters were established: absolute pressure upstream of the filter P1 = 0.5 MPa, mean velocity v of the air immediately upstream of the cartridge v = 0.4 m/s, and oil concentration coil = 0.25 g/m3 (ANR). The air flow rate corresponding to selected parameters P1 and v is approximately 6.4 × 10−3 m3 /s. The oil level in the lubricator is topped up periodically in order to maintain average concentration within 15% of the nominal value. Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
Coalescing Filters
G. Belforte, T. Raparelli, and A. Trivella
Table 3 Oil properties
(a) Oil
Trade name
ISO grade
Viscosity at 40◦ C
Density
Corim 22/L
22
20.2 cSt
859 kg/m3
Oil Aerosol Visualization As a result of the air flow’s turbulent motion, a certain quantity of oil droplets coalesce together and are deposited on the inner surface of the lines. To evaluate the order of magnitude of the droplets dispersed in the air at cartridge inlet and outlet a simple collection system was used to gather droplets which were then viewed under the microscope. A small screen was placed first on the inner surface of the cartridge and then on the inner surface of the collection bowl. The screen consisted of a thin metal foil to ensure that the droplets were not absorbed, and was sufficiently small to cause minimum change in flow conditions. Selected exposure time was such as to collect a minimum number of droplets, whose number and size are heavily dependent on operating conditions. In the test conditions indicated above, selected time was 2–3 s with the screen placed upstream of the cartridge, and approximately 2 min with the screen downstream. Measurements were repeated several times, placing the screen on both the upper and lower portions of the filter. Droplets were viewed using a fiberoptic microscope which made it possible to distinguish particles down to a few microns in diameter. Droplet size distribution upstream and downstream of the cartridge differed significantly, while positioning in terms of height did not entail substantial variations. Upstream of the filter, the droplets coalesce readily because of the higher concentration of oil, producing a broad size distribution from a few microns to several hundred microns, making it difficult to determine the mean value. Downstream of the cartridge, the low oil concentration results in a much more uniform size distribution, with a maximum size of several dozen microns. The behavior was similar for all cartridges. Figure 5 shows two examples of droplet distribution upstream and downstream of one of the cartridges under test in the conditions indicated above. Efficiency Measurement The test method entails introducing a defined quantity of oil into the fluid current by means of the lubricator. A certain amount of this oil is retained by the filter under test. At predetermined operating intervals, Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
(b)
Figure 5 Droplet distribution upstream (a) and downstream (b) of the cartridge.
both the lubricator and the filter are weighed, thus determining operating efficiency. Given the weights ml , mf , and mt of the lubricator, the filter, and the tubing between them, filter efficiency E is: E=
mf ml − mt
with mf = mf,final − mf,initial , ml = ml,initial − ml,final , mt = mt,final − mt,initial . The filter was connected after an initial transient period in which the tubing between the lubricator and the filter was filled with oil to ensure that term mt between two consecutive measurements was negligible. An electronic balance with a sensitivity of 10−4 kg was used for weight measurements. Results and Discussion Efficiencies were plotted versus filter operating time starting with a new cartridge and repeating weighing operations at intervals selected on the basis of oil flow rate (approximately 6 × 10−3 kg/h) and the sensitivity of the analytical balance. The first set of tests compared efficiencies with different types of cartridge. For types A, B, and D, tests were also carried out on specimens from 7
Coalescing Filters
which the outer layer had been removed. These specimens were marked with an asterisk near the corresponding identification layer. For each type, measurements were performed on two specimens in order to evaluate scatter in results, again starting with a new cartridge. Tests were conducted on filter F1 and F2 , using supply system S1 . Total testing time on each cartridge was at least 10 h, with a minimum of seven efficiency measurements. By way of example, Fig. 6(a) shows efficiencies obtained with filter F1 for the different cartridge types on one of the two specimens under test. In all cases, efficiency is highest with a new cartridge (within the first 2 h), then drops significantly and stabilizes in a range linked to the accuracy with which the weight of the oil introduced in the circuit and collected in (a)
(b)
Figure 6 (a) Efficiency versus time, P1 = 0.5 MPa, v = 0.4 m/s, coil = 0.25 g/m3 (ANR), filter F1 , supply system S1 . (b) Pressure drop versus time, P1 = 0.5 MPa, v = 0.4 m/s, coil = 0.25 g/m3 (ANR), filter F1 , supply system S1 .
8
G. Belforte, T. Raparelli, and A. Trivella
the filter bowl is measured. Percentage measurement errors dropped as the variation in oil mass, and thus the time interval between two consecutive weighing operations, increased. However, this time interval must not be excessive, as it was found to have a major impact on the oil level reached in the filter collection bowl. The more oil collected, the closer its free surface will be to the cartridge, making it more likely to be re-entrained by the air flow downstream of the filter. System efficiency drops drastically when the oil level reaches heights close to the cartridge. It is also clear that this effect is more marked with the smaller filter. Because of the lubrication system used, moreover, the set oil concentration varied by 10–15% between measurements. Comparing the performance of the different cartridges calls for appropriate measurement accuracy: to this end, a specific test procedure was used which entails operating intervals of not more than 1.5–2 h to ensure that oil level in the bowl does not exceed approximately one-third of the distance between the base of the cartridge and the bottom of the bowl. After each weighing operation, the filter bowl was emptied and the oil in the lubricator was topped up to its initial level. Pressure drops for the smaller filter are shown in Fig. 6(b). At the start of the test, pressure drop varies from 200 to 370 hPa depending on cartridge type, while pressure drop is higher (increasing by 50–100 hPa) if a drainage layer is provided. With the parameters used, pressure drop increases rapidly along with the weight of the oil retained during the first hour of operation in all cases; after this time, P increases much more slowly, and the pressure drop then stabilizes during the measurement period. To plot the associated curves accurately, values were recorded more frequently during the first hour of operation. The P curves obtained for the larger filter were similar to the preceding ones, and the values were approximately 40 hPa lower for all cartridges. As can be seen from Fig. 6(a), efficiency variation ranges for the different cartridge types overlap, making it necessary to evaluate scatter in results. Mean efficiency in the measurement period was thus calculated for each cartridge. For filters F1 and F2 , Fig. 7(a) compares the mean efficiencies obtained on one of the cartridge specimens together with the corresponding standard deviations. The data indicate that filter performance for any given borosilicate content does not vary significantly with fiber dimensions, probably because of the presence of a pluridimensional oil aerosol in the Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
G. Belforte, T. Raparelli, and A. Trivella
(a)
(b)
(c)
Figure 7 (a) Comparison between mean efficiencies for different cartridges, P1 = 0.5 MPa, v = 0.4 m/s, coil = 0.25 g/m3 (ANR), supply system S1 . (b) Efficiency versus distributor angular position, supply system S1m , cartridge B. (c) Influence of supply system geometry on filter efficiency. P1 = 0.5 MPa, v = 0.4 m/s, coil = 0.25 g/m3 (ANR), cartridge B.
pneumatic system. By contrast, performance varies appreciably with filter size. For filter F1 , mean efficiencies varied from 0.70 to 0.86, depending on cartridge type. For filter F2 , efficiencies were significantly higher, ranging from 0.85 to 0.99. Standard deviations vary by no more Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
Coalescing Filters
than 6% from the mean values, while in several cases scatter does not exceed 3% for the larger filter. For this filter, the advantages of the outer drainage layer were also clearer. When the same measurements were repeated on a second specimen, it was found that mean efficiency did not vary from the earlier measurements by more than 10% for filter F1 and 5% for filter F2 ; measurement uncertainties were similar. An overall examination of results showed no clear differences in behavior for the different types of cartridge. An exception was type D, which had the lowest mean efficiency. Types A, B, and C were preferable to the others, although only slightly. With supply system S1 , cartridge type B and the same operating parameters, further tests were carried out to determine the effect of supply system S1m port orientation relative to the air flow entering the filter. To link the uncertainty of results only to differences in cartridge orientation, the same unit was used on both filter F1 and F2 , repeating measurements five or six times for each configuration. As indicated in Fig. 7(b), the data thus obtained show that efficiency is in general independent of angle α. The influence of supply system geometry was then checked. Tests were carried out using cartridge type B with the same supply pressure and mean velocity employed in the previous tests. For system S3 , the presence of the end plates reduces the effective air passage cross-sectional area by approximately 20% compared with the other supply systems, making it necessary to reduce the air flow rate by a corresponding amount. Efficiencies were determined for filters F1 and F2 , and are shown in the graph in Fig. 7(c), where the results refer to measurements performed on cartridges which had already been used in the previous set of tests (saturated cartridges). When flow velocity is relatively high, performance depends to a significant extent on the type of supply system used. In the cases examined, the best supply system for both filters was S1 , followed by systems S3 and S2 in that order. While the larger filter exhibits good general performance, with mean efficiencies of not less than 0.8, the maximum efficiency of the smaller filter is 0.77 and the effect of supply system type is more marked. The presence of the collector tube on system S1 makes it possible to direct air flow downward, promoting separation of the oil droplets moving out of the cartridge toward the bowl. The absence of this device reduces filter performance: in system S2 , much of the flow exiting from the distributor ports is at the upper part of the cartridge and is directed radially toward the latter’s 9
Coalescing Filters
G. Belforte, T. Raparelli, and A. Trivella
exterior. In system S3 , the flow exiting the distributor is directed radially over a longer length of the cartridge than in the previous case, improving conditions for coalescence. Filter size was again found to have an influence, which was particularly high for systems S2 and S3 . The larger flow passage areas downstream of the cartridge and the greater distance between the cartridge and the level of the oil collected in the bowl appreciably reduce local flow velocities and the resulting entrainment of oil droplets toward the exterior.
However, the smaller filter was also found to reach relatively high efficiencies with systems S2 and S3 when v does not exceed 0.25–0.3 m/s. Tests were then conducted on the system consisting of filter F2 , supply system S1 , and cartridge B to measure efficiency while changing operating parameters one at a time. Measurements were repeated four times on two different units. The data shown in Fig. 8(a) indicate that the system has a certain sensitivity to variations in average oil concentration. Efficiencies are still good, even with
(a)
(b)
(c)
(d)
Figure 8 (a) Efficiency versus oil concentration; (b) efficiency versus pressure upstream of cartridge; (c) efficiency versus air flow velocity; and (d) efficiency and pressure drop versus operating time.
10
Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
Coalescing Filters
G. Belforte, T. Raparelli, and A. Trivella
relatively high oil concentrations. Although variations in pressure P1 upstream of the filter in a normal operating range do not have a significant influence on system behavior (Fig. 8(b)), a certain reduction in efficiency can be observed over 0.5 MPa. By contrast, flow velocity has a major impact on system behavior, as can be seen from Fig. 8(c). With a flow velocity of 0.8 m/s, efficiency drops below 0.5. Finally, the influence of operating time was investigated with two type B cartridges. Efficiencies and pressure drops obtained on one of the two specimens tested for a period of 150 h are shown in Fig. 8(d). In general, it can be seen that there was a slight drop in efficiency compared with the first hours of operation; the average of all measurements was 0.95, as against an average of 0.98 in the first 20 h of operation. Scatter in results was around 3% of the overall mean value. Pressure drop increased rapidly in the first hours of operation, and more gradually thereafter. Conclusions The efficiency of different glass fiber coalescing filters was determined experimentally under operating conditions encountered in industrial pneumatic systems. The results demonstrate that filter efficiency depends significantly on operating conditions as well as on filter element type, canister dimensions, and supply system geometry. The test bench instrumentation and procedure used are such that scatter in results does not exceed 6%. The following conclusions can be drawn from the results: 1. The lubricator and pneumatic circuit contribute to forming an oil aerosol with a high degree of size dispersion, as dimensions are estimated to vary from a few units up to several hundred microns. The size dispersion of the oil particles made it impossible to determine the effect of fiber size on system efficiency. 2. If the mean flow velocity across the cartridge is relatively high, the type of supply system has a significant influence on performance. In the cases examined, system S1 provides the best results thanks to the benefits of using the collector tube. However, systems S2 and S3 can also achieve good performance if flow velocity is reduced to 0.3 m/s. The behavior of supply system S1 is independent of port angular position relative to the direction of air flow entering the filter. 3. The efficiencies achieved with supply system S1 are similar for all cartridge types tested. Except for Experimental Techniques (2012) © 2012, Society for Experimental Mechanics
type D the outer drainage layer on the cartridge significantly improves the coalescence process. Types A, B, C provide the best performance, although the difference is slight. In the future, any effect of cartridge type could be made clearer by testing supply system S2 or S3 . 4. The effect of filter size on system efficiency is significant. Best operation is achieved with the larger filter and when the oil level in the bowl does not exceed approximately one-third of the distance from the edge of the cartridge to the bottom of the bowl. 5. Increases in supply pressure of up to 20% or increases in oil concentration of around four times the value established in the previous series of tests result in a loss of efficiency limited to approximately 10%. By contrast, air velocity has a major effect: increases up to 100% over the value established initially reduce efficiency by more than 50%. 6. Prolonged tests at ambient temperature on supply system S1 and cartridge B indicate that that performance drops by a few percentage points after around 20 h of operation, while the strong binder ensures that the system remains sound and efficient after many hours of operation. The other types, which have similar mechanical properties, are expected to show the same behavior. The tested system is suitable for extended use, although it should be borne in mind that in actual service higher operating temperatures can affect the materials’ mechanical strength, and solid contaminants can drastically reduce the life of the filter material.
Acknowledgments The authors would like to thank Metal Works S.p.A for supplying filter canisters, as well as Julian Boin, Antonio Diluciano, and Aniello Nello De Risi for their help in conducting several of the experimental tests. References 1. Dickenson, C., Filters and Filtration Handbook, 2nd Edition, Elsevier, London (1987). 2. Davies, C.N., Air Filtration, Academic Press, London (1973). 3. Brown, R.C., Air Filtration: An Integrated Approach to the Theory and Application of Fibrous Filters, Pergamon, Oxford (1993). 4. Hutten, M., Handbook of Nonwoven Filter Media, Elsevier-Jordan Hill, Oxford (2007). 11
Coalescing Filters
5. Liew, T.P., and Conder, J.R., ‘‘Fine Mist Filtration by Wet Filters. I. Liquid Saturation and Flow Resistance of Fibrous Filters,’’ Journal of Aerosol Science 16:497–509 (1985). 6. Raynor, P.C., and Leith, D., ‘‘The Influence of Accumulated Liquid on Fibrous Filter Performance,’’ Journal of Aerosol Science 31(1):19–34 (2000). 7. Contal, P., Simao, J., Thomas, D., et al., ‘‘Clogging of Fibre Filters by Submicron Droplets. Phenomena and Influence of Operating Conditions,’’ Journal of Aerosol Science 35:263–278 (2004). 8. Charvet, A., Gonthier, Y., Bernis, A., and Gonze, E., ‘‘Filtration of Liquid Aerosols with a Horizontal Fibrous Filter,’’ Chemical Engineering Research and Design 86:569–576 (2008). 9. Shin, C., and Chase, G.G., ‘‘The Effect of Wettability on Drop Attachment to Glass Rods,’’ Journal of Colloid and Interface Science 272:186–190 (2004). 10. Bansal, S., von Arnim, V., Stegmaier, T., and Planck, H., ‘‘Effect of Fibrous Filter Properties on the Oil-In-Water-Emulsion Separation and Filtration Performance,’’ Journal of Hazardous Materials 190:45–50 (2011). 11. Mullins, B.J., Braddock, R.D., and Kasper, G., ‘‘Capillarity in Fibrous Filter Media: Relationship to Filter Properties,’’ Chemical Engineering Science 62:6191–6198 (2007). 12. Yang, B.W., and Chang, Q., ‘‘Wettability Studies of Filter Media Using Capillary Rise Test,’’ Separation and Purification Technology 60:335–340 (2008). 13. Hajra, M.G., Mehta, K., and Chase, G.G., ‘‘Effects of Humidity, Temperature, and Nanofibers on Drop Coalescence in Glass Fiber Media,’’ Separation and Purification Technology 30:79–88 (2003). 14. Vasudevan, G., and Chase, G.G., ‘‘Performance of B-E-Glass Fiber Media in Coalescence Filtration,’’ Journal of Aerosol Science 35:83–91 (2004). 15. Yun, K.M., Hogan, C.J., Jr., Matsubayashi, Y., Kawabe, M., Iskandar, F., and Okuyama, K., ‘‘Nanoparticle Filtration by Electrospun Polymer Fibres,’’ Chemical Engineering Science 62:4751–4759 (2007). 16. Wertz, J., and Schneiders, I., ‘‘Advantages of Nanofibre Coating Technology,’’ Filtration & Separation 46(4):18–20 (2009). 17. Barhate, R.S., and Ramakrishna, S., ‘‘Nanofibrous Filtering Media: Filtration Problems and Solutions from Tiny Materials,’’ Journal of Membrane Science 296:1–8 (2007).
12
G. Belforte, T. Raparelli, and A. Trivella
18. Shin, C., Chase, G.G., and Reneker, D.H., ‘‘Recycled Expanded Polystirene Nanofibers Applied in Filter Media,’’ Colloids and Surfaces A: Physicochemical and Engineering Aspects 262:211–215 (2005). 19. Shin, C., Chase, G.G., and Reneker, D.H., ‘‘The Effect of Nanofibers on Liquid-Liquid Coalescence Filter Performance,’’ AIChE Journal 51 (12): 3109–3113 (2005). 20. Patel, S.U., and Chase, G.G., ‘‘Gravity Orientation and Woven Drainage Structures in Coalescing Filters,’’ Separation and Purification Technology 75:392–401 (2010). 21. Park, B.H., Lee, M., Kim, S.B., and Jo, Y.M., ‘‘Evaluation of the Surface Properties of PTFE Foam Coating Filter Media Using XPS and Contact Angle Measurements,’’ Applied Surface Science 257:3709–3716 (2011). 22. Hanspal, N.S., Waghode, A.N., Nassehi, V., and Wakeman, R.J., ‘‘Development of a Predictive Mathematical Model for Coupled Stokes/Darcy Flows in Cross-Flow Membrane Filtration,’’ Chemical Engineering Journal 149:132–142 (2009). 23. Belforte, G., Ferraresi, C., and Raparelli, T., ‘‘Testing the Performance of Oil Fog Lubricators, 7th International Fluid Power Symposium, Bath, pp. 359–368 (1986). 24. Belforte, G., D’Alfio, N., and Raparelli, T., ’’Experimental Methodologies and Efficiency Analysis of Compressed Air Filters,’’ International Conference on Fluid Power Control and Robotics, Chengdu, pp. 565–570, October, 1990. 25. Belforte, G., and Raparelli, T., ‘‘Metodologie di Prova eA di Filtri di Scarico per il Trattenimento Dell’olio,’’ Oleodinamica Pneumatica Lubrificazione 6:94–106 (1991). 26. Belforte, G., Ferraresi, C., and Raparelli, T., ‘‘Analisi Della Nebulizzazione Dell’olio in un Circuito Pneumatico,’’ Notiziario Tecnico AMMA 41(9):31–35 (1986). 27. Colcombe, T.P., The Why and How of Coalescing Type Compressed Air Oil Removal Filters. Advances in Filtration and Separation Technology, Volume 13, American Filtration and Separation Society, Northport, Alabama (1999).
Experimental Techniques (2012) © 2012, Society for Experimental Mechanics