- Email: [email protected]
Filtration characteristics of threaded microfiber water filters
Desalination xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
Desalination journal homepage: www.elsevier.com/locate/desal
Filtration characteristics of threaded microfiber water filters Hilla Shemera,⁎, Abraham Sagiva, Marina Holenbergb, Adva Zach Maorb a b
Rabin Desalination Laboratory, Wolfson Faculty of Chemical Engineering, Technion-Israel Institute of Technology, Haifa 32000, Israel Amiad Water Systems Ltd., D.N. Galil Elyon, 1233500, Israel
A R T I C L E I N F O
A B S T R A C T
Keywords: Cake filtration Mechanism Pressure drop Slurry Calcium carbonate
Textile fibers are widely used for fine filtration in the disposable cartridge filter market. In this work the filtration mechanism of threaded microfiber water filters was characterized by testing the effect of filtration velocity, slurry concentration, particles size distribution (PSD) and filter pore size and porosity, on the filter performance. Constant flow rate experiments were conducted with micronized CaCO3 slurry as a model substance. It was found that the predominant filtration mechanism of the studied filters is cake filtration. Very efficient filtration followed by effective automatic cleaning of the filter was obtained as indicated by complete removal of the CaCO3 particles in all the conditions studied and similar clean filter resistance over repetitive cycles. Shorter filtration cycles were obtained at higher velocities, low porosity filter and narrow PSD. Correspondingly, the filter capacity declined as the filtration cycles were shorter. Yet, the filter capacity was found to be independent of the CaCO3 slurry concentration. A criterion of specific consumed energy per unit filtrate volume (Es) was developed. Analyses of the effect of the various studied parameters on Es revealed its dependence on the slurry concentration, velocity and filtration time.
1. Introduction Filtration is a separation process at which suspended solids are removed from liquid on the surface (cake filtration) and/or within the depth of a filter medium. Filtration efficiency depends on many factors including particles characteristics (size, shape, strength, surface chemistry and concentration); filterability characteristics of the liquids (viscosity, volatility, toxicity, corrosiveness, temperature, pH, and ionic strength); filter technology (chemical and thermal resistance), filter design (medium type, pore size and depth); and operating conditions (filtration rate and duration) [1]. In depth filtration the suspended particles are smaller than the medium pores and therefore are trapped within various depths of the filter. The mechanisms of removal include the following forces (alone or combined): hydrodynamic, gravitational, molecular, Brownian, and/or electrostatic [2]. The initial pressure drop across a depth filter is generally higher than a surface filter of equivalent efficiency. However, the pressure drop increase rate, as particles accumulates within the filter, is more gradual [3]. Cake filtration is based on retention of the solid particles on the surface of the porous medium. The driving force for cake filtration includes gravity (hydrostatic pressure), pressure, vacuum or centrifugal. The cake has a complex pore structure determined by the characteristics of the solids. The filtration performance depends on the cake
⁎
permeability, pore size, particle size and compressibility [4]. From operational and energy consumption considerations any filter needs to be cleaned or replaced periodically. Depth filters are backwashed i.e., the flow through the filter is reversed to achieve fluidization. Surface filters are cleaned with high pressure water jets, backflow across the surface of the filter or replaces as in the case of most cartridge filters [5]. Laboratory scale filters testing is most often conducted under constant pressure filtration due to convenience. Nevertheless, most large scale industrial applications operate at a constant filtration rate. During constant rate filtration longer filtration cycles are obtained and the differential pressure is proportional to the filtration time [6]. The basic filtration element in threaded microfiber filter manufactured by Amiad® Water Systems Ltd. (Israel) is a “thread cassette”. Fine threads (10 μm in diameter) are wound over a rigid grooved base plate. Water flows through the thread layers into the grooves and channel the water to specially designed outlets. The rigid base plate supports the thread layers also plays a major role in the cleaning process of the media. The filters are cleaned automatically, at a pre-determined differential pressure level, by boosting highly pressurized water that pass through the thread layers. This creates a powerful spot back flush, which carries with it the trapped particles and the filter cake out of the cassettes thread layers. The performance efficiency of these filters was demonstrated in surface water dealing mainly with
Corresponding author. E-mail address: [email protected] (H. Shemer).
http://dx.doi.org/10.1016/j.desal.2017.07.009 Received 14 May 2017; Accepted 12 July 2017 0011-9164/ © 2017 Elsevier B.V. All rights reserved.
Please cite this article as: Shemer, H., Desalination (2017), http://dx.doi.org/10.1016/j.desal.2017.07.009
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
Filter
Pore size (μm)
Porosity- Ψ
Permeabilityaκ0 (× 10− 14 m2)
A B C D
2 1 2 3
0.18 0.07 0.52 0.62
5.86 5.33 1.11 8.86
a
Volume (%)
Table 1 Filters characterization.
Measured at a velocity of 0.12 cm/s and calculated using Eq. (7).
∑
(1)
40
50
3.1. Filtration equations Quantitative characterization of filters performance is based on Darcy's law which describes filtration through porous media [12].
Q ΔP = A μ (Rm + R c )
(3)
where Q is the flow rate (m /s), A is the filtration area (m ), ΔP is the pressure difference across the porous media (Pa), μ is the dynamic viscosity of the liquid (Pa s), Rm and Rc are the filter and cake resistances respectively (1/m). These resistances relate to filtration parameters according to: 3
2
Rm =
l κ
(4)
RC =
α⋅C⋅Q⋅t A
(5)
here l is the filter thickness (m), κ is the permeability of the liquid through the filter (m2), α is the specific cake resistance (m/kg), and C is the slurry concentration (kg/m3). Combining Eqs. (3), (4) and (5) yields a linear relationship between filtration pressure difference and time:
i=1 n
∑ Di2 vi
30
3. Theory and calculations
D3i vi (2)
i=1
20
particle in the size fraction. The following parameters were monitored during the experiments: differential pressure (ΔP), inlet and outlet turbidities. ΔP was measured with a Piezoresistive Differential Pressure Transmitter (Model NP-300, NOVUS Brazil) with data acquisition system (MyPCLab v. 1.22, Novus Electronics Products, Brazil). Turbidity was measured using Hach 2100P Turbidimeter. As the micronized CaCO3 layer builds up on and in the filter, the pressure differential across the filter increased. Once the differential pressure reached a pre-determined level, a cleaning sequence was conducted using high-pressure water jets spray system. The cleaning was carried out at a flowrate of 1.6 L/min, for 1.1 min at a pressure of 8.5 bar (total of 1.9 L). The filter was used multiple times, as it did not loss its filtration efficiency after cleaning (See Section 4.1).
where VF is the volume of the fibers and VM is the volume of the filtration media. CaCO3 slurry was pumped through the filter (Fig. 1) at a constant flow rate of 0.4, 0.7 and 1.0 L/min (velocities of 0.05, 0.08 and 0.12 cm/s respectively). Initial concentrations of CaCO3 slurry were 35, 47, 75 and 130 mg/L corresponding to turbidities of 50, 75, 150 and 300 NTU respectively. The slurry was prepared in DI water using micronized CaCO3 powder 1 (Socal® Solvay Chemicals International, Belgium). The slurry surface mean diameter D [3,2] was 2.2 μm (Eq. (2), [11]) with a relatively narrow particle size distribution (PSD) of 1.0 μm, as shown in Fig. 2. The width of PSD is defined as the difference between the volume and surface mean diameters (D [4,3] − D [3,2]) [11]. The effect of the size distribution was tested using a second micronized CaCO3, powder 2, of D [3,2] = 4.6 μm (Fluka, Germany) with wider PSD width of 4.9 μm. Size analysis was conducted using Mastersizer 2000 (Malvern, UK).
D[3, 2] =
10
Fig. 2. Particles size distribution of micronized CaCO3 powders 1 and 2.
A laboratory scale filtration system, equipped with 140 cm2 thread cylinders, were obtained from Amiad® Water System Company. Characterization of the filters is displayed in Table 1. The porosities of the filters were determined using the ratio between the volumes of the filters and filtration media (Eq. (1)). The permeability was calculated using equation derived from the Darcy's theory (See Section 3.1).
n
Powder 2
Size ( m)
2. Experimental
VF VM
Powder 1
0
biofouling phenomena [7–10]. Yet, the fundamentals of the filtering mechanism was not studied. The objective of the work presented herein was to characterize the filtration mechanism of Amiad threaded microfiber filters. Effect of various parameters, including filtration velocity, slurry concentration, size distribution and filter porosity, on the filter performance and its consumed energy were investigated using micronized CaCO3 slurry as model substance.
Ψ=1−
10 9 8 7 6 5 4 3 2 1 0
where D is the diameter of a particle (m) and vi is the proportion of
Fig. 1. Schematics of the experimental system.
D/P
Filter
Air relief valve
Filtered water Filter housing CaCO3 slurry 2
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
ΔP =
μ⋅Q ⎛ l α⋅C⋅Q ⎞ + ⋅t A ⎝κ A ⎠
200 180
Filter permeability κ can be experimentally derived from the differential pressure across a clean filter at t = 0 and ΔP = ΔPm bar according to:
160
Q⋅μ⋅l κ= A⋅ΔPm
Δp (mbar)
(6)
SC⋅A2 C⋅Q 2⋅μ
0
ΔPm ε ⋅A⋅Lm 2 m
40
50
experiments). As seen in Fig. 3, the duration of the initial period of filtration - tf was determined by the transition timing from medium clogging to cake filtration defined as the interception of the lines representing the of two filtration stages. The y-axis intercept determined the differential pressure at the beginning of the cake filtration (ΔPf). The clean filter A resistance (ΔPm) at various filtration velocities is displayed in Fig. 4. As seen, the clean filter resistance appears to increase linearly with the filtration velocity according to Darcy's law Eq. (3) regardless of the slurry concentration and size distribution. Similar results were reported in the literature [13,14]. It is important to note that no increase in the clean filter resistance over repetitive cycles was observed. Additionally, in all the experiments, the filtrate turbidity was < 0.2 NTU indicating complete separation of the CaCO3 particles. Both observations indicated very efficient filtration followed by effective cleaning of the filter. As explained previously, the predominant filtration mechanism of the threaded microfiber filter is cake filtration. Short tf times as compared to the cake filtration duration were determined for filter A ranging from 6 to 74 min. Duration of the initial period of filtration was found to depend on the CaCO3 slurry concentration and the velocity of filtration, as shown in Fig. 5. The experimental results indicated the existence of a critical velocity and concentration above which tf was unchanged. As the velocity was increased from 0.05 to 0.08 cm/s and/ or the concentration of the slurry was raised from 35 to 130 mg/L tf became shorter due to faster clogging of the filter medium. Above these critical velocity (ucrit = 0.08 cm/s) and concentration (Ccrit = 130 mg/ L) the initial period of filtration duration was not significantly changed. Once a wider CaCO3 particles size distribution was filtered using filter A (See Section 4.3) or the porosity of the filter was increased (See Section 4.4), at the critical concentration of 130 mg/L, tf becomes immeasurable i.e., very short. The specific cake resistance was found to increase with filtration velocity (Fig. 6). Higher filtration velocity results in faster cake growth and higher cake resistance. Correspondingly, slower cake growth at lower filtration velocity results in lower cake resistance [14].
εs = 1 − εc
(9)
+ ΔPav⋅εc⋅A⋅Lc ,
ΔPav =
ΔP (t ) + ΔPm 2
(10)
C ⎡ ΔPm Lm ⎤ ΔPav⋅εc + εm ρs ⋅εs ⎢ 2 Lc ⎥ ⎦ ⎣
(11)
As the filtration through the filter medium is very short relative to the cake filtration the second term in Eq. (11) becomes negligible therefore,
Es ≈
tf=31 min
20 30 Time (min)
Fig. 3. Typical results of micronized CaCO3 slurry filtration with Filter A.
The specific consumed energy Es defined as the energy E per unit filtrate volume t·Q is obtained by combining Eqs. (9) and (10):
Es =
10
(8)
where εs and ρs are the solidity of the cake and the suspended solids density respectively and εc is the cake porosity. Assuming linear pressure drop ΔP along the cake and filter thicknesses. Then the energy E needed to drive the flow across the cake and the filter (the consumed energy) is,
E (t ) =
Cake filtration
Initial filtration period
40
A criterion of specific consumed energy per unit filtrate volume is proposed for constant flow filtration. This proposed index is independent of the filter type and/or filtration conditions. Therefore, it can be used to compare between various filtration systems and filtration conditions. Assuming that at t = 0 i.e., the moment of filtration starts, all of the pressure drop for filtration acts through the filter medium (ΔPm) since there is no filter cake. After that moment at t > 0, the cake is formed and builds up. The cake thickness Lc is derived via mass balance of the suspended particles neglecting liquid volume in the cake relative the filtrate volume: Q⋅C ⋅t , εs ⋅ ρs ⋅ A
Pm=77 mbar
100 60
3.2. Specific energy consumption
Lc =
120 80
(7)
The slope (Sc) of the filtration pressure against filtrate volume, or time is used to calculate the specific cake resistance α (Eq. (8)), while the intercept is the filter medium resistance ΔPm.
α=
Sc=3.2 mbar/min
Pf=143 mbar
140
εc⋅ΔPav⋅C εs⋅ρs
(12)
4. Results 100
4.1. Filtration mechanism
95 90 ΔPm (mbar)
Fig. 3 displays typical results obtained using Filter A. The filtration follows two regimes: the initial period of filtration and cake dominant filtration. At the initial period of filtration the resistance of filter medium increases exponentially by medium clogging during the initial formation of a filter cake. Once the filtration cake is formed, the increase of the filter medium resistance becomes negligible and the pressure drop acts through the cake. As the cake filtration becomes the predominant mechanism, a linear increase of the ΔP with time is observed; at which its slope (Sc) was used to calculate the specific cake resistance α (Eq. (8)). The experimental data indicated the formation of incompressible cake (until ΔP = 0.7 bar the maximum reached in the
DI C=75mg/L C=200mg/L
C=35mg/L C=130mg/L
85 80 75 70 65 60 0.03
0.05
0.08 0.10 Velocity (cm/s)
0.13
0.15
Fig. 4. Clean filter resistance (ΔP = ΔPm) as a function of filtration velocity.
3
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
250 Filter capacity (mg/cm2)
40 35
tf (min)
30 25 20 15 10 5
C=130 mg/L 200 150 100 50
0 25
50
75
100
125
150
175
C=75 mg/L
0 0.04
200
0.07 0.09 Velocity (cm/s)
Concentration of CaCO3 (mg/L)
Fig. 8. Filter capacity as a function of the filtration velocity at CaCO3 slurry concentration of 75 and 130 mg/L.
Fig. 5. Initial period of filtration (tf) at various slurry concentrations and filtration velocities.
3.0
1.2
C=75mg/L
2.5 SC 10-2 (bar/min)
α
1011 (m/kg)
1.6 1.4
C=130 mg/L
1.0 0.8 0.6 0.4 0.02
0.12
2.0 1.5 1.0 0.5 0.0
0.05
0.07 0.10 velocity (cm/s)
0.12
0
30 60 90 120 150 180 210 Concentration of CaCO3 (mg/L)
Fig. 6. Specific cake resistance at various flow rates at CaCO3 slurry concentrations of 75 and 130 mg/L.
Fig. 9. ΔP increase rate as a function of the inlet CaCO3 slurry concentration at a velocity 0.12 cm/s.
4.2. Effect of velocity and slurry concentration
concentration, less slurry can be filtered during similar filtration time and a higher cake is formed (with the same amount of slurry). For the same cake height, the filtration time has a reverse proportional relationship to the solid concentration i.e., diluted slurry filtration takes more time [1].
At a constant inlet concentration shorter filtration cycles were obtained at higher velocities (Fig. 7) and the filter capacity, up to ΔP = 0.7 bar, declined (Fig. 8). Yet, the filter capacity was found to be independent of the CaCO3 slurry concentration. The filter capacity is defined as:
4.3. Effect of particles size distribution
Filter uptake =
Mass of CaCO3 (mg ) Area of filter (cm2)
(13)
At the higher average diameter CaCO3 particles with the wider PSD longer filtration cycle was obtained resulting in higher filtration capacity of 167.7 compared with 49.9 mg/cm2. The resistance of the cake was found to be lower at a wider PSD (α = 0.45 × 1011 and 1.42 × 1011 m/kg for the wide (powder 2) and narrow (powder 1) PSD respectively) therefore, ΔP increase rate was lower (Fig. 10). In a narrow PSD the particles are smaller and more homogeneous. Reduced particles size increases the specific resistance since the drag force resisting fluid flow through the cake is proportional to the surface area of particles in the cake [17]. Therefore, the narrower PSD exhibit higher resistance compared to wider PSD.
It is evident that higher pressure drops were obtained at higher filtration velocities. Since ΔP rise is smaller at lower velocities the specific cake resistance is lower as shown in Fig. 6. Consequently, thicker cakes are formed at lower filtration velocity, for the same ΔP [14,15]. The cake formed at higher velocity is compact and possesses higher mechanical stability [14,16]. At a constant velocity of 0.12 cm/s, as the slurry concentration was increased, the filtration cycle became shorter (Fig. 9). At a higher solid
1.4
C=130 mg/L
650
1.2
C=75 mg/L
550
1.0
P (mbar)
SC
10-2 (bar/min)
1.6
0.8 0.6
450 350
0.4
250
0.2
150
0.0 0.04
Powder 1 Powder 2
50
0.07 0.09 Velocity (cm/s)
0.12
0
Fig. 7. ΔP increase rate (Sc) as a function of filtration velocity at CaCO3 slurry concentration of 75 and 130 mg/L.
20
40
60 80 100 120 140 Time (min)
Fig. 10. ΔP increase with time for narrow and wide PSD micronized CaCO3 (at a velocity of 0.12 cm/s and C = 130 mg/L).
4
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
700
14 Filter B Filter C Filter D
500
12
3m
10 Es (J/m3)
P (mbar)
600
400 300 200 100
8
2m
6 4 1m
2
0
0 0
10
20
30
40
50
0
Time (min)
5
The pore size effect on the filtration cycle is displayed in Fig. 11. As expected, the smaller pore size filter of 1 μm (i.e., high filter density) exhibited high clean filter resistance and shorter filtration cycle. However, both the 2 and 3 μm pore size filters showed similar clean filter resistance as well as similar filtration cycles. These results indicate that it is possible to use a coarser filter to separate very fine substances when the predominant mechanism is cake filtration. While the pore size and PSD of a filter determine the removal efficiency of the filter, the porosity and permeability are important in determining the rate of fluid flow through the filter. As seen in Fig. 12, the difference between the higher porosity filter of 0.52 (filter C) and the lower porosity filter A (0.18) is restricted to the initial period of filtration. While clogging of the filtration medium was observed in at the lower porosity filter the initial period of filtration was immeasurable at the higher porosity filter. As expected, the cake filtration rate was found to be similar Sc = 12.6 and 12.5 mbar/min filter A and filter C respectively; both filters exhibiting similar clean resistance (ΔPm) of 84 and 81 mbar and close filter capacities of 49.9 ± 2.83 and 45.0 ± 2.75 mg/cm2.
0.1
0.12
Es (J/m3) Filter
A
A
B
C
D
Powder Es (J/m3)
1 1.0–3.0
2 1.5–2.4
1 3.0–5.6
1 2.4–4.9
1 2.2–4.8
compared to the other tested parameters (i.e., PSD and filter pores size). Table 2 displays the specific energy calculated at V = 1 to 3 m for the two PSD powders (Fig. 2) and filters B-D (Fig. 11). As seen, the slurry mean size and PSD as well as the filters pores size had only a minor impacts on Es. This is attributed to the fact that a constant cake porosity of 0.75 was used in the calculations [3]. The difference between the filters is manifested by the clean filter resistance Rm. However, in practical filtration times, Rm is negligible relative to the cake resistance RC, as shown in Eq. (11). Another dominant factor of Es is the filtration time or filtrate volume. As seen in Figs. 13 and 14, a significant increase of Es is obtained with V or with time. Therefore, it is suggested to conduct short filtration cycles in order to save energy. However, it should be realized that short filtration cycles require larger number of filtration units (i.e., increased footprint) to enable constant product water capacity. Additionally the total energy demand also includes the energy required for the filter cleaning. In terms of economic optimization, all these parameters should be considered as well as the product water cost. 5. Concluding remarks
50 20
0.02 0.04 0.06 0.08 Velocity (cm/s)
Table 2 Specific energy for V = 1 to 3 m at a velocity of 0.12 cm/s and 130 mg/L CaCO3.
200
0.52
1m
Fig. 14. Impacts of velocity and filtrate volume per unit filter area on the specific energy consumption energy at slurry concentration of 130 mg/L. Dashed lines are parabolic approximation of Es(Q).
250
100
2
0
300
0.18
2m
3
0
350
150
3m
1
The impacts of filtration parameters on the energy consumed per unit filtrate volume or specific energy consumption, Es, was studied. The cake porosity of CaCO3, used in the calculations, is εc = 0.75 as reported for compressive pressure Ps < 1.3 bar [3]. Figs. 13 and 14 display the specific energy consumption at three cumulative filtrate volumes per unit filter area V of 1, 2 and 3 m. Results display in Fig. 13 show that Es ∝ C2. The specific energy consumption in Eq. (12) is directly proportional to C⋅ΔPav and ΔPav ∝ C, as agreed by Eq. (6). Correspondingly, Es ∝ Q2 (Fig. 14) where Q2 is proportional to the kinetic energy of the slurry flow. The effect of the CaCO3 slurry concentration on Es was found to be the dominant factor
P (mbar)
200
4
4.5. Specific energy consumption
10 15 Time (min)
150
Fig. 13. Impact of slurry concentration and filtrate volume per unit filter area on the specific energy consumption at velocity of 0.12 cm/s. Dashed lines are parabolic approximation of Es(C).
4.4. Effect of filter pore size and porosity
5
100
Concentration of CaCO3 (mg/L)
Fig. 11. Increase in ΔP with time at filter pore size of 1, 2 and 3 μm (velocity of 0.12 cm/s and C = 130 mg/L).
0
50
Comprehensive experimental characterization of Amiad threaded microfiber water filters was carried out. The filters exhibited high potential for treating water containing very fine particles. Repetitive cycles of efficient micronized CaCO3 slurry filtration followed by effective automatic filter cleaning were obtained. The high separation capability
25
Fig. 12. Increase in ΔP with time at filter porosities of 0.18 and 0.52 and pore size of 2 μm (velocity of 0.12 cm/s and C = 130 mg/L).
5
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
[8] Eshel, H. Elifantz, S. Nuriel, M. Holenberg, T. Berman, Microfiber filtration of lake water: impacts on TEP removal and biofouling development, Desalin. Water Treat. 51 (2012) 1043–1049. [9] T. Berman, Transparent exopolymer particles as critical agents in aquatic biofilm formation: implications for desalination and water treatment, Desalin. Water Treat. 51 (2013) 1014–1020. [10] A. Lakretz, H. Elifantz, I. Kviatkovski, G. Eshel, H. Mamane, Automatic microfiber filtration (AMF) of surface water: impact on water quality and biofouling evolution, Water Res. 48 (2014) 592–604. [11] T. Kinnarinen, R. Tuunila, A. Häkkinen, Reduction of the width of particle size distribution to improve pressure filtration properties of slurries, Miner. Eng. 102 (2017) 68–74. [12] C.M. Chew, M.K. Aroua, M.A. Hussain, A practical hybrid modelling approach for the prediction of potential fouling parameters in ultrafiltration membrane water treatment plant, J. Ind. Eng. Chem. 45 (2017) 145–155. [13] N. Hasolli, Y.O. Park, Y.W. Rhee, Experimental evaluation of filter performance of depth filter media cartridge with varying the pleat count and the cartridge assembly arrangement, Par. Aerosol. Res. 8 (2012) 133–141. [14] M. Saleem, G. Krammer, M.S. Tahir, The effect of operating conditions on resistance parameters of filter media and limestone dust cake for uniformly loaded needle felts in a pilot scale test facility at ambient conditions, Powder Technol. 228 (2012) 100–107. [15] J. Sievert, F. Löffler, Dust cake release from non-woven fabrics, Filtr. Sep. 24 (1987) 424–427. [16] E. Schmidt, Experimental investigations into the compression of dust cakes deposited on filter media, Filtr. Sep. 32 (1995) 789–793. [17] R. Wakeman, The influence of particle properties on filtration, Sep. Purif. Technol. 58 (2007) 234–241.
of the filter as well as its multiple reuse point out to the potential of recovering the filtered substance. Evaluation of the consumed filtration energy at various filtration conditions and filters characteristics indicated that the slurry concentration, filtration velocity, and filtration duration are the most energy consumed parameters. Optimization of the filtration process is required, taking into account the specific consumed energy per unit filtrate volume, the cleaning energy and downtime, filters footprint, and cost of the product water. References [1] B.A. Perlmutter, Solid-liquid Filtration: Practical Guides in Chemical Engineering, Elsevier Inc, 2015. [2] C. Ghidaglia, L. de Arcangelis, J. Hinch, É. Guazzelli, Transition in particle capture in deep bed filtration, Phys. Rev. E 53 (1996) R3028–R3031. [3] S.K. Teoh, R.B.H. Tan, C. Tien, Analysis of cake filtration data - a critical assessment of conventional filtration theory, AICHE J. 52 (2006) 3427–3442. [4] K. Sutherland, G. Chas, Filters and Filtration Handbook 5th Edition, Elsevier Science, UK, 2008. [5] D.W. Hendricks, Fundamentals of Water Treatment Unit Processes: Physical, Chemical, and Biological, CRC Press, Taylor and Francis Group, 2010. [6] F.M. Mahdi, R.G. Holdich, Laboratory cake filtration testing using constant rate, Chem. Eng. Res. Des. 91 (2013) 1145–1154. [7] D. Ityel, Ground water: dealing with iron contamination, Filtr. Sep. 48 (2011) 26–28.
6
Contents lists available at ScienceDirect
Desalination journal homepage: www.elsevier.com/locate/desal
Filtration characteristics of threaded microfiber water filters Hilla Shemera,⁎, Abraham Sagiva, Marina Holenbergb, Adva Zach Maorb a b
Rabin Desalination Laboratory, Wolfson Faculty of Chemical Engineering, Technion-Israel Institute of Technology, Haifa 32000, Israel Amiad Water Systems Ltd., D.N. Galil Elyon, 1233500, Israel
A R T I C L E I N F O
A B S T R A C T
Keywords: Cake filtration Mechanism Pressure drop Slurry Calcium carbonate
Textile fibers are widely used for fine filtration in the disposable cartridge filter market. In this work the filtration mechanism of threaded microfiber water filters was characterized by testing the effect of filtration velocity, slurry concentration, particles size distribution (PSD) and filter pore size and porosity, on the filter performance. Constant flow rate experiments were conducted with micronized CaCO3 slurry as a model substance. It was found that the predominant filtration mechanism of the studied filters is cake filtration. Very efficient filtration followed by effective automatic cleaning of the filter was obtained as indicated by complete removal of the CaCO3 particles in all the conditions studied and similar clean filter resistance over repetitive cycles. Shorter filtration cycles were obtained at higher velocities, low porosity filter and narrow PSD. Correspondingly, the filter capacity declined as the filtration cycles were shorter. Yet, the filter capacity was found to be independent of the CaCO3 slurry concentration. A criterion of specific consumed energy per unit filtrate volume (Es) was developed. Analyses of the effect of the various studied parameters on Es revealed its dependence on the slurry concentration, velocity and filtration time.
1. Introduction Filtration is a separation process at which suspended solids are removed from liquid on the surface (cake filtration) and/or within the depth of a filter medium. Filtration efficiency depends on many factors including particles characteristics (size, shape, strength, surface chemistry and concentration); filterability characteristics of the liquids (viscosity, volatility, toxicity, corrosiveness, temperature, pH, and ionic strength); filter technology (chemical and thermal resistance), filter design (medium type, pore size and depth); and operating conditions (filtration rate and duration) [1]. In depth filtration the suspended particles are smaller than the medium pores and therefore are trapped within various depths of the filter. The mechanisms of removal include the following forces (alone or combined): hydrodynamic, gravitational, molecular, Brownian, and/or electrostatic [2]. The initial pressure drop across a depth filter is generally higher than a surface filter of equivalent efficiency. However, the pressure drop increase rate, as particles accumulates within the filter, is more gradual [3]. Cake filtration is based on retention of the solid particles on the surface of the porous medium. The driving force for cake filtration includes gravity (hydrostatic pressure), pressure, vacuum or centrifugal. The cake has a complex pore structure determined by the characteristics of the solids. The filtration performance depends on the cake
⁎
permeability, pore size, particle size and compressibility [4]. From operational and energy consumption considerations any filter needs to be cleaned or replaced periodically. Depth filters are backwashed i.e., the flow through the filter is reversed to achieve fluidization. Surface filters are cleaned with high pressure water jets, backflow across the surface of the filter or replaces as in the case of most cartridge filters [5]. Laboratory scale filters testing is most often conducted under constant pressure filtration due to convenience. Nevertheless, most large scale industrial applications operate at a constant filtration rate. During constant rate filtration longer filtration cycles are obtained and the differential pressure is proportional to the filtration time [6]. The basic filtration element in threaded microfiber filter manufactured by Amiad® Water Systems Ltd. (Israel) is a “thread cassette”. Fine threads (10 μm in diameter) are wound over a rigid grooved base plate. Water flows through the thread layers into the grooves and channel the water to specially designed outlets. The rigid base plate supports the thread layers also plays a major role in the cleaning process of the media. The filters are cleaned automatically, at a pre-determined differential pressure level, by boosting highly pressurized water that pass through the thread layers. This creates a powerful spot back flush, which carries with it the trapped particles and the filter cake out of the cassettes thread layers. The performance efficiency of these filters was demonstrated in surface water dealing mainly with
Corresponding author. E-mail address: [email protected] (H. Shemer).
http://dx.doi.org/10.1016/j.desal.2017.07.009 Received 14 May 2017; Accepted 12 July 2017 0011-9164/ © 2017 Elsevier B.V. All rights reserved.
Please cite this article as: Shemer, H., Desalination (2017), http://dx.doi.org/10.1016/j.desal.2017.07.009
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
Filter
Pore size (μm)
Porosity- Ψ
Permeabilityaκ0 (× 10− 14 m2)
A B C D
2 1 2 3
0.18 0.07 0.52 0.62
5.86 5.33 1.11 8.86
a
Volume (%)
Table 1 Filters characterization.
Measured at a velocity of 0.12 cm/s and calculated using Eq. (7).
∑
(1)
40
50
3.1. Filtration equations Quantitative characterization of filters performance is based on Darcy's law which describes filtration through porous media [12].
Q ΔP = A μ (Rm + R c )
(3)
where Q is the flow rate (m /s), A is the filtration area (m ), ΔP is the pressure difference across the porous media (Pa), μ is the dynamic viscosity of the liquid (Pa s), Rm and Rc are the filter and cake resistances respectively (1/m). These resistances relate to filtration parameters according to: 3
2
Rm =
l κ
(4)
RC =
α⋅C⋅Q⋅t A
(5)
here l is the filter thickness (m), κ is the permeability of the liquid through the filter (m2), α is the specific cake resistance (m/kg), and C is the slurry concentration (kg/m3). Combining Eqs. (3), (4) and (5) yields a linear relationship between filtration pressure difference and time:
i=1 n
∑ Di2 vi
30
3. Theory and calculations
D3i vi (2)
i=1
20
particle in the size fraction. The following parameters were monitored during the experiments: differential pressure (ΔP), inlet and outlet turbidities. ΔP was measured with a Piezoresistive Differential Pressure Transmitter (Model NP-300, NOVUS Brazil) with data acquisition system (MyPCLab v. 1.22, Novus Electronics Products, Brazil). Turbidity was measured using Hach 2100P Turbidimeter. As the micronized CaCO3 layer builds up on and in the filter, the pressure differential across the filter increased. Once the differential pressure reached a pre-determined level, a cleaning sequence was conducted using high-pressure water jets spray system. The cleaning was carried out at a flowrate of 1.6 L/min, for 1.1 min at a pressure of 8.5 bar (total of 1.9 L). The filter was used multiple times, as it did not loss its filtration efficiency after cleaning (See Section 4.1).
where VF is the volume of the fibers and VM is the volume of the filtration media. CaCO3 slurry was pumped through the filter (Fig. 1) at a constant flow rate of 0.4, 0.7 and 1.0 L/min (velocities of 0.05, 0.08 and 0.12 cm/s respectively). Initial concentrations of CaCO3 slurry were 35, 47, 75 and 130 mg/L corresponding to turbidities of 50, 75, 150 and 300 NTU respectively. The slurry was prepared in DI water using micronized CaCO3 powder 1 (Socal® Solvay Chemicals International, Belgium). The slurry surface mean diameter D [3,2] was 2.2 μm (Eq. (2), [11]) with a relatively narrow particle size distribution (PSD) of 1.0 μm, as shown in Fig. 2. The width of PSD is defined as the difference between the volume and surface mean diameters (D [4,3] − D [3,2]) [11]. The effect of the size distribution was tested using a second micronized CaCO3, powder 2, of D [3,2] = 4.6 μm (Fluka, Germany) with wider PSD width of 4.9 μm. Size analysis was conducted using Mastersizer 2000 (Malvern, UK).
D[3, 2] =
10
Fig. 2. Particles size distribution of micronized CaCO3 powders 1 and 2.
A laboratory scale filtration system, equipped with 140 cm2 thread cylinders, were obtained from Amiad® Water System Company. Characterization of the filters is displayed in Table 1. The porosities of the filters were determined using the ratio between the volumes of the filters and filtration media (Eq. (1)). The permeability was calculated using equation derived from the Darcy's theory (See Section 3.1).
n
Powder 2
Size ( m)
2. Experimental
VF VM
Powder 1
0
biofouling phenomena [7–10]. Yet, the fundamentals of the filtering mechanism was not studied. The objective of the work presented herein was to characterize the filtration mechanism of Amiad threaded microfiber filters. Effect of various parameters, including filtration velocity, slurry concentration, size distribution and filter porosity, on the filter performance and its consumed energy were investigated using micronized CaCO3 slurry as model substance.
Ψ=1−
10 9 8 7 6 5 4 3 2 1 0
where D is the diameter of a particle (m) and vi is the proportion of
Fig. 1. Schematics of the experimental system.
D/P
Filter
Air relief valve
Filtered water Filter housing CaCO3 slurry 2
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
ΔP =
μ⋅Q ⎛ l α⋅C⋅Q ⎞ + ⋅t A ⎝κ A ⎠
200 180
Filter permeability κ can be experimentally derived from the differential pressure across a clean filter at t = 0 and ΔP = ΔPm bar according to:
160
Q⋅μ⋅l κ= A⋅ΔPm
Δp (mbar)
(6)
SC⋅A2 C⋅Q 2⋅μ
0
ΔPm ε ⋅A⋅Lm 2 m
40
50
experiments). As seen in Fig. 3, the duration of the initial period of filtration - tf was determined by the transition timing from medium clogging to cake filtration defined as the interception of the lines representing the of two filtration stages. The y-axis intercept determined the differential pressure at the beginning of the cake filtration (ΔPf). The clean filter A resistance (ΔPm) at various filtration velocities is displayed in Fig. 4. As seen, the clean filter resistance appears to increase linearly with the filtration velocity according to Darcy's law Eq. (3) regardless of the slurry concentration and size distribution. Similar results were reported in the literature [13,14]. It is important to note that no increase in the clean filter resistance over repetitive cycles was observed. Additionally, in all the experiments, the filtrate turbidity was < 0.2 NTU indicating complete separation of the CaCO3 particles. Both observations indicated very efficient filtration followed by effective cleaning of the filter. As explained previously, the predominant filtration mechanism of the threaded microfiber filter is cake filtration. Short tf times as compared to the cake filtration duration were determined for filter A ranging from 6 to 74 min. Duration of the initial period of filtration was found to depend on the CaCO3 slurry concentration and the velocity of filtration, as shown in Fig. 5. The experimental results indicated the existence of a critical velocity and concentration above which tf was unchanged. As the velocity was increased from 0.05 to 0.08 cm/s and/ or the concentration of the slurry was raised from 35 to 130 mg/L tf became shorter due to faster clogging of the filter medium. Above these critical velocity (ucrit = 0.08 cm/s) and concentration (Ccrit = 130 mg/ L) the initial period of filtration duration was not significantly changed. Once a wider CaCO3 particles size distribution was filtered using filter A (See Section 4.3) or the porosity of the filter was increased (See Section 4.4), at the critical concentration of 130 mg/L, tf becomes immeasurable i.e., very short. The specific cake resistance was found to increase with filtration velocity (Fig. 6). Higher filtration velocity results in faster cake growth and higher cake resistance. Correspondingly, slower cake growth at lower filtration velocity results in lower cake resistance [14].
εs = 1 − εc
(9)
+ ΔPav⋅εc⋅A⋅Lc ,
ΔPav =
ΔP (t ) + ΔPm 2
(10)
C ⎡ ΔPm Lm ⎤ ΔPav⋅εc + εm ρs ⋅εs ⎢ 2 Lc ⎥ ⎦ ⎣
(11)
As the filtration through the filter medium is very short relative to the cake filtration the second term in Eq. (11) becomes negligible therefore,
Es ≈
tf=31 min
20 30 Time (min)
Fig. 3. Typical results of micronized CaCO3 slurry filtration with Filter A.
The specific consumed energy Es defined as the energy E per unit filtrate volume t·Q is obtained by combining Eqs. (9) and (10):
Es =
10
(8)
where εs and ρs are the solidity of the cake and the suspended solids density respectively and εc is the cake porosity. Assuming linear pressure drop ΔP along the cake and filter thicknesses. Then the energy E needed to drive the flow across the cake and the filter (the consumed energy) is,
E (t ) =
Cake filtration
Initial filtration period
40
A criterion of specific consumed energy per unit filtrate volume is proposed for constant flow filtration. This proposed index is independent of the filter type and/or filtration conditions. Therefore, it can be used to compare between various filtration systems and filtration conditions. Assuming that at t = 0 i.e., the moment of filtration starts, all of the pressure drop for filtration acts through the filter medium (ΔPm) since there is no filter cake. After that moment at t > 0, the cake is formed and builds up. The cake thickness Lc is derived via mass balance of the suspended particles neglecting liquid volume in the cake relative the filtrate volume: Q⋅C ⋅t , εs ⋅ ρs ⋅ A
Pm=77 mbar
100 60
3.2. Specific energy consumption
Lc =
120 80
(7)
The slope (Sc) of the filtration pressure against filtrate volume, or time is used to calculate the specific cake resistance α (Eq. (8)), while the intercept is the filter medium resistance ΔPm.
α=
Sc=3.2 mbar/min
Pf=143 mbar
140
εc⋅ΔPav⋅C εs⋅ρs
(12)
4. Results 100
4.1. Filtration mechanism
95 90 ΔPm (mbar)
Fig. 3 displays typical results obtained using Filter A. The filtration follows two regimes: the initial period of filtration and cake dominant filtration. At the initial period of filtration the resistance of filter medium increases exponentially by medium clogging during the initial formation of a filter cake. Once the filtration cake is formed, the increase of the filter medium resistance becomes negligible and the pressure drop acts through the cake. As the cake filtration becomes the predominant mechanism, a linear increase of the ΔP with time is observed; at which its slope (Sc) was used to calculate the specific cake resistance α (Eq. (8)). The experimental data indicated the formation of incompressible cake (until ΔP = 0.7 bar the maximum reached in the
DI C=75mg/L C=200mg/L
C=35mg/L C=130mg/L
85 80 75 70 65 60 0.03
0.05
0.08 0.10 Velocity (cm/s)
0.13
0.15
Fig. 4. Clean filter resistance (ΔP = ΔPm) as a function of filtration velocity.
3
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
250 Filter capacity (mg/cm2)
40 35
tf (min)
30 25 20 15 10 5
C=130 mg/L 200 150 100 50
0 25
50
75
100
125
150
175
C=75 mg/L
0 0.04
200
0.07 0.09 Velocity (cm/s)
Concentration of CaCO3 (mg/L)
Fig. 8. Filter capacity as a function of the filtration velocity at CaCO3 slurry concentration of 75 and 130 mg/L.
Fig. 5. Initial period of filtration (tf) at various slurry concentrations and filtration velocities.
3.0
1.2
C=75mg/L
2.5 SC 10-2 (bar/min)
α
1011 (m/kg)
1.6 1.4
C=130 mg/L
1.0 0.8 0.6 0.4 0.02
0.12
2.0 1.5 1.0 0.5 0.0
0.05
0.07 0.10 velocity (cm/s)
0.12
0
30 60 90 120 150 180 210 Concentration of CaCO3 (mg/L)
Fig. 6. Specific cake resistance at various flow rates at CaCO3 slurry concentrations of 75 and 130 mg/L.
Fig. 9. ΔP increase rate as a function of the inlet CaCO3 slurry concentration at a velocity 0.12 cm/s.
4.2. Effect of velocity and slurry concentration
concentration, less slurry can be filtered during similar filtration time and a higher cake is formed (with the same amount of slurry). For the same cake height, the filtration time has a reverse proportional relationship to the solid concentration i.e., diluted slurry filtration takes more time [1].
At a constant inlet concentration shorter filtration cycles were obtained at higher velocities (Fig. 7) and the filter capacity, up to ΔP = 0.7 bar, declined (Fig. 8). Yet, the filter capacity was found to be independent of the CaCO3 slurry concentration. The filter capacity is defined as:
4.3. Effect of particles size distribution
Filter uptake =
Mass of CaCO3 (mg ) Area of filter (cm2)
(13)
At the higher average diameter CaCO3 particles with the wider PSD longer filtration cycle was obtained resulting in higher filtration capacity of 167.7 compared with 49.9 mg/cm2. The resistance of the cake was found to be lower at a wider PSD (α = 0.45 × 1011 and 1.42 × 1011 m/kg for the wide (powder 2) and narrow (powder 1) PSD respectively) therefore, ΔP increase rate was lower (Fig. 10). In a narrow PSD the particles are smaller and more homogeneous. Reduced particles size increases the specific resistance since the drag force resisting fluid flow through the cake is proportional to the surface area of particles in the cake [17]. Therefore, the narrower PSD exhibit higher resistance compared to wider PSD.
It is evident that higher pressure drops were obtained at higher filtration velocities. Since ΔP rise is smaller at lower velocities the specific cake resistance is lower as shown in Fig. 6. Consequently, thicker cakes are formed at lower filtration velocity, for the same ΔP [14,15]. The cake formed at higher velocity is compact and possesses higher mechanical stability [14,16]. At a constant velocity of 0.12 cm/s, as the slurry concentration was increased, the filtration cycle became shorter (Fig. 9). At a higher solid
1.4
C=130 mg/L
650
1.2
C=75 mg/L
550
1.0
P (mbar)
SC
10-2 (bar/min)
1.6
0.8 0.6
450 350
0.4
250
0.2
150
0.0 0.04
Powder 1 Powder 2
50
0.07 0.09 Velocity (cm/s)
0.12
0
Fig. 7. ΔP increase rate (Sc) as a function of filtration velocity at CaCO3 slurry concentration of 75 and 130 mg/L.
20
40
60 80 100 120 140 Time (min)
Fig. 10. ΔP increase with time for narrow and wide PSD micronized CaCO3 (at a velocity of 0.12 cm/s and C = 130 mg/L).
4
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
700
14 Filter B Filter C Filter D
500
12
3m
10 Es (J/m3)
P (mbar)
600
400 300 200 100
8
2m
6 4 1m
2
0
0 0
10
20
30
40
50
0
Time (min)
5
The pore size effect on the filtration cycle is displayed in Fig. 11. As expected, the smaller pore size filter of 1 μm (i.e., high filter density) exhibited high clean filter resistance and shorter filtration cycle. However, both the 2 and 3 μm pore size filters showed similar clean filter resistance as well as similar filtration cycles. These results indicate that it is possible to use a coarser filter to separate very fine substances when the predominant mechanism is cake filtration. While the pore size and PSD of a filter determine the removal efficiency of the filter, the porosity and permeability are important in determining the rate of fluid flow through the filter. As seen in Fig. 12, the difference between the higher porosity filter of 0.52 (filter C) and the lower porosity filter A (0.18) is restricted to the initial period of filtration. While clogging of the filtration medium was observed in at the lower porosity filter the initial period of filtration was immeasurable at the higher porosity filter. As expected, the cake filtration rate was found to be similar Sc = 12.6 and 12.5 mbar/min filter A and filter C respectively; both filters exhibiting similar clean resistance (ΔPm) of 84 and 81 mbar and close filter capacities of 49.9 ± 2.83 and 45.0 ± 2.75 mg/cm2.
0.1
0.12
Es (J/m3) Filter
A
A
B
C
D
Powder Es (J/m3)
1 1.0–3.0
2 1.5–2.4
1 3.0–5.6
1 2.4–4.9
1 2.2–4.8
compared to the other tested parameters (i.e., PSD and filter pores size). Table 2 displays the specific energy calculated at V = 1 to 3 m for the two PSD powders (Fig. 2) and filters B-D (Fig. 11). As seen, the slurry mean size and PSD as well as the filters pores size had only a minor impacts on Es. This is attributed to the fact that a constant cake porosity of 0.75 was used in the calculations [3]. The difference between the filters is manifested by the clean filter resistance Rm. However, in practical filtration times, Rm is negligible relative to the cake resistance RC, as shown in Eq. (11). Another dominant factor of Es is the filtration time or filtrate volume. As seen in Figs. 13 and 14, a significant increase of Es is obtained with V or with time. Therefore, it is suggested to conduct short filtration cycles in order to save energy. However, it should be realized that short filtration cycles require larger number of filtration units (i.e., increased footprint) to enable constant product water capacity. Additionally the total energy demand also includes the energy required for the filter cleaning. In terms of economic optimization, all these parameters should be considered as well as the product water cost. 5. Concluding remarks
50 20
0.02 0.04 0.06 0.08 Velocity (cm/s)
Table 2 Specific energy for V = 1 to 3 m at a velocity of 0.12 cm/s and 130 mg/L CaCO3.
200
0.52
1m
Fig. 14. Impacts of velocity and filtrate volume per unit filter area on the specific energy consumption energy at slurry concentration of 130 mg/L. Dashed lines are parabolic approximation of Es(Q).
250
100
2
0
300
0.18
2m
3
0
350
150
3m
1
The impacts of filtration parameters on the energy consumed per unit filtrate volume or specific energy consumption, Es, was studied. The cake porosity of CaCO3, used in the calculations, is εc = 0.75 as reported for compressive pressure Ps < 1.3 bar [3]. Figs. 13 and 14 display the specific energy consumption at three cumulative filtrate volumes per unit filter area V of 1, 2 and 3 m. Results display in Fig. 13 show that Es ∝ C2. The specific energy consumption in Eq. (12) is directly proportional to C⋅ΔPav and ΔPav ∝ C, as agreed by Eq. (6). Correspondingly, Es ∝ Q2 (Fig. 14) where Q2 is proportional to the kinetic energy of the slurry flow. The effect of the CaCO3 slurry concentration on Es was found to be the dominant factor
P (mbar)
200
4
4.5. Specific energy consumption
10 15 Time (min)
150
Fig. 13. Impact of slurry concentration and filtrate volume per unit filter area on the specific energy consumption at velocity of 0.12 cm/s. Dashed lines are parabolic approximation of Es(C).
4.4. Effect of filter pore size and porosity
5
100
Concentration of CaCO3 (mg/L)
Fig. 11. Increase in ΔP with time at filter pore size of 1, 2 and 3 μm (velocity of 0.12 cm/s and C = 130 mg/L).
0
50
Comprehensive experimental characterization of Amiad threaded microfiber water filters was carried out. The filters exhibited high potential for treating water containing very fine particles. Repetitive cycles of efficient micronized CaCO3 slurry filtration followed by effective automatic filter cleaning were obtained. The high separation capability
25
Fig. 12. Increase in ΔP with time at filter porosities of 0.18 and 0.52 and pore size of 2 μm (velocity of 0.12 cm/s and C = 130 mg/L).
5
Desalination xxx (xxxx) xxx–xxx
H. Shemer et al.
[8] Eshel, H. Elifantz, S. Nuriel, M. Holenberg, T. Berman, Microfiber filtration of lake water: impacts on TEP removal and biofouling development, Desalin. Water Treat. 51 (2012) 1043–1049. [9] T. Berman, Transparent exopolymer particles as critical agents in aquatic biofilm formation: implications for desalination and water treatment, Desalin. Water Treat. 51 (2013) 1014–1020. [10] A. Lakretz, H. Elifantz, I. Kviatkovski, G. Eshel, H. Mamane, Automatic microfiber filtration (AMF) of surface water: impact on water quality and biofouling evolution, Water Res. 48 (2014) 592–604. [11] T. Kinnarinen, R. Tuunila, A. Häkkinen, Reduction of the width of particle size distribution to improve pressure filtration properties of slurries, Miner. Eng. 102 (2017) 68–74. [12] C.M. Chew, M.K. Aroua, M.A. Hussain, A practical hybrid modelling approach for the prediction of potential fouling parameters in ultrafiltration membrane water treatment plant, J. Ind. Eng. Chem. 45 (2017) 145–155. [13] N. Hasolli, Y.O. Park, Y.W. Rhee, Experimental evaluation of filter performance of depth filter media cartridge with varying the pleat count and the cartridge assembly arrangement, Par. Aerosol. Res. 8 (2012) 133–141. [14] M. Saleem, G. Krammer, M.S. Tahir, The effect of operating conditions on resistance parameters of filter media and limestone dust cake for uniformly loaded needle felts in a pilot scale test facility at ambient conditions, Powder Technol. 228 (2012) 100–107. [15] J. Sievert, F. Löffler, Dust cake release from non-woven fabrics, Filtr. Sep. 24 (1987) 424–427. [16] E. Schmidt, Experimental investigations into the compression of dust cakes deposited on filter media, Filtr. Sep. 32 (1995) 789–793. [17] R. Wakeman, The influence of particle properties on filtration, Sep. Purif. Technol. 58 (2007) 234–241.
of the filter as well as its multiple reuse point out to the potential of recovering the filtered substance. Evaluation of the consumed filtration energy at various filtration conditions and filters characteristics indicated that the slurry concentration, filtration velocity, and filtration duration are the most energy consumed parameters. Optimization of the filtration process is required, taking into account the specific consumed energy per unit filtrate volume, the cleaning energy and downtime, filters footprint, and cost of the product water. References [1] B.A. Perlmutter, Solid-liquid Filtration: Practical Guides in Chemical Engineering, Elsevier Inc, 2015. [2] C. Ghidaglia, L. de Arcangelis, J. Hinch, É. Guazzelli, Transition in particle capture in deep bed filtration, Phys. Rev. E 53 (1996) R3028–R3031. [3] S.K. Teoh, R.B.H. Tan, C. Tien, Analysis of cake filtration data - a critical assessment of conventional filtration theory, AICHE J. 52 (2006) 3427–3442. [4] K. Sutherland, G. Chas, Filters and Filtration Handbook 5th Edition, Elsevier Science, UK, 2008. [5] D.W. Hendricks, Fundamentals of Water Treatment Unit Processes: Physical, Chemical, and Biological, CRC Press, Taylor and Francis Group, 2010. [6] F.M. Mahdi, R.G. Holdich, Laboratory cake filtration testing using constant rate, Chem. Eng. Res. Des. 91 (2013) 1145–1154. [7] D. Ityel, Ground water: dealing with iron contamination, Filtr. Sep. 48 (2011) 26–28.
6