You are on page 1of 9

South African Journal of Chemical Engineering 39 (2022) 97–105

Contents lists available at ScienceDirect

South African Journal of Chemical Engineering


journal homepage: www.elsevier.com/locate/sajce

Hydrodynamic evaluation of a filter bed of porous material from stratified


sedimentary rocks for the removal of turbidity in surface waters
Angel Villabona-Ortíz , Candelaria Tejada-Tovar *, David López-Barbosa
Chemical Engineering Department, Process Design, and Biomass Utilization Research Group (IDAB), Universidad de Cartagena, Avenida del Consulado St. 30. Cartagena
de Indias, 130015, Colombia

A R T I C L E I N F O A B S T R A C T

Keywords: In this study, the behavior of a filter bed of stratified sedimentary rocks was evaluated through the analysis of
Bed turbidity removal efficiency and head losses. Two types of sands were used as materials, fine and medium, as well
Stratification as crushed limestone. The materials were characterized to determine the average particle diameter, porosity, and
Filtration
permeability coefficient, obtaining respectively 0.17 mm, 0.39, and 1 × 10− 3 cm/s for fine sand, 0.33 mm, 0.41,
Sand
Limestone
and 3 × 10− 3 cm/s for medium sand, and 1.26 mm, 0.45, and 1 × 10− 2 cm/s for crushed limestone. Hydro­
Turbidity dynamic analyses of the clean filter under conditions of rapid filtration were carried out, by which the different
Surface water rates of filtration to be used in each experiment were determined, establishing a direct correlation between the
rate of filtration and the thickness of the bed; in this way, it was possible to determine the maximum bed
thickness that could be used to maintain a rate of rapid filtration higher than 120 m/day. A hydrodynamic study
with synthetic turbid water prepared at 8 NTU using tap water and bentonite was also carried out. Total head
losses and turbidity removal were studied for each material at a rate of 153 m/day with removal efficiencies of
93.7% for medium sand with a loss of 23 cm and 67.8% for limestone with a loss of 4.5 cm and a rate of 389 m/
day. It is recommended to use the medium sand and limestone type material for use in surface water filtration
processes.

1. Introduction filtration of water through filters with porous media; these technologies
allow obtaining effluent water with good physical-chemical and bacte­
Drinking water is an increasingly scarce resource whose supply is riological quality (Verma et al., 2017). In this sense, several studies have
diminishing due to the depletion of water sources and pollution, while been carried out on different porous materials, of mineral or sedimen­
demand is rapidly increasing due to population growth, industrializa­ tary rock type, as alternatives to conventional sand filters. The potential
tion, mechanization, and urbanization (Sahu et al., 2019). Thus, the for water filtration that these materials have is due to the fact that they
processes of treatment and potabilization of raw water, extracted from have two levels of porosity, one of the materials itself and the other of
surface and underground sources, are necessary to provide safe quality the filter bed. Among the most relevant materials are granular activated
water for consumption (Torres-Lozada et al., 2018). Much of the water carbon, dolomite, calcite, limestone, sandstone, pumice, and zeolites
used in rural and urban areas is extracted from surface sources such as (Farhaoui and Derraz, 2016). These porous materials can be found in
rivers, streams, lakes, ponds, and springs, which is usually treated by different soil layers, which gives it the ability to transport and store
clarification by flocculation or sedimentation, filtration, and disinfec­ water (Sabiri et al., 2017).
tion (Ma et al., 2017). One of the main surface water contaminants is The objective of this research was to evaluate, through the analysis of
associated with the presence of turbidity, which is conventionally the efficiency of turbidity removal and head losses, the behavior of a
removed using filtration by deposition of the particles in the filter media stratified sedimentary rock porous material taken from soil in the village
(Sutherland, 2014). of Galerazamba, Bolívar. The material was characterized by granulo­
The development and implementation of appropriate water purifi­ metric analysis, and density and permeability determination. The
cation technologies at rural community level in developing countries is technology proposed in this work is an alternative treatment for surface
necessary due to the lack of economic resources. Among these is the waters that have issues with turbidity, microorganisms, and organic

* Corresponding author.
E-mail address: ctejadat@unicartagena.edu.co (C. Tejada-Tovar).

https://doi.org/10.1016/j.sajce.2021.10.002
Received 11 March 2021; Received in revised form 27 August 2021; Accepted 20 October 2021
Available online 25 October 2021
1026-9185/© 2021 The Authors. Published by Elsevier B.V. on behalf of Institution of Chemical Engineers. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
A. Villabona-Ortíz et al. South African Journal of Chemical Engineering 39 (2022) 97–105

Fig. 1. PFD of the pilot filtration plant.

matter, it is low-cost, very efficient, easy to operate, suitable for small N.V. E-104-13. Afterward, they were dried in an oven at 110 ± 5 ◦ C to a
communities, of application at both domestic and public scales, and it constant mass and then reduced in size. The granulometric analysis was
can be improved by a process of disinfection (Zuluaga-Gomez et al., determined following the I.N.V. E-123-13 Standard (Instituto, 2014).
2020). The bulk density was determined by the cylinder method (Eq. (1)). A soil
sample was selected, and the auger was introduced with the cylinder,
2. Methodology taking care not to compact the soil so it would keep the same structure
and known volume. The cylinder was hermetically sealed so that there
2.1. Materials was no loss of moisture and weight. The sample was dried at 110 ◦ C for
24 h or until constant weight, and the moisture was determined by loss
The granulometric analysis of the sedimentary rock samples was of mass. The volume of the cylinder was calculated according to the
performed using Standard A.S.T.M.E-11 sieves. The turbidity of the diameter and height of the cylinder.
samples was determined using a VELP TB1 turbidimeter. Synthetic
ρ = Ws /Vt (1)
turbid water was prepared at a concentration that emulated a real sur­
face water reservoir using tap water and bentonite. In this work, bed Where Vt is the cylinder volume, and Ws is the dry soil weight.
depth (cm) and initial turbidity (NTU) were considered as independent The density of the samples was determined according to the standard
variables; and flow rate (cm3/s), head losses (cm), removal efficiency I.N.V. E-128-13 using a VIDRIOLAB 250 mL pycnometer (Instituto,
(%), and the type of filter bed material were considered as response 2015). Porosity (ε) was calculated according to Eq. (2):
variables.
ε = ρr − ρa / ρr (2)

2.2. Design of experiments Where ρr and ρa are the apparent and real densities, respectively, in
g/cm3. The real density of soils corresponds to the density of the total
The hydrodynamic analysis was performed following a multilevel particles, excluding the pore spaces between particles (Xu and Bezuijen,
factorial experimental design 2 × 3 × 5, where three factors are taken 2019). The apparent density was obtained as a relation between the
into account: the initial turbidity (0 and 8 NTU), the bed material (sand volume occupied in a specimen by a certain mass of material, the real
1, sand 2, limestone), and the bed depth (1 – 5 cm), for a total of 60 density was obtained by a specific gravity test according to Eqs. (3) and
experiments. All experiments were performed in duplicate. (4) using a 250 mL pycnometer and 90.4 g and taking as reference the
density of distilled water at 4 ◦ C (1 g/cm3); both tests were performed in
triplicate for each material and the result was averaged. Specific gravity
2.3. Filter bed characterization (SG) and real density (ρr) are calculated using the following ratios:

The sand and limestone samples were taken in the village of Galer­ SG = Wm /Wm + Wwater + Wm+water (3)
azamba (Bolivar), using a sharp-edged cylinder according to Standard I.

98
A. Villabona-Ortíz et al. South African Journal of Chemical Engineering 39 (2022) 97–105

Fig. 2. Flow diagram of the pilot filter calibration process for constant level control.

SG = ρr /ρw (4) Table 1


Specifications of the supporting gravel.
Where Wm, Wwater, and Wm+water are the mass of material, water, and
material+water in g, respectively. Layer Gravel size (mm) Approximate thickness of the medium (cm)
Finally, the permeability of air-dried soil samples containing less 1 9 – 10 1.0 – 1.5
than 10% of soil passing 75 µm (No.200) sieve and removing sizes larger 2 2–9 0.5 – 1.0
than 19.0 mm (3/4′′ ) was determined using a 58 cm head, a 6.5 cm 3 1–2 0.5

diameter packed bed, and 5 cm depth. During the test, five times the
flow rate or volume discharged by the equipment was measured in a plug or clog, decreasing the filtration rate and causing an increase in the
given amount of time, the average flow rate was used to calculate the water level. To do this, V-103 was gradually opened as the filtration run
permeability coefficient (κ) according to Eq. (5) (Zhou et al., 2019). (I. progressed in order to keep the water level at a constant C-101.
Instituto Nacional de Vías, 2015). A system consisting of three layers was used as drainage: a High-
k = Q ∗ L/A ∗ t ∗ h (5) Density Polyethylene geogrid covered on both sides by a Poly­
propylene non-woven geotextile, the geotextile serves the function of
Where Q is the amount of water discharged in L, L is the distance preventing fine sand particles to pass into the effluent without
between the gauges from the permeameter in mm, A is the cross- obstructing the water flow. As a support for the sand bed, a spherical
sectional area of the specimen in mm2, t is the total drainage time in s gravel bed of approximately 2.5 cm deep was used, stratified in several
and h is the head difference between the gauges in mm. layers according to the particle size, following the recommendations of
the Colombian Regulations (C. y T., 2013); the stratification used is
2.4. Construction of the pilot filtration plant presented in Table 1.
A calibration curve was made to relate the concentration of solids in
As a filtration column, an acrylic cylinder of 35 cm height and 4.1 cm mg/L with the turbidity of the water in NTU. Initially, the synthetic
internal diameter was used, coupled at each end with a universal PVC water was prepared, taking into account the bentonite concentration in
fitting of 1½”; these fittings allow the column to be easily assembled and mg/L to obtain the desired turbidity according to the calibration curve;
disassembled from the system for easy cleaning. Once the cylinder was then the total volume of water to be prepared was taken and the total
connected to the fittings, a cm-scale was placed and holes were drilled amount of bentonite in mg was calculated. With this amount the sus­
on the side of the column, from the base and 1 cm away from each other, pension was prepared and diluted to the total volume, taking into ac­
which were used to couple a system of piezometers and sampling points count the dead volumes that were used both to keep the P-101 pump
to the column, similar to the pilot filter used by (Farizoglu et al., 2003). submerged in water and to keep the level constant in the T-102 tank,
Once the column was completed, it was attached to a system of PVC which was calculated with the dimensions of the tank to a total of 4.2 L.
piping, ½” valves, and a flow meter that conducts and controls the flow
of water through the system. Three tanks were installed at the plant: a 2.5. Clean filter analysis
60 L tank for storing turbid water equipped with a 17 W submersible
pump, a 7 L tank for feeding water to the system, and a 9 L tank for The clean filter hydrodynamic analysis was performed by filtering
collecting treated water, as shown in the pilot plant’s Process Flow Di­ clean water (T0 ≈ 0 NTU), using the three filter materials under study
agram (PFD) Fig. 1. (sand 1, sand 2, and limestone). The discharge rate was measured in
The filtration pilot plant was placed in operation, loading T-101 with triplicate using a 100 mL test tube and the average value was used to
the water to be treated, then P-101 was turned on to pump water to T- calculate the filtration rate. Head losses were measured at different
102, in this tank the water level rises to an overflow placed at 8 cm and depths of the bed using the system of piezometers attached to the col­
then recirculates to T-101; once the level in T-102 is stabilized, V-101 is umn, located 1 cm away from each other, these losses represent the
opened to allow the flow to the filtration column, V-102 was kept initial or clean filter losses h0 (Fig. 3).
completely open during the filtration process and it is closed to measure
the water flow through V-101; when the water level in C-101 reached 2.6. Soiled filter analysis
the desired level, V-103 was opened until the desired rate was obtained
and V-101 was manipulated in order to calibrate the plant with control The hydrodynamic analysis of the soiled filter was done by carrying
at a constant level. When the flow through V-101 and V-103 was out the filtration process using turbid water with initial turbidity of 8
equalized, the water level was kept constant and the filtration run was NTU, this analysis was done for each of the materials under study (sand
started (Fig. 2). When working with cloudy water, constant rate control 1, sand 2, and limestone), considering the same assembly done for the
was used, because of the retention of material in the bed, which tends to clean filter (Fig. 3). The losses obtained at this stage represent the total

99
A. Villabona-Ortíz et al. South African Journal of Chemical Engineering 39 (2022) 97–105

Fig. 3. Graphic representation of the assembly and the response variables, to an initial turbidity T0 and a specific bed material with thicknesses of (a) 1 cm, (b) 2 cm,
(c) 3 cm, (d) 4 cm, (e) 5 cm.

losses or hT ; by subtracting from the total losses the initial losses h0 , the 2.7. Evaluation of the stratified filter of sedimentary rocks
final or by clogging losses hf are obtained. The turbidity was measured at
different levels of bed depth by taking a water sample using a 15 mL The operating parameters, such as stratification and filtration rate to
syringe, through sampling points coupled to the piezometer system. be used were determined based on the results obtained in the hydro­
Both turbidity and losses were measured every 15 min during a 2-hour dynamic analysis of each material. The thickness of each material that
filtration run. obtained the best results for removal percentages and head losses was
The head losses hi , were obtained for each value of bed thickness L established as the thickness of each layer of the stratified bed, placing
through piezometers located at 1 cm distance between each one, these the material with the highest particle diameter in the upper layer and the
losses were added to obtain the total head loss or hT , using a total head h one with the lowest particle diameter in the lower layer.
of 30 cm for each experiment; with this information, the hydraulic pa­ Head losses and turbidity removal percentages were obtained by
rameters described in Darcy’s Law, such as the energy available Δh, and stratification evaluation, using turbid water with initial turbidity of 8
the hydraulic gradient i, were calculated as described in Eq. (6): NTU, and response variables were measured in a similar manner to the
hydrodynamic analysis. Turbidity profiles and time-dependent head
Δh = h − h0 ; i = Δh/L (6)
losses were obtained during a 2-hour filtration run. The turbidity
removal efficiency (TRE) was calculated with Eq. (7).
TRE (%) = Ti − Te /Ti × 100 (7)

100
A. Villabona-Ortíz et al. South African Journal of Chemical Engineering 39 (2022) 97–105

Table 2
Data for differential and accumulative granulometric analysis of the sand sample 1.
Differential analysis Accumulative analysis
Material No. sieve Opening (mm) Retained mass (g) Retained fraction (Δφi Average diameter Dpi (mm) Accumulative fraction Passing fraction

Sand 1 30 0.6 0 0 — 0 1
50 0.3 0.8 0.0035 0.45 0.0035 0.9965
100 0.15 163.9 0.7223 0.225 0.7259 0.2741
200 0.075 59.9 0.2640 0.1125 0.9899 0.0101
Fondo 0 2.3 0.0101 0.035 1 0
Total (g) 226.9 1 — — —
Sand 2 20 0.85 0 0 — 0 1
30 0.6 20.7 0.1035 0.725 0.1035 0.8965
40 0.425 43.6 0.218 0.5125 0.3215 0.6785
60 0.25 82.5 0.4125 0.3375 0.734 0.266
100 0.15 43.1 0.2155 0.2 0.9495 0.0505
200 0.075 10.1 0.0505 0.1125 1 0
Total (g) 200 1 — — —
Limestone 8 2.36 0 0 — 0 1
10 2 37.6 0.1705 2.18 0.1705 0.8295
20 0.85 149.6 0.6785 1.425 0.849 0.151
30 0.6 20.6 0.0934 0.725 0.9424 0.0576
40 0.425 12.7 0.0576 0.5125 1 0
Total (g) 220.5 1 — — —

Table 3 Table 4
Granulometric parameters calculated from differential and accumulative Porosity calculation.
analysis.
Fine Medium Crushed
Material Average d60 d10 Effective Uniformity sand sand limestone
diameter dM (mm) (mm) size de coefficient, Cu
Average apparent density (g/ 1.39 1.49 1.33
(mm) (mm)
cm3)
Sand 1 0.17 0.2 0.1 0.1 2 Average real density (g/cm3) 2.65 2.66 2.42
Sand 2 0.33 0.39 0.19 0.19 2.05 Porosity 0.47 0.44 0.45
Limestone 1.26 1.6 0.71 0.71 2.25

Where Ti and To are the turbidity of the affluent and effluent in NTU, Table 5
Constant head permeability test results.
respectively.
Fine sand Medium sand Crushed limestone
3. Results Test Output Time Output Time Output Time
(cm3) (s) (cm3) (s) (cm3) (s)

3.1. Characterization of the filter bed material 1 12.5 30 37 30 65 15


2 11 30 36 30 65 15
3 10.5 30 35 30 64 15
Four materials were collected: a clayey material, two sandy materials 4 10.5 30 35 30 64 15
(sand 1 and sand 2), and a sedimentary rock type material of calcareous 5 10.5 30 36 30 65 15
origin (limestone). Of these, the clayey material was not taken into ac­ Average 11 30 35.8 30 64.6 15
Flow 0.37 1.19 4.31
count because it was a plastic clay. The granulometric analysis was
(cm3/s)
carried out using a different series of sieves depending on the nature of κ (cm/s) 9 × 10− 4
3 × 10− 3
1 × 10− 2

the material. With the data of the mass retained in each sieve, the
respective differential and accumulative analyses were made for each
material (Table 2). In the differential analysis, the fractions of the sample respectively.
sample Δϕi , retained in sieve i, are represented, assuming that the To calculate the porosity according to Eq. (2), the apparent and real
retained particles have a size equal to the arithmetic average of the density values of each material were first determined. Average values of
meshes between which said fraction is retained; therefore, ϕi represents apparent density and real density for each filter material were obtained,
the fraction in weight of the sample constituted by particles of diameter with deviations of 0.01. The summary of porosity calculation is pre­
dpi . With these values, parameters such as the average diameter of the sented in Table 4.
sample were calculated. The accumulative analysis is made from the From the results obtained, it is deduced that the fine sand is a loamy
differential analysis by the accumulative sum of the individual differ­ soil material, and the medium sand is a sandy-loam soil material;
ential increments, from the one retained in the mesh with the largest however, there is no point of comparison for the limestone bed since it
opening, which allows obtaining parameters such as the effective par­ was crushed to a specific particle size (Hoslett et al., 2018). For lime­
ticle size and the uniformity coefficient (Andreoli and Sabogal-Paz, stone, a range of 2.4–2.6 g/cm3 was defined (González de Corbella and
2020). Ruiz Abrio, 2015). According to the final results of porosity, it is
A summary of the granulometric parameters for the materials found observed that the greater porosity was obtained for the fine sand, which
is presented in Table 3. According to ASTM D2487-17, sand 1 can be means that this one has a higher amount of empty space, nevertheless,
classified according to particle size as fine sand, sand 2 as medium sand, this does not mean that it is the most appropriate for retention or
and limestone as very coarse sand; this granulometric distribution for transport of water since this test does not establish what percentage of
limestone is due to the fact that it was previously crushed (ASTM In­ these pores are interconnected. A range of total porosity has been
ternational, 2017). The values of d60 and d10 obtained, represent the established for both fine and coarse sands between 0.2 and 0.5, but only
opening of the dummy mesh that would let pass 60% and 10% of the between 0.1 to 0.35 is effective porosity, that is, they are interconnected

101
A. Villabona-Ortíz et al. South African Journal of Chemical Engineering 39 (2022) 97–105

Table 6
Calculation of filtration rate.
Material Bed thickness (cm) Discharged flow (cm3/s) Filtration rate (cm/s) Filtration rate (m/day)
1 2 3 Average Deviation

Sand (1) 1.00 1.00 1.00 1.00 1.00 0.00 0.076 65.60
2.00 0.65 0.65 0.66 0.65 0.01 0.05 43.20
3.00 0.33 0.35 0.33 0.34 0.01 0.026 22.50
4.00 0.25 0.25 0.25 0.25 0.00 0.019 16.4
5.00 0.19 0.20 0.20 0.20 0.01 0.015 13.0
Sand (2) 1.00 8.50 8.50 8.50 8.50 0.00 0.64 553.00
2.00 7.00 6.80 6.90 6.90 0.10 0.52 449.30
3.00 4.6 4.50 4.50 4.53 0.06 0.34 293.80
4.00 2.30 2.40 2.30 2.33 0.06 0.17 152.70
5.00 1.10 1.00 1.00 1.03 0.06 0.078 67.40
Crushed Limestone 1.00 10.00 10.00 10.00 10.00 0.00 0.76 656.60
2.00 8.60 8.80 8.30 8.57 0.25 0.65 561.60
3.00 6.70 6.70 6.50 6.63 0.12 0.50 432.00
4.00 6.00 6.00 6.00 6.00 0.00 0.45 388.8
5.00 5.70 5.50 5.30 5.50 0.20 0.41 354.2

During the clean filter analysis, the filtration velocity or rate (m/day)
was determined by the correlation between the flow discharged by the
filter and the surface area of the bed. The discharged flow was calculated
as the volume discharged in 10 s, this was measured in triplicate and
averaged; the bed diameter is equivalent to the internal diameter of the
column (4.1 cm), therefore the surface area was calculated at 13.2 cm2.
The experimental data of measured flow and calculated rate are pre­
sented in Table 6. Fig. 4 shows the filtration rate profile as a function of
the bed thickness used.
This research studies the behavior of two sandy materials and one
limestone type, under conditions of rapid filtration (this condition oc­
curs when the filtration rate is greater than 120 m/day) (C. y T., 2013).
From this and the data recorded in Table 6, it can be seen that the rate at
which sand 1 operates does not meet the requirement for rapid filtration,
so this material was not taken into account for the rest of the research. In
the same way, it was determined by linear interpolation that sand 2 has a
thickness limit of 4.6 cm in which the rate equals 120 m/day, therefore it
was decided to use a maximum thickness of 4 cm for the rest of the
Fig. 4. Filtration rate profile as a function of bed thickness. investigation.
Regarding the filtration time, a filtration rate that varied between 1.7
pores. Previous studies report a porosity of 0.41 for coarse gravel and and 7.2 m/day was previously reported in a slow sand filter for use at
0.45 for fine gravel (Andreoli and Sabogal-Paz, 2020). home with a turbidity removal of. The most influential variable in the
The permeability coefficient (κ) was obtained through a constant establishment of the filtration period was the quality of the fed water
head permeability test; the results of this test are shown in Table 5. (Maciel and Sabogal-Paz, 2018). On the other hand, when using sand
According to the permeability values obtained, it can be established and anthracite as a filter medium in two rapid filter columns, a constant
that fine sand (sand 1) has a low degree of permeability, which means filtration rate of 72 m/day was established; It was found that the quality
that, despite having a high porosity, a large part of these pores are not of the effluent of the anthracite obtained was better for all the filtering,
interconnected; therefore, it is considered that this material is not the development of the pressure drop over time was greater for the sand
appropriate for the process of infiltration or runoff, since these non- filter in all the filtering cycles (Sabiri et al., 2017).
interconnected voids cannot contribute to the transport of water From the information in Fig. 4, it can be seen that the behavior of
through the porous media. (Lee and Kim, 2018). On the other hand, both sands 1 and 2 show an approximately linear trend, which indicates a
the medium sand (sand 2) and the crushed limestone, have a medium proportionality between the filtration rate and bed thickness, that is, the
permeability level, being higher for the calcareous material, which initiation of flow in these materials is given as a free or gravitational
means that these materials have a better interconnection between their flow that depends on a combination of external hydraulic load, flow
pores and allow a better water transport through the porous media. length and permeability coefficient. On the other hand, limestone does
Permeability coefficient values have been determined for fine sands not show this proportionality reason why it is not possible to establish a
between 10− 6 and 10− 4 cm/s, for coarse sands between 10− 4 and free flow in this material, which is given by the effect of the interaction
10− 2 cm/s, and for gravel >10− 2 cm/s (Fujikura, 2019), which is between electrical charges present in the material and the electrical
consistent with the results obtained in this research. charges of the water; in these materials, the flow is generated by surface
tension, in the case of capillary flow, or must be forced by the applica­
tion of energy different from that associated exclusively with the
3.2. Hydrodynamic analysis of filter bed materials terrestrial gravitational field (Rybakov et al., 2019), (Vecherkovskaya
and Popereshnyak, 2017).
The pilot filtration equipment was calibrated to control the level and
filtration rate by establishing an initial opening of V-103 at half a turn of
the valve (50%), and a level of 30 cm on the scale located on the 3.3. Obtaining initial head losses
filtration column. All experiments were standardized for these
conditions. The experimental data of the measured head losses for each bed

102
A. Villabona-Ortíz et al. South African Journal of Chemical Engineering 39 (2022) 97–105

Table 7
Initial or clean filter head losses.
Material Bed thickness (cm) Filtration rate (m/day) Head losses (cm)
h1 h2 h3 h4 h5 h0 Δh i

Medium sand 1 553 1.1 - - - - 1.1 28.9 28.9


2 449.3 2.3 1.2 - - - 3.5 26.5 13.25
3 293.8 2.5 4.2 1.7 - - 8.4 21.6 7.2
4 152.7 4.5 5.8 3.7 2.6 - 16.6 13.4 3.35
5 67.4 3.8 5.9 7 4.9 3.4 25 5 1
Crushed Limestone 1 656.6 0.1 - - - - 0.1 29.9 29.9
2 561.6 0.4 0.5 - - - 0.9 29.1 14.55
3 432 0.9 1 0.4 - - 2.3 27.7 9.23
4 388.8 1 1.5 0.9 0.4 - 3.8 26.2 6.6
5 354.2 0.5 1.3 2 1.7 1 6.5 23.5 4.7

Fig. 5. Initial head loss profile as a function of filtration rate.


Fig. 6. Total head loss profile as a function of time at different bed thicknesses.

thickness value and their respective rate are shown in Table 7. Based on
Darcy’s Law ratio is not met and then the non-linear behavior is
the information in Table 7, it is observed that the losses are considerably
observed (Daneshi et al., 2018).
greater for sand than for limestone, this is due to the fact that sand
opposes greater resistance to the viscous forces that drive the flow due to
3.4. Soiled filter analysis
the relationships between different parameters of the material such as
the sphericity of the particles, permeability, porosity and particle
The total head losses hi , in cm, were obtained for each bed thickness
diameter, causing the filtration rate produced by sand to be lower
value L through piezometers located at 1 cm distance from each other
compared to limestone (Singh, Singh & Gangacharyulu, 2016).
during a 2 h filtration run. These losses were added up to obtain the total
Fig. 5 shows the profile of initial head losses as a function of filtration
head loss (hT ); It was used a total head h equal to 30 cm for each
rate, observing a behavior initially linear for sand that then becomes
experiment. The total head loss profiles as a function of the filtration run
non-linear, while the limestone presents a completely non-linear
are shown in Fig. 6.
behavior of initial head losses with respect to filtration rate, this is
The variation of the total head losses is evident as a function of the
consistent with the results reported by (Khelladi et al., 2020), which
bed thickness and the filtration run. For sand 2, it is observed that the
studied filter losses with three varieties of sand of different particle size
thickness of the bed is the most significant variable in the filtration run,
and porosity.
obtaining only 15 min of the race when the thickness was 1 cm before
It was determined that the behavior presented is due to the high
the bed is completely plugged, and the head losses reach 30 cm of total
filtration rates due to the hydraulic conditions of the filter media
head, until a race greater than 2 h when the bed of 4 cm was used, this
described by the Reynolds Number. The Reynolds Number calculated in
behavior was also observed by (Mahanna et al., 2015), in which it is
this experiment varies for sand from 0.28 to 2.37, which explains the
attributed to the capacity of penetration of the material suspended in the
initially linear behavior for the lower rates, for limestone the Reynolds
bed due to the high filtration rates and the deposition of this material in
Number varies from 5.79 to 10.74, which explains the non-linear
the pores of the bed; in the case of limestone, there is no significant
behavior; this difference of the Reynolds Number is due to the me­
relationship in the filtration run in relation to head losses or bed
dium particle size being larger for limestone than for sand. According to
thickness, due to the fact that the flow through the limestone bed is not
(Taghavijeloudar et al., 2019), the non-linear behavior of the head losses
driven in the same way as in a sand bed by the possible interaction
as a function of the filtration rate exhibited by the pilot filtration
between electrical charges present in the material and the capillary flow
equipment when using the two filter materials could be due to the large
by the action of surface tension. The linear increase of total head losses
initial flow handled in the present study; this considering that it has been
when using the sand at different bed heights, due to the constant
reported as a significant part of the total permeate flow (approximately
accumulation of material in the bed, decreasing the porosity and
30–50%) (Yao et al., 2018). Likewise, for Reynolds Number values less
restricting the flow, which indicates that the phenomenon of filtration in
than or equal to one (Re ≤ 1) the head losses behave in a linear manner
the sand occurs by sifting; however, in the limestone is also observed a
with respect to the filtration rate, following Darcy’s Law ratio; for higher
linear behavior in time but tends to remain constant, suggesting that the
rates where the Reynolds Number is greater than one (Re > 1) the
mechanism of filtration in the limestone is not given by retention of

103
A. Villabona-Ortíz et al. South African Journal of Chemical Engineering 39 (2022) 97–105

respectively. For the limestone bed, a period for filter column wash was
not determined, as this bed was only left running for up to 4 h, in which
time it did not get clogged. Table 8 presents a comparison between the
results of this research with other studies carried out, in which it is
observed that the different studies of rapid sand filters have obtained
similar turbidity removal percentage results, which gives credibility to
the results presented in this research.
From Fig. 7 it is established that the behavior of the percentage of
turbidity removal as a function of time for the sand presents a linear
trend in accordance with the trend of the head losses since as the losses
increase there is a reduction in the removal efficiency, this is due to the
fact that as the filtration run passes the suspended material accumulates
in the bed diminishing the porosity and therefore limiting the capacity of
adhesion of the suspended solids causing that these solids penetrate
through the bed towards the effluent (Davies and Wheatley, 2012). For
the limestone it is observed that the profile does not present a defined
trend, but tends to oscillate in time and to stabilize in a final value of
turbidity, this added to the fact that the total head losses tend to remain
Fig. 7. Turbidity removal profile as a function of time for a bed thickness
constant in time allows supposing that the phenomenon by which the
of 1 cm.
limestone removes turbidity is not by the deposition of particles in the
middle or sifting, which occurs with the sand. Previously, it was
particles in the bed (Yiǧit Hunce et al., 2016). concluded that the removal potential of limestone is due to an adsorp­
tion phenomenon that occurs on the surface of the solid thanks to the
3.5. Turbidity removal percentage profiles presence of calcium ions in its composition (Affam and Ezechi, 2019).
It has been previously reported that the turbidity removal efficiency
The turbidity values Ti , in NTU, were obtained similarly for each bed increased with increasing initial turbidity of the water sample; and that
thickness value L through sampling points located 1 cm away, with decreases at high flow rates. Thus, removal efficiencies of 57% and 22%
initial turbidity of approximately 8 NTU. Fig. 7 shows the turbidity were achieved when a filtration rate of 388.8 m/day and 1719.4 m/day
removal percentage profiles as a function of the filtration run. It is noted was used. (Senaratne et al., 2001). Likewise, when evaluating sand fil­
that at the beginning of each race of filtration, the turbidity in the ters with different bed height, a decrease in turbidity from 1.6 NUT to
effluent is relatively high and it takes a few minutes to reach the mini- 0.0724 NTU was achieved, finding that the turbidity of the fluctuating
mum turbidity in a period called maturation or "ripening" (Mahanna effluent did not affect the performance of the filter with respect to the
et al., 2015); in the present study the maturation period was located removal of turbidity. (Napotnik et al., 2017). Previously, it was reported
between 5 and 10 min and the filtration run was timed after the end of that the increase in turbidity in the feed water has a positive effect on the
this period. removal efficiency, the above taking into account that there is greater
The variation of the turbidity according to the thickness of the bed availability of dissolved and suspended solids to be retained in the
and the filtration run is evident. For sand 2 it was presented that in each packed filter bed. (Maciel and Sabogal-Paz, 2018).
case the requirement of Colombian Regulations for treated drinking
water is fulfilled, which establishes a maximum permissible turbidity 4. Conclusion
value of 2 NTU (Martínez-Orjuela et al., 2020), (Martínez-Orjuela et al.,
2020), obtaining minimum values of 0.71 NTU for the 1 cm bed, 0.6 This study concludes that: (i) the use of medium sand and calcareous
NTU for the 2 cm bed, 0.54 NTU for the 3 cm bed, and 0.5 NTU for the type material is recommended as a filter bed material. (ii) From the
4 cm bed, with average values in time of 0.8 NTU, 0.84 NTU, 0.96 NTU, characterization, it was obtained that these materials have an average
and 0.73 NTU, respectively. It was also presented that, only for the 4 cm particle diameter, porosity, and permeability coefficient of 0.17 mm,
bed, the 2-hour filtration run was completed without complete filter 0.39, and 1 × 10− 3 cm/s for fine sand, 0.33 mm, 0.41, and 3 × 10− 3 cm/
plugging, because it allowed a better distribution of the suspended s for medium sand, and 1.26 mm, 0.45 and 1 × 10− 2 cm/s crushed
material in the bed due to the lower filtration rates increase the removal limestone, respectively. (iii) The hydrodynamic analysis of the clean
efficiency by increasing the time of water retention in the bed (Mahanna filter made it possible to identify a direct ratio between the filtration rate
et al., 2018). For the 4 cm sand bed a maximum filtration run of and the bed thickness, making it possible to determine which maximum
approximately 2 h and 30 min was achieved before the filter got bed thickness could be used to maintain a rapid filtration rate higher
completely clogged, which means that after this period of time the filter than 120 m/day. (iv) From the soiled filter analysis, the total head loss
column would need to be washed. In the case of limestone, the standards and turbidity removal values were obtained for each material, using
are not met, obtaining minimum values of 4.36 NTU for the 1 cm bed, turbid water of 8 NTU; at a rate of 153 m/day presenting removal effi-
4.02 NTU for the 2 cm bed, 3.27 NTU for the 3 cm bed, and 2.59 NTU for ciencies of 93.7% for the medium sand with a loss of 23 cm and
the 4 cm bed, with amounts that vary over time and tend to stabilize at the limestone a removal of 67.8% with a loss of 4.5 cm and a rate of 389
approximately 5.4 NTU, 4.56 NTU, 4.24 NTU, and 4.15 NTU, m/day.

Table 8
Turbidity removal percentage result comparison.
Bed Material Bed thickness (cm) Filtration rate (m/día) Turbidity (NTU) Removal efficiency (%) Reference

Coarse sand 75 183 30 90 (Farizoglu et al., 2003)


Sand 60 206 9.27 88.4 (Davies and Wheatley, 2012)
Limestone - 12 150 79
Course sand 80 144 10 95 (Mahanna et al., 2018)
Medium sand 4 153 8.05 93.8 This study
Crushed linestone 4 389 8.05 67.8

104
A. Villabona-Ortíz et al. South African Journal of Chemical Engineering 39 (2022) 97–105

I. Instituto Nacional de vías, norma I.N.V. E-128-13- Determinación del peso específico
de los suelos y del llenante mineral. 2015, pp. 1–8.
Khelladi, R.M.B., Fellah, A.C., Pontié, M., Guellil, F.Z., 2020. Influence of particle and
Funding grain size on sand filtration: effect on head loss and turbidity. Aquat. Sci. Technol 8
(2), 36. https://doi.org/10.5296/ast.v8i2.17512.
This research did not receive any specific grant from funding Lee, J.S., Kim, C., 2018. Study on permeability, optimum yield and long-term stability in
alluvial well with filter layer change. J. Eng. Geol 28 (1), 101–115.
agencies in the public, commercial, or not-for-profit sectors. Ma, L., et al., 2017. Fabrication and water treatment application of carbon nanotubes
(CNTs)-based composite membranes: a review. Membranes (Basel) 7 (1). https://doi.
Disclosure statement org/10.3390/membranes7010016.
P.M.F. Maciel and L.P. Sabogal-Paz, “Household slow sand filters with and without water
level control: continuous and intermittent flow efficiencies,” https://doi.org/
The authors reported no potential conflict of interest 10.1080/09593330.2018.1515988, vol. 41, no. 8, pp. 944–958, Apr. 2018, doi:
10.1080/09593330.2018.1515988.
Mahanna, H., Fouad, M., Radwan, K., Elgamal, H., 2015. Predicting of effluent turbidity
CRediT authorship contribution statement from deep bed sand filters used in water treatment. Int. J. Sci. Eng. Res 6 (9),
621–626. https://doi.org/10.14299/ijser.2015.09.006.
Angel Villabona-Ortíz: Conceptualization, Visualization, Method­ Mahanna, H., Radwan, K., Fouad, M., Elgamal, H., 2018. Effect of operational conditions
on performance of deep sand filter in turbidity removal. Trends Tech. Sci. Res 2 (5),
ology, Software. Candelaria Tejada-Tovar: Data curation, Writing – 1–7.
original draft. David López-Barbosa: Investigation, Software, Formal Martínez-Orjuela, M.R., Mendoza-Coronado, J.Y., Medrano-Solís, B.E., Gómez-Torres, L.
analysis, Validation, Writing – review & editing. M., Zafra-Mejía, C.A., 2020. Evaluación de la turbiedad como parámetro indicador
del tratamiento en una planta potabilizadora municipal. Rev. UIS Ing 19 (1), 15–24.
https://doi.org/10.18273/revuin.v19n1-2020001.
Declaration of Competing Interest Napotnik, J.A., Baker, D., Jellison, K.L., 2017. Effect of sand bed depth and medium age
on escherichia coli and turbidity removal in biosand filters. Environ. Sci. Technol 51
The authors declared that they have no conflicts of interest to this (6), 3402–3409. https://doi.org/10.1021/ACS.EST.6B05113. Mar.
Rybakov, Y.P., Semenova, N.V., Safarov, J.S., 2019. Generalizing Darcy’s law for
work. filtration radial flows through porous media. IOP Conf. Ser. Mater. Sci. Eng 675 (1).
https://doi.org/10.1088/1757-899X/675/1/012064.
Acknowledgements Sabiri, N.E., Monnier, E., Raimbault, V., Massé, A., Séchet, V., Jaouen, P., 2017. Effect of
filtration rate on coal-sand dual-media filter performances for microalgae removal.
Environ. Technol. (United Kingdom) 38 (3), 345–352. https://doi.org/10.1080/
The authors express their thanks to the University of Cartagena for 09593330.2016.1193224.
providing the equipment and reagents required to complete this Sahu, R.L., Dash, R.R., Pradhan, P.K., Das, P., 2019. Effect of hydrogeological factors on
removal of turbidity during river bank filtration: laboratory and field studies.
research. Groundw. Sustain. Dev 9 (September 2018). https://doi.org/10.1016/j.
gsd.2019.100229.
References A. Senaratne, G.W.A.R. Fernando, K.M.D.P. Kodithuwakku1, T.H.N.G. Amaraweera2, G.
W.A.R. Fernando3, and K.W.S.N. Kumari2, “Performance of a garnet filter media
developed from red beach sand in Sri Lanka in removal of turbidity,” 2001,
A.C. Affam and E.H. Ezechi, “Application of graded limestone as roughing filter media for
Accessed: Aug. 26, 2021. [Online]. Available: https://www.researchgate.net/pu
the treatment of leachate,” in Handbook of Research on Resource Management For
blication/249325474.
Pollution and Waste Treatment, 2019, pp. 176–219.
Sutherland, K., 2014. Drinking & other pure water production: filtration and
Andreoli, F.C., Sabogal-Paz, L.P., 2020. Household slow sand filter to treat groundwater
sedimentation in clean water production. Filtr. Sep 51 (1), 24–27. https://doi.org/
with microbiological risks in rural communities. Water Res 186. https://doi.org/
10.1016/S0015-1882(14)70032-1.
10.1016/j.watres.2020.116352.
Taghavijeloudar, M., Park, J., Han, M., Taghavi, A., 2019. A new approach for modeling
ASTM International, ASTM D2487-17: standard practice for classification of soils for
flux variation in membrane filtration and experimental verification. Water Res 166.
engineering purposes (unified soil classification system), vol. 04. 2017, pp. 1–12.
https://doi.org/10.1016/j.watres.2019.115027.
CyT. Ministerior de vivienda, Reglamento Técnico del Sector de Agua Potable y
Torres-Lozada, P., Amezquita-Marroquín, C.P., Agudelo-Martínez, K.D., Ortiz-
Saneamiento Básico. 2013, p. 198.
Benítez, N., Martínez-Ducuara, D.S., 2018. Evaluation of turbidity and dissolved
Daneshi, N., Nafchi, R.F., Banejad, H., 2018. Prediction of Temporal Changes of Head
organic matter removal through double filtration technology with activated carbon.
Loss in Sand Filter. Predict. Temporal Chang. Head Loss Sand Filter 29 (3), 78–87.
Dyna (Medellin) 85 (205), 234–239 [Online]Available. http://www.scielo.org.co/sc
https://doi.org/10.22093/wwj.2017.43597.2068.
ielo.php?script=sci_arttext&pid=S0012-73532018000200234&lang=pt%0Ahttp://
Davies, P.D., Wheatley, A.D., 2012. Pilot plant study of alternative filter media for rapid
www.scielo.org.co/pdf/dyna/v85n205/0012-7353-dyna-85-205-00234.pdf.
gravity filtration. Water Sci. Technol 66 (12), 2779–2784. https://doi.org/10.2166/
Vecherkovskaya, A., Popereshnyak, S., 2017. Mathematical modeling of the process of
wst.2012.517.
fluid filtration through a multi-layer filtering element. Technol. Audit Prod. Reserv 4
Farhaoui, M., Derraz, M., 2016. Review on Optimization of Drinking Water Treatment
(36), 9–13. https://doi.org/10.15587/2312-8372.2017.109309, 3.
Process. J. Water Resour. Prot 08 (08), 777–786. https://doi.org/10.4236/
Verma, S., Daverey, A., Sharma, A., 2017. Slow sand filtration for water and wastewater
jwarp.2016.88063.
treatment–a review. Environ. Technol. Rev 6 (1), 47–58. https://doi.org/10.1080/
Farizoglu, B., Nuhoglu, A., Yildiz, E., Keskinler, B., 2003. The performance of pumice as a
21622515.2016.1278278.
filter bed material under rapid filtration conditions. Filtr. Sep 40 (3), 41–47. https://
Xu, T., Bezuijen, A., 2019. Bentonite slurry infiltration into sand: filter cake formation
doi.org/10.1016/S0015-1882(03)80137-4.
under various conditions. Geotechnique 69 (12), 1095–1106. https://doi.org/
Fujikura, Y., 2019. Estimation of permeability for sand and gravel based on pore-size
10.1680/jgeot.18.P.094.
distribution model. J. Mater. Civ. Eng 31 (12), 04019289. https://doi.org/10.1061/
Yao, W., Wang, Z., Song, P., 2018. The cake layer formation in the early stage of filtration
(asce)mt.1943-5533.0002945.
in MBR: mechanism and model. J. Memb. Sci 310, 393–401. https://doi.org/
González de Corbella, A., Ruiz Abrio, M.T., 2015. Estudio y análisis del comportamiento
10.1016/j.memsci.2018.04.042.
de la piedra caliza de la sierra de esparteros de morón de la frontera (sevilla) durante
Yiǧit Hunce, S., Soyer, E., Akgiray, Ö., 2016. Characterization of granular materials with
el proceso de cocción y apagado posterior. Tecnol. y Desarro. Rev. Ciencia, Tecnol. y
internal pores for hydraulic calculations involving fixed and fluidized beds. Ind. Eng.
medio Ambient 13, 1–29 [Online]Available. http://www.uax.es/publicacion/es
Chem. Res 55 (31), 8636–8651. https://doi.org/10.1021/acs.iecr.6b00953.
tudio-y-analisis-del-comportamiento-de-la-piedra-caliza-de-la-sierra.pdf.
Zhou, W., Zhang, L., Wu, P., Cai, Y., Zhao, X., Yao, C., 2019. Study on permeability
Hoslett, J., et al., 2018. Surface water filtration using granular media and membranes: a
stability of sand-based microporous ceramic filter membrane. Materials (Basel) 12
review. Sci. Total Environ 639, 1268–1282. https://doi.org/10.1016/j.
(13). https://doi.org/10.3390/ma12132161.
scitotenv.2018.05.247.
Zuluaga-Gomez, J., Bonaveri, P., Zuluaga, D., Álvarez-Peña, C., Ramírez-Ortiz, N., 2020.
I. Instituto Nacional de Vías, Norma I.N.V. E - 130- Permeabilidad de suelos granulares.
Techniques for water disinfection, decontamination and desalinization: a review.
2015, pp. 1–9.
Desalin. Water Treat 181, 47–63. https://doi.org/10.5004/dwt.2020.25073.
I. Instituto Nacional de vías, norma I.N.V. E-123-13-Análisis granulométrico de suelos
por tamizado. 2014, pp. 1–7.

105

You might also like