Sie sind auf Seite 1von 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230040977

Crushed Recycled Glass as a Filter Medium and Comparison with Silica Sand

Article  in  CLEAN - Soil Air Water · October 2010


DOI: 10.1002/clen.201000217

CITATIONS READS

14 1,064

4 authors, including:

Elif Soyer Ömer Akgiray


Marmara University Marmara University
31 PUBLICATIONS   221 CITATIONS    43 PUBLICATIONS   226 CITATIONS   

SEE PROFILE SEE PROFILE

Ahmet M. Saatci
Turkish Water Institute SUEN
51 PUBLICATIONS   567 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Wall-effects in liquid-solid fluidization. View project

All content following this page was uploaded by Ömer Akgiray on 05 November 2018.

The user has requested enhancement of the downloaded file.


Clean – Soil, Air, Water 2010, 38 (10), 927–935 927

Elif Soyer1
Ömer Akgiray2
Research Article
Nursen Öz Eldem1
Ahmet Mete Saatçı2 Crushed recycled glass as a filter medium and
1
comparison with silica sand
Istanbul Technical University, Faculty
of Civil Engineering, Department of
Environmental Engineering, Istanbul, The objective of this work was to evaluate crushed recycled glass as a medium for rapid
Turkey filtration. In the first part of this work, physical and hydraulic characteristics of the
2
Marmara University, Faculty of glass medium were studied. In the second part, pilot scale inline filtration experiments
Engineering, Department of
were carried out using raw waters from three different water sources. Two physically
Environmental Engineering, Istanbul,
Turkey identical filter columns were operated in parallel in all the experiments. One filter
contained a silica sand medium that is widely used in Turkey, whereas the other filter
contained crushed recycled glass. Experiments were repeated five times as follows: (i)
Without the use of a coagulant, (ii–iii) with 5 mg/L and 10 mg/L of alum, and (iv–v) with
5 mg/L and 10 mg/L of ferric chloride. Turbidity, particle counts, and head losses were
measured and compared as functions of time. The following were observed: (1) Provided
that a coagulant was used, the filter containing crushed glass produced effluent
turbidities and particle counts similar to those obtained with the sand filter. (2) The
crushed glass medium generated both a smaller clean-bed head loss and smaller
clogging head losses than those of the sand filter. It is concluded that crushed glass
shows significant promise as an alternative to silica sand in rapid filtration.
Keywords: Crushed Glass; Filter Media; Head Loss; Particle Removal; Rapid Filtration
Received: June 9, 2010; accepted: July 17, 2010
DOI: 10.1002/clen.201000217

1 Introduction example, a coarse anthracite layer is usually placed above a relatively


fine sized sand layer. The pore sizes of coarser materials are larger in
Silica sand has been the most widely used rapid filter medium
absolute magnitude. Furthermore, anthracite has porosity values
around the world. Other commonly used granular media for water
(0.56–0.60) higher than the typical values (0.40–0.47) reported for
filtration include anthracite coal, garnet sand, and ilmenite [1].
silica sand [1]. This means that not only the pore size of the anthra-
Granular activated carbon has been used in the so called ‘‘filter-
cite layer is larger than that of a finer sand layer, but also the volume
adsorbers’’ for reducing taste and odor as well as turbidity [1].
fraction of pores (i.e. volume fraction available for solids deposition)
Manganese greensand is sometimes used to remove soluble iron
in the anthracite layer is greater. Anthracite and other coarse media
and manganese [2]. Proper selection of filter media is of great
are used to prevent surface clogging by encouraging deep penetra-
importance for filter performance. Different types of media can
tion of solids into the filter bed. Deep penetration of solids delays the
be used alone or in combination with one another in dual or
development of head loss in the bed and thereby leads to relatively
multi-media filters. Media size distribution, density, shape, and
longer filter runs. Bottom fine layers help achieve high water qual-
porosity are among the properties which are important in establish-
ities by capturing those particles that pass through the upper coarse
ing filter performance characteristics.
layers.
Grain shape is an important factor because it affects fixed-bed
The traditional filter materials mentioned above may not always
porosity, head loss during filtration, and backwash flow require-
be readily available. For example, anthracite coal with suitable
ments. Sphericity is the most commonly used measure of grain
properties is not available in some countries and has to be imported.
shape for granular media [1]. It is defined as the ratio of the surface
Materials that can be used instead of anthracite coal in dual media
area of an equal volume sphere to the surface area of the grain [3].
filters have been considered by a number of investigators [5–7] and a
Materials which are more angular such as anthracite and other
variety of indigenous materials such as different kinds of coal,
crushed materials have lower sphericity values compared to rounder
coconut shells, berry seeds, and various sorts of fruit stones have
materials such as silica sand [1]. Materials with lower sphericity
been applied in countries such as Brazil, Chile, Colombia, India,
values have higher porosity values in the fixed-bed arrangement [4].
Philippines, and Korea [8]. In a study carried out in Turkey, the use of
In dual and multi-media filter beds, coarser materials are placed
crushed shells of apricot stones as the upper layer in dual media
above finer and denser materials. In dual media filtration, for
filters was investigated [9]. Similarly, certain regions in the world do
not have access to silica sand either because of local geology or
economic considerations. Basalt and tuff [10], perlite [11], quartz
Correspondence: Ömer Akgiray, Marmara University, Faculty of [12], and wood charcoal [13] are among the materials considered
Engineering, Department of Environmental Engineering, Istanbul,
Turkey. as alternatives to silica sand in rapid filtration. Pulverized glass
E-mail: omer.akgiray@marmara.edu.tr produced by crushing and sieving recycled glass was employed

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clean-journal.com


928 E. Soyer et al. Clean – Soil, Air, Water 2010, 38 (10), 927–935

successfully in wastewater filtration [14]. It has been demonstrated same size distribution. Because the fraction of volume available for
that crushed glass can be an effective medium in recirculating solids deposition in a crushed glass bed is larger than that in a sand
biofilters [15]. There is also some research regarding the use of this medium, it may also be expected that the rate of clogging head loss
material in slow sand filtration [16]. The use of pulverized glass in development will be smaller in a crushed glass bed. These expec-
rapid filters is a relatively new subject [17–19]. tations are put to test in the present work.
Evans et al. [17] compared the performance of recycled glass to that Hudson [24] stated that ‘‘rounded particles produce clearer water
of a typical sand filter by pilot scale experiments. The glass medium than angular ones. This is because angular materials give greater
had an effective size of 0.97 mm whereas the effective size of the sand porosity’’. This opinion was re-iterated by Baylis et al. in the third
medium was 0.98 mm. The depths of both filters were 90 cm. edition of Water Quality and Treatment [25] and again by Hudson
Filtration rates between 7.5 and 12.5 m/h were tested. They con- [26]. On the other hand, Trussell et al. [27] noted that ‘‘use of
cluded that the glass medium reached particle breakthrough in a angular media results in an improved performance from every
time 10–15% longer than the sand medium. The rate of head loss aspect’’. Similarly, Kawamura [28] stated that ‘‘angular grains
development for the glass medium was equal to or slightly lower usually perform better than rounded and worn grains do when
than that of the sand filter, whereas both filters generally produced polymer is not used as a filter aid’’. The effluent quality obtained
effluents of equal quality. Clarified water from only one water source with an angular medium (crushed glass) and that obtained with a
was used in this particular study and attention was restricted to relatively round medium (silica sand) are compared in the present
single-medium (either sand or glass) filtration. work. Since the two media have approximately the same size
In a different study, Rutledge and Gagnon [18] compared the distribution and the two filters are operated in parallel under
performance of a pressure dual-media filter which was composed of exactly identical conditions and with the same influent water,
pulverized recycled glass and anthracite layers with that of a filter the new data presented here allows a re-examination of
containing silica sand and anthracite. Both filters contained a Hudson’s statement.
60 cm deep layer of anthracite over 40 cm of either glass or silica Crushed recycled glass is not used as a filter medium in Turkey yet
sand. Filtration rate was 5 m/h. They found that particle removal in and a local market price for crushed glass with gradation suitable for
the glass filter was slightly poorer than that in the sand filter. water filtration cannot be reported at the present time. In a study
Additional testing was recommended by the authors. It should carried out in Canada on the use of crushed glass in recirculating
be noted that the effective sizes of the sand and glass media used biofilters, Hu and Gagnon [15] noted that crushed glass was less
were significantly different in this work. Sand had an effective size expensive than silica sand. The d10 (effective size) and UC (uniformity
of 0.33 mm whereas this value was 0.59 mm for the glass medium. coefficient) values for both sand and glass used in their study were
From filtration theory, it is known that finer media sizes lead to 1.5 mm and 2.0, respectively. Based on a study carried out in the USA,
better particle removals in rapid filters [20]. In order to isolate the Elliot [14] reported that crushed glass suitable for intermittent
effects of using a different type of filter medium (i.e. glass instead of wastewater filters was acquired at about 38% of the price (per unit
sand), it is preferable if the two media being compared are of about weight) of silica sand meeting the required gradation. The reported
the same size. UC values for glass and sand in Elliot’s work were 5.6 and 1.7,
Fitzpatrick [19] studied the attrition of crushed glass during respectively. A UC of 1.5 or less is normally specified in potable water
backwash and its friability. The backwash tests were performed filtration [1]. Additional sieving to achieve smaller UC values
with and without simultaneous air backwash. The friability test increases the cost of any filter medium. In a study in Australia on
employed an apparatus ‘‘to exert a force equivalent to the weight of potable water filtration with crushed glass, Evans et al. [17] stated
a person treading on the media.’’ It was reported that, when that ‘‘the glass medium is of comparable cost to sand medium ex
compared to silica sand, crushed glass showed similar levels of works, while it is expected that increased use of glass in filters has
fine particles produced by attrition during backwash and similar the potential to lower the unit cost’’. The UC values of the sand and
friability. the crushed glass employed in their work were 1.27 and 1.31,
Hu and Gagnon [15] studied the use of crushed glass as a medium respectively. In general, crushed glass appropriate for potable water
in recirculating biofilters and noted that crushed glass has the treatment with rapid filtration can be expected to be more expensive
following advantages over traditional silica sand: (i) Crushed than glass for other applications (e.g. intermittent wastewater treat-
glass is less expensive than silica sand, (ii) crushed glass is more ment, swimming pool filtration, slow sand filtration, etc.) because of
environmentally friendly as it is a recycled product, and (iii) crushed required additional processing to satisfy filter media standards for
glass can be pulverized into different sizes for specific design drinking water treatment. When comparing the costs of silica sand
requirements. and crushed glass, the differences in the porosities and the densities
As noted above, the sphericity of the grains affects the porosity of a of the two materials should be taken into account. For the media
bed. Due to this particle shape effect, the porosity of silica sand is used in this work (see below), the weight of sand required to fill a
smaller than the porosity of an anthracite medium regardless of certain volume of filter bed is 27.4% higher than the weight of
particle size. This also means that it should be possible to reduce crushed glass that would fill the same volume. A cost-benefit analysis
head losses in single-medium filters by employing media of the same should also consider the benefits associated with reductions in waste
size but with lower sphericity values. It was shown in previous work glass and sand quarrying as a result of employing recycled glass
that crushed glass is considerably more angular (i.e. has lower instead of silica sand.
sphericity values) than silica sand [21–23]. This finding is consistent This paper considers the use of crushed recycled glass instead of
with the visual observations made during the course of the present silica sand in rapid filters. The following issues were experimentally
work and with the SEM micrographs presented by Rutledge and investigated: (1) Physical characteristics and hydraulic properties of
Gagnon [18]. It is therefore expected that a crushed glass medium crushed glass. (2) Filtrate quality as compared with a filter employing
should lead to smaller clean-bed head losses than a sand bed with the silica sand. (3) Head loss development during filtration.

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clean-journal.com


Clean – Soil, Air, Water 2010, 38 (10), 927–935 Crushed recycled glass as a filter medium 929

2 Experimental as column diameter and how the column is filled. In this work, two
parallel filter columns each with an internal diameter of 10 cm were
used. After backwashing with fluidization, the backwash valves were
2.1 Physical and Hydraulic Properties
closed slowly letting the beds settle. The columns were then tapped
The sand medium utilized in the experiments is the sand currently slightly when necessary to achieve the predetermined bed height of
used in all the municipal water treatment plants in Istanbul. 104 cm exactly. This height remained constant (there was no bed
Crushed glass was obtained from Anadolu Şişecam, Inc. which is compaction) during the filtration experiments. Under these con-
one of the glass factories in Istanbul. The following steps were ditions, the clean-bed porosity of the sand filter was measured to
performed before using this material in the filtration experiments: be 0.38, whereas that of the crushed glass filter was 0.49.
(i) Removal of impurities from all glass pieces (which were about 3– Sphericities of the sand and crushed glass fractions were deter-
4 cm in size), (ii) crushing (Retsch Jaw Crusher was used for the size mined by measuring fixed-bed head losses (h versus V values) for
reduction of glass pieces), (iii) washing the crushed material to different fractions and using the Ergun equation [29] which can be
eliminate fines which can lead to difficulties in the subsequent rearranged as a quadratic equation in sphericity c:
sieving process, (iv) sieving the bulk material to the desired range.
      "   #
The glass medium was prepared such that its effective size (0.77 mm) h 2 k2 ð1"0 Þ 6 2 k1 m ð1"0 Þ2 6 2
c  V c V ¼0 (1)
and uniformity coefficient (1.41) were sufficiently close to those (0.79 L0 g "30 deq rg "30 deq
and 1.33 mm, respectively) of the sand medium. As noted before, this
is needed for the comparison to be meaningful: If the sand and glass Once the quantities within the square brackets are measured, the
media being compared have significantly different sizes, then any sphericity is determined by using the well-known quadratic formula.
observed effluent quality differences due to size differences may not The sphericity of each fraction was calculated as the mean value for
be easily differentiated from effluent quality differences resulting several measurements. Sphericities of sand fractions were found to
from using different types of media. Resulting sieve analysis curves be in the range 0.70–0.76 which is consistent with the typical values
for the glass and the sand can be seen in Fig. 1. reported in the literature [1, 30]. It was visually observed that the
A water-displacement technique was used to measure the crushed glass particles were highly non-spherical and the low spher-
densities of the materials. The density of glass was found to be icity values measured (0.41–0.43) are consistent with this obser-
2.50 g/cm3 which is slightly less than the density (2.62 g/cm3) of vation. Figs. 2a and 2b display the measured head loss values for
the silica sand used in this work. a sand fraction and a glass fraction, respectively. These particular
Hydraulic properties such as porosity, sphericity, and bed expan- fractions were collected between the two adjacent sieve sizes 0.84
sions during backwashing were also investigated. Filter bed poros- and 1.00 mm. It is seen that the clean-bed head loss in this sand
ities were calculated from bed weight, bed height, and density fraction was significantly larger than that in the crushed glass
values. It should be remembered that the porosity of a granular fraction. The difference increases as rate increases. Similar results
medium does not have a unique value and it depends on factors such were obtained for all other fractions.

100 100
Percent media passing
Percent media passing

sand crushed glass


80 80

60 60
(%)
(%)

40 40

20 20

0 0
0.5 0.8 1.1 1.4 1.7 0.5 0.8 1.1 1.4 1.7
Sieve size (mm) Sieve size (mm)

Figure 1. Sieve analysis of the media used in filtration experiments.

a) 0.4 b) 0.2
Crushed glass
Sand
0.16 Ergun
Head loss (m)

Ergun
Head loss (m)

0.3
0.12
0.2
0.08
0.1 0.04

0 0
0 0.005 0.01 0 0.005 0.01 0.015 0.02

Velocity (m/s) Velocity (m/s)

Figure 2. Fixed-bed head loss data for 0.84  1.00 mm silica sand and crushed glass fractions. Solid curve was drawn with c ¼ 0.74 for sand and c ¼ 0.41
for the glass medium.

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clean-journal.com


930 E. Soyer et al. Clean – Soil, Air, Water 2010, 38 (10), 927–935

The sphericity of crushed glass was found to depend on the In the absence of data for the specific filter medium that will be
number of times it was crushed. The sphericity increases when a used in the full-scale plant, it is recommended that this equation is
batch of crushed glass is crushed for a second time. Glass crushed used to predict filter expansion. The expansion of a non-uniform bed
once had sphericity values in the range 0.41–0.43, whereas glass (such as the glass and sand beds used in this work) can be predicted
crushed (and sieved) for a second time had sphericity values in the as follows: The bed is considered to consist of several layers of
range 0.51–0.56. This finding is important because sphericity affects approximately uniform size according to the sieve analysis data,
porosity which in turn influences a number of dynamic properties of and the expansion of each layer is separately calculated using Eq. (2).
the medium: Clean bed head loss, capacity of the bed for retaining The total expansion is then calculated by adding the expansions of
suspended solids, the morphology of the retained solids (e.g. surface all the layers [1, 30].
filtration versus depth filtration), and therefore the resulting clog-
ging head loss are among the properties affected by porosity. In this
2.2 Filtration Experiments
study, the glass medium which was produced by crushing the
material once (sphericity: 0.41–0.43) was utilized. Two physically identical parallel filters were operated to compare
Fluidization experiments were carried out in the two 10 cm I.D. the performance of crushed glass to that of silica sand as a rapid
filter columns described below. Expanded bed height versus back- filter medium. The filters were made of plexiglass and had an inner
wash velocity was recorded for each medium. The water temperature diameter of 100 mm and a height of 2.5 m. The top of each filter
remained between 18 and 198C during these measurements. These column was open to the atmosphere and the water level above the
experiments were repeated twice for the sand bed and three times beds increased throughout the filter run due to increasing clogging
for the glass bed. All of the recorded measurements are displayed in head losses. Each filter contained 104 cm of a single medium. The
Fig. 3. It is seen that the two beds expanded to the same extent in the porosity of the sand filter was 0.38, whereas that of the glass filter
range of expansions (30–50%) used in practice, although the glass was 0.49. Filter runs were terminated when the available head was
bed expanded slightly more than sand at higher backwash rates. exhausted. Three different surface water sources were used in the
It should be remembered that expansion data such as those in study. These waters were obtained from the intakes at the Omerli,
Fig. 3 apply only to the specific beds studied and do not have Ikitelli, and Haci Osman water treatment plants of Istanbul. Raw
universal applicability. A crushed glass bed with the same initial waters from these sources were brought to the laboratory by means
height but with different physical properties (i.e., size distribution, of trucks and transferred to two connected 5 m3-tanks. The contents
density, sphericity, fixed-bed porosity) will behave differently. A of these tanks were continuously mixed to prevent settling. Water
large number measurements were carried out by Soyer and was pumped from these tanks to a single 1.5 m3-tank located near
Akgiray [22] using several sieved fractions of crushed glass as well the filters. Raw water was pumped from this tank into a constant
as silica sand, garnet, perlite, and several different sizes of plastic level overhead tank (located 3.5 m above the floor level and had a
and glass balls to develop a general equation for filter media: volume of 20 L) using a centrifugal pump. The water then flowed by
  gravity through the filter media in the two columns. Chemicals
log’ ¼ log 3:137 Re1 þ 0:673 Re1 1:766 (alum or ferric chloride) were dosed in-line with peristaltic pumps.
(2)
 ð0:930 þ 0:274 logRe1 ÞðlogcÞ1:262 Filters were operated at a constant rate of 11.5 m/h (which is close to
the filtration rates used in many municipal water treatment plants
The dimensionless groups Re1 and w in this equation are defined as in Turkey). The velocity was controlled using rotameters and globe
follows: valves located at the inlet of the filters. On-line instrumentation on
the setup provided a continuous monitoring of turbidity (Hach
cdeq rV
Re1 ¼ (3) 1720D Low Range Turbidimeter, Hach Company, Loveland-
6mð1"Þ
Colorado) and particle counts (ARTI WPC-22 Particle Counter) of
  both the influent and the effluent. Changes in these parameters
"3 c3 d3eq r rp r g were recorded using a data logger and transferred to a PC with 1-min
’¼ (4)
ð1"Þ2 216m2 intervals. The head loss development through each filter bed was
monitored by means of piezometer tubes installed at various heights
of the filter columns.
Several filtration runs were completed with each of the three raw
200
190
waters. The two filters were operated simultaneously and then back-
180
washed and cleaned under identical conditions after each run. Data
170 (input and effluent turbidities, particle counts, head losses) were
Bed Height (cm)

160 recorded carefully in five separate runs: (i) Without the use of a
150 coagulant, (ii–iii) with 5 mg/L and 10 mg/L of alum (expressed as
140 Al2(SO4)3  18 H2O), and (iv–v) with 5 mg/L and 10 mg/L of ferric
130 chloride (expressed as FeCl3  6 H2O). Raw water turbidities and
120 Sand
particle counts recorded during these experiments are given in
110 Crushed Glass
Table 1. The two filters were operated in parallel, received exactly
100 the same water, and the effluent parameters and head losses were
40 60 80 100 120 140 160
recorded simultaneously. The main goal here was to ascertain the
Velocity (m/h)
differences in the performances of the two filters when receiving the
Figure 3. Expanded bed height versus backwash velocity at 18–198C. same input water and operated at the same rate. It is believed that
Initial bed height was 104 cm. better effluent qualities could be obtained with each filter by

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clean-journal.com


Clean – Soil, Air, Water 2010, 38 (10), 927–935 Crushed recycled glass as a filter medium 931

Table 1. Turbidities and particle counts of the raw waters used in the a) 5
filtration experiments. Raw water
4

Turbidity (NTU)
Water Source Turbidity Particle Counts
3
(NTU) (particles/mL) Crushed glass
2
Ikitelli 3.2–4.5 5000–6000
H. Osman 7.5–8.0 6000–7000 1
Sand
Omerli 10–15 18 000–20 000
0
00:00 01:12 02:24 03:36 04:48 06:00 07:12 08:24

Filter run time (h)


optimizing each separately (e.g. by using different coagulant dos-
b)

Particle Counts (particles/mL)


5000
ages, using a filter aid polymer, perhaps by trying other types of
coagulants, by applying pre-ozonation to enhance particle removal, 4000
Crushed glass
etc.). The purpose in these experiments was, however, to compare the 3000
two filters under identical conditions and not to optimize the per-
2000
formance of each individually.
Sand
1000

3 Results 0
0 100 200 300 400 500
Variations in filter effluent turbidities and particle counts as well as
Filter run time (min)
head loss values were recorded during the experiments. Raw waters
used in the experiments had different qualities in terms of turbidity Figure 4. Filtered water qualities in terms of turbidity and particle counts
and particle counts (Table 1). The use of raw waters from different (Ikitelli raw water without a coagulant).
sources was important so as to see how the two filters performed
with waters with different characteristics. The Ikitelli and Omerli
waters are two of the most important water sources in Istanbul. In batch of water brought in from Ikitelli was mixed with the raw water
previous work, water from Haci Osman was found to be more already in the storage tank. This new water had lower turbidity than
difficult to treat with direct filtration [31]. A decision was therefore the previous batch received a few days ago, and the resulting lower
made to include it in this study despite the fact that it is an insig- turbidity of the mixture lead to a drop in effluent turbidity as well.
nificant source for the city of Istanbul as a whole. Turbidity removal percentages after the initial ripening period have
been calculated and shown in Table 2. The first row displays the
results obtained when no coagulant was used. As noted, both filters
3.1 Experiments without a coagulant
yielded approximately the same effluent turbidity values. The sand
It is generally accepted that pretreatment of surface waters by filter produced slightly better particle removals with all the three
chemical coagulation is essential to achieve efficient removal of raw waters (e.g. Ikitelli water, cf. Fig. 4b) treated in this study. As
particulates in rapid filtration [1]. On the other hand, groundwater expected, the clean bed head loss (about 33 cm) in the crushed glass
treated for iron and manganese removal by oxidation, precipitation, bed was less than that (39 cm) in the sand bed. Since no coagulant
and filtration, generally do not need other chemical pretreatment was used, the head loss increases due to clogging were small in both
[1]. Another application of rapid filtration without coagulation is filters and therefore the differences between the head loss increases
the pretreatment of raw waters ahead of slow sand filters [1]. This in the two filters were negligible.
latter application is sometimes referred to as ‘‘roughing filtration’’.
Although a coagulant (either alum or ferric chloride) was employed
3.2 Experiments with alum and iron
in the bulk of this study, a number of experiments were carried out
without adding a coagulant and the results of these experiments are Sample results are displayed in Figs. 5 to 7. Figure 5a shows the
reported presently. turbidity and particle count values obtained with the two filters
The two filters produced nearly equal effluent turbidities with all when filtering water from Haci Osman using 10 mg/L alum. Results
three raw waters. For example, the influent and effluent turbidities with 10 mg/L iron and the same raw water can be seen in Fig. 5b.
recorded during an experiment with the Ikitelli water can be seen in Figure 6 shows the total head loss measurements for the two filters
Fig. 4a. About four hours after the start of this particular run, a new with the Haci Osman water. As illustrated in this figure, significantly

Table 2. Percent reductions in turbidity.

Water source Reduction in turbidity (NTU), sand filter Reduction in turbidity (NTU), glass filter

Omerli Ikitelli H. Osman Omerli Ikitelli H. Osman

No coagulant 82–85 58–61 47–50 82–85 58–60 48–51


Alum 5 mg/L 95–97 88–89 60–77 95.0–97.5 88–90 60–82
Alum 10 mg/L 98.8–99.1 94–95 96.0–98.5 99.0–99.2 94–96 97.0–98.7
Iron 5 mg/L 98.8–98.9 92 60–75 98.8–98.9 92.5 63–82
Iron 10 mg/L 99.5–99.7 94–95 78–91 99.5–99.7 94–97 82–93

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clean-journal.com


932 E. Soyer et al. Clean – Soil, Air, Water 2010, 38 (10), 927–935

1.0 1000
a)

Particle counts (particles/ml)


0.8 800 sand filter
crushed
Turbidity (NTU)

glass filter
0.6 600
sand filter
0.4 400

0.2 crushed
200
glass filter
0.0 0
00:00 02:24 04:48 07:12 0 100 200 300 400 500

Filter run time (h) Filter run time (min)

b) 3.0 2000

Particle counts (particles/ml)


2.5
1600
Turbidity ( NTU)

2.0 sand filter


sand filter crushed glass 1200
1.5 filter crushed glass
filter
800
1.0

0.5 400

0.0 0
00:00 01:55 03:50 05:45 0 100 200 300 400

Filter run time (h) Filter run time (min)

Figure 5. Effluent turbidity and particle counts (Haci Osman water, a) coagulated with 10 mg/L alum, b) 10 mg/L ferric chloride).

lower head losses were obtained with the use of crushed glass 5 mg/L and then to 10 mg/L with both alum and iron, and with all of
instead of silica sand. Similar results were obtained with the the three waters. The most important result here is that very similar
Ikitelli and Omerli waters. Figure 7 shows the effluent quality and turbidity and particle removals were achieved with the two filter
filter head losses for Ikitelli water coagulated with 10 mg/L alum. It media, regardless of the coagulant used. Results of the filtration
may be noted that clogging head loss in the sand filter was about 50% experiments can be outlined as follows:
higher than that in the crushed glass filter. Again, lower head losses
resulted with the use of crushed glass instead of silica sand.  As expected, the performances of both filters improved signifi-
Turbidity and particle removals observed during these and all the cantly with the use of a coagulant.
other experiments with alum and iron and with all three raw waters  The differences between the effluent qualities of the two filters
are summarized in Tables 2 and 3. The data in these tables is became negligibly small whenever a coagulant (alum or iron) was
reproduced in Fig. 8 for ease of comparison of the performances used. In other words, with the glass medium, it was possible to
of the two filters. It is seen that turbidity and particle removals obtain effluent turbidities and particles counts similar to those of
always increased as coagulant dosage was increased from zero to the sand filter. The crushed glass medium gave slightly better

a) 160 b) 160
sand filter
140 140
120 120
Head loss (cm)

Head loss (cm)

100 100
80 80
60 60
crushed
40 glass filter 40
20 20
0 0
00:00 02:24 04:48 07:12 09:36 00:00 01:12 02:24 03:36 04:48 06:00

Filter run time (h) Filter run time (h)

Figure 6. Total head loss variations (Haci Osman raw water, a) 10 mg/L alum, b) 10 mg/L ferric chloride).

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clean-journal.com


Clean – Soil, Air, Water 2010, 38 (10), 927–935 Crushed recycled glass as a filter medium 933

a) 1.4 b)
200

Particle counts (particles/ml)


1.2
160 crushed
1.0 glass filter
Turbidity (NTU)

0.8 120
crushed
0.6 glass
sand filter 80
0.4
40
0.2 sand filter
0.0 0
00:00 02:24 04:48 07:12 0 100 200 300 400 500
Filter run time (h) Filter run time (min)
c) 150 d) 150

05:07 120 07:37


04:12 120
06:16
Head loss (cm)

Head loss (cm)


90 90
03:07
05:07 04:12
02:12
60 60
03:07 02:12
01:12
H0 30 30
01:12 H0
0 0
100 80 60 40 20 0 100 80 60 40 20 0
Filter media depth (cm) Filter media depth (cm)

Figure 7. Ikitelli water coagulated with 10 mg/L alum (a) effluent turbidity, b) particle counts, c) head loss increase through the sand filter, d) head loss
increase through the glass filter. Numbers on the curves show the time in hours).

effluent quality in some of the experiments. The differences, that in the crushed glass filter. There was again only one exception
however, were always small. and that occurred when 5 mg/L ferric chloride was used. In that
 For the particular filtration rate used (11.5 m/h), the clean bed case, the clogging head loss in the sand filter was much higher, by
head loss was reduced by about 18% when crushed glass was used about 400%.
instead of sand.  For the experiments carried out using a coagulant to treat the raw
 Whether a coagulant was used or not, the head loss increase due to water from the Omerli Treatment Plant, the clogging head loss in
clogging was smaller with the crushed glass medium. The differ- the sand filter was about 80–100% higher than that in the crushed
ence was small when a coagulant was not used, because the head glass filter.
loss increase itself was small for each filter in such a case.
 For the experiments carried out using a coagulant to treat the raw
water from the Ikitelli Treatment Plant, the clogging head loss in
the sand filter was about 50% higher than that in the crushed glass
filter. There was only one exception which occurred when 5 mg/L
4 Summary and Conclusions
alum was added, and in that case the clogging head loss in the Crushed glass was evaluated as a rapid filter medium. The glass
sand filter was again higher, but only by about 10%. medium was obtained by pulverizing and sieving recycled glass.
 Similarly, for the experiments carried out using a coagulant to In the first part of this work, physical and hydraulic characteristics
treat the raw water from the Haci Osman Treatment Plant, the of crushed glass were studied. It was found that the sphericity of the
clogging head loss in the sand filter was about 50% higher than crushed glass particles increases when a batch of crushed glass is

Table 3. Percent reductions in particle counts.

Water source Reduction in particle counts, sand filter Reduction in particle counts, glass filter

Omerli Ikitelli H. Osman Omerli Ikitelli H. Osman

No coagulant 71–72 65–77 25–40 66–67 60–72 15–34


Alum 5 mg/L 89–91 97–97.5 72–78 89–91 97.0–97.5 73–78
Alum 10 mg/L 93 98.4–98.8 87–91 93–94 98.4–98.8 89–93
Iron 5 mg/L 92–93 98.2–98.8 73–78 91.5–92.5 98.0–98.5 73–80
Iron 10 mg/L 97.0–97.5 97.6–98.6 88–93 96.5–97.5 96.8–98.8 88–94

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clean-journal.com


934 E. Soyer et al. Clean – Soil, Air, Water 2010, 38 (10), 927–935

Omerli Ikitelli Haci Osman


a) 100
90
Percent Reducons in

80 Crushed Glass filter


70
Turbidity

60
50
40

Glass
Glass

Glass

Glass
Sand

Glass
Sand

Sand

Sand
Glass

Sand
Sand
30
Sand

20

Glass

Glass

Glass
Sand

Sand

Sand
Glass

Glass

Glass

Glass
Sand

Sand

Sand

Sand

Glass
Sand
10
0
No coagulant Alum 5 mg/L Alum 10 mg/L Iron chloride 5 mg/L Iron chloride 10 mg/L

b) 100
90
Percent Reducons in

80
Parcle Counts

70
60
50
40
Crushed Glass

30
Sand filter

Glass

Glass

Glass
Sand

Sand

Sand

Glass
Glass

Glass
Sand
Sand

Sand
20
Glass
Glass

Glass

Glass

Glass

Glass
Sand
Glass
Sand

Sand

Sand

Sand

Glass
Sand
Sand

Sand
10
0
No coagulant Alum 5 mg/L Alum 10 mg/L Iron chloride 5 mg/L Iron chloride 10 mg/L

Figure 8. Percent reductions in turbidity (a) and particle counts (b).

crushed for a second time. Glass crushed once had sphericity values 5 Notation
in the range 0.41–0.43, whereas fractions of glass crushed for a
second time had sphericity values in the range 0.51–0.56. This find- deq Equivalent diameter defined as the diameter of the sphere with
ing is important because sphericity affects porosity which in turn the same volume as the particle
influences a number of dynamic properties of the medium: Clean Re1 Modified Reynolds number
bed head loss, capacity of the bed for retaining suspended solids, the L Depth of the fluidized bed
morphology of the retained solids (e.g. surface filtration versus depth L0 Fixed-bed depth
filtration), and the resulting clogging head loss are among the V Superficial velocity based on empty cross-section of filter bed
properties effected by porosity. In the second part, inline filtration c Sphericity defined as the surface area of the equivalent volume
experiments were carried out with three different water sources and sphere divided by the actual surface area of the particle
with and without alum and iron coagulants. m Fluid viscosity
In all cases and with all three raw waters, the clean bed head loss e Porosity of the fluidized bed
and the clogging head loss in the crushed glass medium were both e0 Fixed-bed porosity
smaller than those in the sand filter. In some of the experiments, the r Fluid density
raw water turbidity was considerably above the 5 NTU maximum rp Particle density
recommended limit for direct filtration. In such cases, the filters w Dimensionless group defined as w ¼ f Re21, where f is a friction
clogged quickly as expected and filtration runs had to be ended factor
rather early because the maximum allowed head loss was reached. In
such cases, it was observed that the filter containing crushed glass
clogged more slowly and allowed longer filter runs compared to the Acknowledgments
sand filter.
For all of the three waters and for both of the coagulants tested This research was supported by The Scientific and Technological
in this work, the crushed glass and sand filters gave very similar Research Council of Turkey (TÜBİ TAK) project number 104M435.
particle and turbidity removal efficiencies, while significantly
reduced head losses were obtained with the use of crushed glass. The authors have declared no conflict of interest.
For the filter media used in this work, the expansion versus back-
wash rate curves for the crushed glass and sand filters were very References
close. This means that the available backwash facilities of an existing
sand filter can be used with little or no modification if the sand [1] AWWA, Water Quality and Treatment, A Handbook of Community
Water Supplies, 5th Ed., McGraw-Hill, New York 1999, Chapter 8.
medium is replaced by crushed glass with a similar size distribution.
[2] M. Cakmakci, I. Koyuncu, C. Kinaci, The Effects of Iron
In conclusion, it is believed that crushed recycled glass can be a good
Concentrations, Filter Hydraulic Loading Rates and Porosities on
alternative to silica sand in rapid filtration in places where it can be Iron Removal by Rapid Sand Filtration, Environ. Eng. Sci. 2008, 25 (5),
acquired at a competitive cost. 669–675.

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clean-journal.com


Clean – Soil, Air, Water 2010, 38 (10), 927–935 Crushed recycled glass as a filter medium 935

[3] H. Wadell, The Coefficient of Resistance as a Function of Reynolds Water Treatment Plants, Water Sci. Technol. Water Supply 2002, 2
Number for Solids of Various Shapes, J. Franklin Inst. 1934, 217 (4), (5–6), 9–16.
459–490. [18] S. O. Rutledge, G. A. Gagnon, Comparing Crushed Recycled Glass to
[4] A. S. Foust, L. A. Wenzel, C. W. Clump, L. Maus, et al., Principles of Unit Silica Sand for Dual Media, J. Environ. Eng. Sci. 2002, 1 (5), 349–
Operations, Corrected Second Printing, Wiley, Japan 1960, p. 537. 358.
[5] A. G. Bhole, J. T. Nashikkar, Berry Seed Shell as Filter Media, J. Inst. [19] C. S. B. Fitzpatrick, Crushed Glass as a Filter Medium, Proceedings of
Eng. 1974, 54 (Feb), 45–47. IWA International Conference on Particle Separation, Seoul, S. Korea
[6] R. Paramasivam, S. K. Gadkari, N. S. Joshi, A. W. Deshpande, 2005, 399–404
Bituminous Coal – a Substitute for Anthracite Filter Media in [20] J. C. Crittenden, R. R. Trussell, D. W. Hand, K. J. Howe, G.
Two Layer Filtration of Water, Indian J. Environ. Health 1973, 16 (3), Tchobanoglous, in Water Treatment: Principles and Design (Ed.:
178–188. MWH, Inc.), 2nd Ed., John Wiley & Sons, Hoboken, New Jersey 2005.
[7] S. V. Ranade, J. M. Gagdil, Design of Dual Media Filters to Suit [21] Ö. Akgiray, E. Soyer, E. Yüksel, Prediction of Filter Expansion during
Existing Water Treatment Plants in India, J. Indian Water Works Backwashing, Water Sci. Technol. Water Supply 2004, 4 (5–6), 131–138.
Assn. 1981, 13 (1), 81–85. [22] E. Soyer, Ö. Akgiray, A New Simple Equation for the Prediction of
[8] C. R. Schulz, D. A. Okun, Surface Water Treatment for Communities in Filter Expansion during Backwashing, J. Water Supply Res. Technol.
Developing Countries, John Wiley & Sons, New York 1984. AQUA 2009, 58 (5), 336–345.
[9] S. Aksogan, A. Basturk, E. Yuksel, Ö. Akgiray, The Use of Crushed [23] E. Soyer, Ö. Akgiray, Expansion of Non-Spherical Media during
Shells of Apricot Stones as the Upper Layer in Dual Media Filters, Fluidization, Proceedings of IWA International Conference on Particle
Water Sci. Technol. 2003, 48 (11–12), 497–503. Separation, Seoul, S. Korea 2005, 579–586
[10] A. Adin, S. Hatukai, Optimization of Multilayer Filter Beds, Filtr. Sep. [24] H. E. Hudson, Operating Characteristics of Rapid Sand Filters, J. Am.
1991, 28 (1), 33–36. Water Works Assn. 1959, 51 (1), 115–122.
[11] S. S. Uluatam, Assessing Perlite as a Sand Substitute in Filtration, [25] AWWA, Water Quality and Treatment, A Handbook of Community
J. Am. Water Works Assn. 1991, 83 (6), 70–71. Water Supplies, 3rd Ed., McGraw-Hill, New York 1971, Chapter 7a,
[12] S. Suthaker, D. W. Smith, S. J. Stanley, Evaluation of Filter Media for p. 253.
Upgrading Existing Filter Performance, Environ. Technol. 1995, 16 (7), [26] H. E. Hudson, Water Clarification Processes, Van Nostrand Reinhold
625–643. Company, New York 1981, pp. 175–176.
[13] R. O. Agbanobi, Using Granulated Wood Charcoal as a Filter [27] R. R. Trussell, A. R. Trussell, J. S. Lang, C. H. Tate, Recent
Medium, J. Environ. Qual. 1999, 28 (3), 1038–1040. Developments in Filtration System Design, J. Am. Water Works
[14] R. W. Elliot, Evaluation of the Use of Crushed Recycled Glass as a Assn. 1980, 72 (12), 705–710.
Filter Medium: Part 1, Water Eng. Manage. 2001, 148 (7), 13–18. [28] S. Kawamura, Design and Operation of High-Rate Filters, J. Am. Water
[15] Z. Hu, G. A. Gagnon, Impact of Filter Media on the Performance of Works Assn. 1999, 91 (12), 77–90.
Full-Scale Recirculating Biofilters for Treating Multi-Residential [29] S. Ergun, Fluid Flow through Packed Columns, Chem. Eng. Prog. 1952,
Wastewater, Water Res. 2006, 40 (7), 1474–1480. 48 (2), 89–94.
[16] J. B. Piccirillo, R. D. Letterman, Examination of Pulverized Waste [30] G. M. Fair, J. C. Geyer, D. A. Okun, Elements of Water Supply and
Recycled Glass as Filter Media in Slow Sand Filtration, New York Wastewater Disposal, 2nd Ed., John Wiley and Sons, Inc., New York
State Energy Research & Development Authority, Albany, New 1971.
York 1997. [31] E. Yüksel, Ö. Akgiray, E. Soyer, Direct Filtration with Preozonation
[17] G. Evans, P. Dennis, M. Cousins, R. Campbell, Use of Recycled for Small Water Treatment Systems, Water Sci. Technol. 2003, 48 (11–
Crushed Glass as a Filtration Medium in Municipal Potable 12), 473–479.

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clean-journal.com

View publication stats

Das könnte Ihnen auch gefallen