United States
Environmental Protection
Agency
EPA/600/R-13/342 | May 2014 | www.epa.gov/research
Evaluation of an
Innovative Sand Filter for
Small System Drinking
Water Treatment
Office of Research and Development
-------
EPA/600/R-13/342
May 2014
Evaluation of an Innovative Sand Filter for Small System
Drinking Water Treatment
by
Christopher A. Impellitteri, Ph.D. and Craig L. Patterson, P.E.
U.S. Environmental Protection Agency
Cincinnati, OH
Nur Muhammad, Ph.D., P.E. and Rajib Sinha, P.E.
Shaw Environmental & Infrastructure (E&l), Inc.
Cincinnati, OH
Fred Stottlemyer
International Rural Water Association
White Hall, MD
This report was compiled in cooperation with Shaw Environmental & Infrastructure (E&l), Inc.
Under EPA Contract EP-C-09-041, Work Assignment No. 1-04 and 2-04
Submitted to
National Risk Management Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
-------
Disclaimer
The U.S. Environmental Protection Agency, through its Office of Research and Development,
funded and managed, or partially funded and collaborated in, the research described herein. It
has been subjected to the Agency's peer and administrative review and has been approved for
publication. Any opinions expressed in this report are those of the author(s) and do not
necessarily reflect the views of the Agency, therefore, no official endorsement should be
inferred. Any mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
11
-------
Abstract
Results of evaluation of an innovative sand filter that uses the concepts of both slow and rapid
sand filtration are presented in this article. The system uses a low-cost "Drum Sand Filter" (DSF)
that consists of a 55-gallon drum filled with layers of sand of varying size. A low-cost "Drum
Flocculator" (DF) and tablet chlorination are incorporated before and after the DSF,
respectively, to enhance the filter performance and to provide a final barrier against microbial
contamination. The results of the evaluation demonstrated that the DSF with the DF and tablet
chlorination is very effective in removing turbidity and the selected microbiological
contaminants including Escherichia coli, Bacillus subtilis, Total and Fecal Coliform and
Polystyrene Latex (PSL) beads, a surrogate for Cryptosporidium. The DF/DSF system evaluated
in this work is meant to provide a low-cost system in order to provide basic water treatment for
locations where such treatment is not available.
Keywords
Drum flocculator; Drum sand filter; Microbiological contaminant; Small community; Tablet
chlorinator; Turbidity
in
-------
Table of Contents
Disclaimer ii
Abstract iii
Table of Contents iv
List of Tables v
List of Figures v
Acronyms and Abbreviations vi
1.0 INTRODUCTION 1
2.0 METHODS AND MATERIALS 1
2.1 System Description 1
2.1.1 Drum Sand Filter 1
2.1.2 Chemical Coagulation and Drum Flocculator 4
2.1.3 Disinfection 5
2.2 Test Plan and Conditions 6
2.2.1 Turbidity Challenges 6
2.2.2 Microbial Challenges 6
2.2.3 Data Analysis 8
3.0 RESULTS AND DISCUSSION 8
3.1 Estimation of Turbidity Removal Performance of the DF-DSF System 8
3.1.1 Turbidity Challenges 8
3.2 Estimation of Cryptosporidium Removal Performance of the DF-DSF System Using
Surrogates 10
3.2.1 PSL Beads Challenges 10
3.2.2 Particle Removal 12
3.2.3 B.subtilis Challenges 12
3.3 Estimation of Bacteria Removal Performance of the DF-DSF System 15
3.3.1 HPC Removal 15
3.3.2 f. co//Challenges 16
3.3.3 Total and Fecal Coliform Challenges 18
3.4 Maintenance of the DSF 21
4.0 CONCLUSIONS 21
5.0 REFERENCES 23
iv
-------
List of Tables
Table 1 - Media Configuration of the 'Drum Sand Filter' 2
List of Figures
Figure 1 - Schematic Layout of the DF-DSF System 3
Figure 2 - Drum Sand Filter Set-up 4
Figure 3 - Drum Flocculator Setup 5
Figure 4 - Chlorine Tablet Feeder 5
Figure 5-a) Influent and Effluent Turbidity; b) Percent Removal of Turbidity at Different
Coagulant Conditions 9
Figure 6 - a) Influent and Effluent Concentrations of PSL Beads; b) Log Removal of PSL Beads at
Different Coagulant Conditions 11
Figure 7 - Percent Removal of Particles at Different Coagulant Conditions 12
Figure 8 - a) Influent and Effluent Concentrations of B. subtilis; b) Log Removal of B. subtilis at
Different Coagulant Conditions 14
Figure 9 - Removal of HPC at Different Coagulant Conditions 15
Figure 10 - a) Influent and Effluent Concentrations of E. coli; b) Log Removal of E. coli at
Different Coagulant Conditions 17
Figure 11 - a) Influent and Effluent Concentrations of Total Coliform; b) Log Removal of Total
Coliform at Different Coagulant Conditions 19
Figure 12 - a) Influent and Effluent Concentrations of Fecal Coliform; b) Log Removal of Fecal
Coliform at Different Coagulant Conditions 20
-------
Acronyms and Abbreviations
B. subtilis Bacillus subtilis
DF drum flocculator
DSF drum sand filter
E. coli Escherichia coli
EPA U.S. Environmental Protection Agency
gpm gallon/minute
HOCI hypochlorous acid
HPC heterotrophic plate counts
IRWA International Rural Water Association
L liter
Ipm liter/minute
LRV log removal value
LT2ESWTR Long Term 2 Enhanced Surface Water Treatment Rule
m meter
m3/rn2/hr cubic meter/square meter/hour
mg milligram
ml milliliter
mm millimeter
MLSS mixed liquor suspended solids
MSD Metropolitan Sewer District
NTU Nephelometric Turbidity Unit
POE point-of-entry
POU point-of-use
PSL Polystyrene Latex
PVC Polyvinyl Chloride
PWS Public Water System
SDWA Safe Drinking Water Act
T&E Test and Evaluation
u.m micron
VI
-------
1.0 INTRODUCTION
Safe drinking water is essential for good public health. Contamination of drinking water by
microorganisms is a major cause of human illness world-wide. Between 1971 and 2003, more
than 600 waterborne disease outbreaks were recorded in the United States and most of the
outbreaks resulted in serious illness and even death (U.S. EPA, 2003). Therefore, properly
designed water treatment systems, either large or small, are of critical importance to protect
public health. Drinking water regulations present a challenge to all U.S. water utilities,
especially for communities with small systems where resources and funding are limited. Small
water treatment systems are larger than Point-of-Use (POU) (2-8 L/person/day) and Point-of-
Entry (POE) (100-150 L/person/day) units, but with a distinctly smaller capacity than centralized
public water systems (PWS) (Peter-Varbanets et a/., 2009). EPA defines small systems as those
serving between 25 and 10,000 people (U.S. EPA, 2003). The capacity of a small system cannot
be unequivocally defined, but usually varies between 1,000 L/day and 10,000 L/day (Peter-
Varbanets et at., 2009). The Safe Drinking Water Act (SDWA) established standards for drinking
water systems and required EPA to assess treatment technologies relevant to small systems.
Many small water utilities are interested in evaluating low-cost technologies for drinking water.
One possible technology is a low cost "Drum Sand Filter" (DSF). With this technology, a simple
drum is filled with layers of fine sand supported by coarser sand and associated pipe work
arranged to force the water to flow downward through the filter. A low-cost flocculation unit
using a simple "Drum Flocculator" (DF) is incorporated before the sand filter to enhance the
filtration performance of the system. A low-cost disinfection process using tablet chlorination
is incorporated into the effluent stream as a final barrier against microbial contamination. This
innovative sand filtration system utilizes concepts of both slow and rapid sand filtration. This
technology could be used for short-term measures to provide a safe supply of drinking water
from unsafe polluted water sources. The small basic treatment system described here could
also serve small systems for longer periods of time if adequately maintained and operated. This
option should be sustainable until a longer-term safe and cost-effective supply is available to
the population. A series of turbidity and microbial challenge tests were conducted on a pilot
system set up at the U.S. EPA Test & Evaluation (T&E) Facility in Cincinnati, Ohio. The objective
of the tests was to evaluate the performance of the system and the efficacy of different
chemical coagulants. This paper summarizes the results of the tests conducted on the
innovative coagulation/sand filtration system and critically evaluates its potentials for drinking
water treatment for a small community.
2.0 METHODS AND MATERIALS
2.1 System Description
2.1.1 Drum Sand Filter
The DSF consists of fine sand supported by different layers of coarser sand and gravel in a 208-L
(55-gallon) drum. Table 1 presents the filter configuration established by a series of tests
1
-------
during preliminary studies. The overall configuration of the DSF utilizes sand sizes typical of
those used in slow sand filtration systems (0.15 mm - 0.45 mm) (Ellis, 1985; Muhammad et al.,
1996). The bottom layer consists of coarse sand that serves a dual role; filtration and support
for the finer sand. The rate of filtration of the DSF is 0.75 m3/rn2/hr (i.o gpm) which is faster
than the recommended range of 0.10 m/hr to 0.30 m/hr for slow sand filtration (Ellis, 1985;
Muhammad et al., 1996). The depth of the DSF (0.6 m) is noticeably lower than the
recommended range of 1.2 m to 1.4 m for slow sand filters (Ellis, 1985), and close to the
minimum possible depth of 0.48 m for a slow sand filter (Bellamy et al., 1985). The depth of the
DSF is within the recommended range of 0.6 m to 0.75 m for a rapid sand filter (Schultz & Okun,
1984). The DSF system consists of two drum sand filters in series.
Table 1 - Media Configuration of the 'Drum Sand Filter'
Media
Sand
Sand
Sand
Sand
Size
Global No. 560 (Effective
size = 0.15 mm)
Global No. 8 (Effective size
= 0.20 mm)
Global No. 7 (Effective Size
= 0.48 mm)
Global No. 5 (Effective Size
= 1.10 mm)
Depth (mm)
175
175
100
150
Figure 1 shows the schematic layout of the DSF System. If the inlet valve of the first filter is fully
open and the design flow rate is not achieved, the filter is becoming clogged. At this stage, the
order of the filters is reversed by placing the cleaner second filter as the lead filter and the
dirtier first filter as the lag filter that allows longer operation of the system prior to cleaning.
The blue lines in the schematic layout show the reversing of the filters. When both filters
become dirty, they are cleaned one at a time keeping the other one operational. When both
filters are cleaned and re-installed, the new operational cycle starts.
-------
20
Legend: 1 - Feed Water Tube; 2 - Coagulant Tube; 3 - Drum Flucculator (DF); 4 - Microbial Injection Tube; 5 - Outlet Tube from DF; 6 - Flow
Meter; 7 - Influent Tube for Drum Sand Filter (DSF) 1 (Series Configuration); 8 - DSF 1; 9 - Effluent Tube from DSF 1; (Series Configuration)
10 - Influent Tube for Filter 2 (Series Configuration); 11 - DSF 2; 12 - Effluent Tube from DSF 2 (Series Configuration); 13 - Feed Tube to
Chlormator from DSF 2 (Series Configuration); 14 - Influent Tube for DSF 2 (Reversed Series Configuration); 15 - Effluent Tube from DSF 2
(Reversed Series Configuration); 16 - Influent Tube for DSF 1 (Reversed Series Configuration); 17 - Effluent Tube from DSF 1 (Reversed
Series Configuration); 18 - Feed Tube to Chlormator from DSF 1; 19 - Chlormator; 20 - Final Effluent
Figure 1 - Schematic Layout of the DF-DSF System
A 50 mm diameter effluent PVC pipe connected to a fine stainless steel strainer (20 u.m pore
size) was inserted into the coarser sand media to extract the effluent water. The stainless steel
strainer was fine enough to prevent the finer sand media from entering the effluent pipe. A
tablet chlorinator was connected to the outlet pipe to disinfect the filtered water. The final
filter set-up is depicted in Figure 2.
-------
Tablet
Feeder
Outlet Pipe
Filter 2
Flow Meter
Inlet Pipe
Filter 1
Figure 2 - Drum Sand Filter Set-up
2.1.2 Chemical Coagulation and Drum Flocculater
Three different coagulants were used during this study: alum, ferric chloride and chitosan. The
doses of alum (10 mg/L), ferric chloride (4 mg/L) and chitosan (3 mg/L) were determined by jar
tests conducted on source water with a turbidity of 10 nephelometric turbidity units (NTU).
Turbidity challenges that included monitoring heterotrophic plate counts (HPC) and natural
particles (2-5 u.m) were conducted using all three selected coagulants. Microbial/surrogate
tests were conducted using one conventional coagulant, alum and the emerging coagulant,
chitosan; no microbial challenge tests were conducted using ferric chloride. A drum flocculator
was used to mix the coagulants with the influent water in this study. This low-cost and
sustainable method of flocculation in a simple drum was developed by the International Rural
Water Association (IRWA) (Maryland, U.S.A.). A stock solution of the selected coagulant was
mixed with deionized water and added to the flocculator at a specific rate to achieve the
desired concentration. The feed water mixed with coagulant flows through an inlet tube to the
sand filter. The flocculator was located at 3.7 m (12 feet) above the sand filter to provide
enough head to achieve a feed flow rate of 3.8 Lpm (1.0 gpm). Microbial contaminants were
injected into the inlet tube using a peristaltic pump to achieve the desired inlet concentrations.
Figure 3 shows the DF setup at the T&E Facility. The contact time of the coagulants in the DF
was 55 minutes.
-------
Motor for
Mixer
Drum
Flocculator
Coagulant
Injection
Microbial
Injection
Sand Filter
Inlet Tube
Figure 3 - Drum Flocculator Setup
2.1.3 Disinfection
A tablet chlorinator, that generates chlorine by dissolving a chlorine tablet, (Severn Trent
Services, model: 200) was connected to the effluent pipe to disinfect the final product. The
generation of chlorine depends on a number of factors including the contact surface, flow rate
and the age of the tablet. Figure 4 shows the chlorine tablet feeder.
Tablet
Feeder
Contact Tank
Figure 4 - Chlorine Tablet Feeder
-------
An automatic chlorine injector that uses a pre-set injection ratio to draw a hypochlorite solution
was also considered as an alternate option for disinfection. However, as the sand filter was fed
from the drum flocculator by gravity, the pressure in the influent line was not enough to extract
the chlorine solution from the injector. Therefore, the tablet chlorinator was used in the
present study.
2.2 Test Plan and Conditions
2.2.1 Turbidity Challenges
For evaluating turbidity removal performance of the DSF, surface water was obtained from the
Creek (source water adjacent to the U.S. EPA T&E Facility) and mixed with dechlorinated
potable water in an 18,927-liter 5000-gallon) tank to produce matrix water with an influent
target turbidity level of 10 NTU. Grab samples (100 ml) from the influent and effluent streams
were collected at hourly intervals for approximately 5 hours. A HACH turbidity meter (Model
2100P) was used to measure the turbidity of the grab samples.
2.2.2 Microbial Challenges
To evaluate specific bacteria removal, the system was challenged with three different species:
1) Escherichia coli (E. coli), a common normal flora in the gut, 2) Bacillus subtilis (B. subtilis), a
bacteria that goes from a vegetative stage to a spore stage depending on environmental
conditions, and 3) Total and Fecal Coliform, indicator organisms for fecal pollution. For
establishing the performance of the system in removing heterotrophic bacteria, HPC of the
influent and effluent samples were determined during the turbidity challenges.
Freeze-dried E. coli (ATCC 15222™) was obtained from American Type Culture Collection
(Manassas, Virginia) and the stock was re-constituted by adding 1 ml nutrient broth. The
mixture was then transferred into 100 ml nutrient broth and incubated in a shaker incubator at
36 °C and 170 rpm for 20 - 24 hours. The culture was analyzed using IDEXX Colilert-18 Method
(IDEXX, 2003) and the concentration was ~ 1.0 x 109/mL. A sub-culture, using 1 ml of the stock
in 100 ml of nutrient broth, was prepared by incubating the mixture as described above for the
challenge tests. Grab samples for E. coli were analyzed using the IDEXX Colilert-18 Method
(IDEXX, 2003).
B. subtilis aerobic spores were used to fulfill two roles in the testing, both as a chlorine-resistant
surrogate and as a Cryptosporidium surrogate. B. subtilis spores were obtained from Raven
Laboratories (Omaha, Nebraska). Grab samples for B. subtilis were analyzed in accordance with
methods described by Rice et al. (1994) using heat shock and standard membrane filtration.
For total and fecal coliform tests, the system was challenged with mixed liquor suspended
solids (MLSS) collected from the aeration basin of the Metropolitan Sewer District (MSD)
wastewater treatment plant located adjacent to the U.S. EPA T&E Facility. The concentrations
of total and fecal coliform of the MLSS were ~1.0 x 106/rnL and ~ 5.0 x 104/mL, respectively.
-------
Grab samples for the total and fecal coliform were analyzed using Standard Methods 9222B and
9222C, respectively (APHA, AWWA and WEF, 2005).
To provide indirect and secondary measures of microbial removal, the removal of heterotrophic
bacteria was monitored during turbidity challenges. Grab samples (100 ml) for HPC from the
influent and effluent streams were collected twice during each experiment and analyzed using
the IDEXXSimPlate method (IDEXX, 2002).
To evaluate the performance of the DF/DSF in removing Cryptosporidium, the system was
challenged with PSL beads having a mean size of 2.83 u.m. PSL bead stock was obtained from
Polysciences Inc. (Warrington, PA). Grab samples (1 L) were centrifuged directly without
filtration and analyzed using a hemacytometer following U.S. EPA Method 1622 (U.S. EPA,
2001). As the influent bead concentrations were reasonably high, the samples were centrifuged
directly. Although the effluent beads concentrations were low, the grab samples were
centrifuged to keep the method consistent with that used for influent samples. A 1 u.m
membrane filter was tested for collection of beads from the effluent as an alternative sampling
method. However, the effluent pressure was not adequate to flow through the membrane.
To provide an indirect and secondary measure of protozoa removal, the removal of particles in
the size range of 2-5 u.m (that encompasses the size of Cryptosporidium parvum) was also
determined during the turbidity challenges. Grab samples (100 ml) for particle counts from the
influent and effluent streams were collected twice during each experiment and analyzed using
a HIAC Royco (Model 9703) Particle Analyzer. The particle count data in the size range of 2 - 5
u.m for the influent and effluent were compared.
For B. subtilis, E. coli and PSL beads, 2 ml of stock suspension with an approximate
concentration of 109 cells or surrogates per ml was mixed with 1000 ml of 0.01% Tween 20 in a
2-L glass beaker. A sub-sample was collected to determine the actual concentration of the
injection suspension. For total and fecal coliform, 1000 ml MLSS from the adjacent MSD
wastewater treatment plant was used as the stock suspension. The 1000 ml MLSS suspension
and the rinseate (1000 ml Dl water) were added into the influent stream of the system using a
peristaltic pump. The total injection time was approximately 120 minutes; the sand filter was
operated for an additional 60 minutes to observe the presence of contaminant after stopping
the injection. Samples from the influent stream were collected at 0, 5, 15, 30 and 60 minutes
after the start of the injection. Initially, effluent samples were collected at the same sampling
events as the influent samples and maximum effluent concentrations were observed at 60
minutes and remained consistent until 180 minutes. This led to the assumption of a lag time of
60 minutes between the influent and effluent, and effluent samples were collected at 60, 90,
120 and 180 minutes after the start of the injection for the subsequent experiments. The
volume of grab samples for B. subtilis and E. coli was 100 ml and that for PSL beads was 1000
ml. The total and free chlorine concentrations of the effluent samples were determined using a
HACH Spectrophotometer (Model 2400).
-------
2.2.3 Data Analysis
The average influent and effluent concentrations and percent/log removal of contaminants are
presented graphically with standard error. Student's t-tests were conducted to compare the
significance of the difference in performance of the sand filter at different coagulant conditions.
Log removal values for challenge contaminants are presented as DSF log removal value (LRV)
which is the removal by the DSF only, and total LRV which represents combined removal by the
DSF and chlorination.
3.0 RESULTS AND DISCUSSION
3.1 Estimation of Turbidity Removal Performance of the DF-DSF System
3.1.1 Turbidity Challenges
Figure 5 shows the influent and effluent turbidity and percent removals of turbidity under
different coagulant conditions. The influent and effluent data for turbidity represent the
average of six grab samples collected during two turbidity challenges at each coagulant
condition. For approximately similar influent turbidity (10.3 - 11.4 NTU), the average effluent
turbidity was 0.45 NTU, 0.25 NTU, 0.25 NTU and 0.14 NTU for tests conducted with no
coagulant, 10 mg/L alum, 4 mg/L ferric chloride and 3 mg/L chitosan, respectively. Statistical
analysis indicates significant improvement of effluent quality due to the use of 10 mg/L alum (p
value 0.0003), 4 mg/L ferric chloride (p value 0.0007) and 3 mg/L chitosan (p value 0.00008) as
chemical coagulants. The use of 3 mg/L chitosan produced effluent with significantly lower
turbidity than 10 mg/L alum (p value 0.00005) and ferric chloride (p value 0.0009). No
statistically significant differences in removal performance were observed between 10 mg/L
alum and 4 mg/L ferric chloride (p value 0.18).
-------
I Influent
I Effluent
No 10 mg/L 4 mg/L Ferric 3 mg/L
Coagulant Alum Chloride Chitosan
a)
+j
-------
The effluent quality produced due to the addition of alum, ferric chloride and chitosan satisfied
the Long Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR) requirement of <0.30
NTU effluent turbidity (U.S. EPA, 2006). Chitosan was found to be very effective in reducing
turbidity. This concurs with the findings reported in another study (Brown and Emelko, 2009).
3.2 Estimation of Cryptosporidium Removal Performance of the DF-DSF System Using
Surrogates
3.2.1 PSL Beads Challenges
PSL beads (2.83 u.m) were used as a surrogate for removal of Cryptosporidium. The LT2ESWTR
dictates that a surrogate must have an effective size of 3 u.m or smaller to demonstrate
Cryptosporidium removal. Figure 6 shows the influent and effluent concentrations and LRVs of
PSL beads at different coagulant conditions. The influent and effluent data represent the
average of six grab samples collected during two PSL beads challenges at each coagulant
condition. The sand filter performed effectively in removing 2.83 u.m PSL beads with an average
LRV of 1.85 that is close to the LT1ESWTR requirement (2.0 log) for Cryptosporidium removal
(U.S. EPA, 2002). The average LRVs achieved by the DSF based on influent beads concentrations
were >6.06 and >6.10 during challenges with 10 mg/L alum and 3 mg/L chitosan, respectively,
that satisfied the LT2ESWTR requirement (> 5.5 log) for Cryptosporidium removal at the highest
category of Bin 4 (U.S. EPA, 2006). The improvement in removal performance was statistically
significant due to the addition of 10 mg/L alum (p value 0.000005) and 3 mg/L chitosan (p value
0.000002). No statistically significant difference in removal performance was observed between
10 mg/L alum and 3 mg/L chitosan (p value 0.47). The overall performance of the DF-DSF
system in removing Cryptosporidium is either similar or superior to that of the conventional
slow sand filter reported in another study (Bellamy et a/., 1985).
10
-------
Influent
DSF Effluent
No Coagulant 10 mg/L Alum 3 mg/L Chitosan
a)
No Coagulant
10 mg/L Alum
b)
3 mg/L Chitosan
Figure 6 - a) Influent and Effluent Concentrations of PSL Beads; b) Log Removal of PSL Beads
at Different Coagulant Conditions
11
-------
3.2.2 Particle Removal
The particle count data were evaluated to obtain secondary information on the performance of
the DSF. The influent particle counts were measured as available in matrix water. Figure 7
shows the percent removal of 2-5 u.m size particles that mimic the size of Cryptosporidium
oocysts, at different coagulant conditions. The data represent the average of four grab samples
collected during two turbidity challenges at each coagulant condition. The sand filter achieved
an overall 74.2% particle (2-5 u.m) removal performance. Percent removal of particle counts
increased significantly due to the addition of 10 mg/L alum (92.9%; p value 0.013), 4 mg/L ferric
chloride (94.9%; p value 0.014) and 3 mg/L chitosan (96.5%; p value 0.006).
The removal of natural particles (2-5 u.m) is noticeably lower than that for PSL beads indicating
Cryptosporidium size natural particles to be a conservative surrogate for Cryptosporidium
removal. This concurs with the findings of another study (Emelko et o\., 2005).
100
No Coagulant 10 mg/L Alum 4 mg/L Ferric 3 mg/L Chitosan
Chloride
Figure 7 - Percent Removal of Particles at Different Coagulant Conditions
3.2.3 B. subtilis Challenges
Figure 8 shows the influent and effluent concentrations and LRVs of B. subtilis at different
coagulant conditions. The influent and effluent data represent the average of four grab samples
collected during the single B. subtilis challenge at each coagulant condition. Results showed
that the performance of the sand filter was not adequate in removing B. subtilis; the LRV
achieved was 1.19. Addition of 10 mg/L alum enhanced the performance of the sand filter
significantly (p value 0.002); the LRV increased to 2.12. Significant improvement of the removal
12
-------
performance was also observed due to addition of 3 mg/L chitosan (p value 0.008); the average
LRV achieved was 2.85. The 3 mg/L chitosan was observed to be more effective than the 10
mg/L alum (p value 0.0002) in enhancing the removal performance. The free residual chlorine
concentrations during the B. subtilis challenges varied between 0.6 and 2.0 mg/L. Post-
chlorination was observed to be ineffective at inactivating B. subtilis. As sodium thiosulfate
tablets were added to the sample bottles during sampling, no additional contact time was
available for disinfection of the spores.
The LRVs for B. subtilis by the sand filter were noticeably lower than that for PSL beads
indicating B. subtilis to be a conservative surrogate for Cryptosporidium. This concurs with the
findings of another study (Muhammad et al., 2008) conducted on different drinking water
treatment systems. The overall performance of the DF-DSF system in removing bacterial spores
is superior to that of the conventional slow sand filter reported in another study (Heller et al.,
2007).
13
-------
g,5
O
O
(/>
= 3
4-*
Si
D
V)
0)2
I Influent
IDSF Effluent
CI2 Effluent
No Coagulant 10 mg/L Alum 3 mg/L Chitosan
a)
I DSF LRV
ITotalLRV
No Coagulant 10 mg/L Alum 3 mg/L Chitosan
b)
Figure 8 - a) Influent and Effluent Concentrations of B. subtilis; b) Log Removal of B. subtilis at
Different Coagulant Conditions
14
-------
3.3 Estimation of Bacteria Removal Performance of the DF-DSF System
3.3.1 HPC Removal
The HPC data were evaluated to obtain secondary information on the performance of the DSF
in removing bacteria. The influent HPC concentrations were measured as available in matrix
water. Figure 9 shows the removal of HPC at different coagulant conditions. The data represent
the average of four grab samples collected during two turbidity challenges at each coagulant
condition. The performance of the sand filter was not adequate in removing HPC; the LRV
achieved was 0.13. Removal of HPC increased significantly due to the addition of 10 mg/L alum
(0.705 log; p value 0.04) and 4 mg/L ferric chloride (1.25 log; p value 0.0007). No significant
improvement of HPC removal was observed with the addition of 3 mg/L chitosan (0.87 log; p
value 0.05). Post-chlorination was very effective in removing heterotrophic bacteria. The
average free chlorine concentrations (HOCI) for turbidity challenges conducted with no
coagulant, 10 mg/L alum, 4 mg/L ferric chloride and 3 mg/L chitosan were 5.15 mg/L, 2.50
mg/L, 2.15 mg/L and 0.98 mg/L, respectively. The overall LRVs during turbidity challenges at 3
mg/L chitosan were low due to relatively lower residual chlorine concentrations. The residual
chlorine varied among different tests due to age of the tablet, contact surface and
accumulation of tablet residue in the chlorinator effluent tube. The overall performance of the
DF-DSF system in removing heterotrophic bacteria is similar to that of the conventional slow
sand filter reported in another study (Palmateer et a/., 1999).
IDSFLRV
I Total LRV
No 10 mg/L 4 mg/L 3 mg/L
Coagulant Alum Ferric Chitosan
Chloride
Figure 9 - Removal of HPC at Different Coagulant Conditions
15
-------
3.3.2 E. coli Challenges
Figure 10 shows the influent and effluent concentrations and LRVs of E. coli at different
coagulant conditions. The influent and effluent data represent the average of eight grab
samples collected during two challenges at each condition. Results showed that the
performance of the sand filter was moderate in removing E. coli; the LRV achieved was 2.4.
Addition of 10 mg/L alum enhanced the performance of the sand filter significantly (p value
0.0002); the LRV increased to 3.70. No significant improvement in removal of E. coli was
observed due to the addition of 3 mg/L chitosan (p value 0.29); the average LRV achieved was
2.6. Backwashing followed by prolonged flushing of the sand filter between the E. coli
challenges conducted with 3.0 mg/L chitosan did not result in noticeable improvement of
performance. The sand filter was re-packed for the total and fecal coliform challenges.
The free residual chlorine concentrations during the E. coli challenges varied between 0.5 and
1.9 mg/L. Nearly complete inactivation of E. coli was achieved by post-chlorination despite the
addition of sodium thiosulfate tablets in the samples.
The performance of the DF-DSF system in removing E. coli is similar to that of conventional slow
sand filter reported in another study (Ellis, 1985); however, the overall performance of the
system including chlorine disinfection is noticeably superior than the conventional slow sand
filter.
16
-------
7
6
5
UJ
I Influent
IDSF Effluent
CI2 Effluent
No Coagulant lOmg/LAIum 3 mg/L Chitosan
a)
IDSFLRV
ITotalLRV
No Coagulant 10 mg/L Alum 3 mg/L Chitosan
b)
Figure 10 - a) Influent and Effluent Concentrations of E. coli; b) Log Removal of E. coli at
Different Coagulant Conditions
17
-------
3.3.3 Total and Fecal Coliform Challenges
Figure 11 shows the influent and effluent concentrations and log removals of total coliform at
different coagulant conditions. Figure 12 shows the influent and effluent concentrations and log
removals of fecal coliform at different coagulant conditions. The influent and effluent data
represent the average of eight grab samples collected during two challenges at each condition.
Results show that the performance of the sand filter was moderate in removing total coliform
and fecal coliform; the LRV for total coliform and fecal coliform were 1.80 and 1.45,
respectively. Addition of 10 mg/L alum enhanced the performance of the sand filter
significantly in removing total coliform ((p value 0.004) and fecal coliform (0.005); the LRV of
total coliform increased to 2.60 and that for fecal coliform increased to 2.75. Addition of 3 mg/L
chitosan improved the performance of the sand filter significantly in removing total coliform (p
value 0.003) and fecal coliform (p value 0.02); the average LRV for total coliform increased to
2.45 and that for fecal coliform increased to 2.62. No significant differences were observed
between alum and chitosan at the selected doses in removing total coliform (p value 0.12) and
fecal coliform (p value 0.30).
The free residual chlorine concentrations during the total and fecal coliform challenges varied
between 0.5 and 1.9 mg/L. Nearly complete inactivation of total and fecal coliform was
achieved by post-chlorination despite the addition of sodium thiosulfate tablets in the samples.
The performance of the DF-DSF system in removing total and fecal coliform is similar to that of
the conventional slow sand filter reported in several other studies (Muhammad et al., 1996;
Eiilis, 1985); however, the overall performance of the system including chlorine disinfection is
noticeably superior than the conventional slow sand filter.
18
-------
I Influent
IDSF Effluent
CI2 Effluent
No Coagulant 10 mg/L Alum 3 mg/L Chitosan
a)
I5
o
O
$ 4
i2
n
52
o
i
0)1
o
• DSF LRV
• Total LRV
No Coagulant 10 mg/L Alum 3 mg/L Chitosan
b)
Figure 11 - a) Influent and Effluent Concentrations of Total Coliform; b) Log Removal of Total
Coliform at Different Coagulant Conditions
19
-------
5
4.5
_ 4
d.3.5
6
§ 3
o
E2.5
£
'o 2
o
8 1-5
v
1L.
1
0.5
0
I Influent
IDSF Effluent
CI2 Effluent
No Coagulant 10 mg/L Alum 3 mg/L Chitosan
a)
IDSFLRV
ITotalLRV
No Coagulant 10 mg/L Alum 3 mg/L Chitosan
b)
Figure 12 - a) Influent and Effluent Concentrations of Fecal Coliform; b) Log Removal of Fecal
Coliform at Different Coagulant Conditions
20
-------
3.4 Maintenance of the DSF
The treatment mechanism of the DSF is not biological in nature; the performance of the system
was noticeably enhanced by the addition of coagulant in this study. Therefore, it is necessary to
maintain the correct dose and proper mixing of coagulant during the operation of the system.
It is necessary to check the raw and finished water turbidity daily. If the raw water turbidity is
>10 NTU, a pre-treatment unit is recommended to be incorporated before the DF-DSF system
to avoid rapid clogging and frequent cleaning of the filter. If the finished water turbidity is >1
NTU, a full inspection of the system that include checking the flow rate, feed water turbidity,
coagulant dose and short-circuiting or channeling of the sand bed is necessary.
Backwashing of the DSF is necessary when the flow rate declines due to clogging of the sand
and the desired flow is no longer achieved. Although the flow rate did not decline during the
operation of the unit in this study, backwashing was conducted after the completion of a
specific contaminant challenge test to create clean conditions for the next contaminant
challenge tests. Backwashing of the unit was conducted using clean water at a flow rate of 1
gpm for 30 minutes. Backwashing and flushing was conducted on the DSF to create clean
conditions for challenge tests. Backwashing flow rate is recommended to keep to a minimum to
avoid vigorous fluidization and loss of fine sand. After the completion of each challenge test,
the DSF is recommended to be flushed using at least 3 bed volumes of water to maintain a
stable sand bed condition.
Maintaining a uniform flow rate is very important to ensure better performance of the DSF.
Sudden increase of flow rate may create short-circuiting resulting in a poor filtrate quality. It
will also impact the generation of free chlorine in the tablet chlorinator.
As the concentration of residual chlorine generated using the tablet chlorinator depends on a
number of factors including contact surface, flow rate and age of the tablet, it is recommended
that the residual chlorine be checked at least twice a day. It is also recommended to check the
accumulation of tablet residue in the outlet tube as it may cause high residual chlorine
concentration.
4.0 CONCLUSIONS
Based on the results of turbidity challenges, the sand filter with the current configuration was
effective in removing turbidity; for an influent turbidity of ~ 10.0 NTU, the effluent turbidity was
consistently below 0.5 NTU without chemical coagulant. Significant improvement of effluent
quality was observed due to use of chemical coagulants; the effluent turbidity was consistently
below 0.3 NTU that satisfied the LT2ESWTR requirement (< 0.3 NTU) for turbidity removal.
The sand filter performed effectively in removing 2.83 u.m PSL beads with an average LRV of
1.85 that was close to the LT1ESWTR requirement (2.0 log) of Cryptosporidium removal. The
21
-------
system achieved >6.0 log removal of 2.83 jam PSL beads during challenges with chemical
coagulants that satisfied the LT2ESWTR requirement (> 5.5 log) for Cryptosporidium removal at
the highest category of Bin 4 for the DSF. The sand filter was found to be effective in removing
Cryptosporidium size natural particles (2-5 u.m), and the addition of chemical coagulants
enhanced the performance significantly. The performance of the sand filter alone was not
adequate (1.19 log removal) in removing B. subtilis. However, the addition of 10 mg/L alum
(2.12 log removal) and 3 mg/L chitosan (2.85 log removal) enhanced the performance of the
system significantly. The LRVs for B. subtilis and particle counts (2-5 u.m) by the sand filter
were noticeably lower than those for PSL beads indicating that the potential of B. subtilis and
particle counts may be too conservative for consideration as a surrogate for Cryptosporidium in
comparison with the PSL beads.
The performance of the sand filter was not adequate in removing heterotrophic bacteria
without chemical coagulant; significant improvement was observed due to the addition of
chemical coagulants. The system demonstrated moderate performance in removing E. coli (2.40
log), total coliform (1.80 log) and fecal coliform (1.45 log) without chemical coagulant.
Significant improvements in the removal of E. coli were observed due to the addition of 10
mg/L alum (3.70 log) in feed water. No significant improvement was observed with the addition
of 3 mg/L chitosan (2.60 log). Addition of 10 mg/L alum enhanced the performance of the sand
filter significantly in removing total and fecal coliform; the LRV for total coliform (1.80 log)
increased to 2.60 and that for fecal coliform (1.45 log) increased to 2.75. Addition of 3 mg/L
chitosan improved the performance of the sand filter significantly in removing total and fecal
coliform; the LRV for total coliform (1.80 log) increased to 2.45 and that for fecal coliform (1.45
log) increased to 2.62.
Post-chlorination was very effective in inactivating heterotrophic bacteria, E. coli, fecal coliform
and total coliform; nearly complete inactivation of the microorganisms was achieved at 0.9
mg/L - 1.9 mg/L free chlorine with a low contact time. Post-chlorination was not effective in
inactivating B. subtilis spores at concentrations of 1.0 - 2.0 mg/L, as the sampling strategy with
sodium thiosulfate tablet in the bottle provided minimum contact time.
The performance of the DSF in combination with the selected coagulants and tablet disinfection
process was excellent in removing turbidity and the selected microorganisms/surrogates,
except B. subtilis spores. The system was easy to operate with no power and negligible
chemical requirements. Based on the performance, ease of operation, and cost, the DF-DSF
appeared to demonstrate good potential for drinking water treatment in very small
communities (serving less than 100 people) and for emergency situations.
22
-------
5.0 REFERENCES
APHA, AWWA and WEF (2005) 'Standard Methods for the Examination of Water and
Wastewater', 21st Ed., APHA, AWWA & WEF Publication, Washington D.C., U.S.A.
Bellemy, W.D., Hendricks, D.W. and Longsdon, G.S. (1985) 'Slow sand filtration: Influences of
selected process variables', J. AWWA, 77(12), 62-66.
Brown, J.T. and Emelko, B.M. (2009) 'Chitosan and metal salt coagulant impacts on
Cryptosporidium and microsphere removal by filtration'. Wat. Res., 43, 331-338.
Ellis, K.V. (1985) 'Slow Sand Filtration', CRC Crit. Rev. in Environ. Controls, 15(4), 315-354.
Emelko, M. B., Huck, P. M. and Coffey, B. M. (2005) 'A Review of Cryptosporidium Removal by
Granular Media Filtration', Jour. AWWA, 97(12), 101-115.
Heller, L, Viera, M. B. C. M., Brito, L. L. A. and Salvador, D. P. (2007) 'Association between the
Concentration of Protozoa and Surrogates in Effluents of the Slow Sand Filtration for Water
Treatment', Brazilian Jour, of Microbiol., 38, 337-345.
IDEXX (2003)'Instruction Manual for Colilert-18 for E. coll Analysis', Maine, U.S.A.
IDEXX (2002) 'Instruction Manual for SimPlate for HPC Multi Dose', Maine, U.S.A.
Muhammad, N., Sinha, R., Krishnan, R., Piao, H., Patterson, C. L., Cotruvo, J., Cumberland, S. L.,
Nero, V. P. and Delandra, C. (2008) 'Evaluating Surrogates for Cryptosporidium Removal in Point
of Use System', Jour. AWWA, 100(12), 98 - 104.
Muhammad, N., Ellis, K.V., Parr, J. and Smith, M, D. (1996) 'Optimization of slow sand filtration.
In Reaching the Unreached: Challenges for the 21st Century', Proceeding of the 22nd WEDC
conference. New Delhi, India.
Palmateer, G., Manz, D., Jurkovic, A., Mclnnis, R., Unger, S., Kwan, K. K. and Dutka, B. J. (1995)
'Toxicant and Parasite Challenge of Manz Intermittent Slow Sand Filter', Environ. Toxicol., 14,
217-225.
Peter-Varbanets, M., Zurbrugg, C. Swartz, C. and Pronk, W. (2009) 'Decentralized Systems for
Potable Water and Potential of Membrane Technology', Wat. Res., 43, 245-265.
Rice, E. W., Fox, K. R., Miltner, R. J., Lytle, D. A. and Johnson, J. H. (1994) 'A Microbiological
Surrogate for Evaluating Treatment Efficiency', In Proc. AWWA Water Quality Technology
Conference. San Francisco, California, U.S.A
Schultz, C. and Okun, D. (1984) 'Surface Water Treatment for Communities in Developing
Countries', Willey Publications, New York, U.S.A.
U.S. EPA (2006) 'The Long-Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR)
Implementation Guidance', EPA-816-F-06-019, Washington, D.C.
U.S. EPA (2003) 'Small Systems Guide to Safe Drinking Water Act Regulations', EPA 816-R-03-
016, Washington D.C., U.S.A.
U.S. EPA (2002) 'The Long-Term 1 Enhanced Surface Water Treatment Rule (LT1ESWTR)
Implementation Guidance', EPA-816-R-04-008, Washington, D.C.
U.S. EPA (2001) 'Method 1622: Cryptosporidium in Water by Filtration/IMS/FA', Washington,
DC, U.S.A.
23
------- |