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

-------