United States
 Environmental Protection
 Agency
 Water Engineering
 Research Laboratory
 Cincinnati OH 45268
 Research and Development
 EPA/600/S2-85/027 Sept. 1985
Project  Summary
                                                                    r/l
 Filtration  of Giardia
 Cysts  and Other Substances:
Volume 3. Rapid-Rate Filtration
Mohammed AI-Ani, John M. McElroy,
Charles P. Hibler, and David W. Hendricks
  Rapid-rate filtration was evaluated
for a range of operating conditions
using waters having turbidity levels of
less than 1 NTU and temperatures rang-
ing from 0° to 17°C. The object was to
determine its effectiveness as a process
in drinking water treatment for removal
of Giardia lamblia cysts, total coliform
bacteria, standard plate count bacteria,
turbidity, and  particles from  low-
turbidity, low-temperature water.
  Results showed that when the filter is
operated as a  strainer (i.e., when no
chemical coagulation  is used); remo-
vals  ranged from 0 to  50 percent.
Improvement was not significant when
ineffective coagulants or improper dos-
ages were used. Effective coagulation
(that adequate to reduce turbidity from
about 0.5 NTU to about 0.1 NTU) was
capable of removing 95 to 99.9 percent
of Giardia cysts and 95 to 99.9 percent
of total coliform bacteria. Two coagu-
lant aids were found in this research
that provided effective coagulation.
  The filtration efficiency was unaf-
fected by mode of filtration. The in-line
mode was as efficient as the direct fil-
tration mode. Testing at 3° and at 17°C
showed no discernible differences in
percent  removals.  Increasing the
hydraulic loading rate from 8 to 41
cm/mm (2 to 10 gpm/ft2) showed no
discernible effect until the latter rate
was reached. The work showed that
efficient filtration of Giardia cysts and
other substances  present in low-
turbidity waters requires careful selec-
tion of coagulants and the use of proper
dosages. Turbidity  reduction can be
used as a measure of efficiency. Rou-
tine use of pilot plants, operated side-
by-side with full-scale  plants,  is
advocated for this purpose. The pilot
plant can be spiked with  bacteria as
another means to evaluate the effec-
tiveness of coagulation.
  This report is the third (last) describ-
ing the research conducted under EPA-
CSU  Cooperative  Agreement No.
CR808650-02. The  first was entitled
"Filtration of Giardia Cysts and Other
Substances. Volume 1: Diatomaceous
Earth  Filtration."   EPA/600/S2-
84/114,  September 1984, and  the
second was entitled "Filtration of Giar-
dia Cysts and  Other  Substances.
Volume 2:  Slow Sand  Filtration,"
EPA/600/S2-85/026, April 1985.
  This  Project Summary was devel-
oped  by  EPA's  Water  Engineering
Research  Laboratory, Cincinnati. OH.
to announce  key  findings  of  the
research project that is fully  docu-
mented in a separate report of the same
title (see Project Report ordering infor-
mation at back).

Introduction
Background
  Giardiasis  is an intestinal disease
caused by ingestion of cysts of the proto-
zoan  Giardia lamblia. In  recent  years,
reports of waterborne outbreaks of giar-
diasis in the United States have become
increasingly frequent. Some 53 water-
borne outbreaks and 20,039 cases were
reported during the period 1965 through
1981.   Most  outbreaks  have  been

-------
reported in small mountain communities
in the western and northeastern United
States,  but during the  period 1983-84,
cases were reported in Pennsylvania as
well.
  In the western United States, the cyst
is known to occur in ambient waters with
turbidity levels  of  less  than 1  NTU.
Though there is no reason to believe that
cysts  are  not found pervasively in all
kinds  of waters, it is important to point
out that they do  occur in low-turbidity
waters  and  that  many outbreaks have
been associated with these clear waters.
The conventional wisdom is that such
waters are likely to be benign, since even
without treatment, they may nearly con-
form to well-established standards—the
1-NTU turbidity standard and  the coli-
form standard,  for example. Yet public
systems using low-turbidity raw waters
may meet all of these standards and still
the source of a giardiasis outbreak.
  A major process problem for treatment
plants using low-turbidity waters is that it
is common practice not to use chemical
pretreatment,  though polymers  are
sometimes used  as  filter aids.  These
waters may already meet the 1  NTU tur-
bidity standard, and coliform counts are
often  very low, e.g. 10 organisms/100
ml. Thus, they are not as easily used mea-
sures of process efficiency as they are in
other  surface waters, such  as those
found at lower elevations. Also, the low
turbidity waters  are more difficult  to
treat. Without chemical treatment, how-
ever, the rapid-rate filtration process is
simply a method of straining and its use
deviates from one of the basic tenets of
rapid-rate filtration-that  chemical  pre-
treatment be  integral  to the  process.
Without such chemical treatment, sub-
stances in the raw water such as Giardia
cysts  and bacteria  can  and  do pass
through the  filter. Though such waters
have ocassionally been treated success-
fully, chemical pretreatment knowledge
is not adequate to effect rapid-rate filtra-
tion of low-turbidity,  low-temperature
waters containing Giardia cysts.


Objectives
  The purpose  of the  research was  to
determine how to remove Giardia lamblia
cysts from water  supplies by rapid rate
filtration when raw waters have turbidity
levels of less than 1 NTU. The main objec-
tives were (1) to determine a chemical
pretreatment for low-turbidity water that
results  in efficient  rapid-rate  filtration.
and (2) to determine the respective roles
of process variables on  removal efficien-
cies of Giardia cysts, turbidity, and bacte-
ria. Process variables included chemical
pretreatment (coagulant  selection and
dosages), filtration mode (conventional,
direct, or in-line), media, hydraulic load-
ing rate, and temperature. A third objec-
tive was  to determine  whether  a
surrogate   indicator  was  feasible  to
assess the treatment efficiencies for
removal of  Giardia cysts.

Methods
Pilot Plants
  The research was based on two physi-
cal models—laboratory-scale and field-
scale, rapid-rate filtration  pilot plants.
The laboratory-scale pilot plant was a
dual-train, conventional, rapid-rate filtra-
tion plant built to be operated under pres-
sure. The raw water to be processed was
from the Cache La Poudre River when
water with  less than  1  NTU  turbidity
could be obtained. When this was not
possible,  and  artificial low-turbidity
water was prepared by treating water
from the Horsetooth Reservoir with dia-
tomaceous earth filtration  to  remove
turbidity-causing particles  without
changing  chemical  quality.  The  raw
water was  stored  in  a  1400-L,
temperature-controlled milk cooler and
then pumped by a positive displacement
pump (with dampenertocontrol pressure
surges) to three rapid-mix basins in ser-
ies.-Each  basin was a  12.7-cm cube
(inside dimensions). The stirring paddles
had four  rectangular  blades 1.25 cm
wide, 1.25 cm high, and 2.54 cm from the
center of stirring shaft to outer edge  of
blade. The maximum rotational speed of
the shaft was 600 rpm, yielding a calcu-
lated G value of 400 sec' at 20°C. The
rotational  speed  could  be varied by
means of a rheostat control I ing the motor
speed. Chemicals were metered by posi-
tive  displacement pumps  capable  of
metering flows  as low as 0.2 ml/min.
Flows  were  measured  volumetrically
using 50-ml graduated burettes, which
served also for chemical storage.
  Four filter columns 183 cm long were
installed from a manifold that permitted
one to four filters to be operated in any
combination. Two of the filters were 5 cm
in diameter, and two were 10 cm. Copper
coils were installed in the top of the filters
for temperature control. Tailwater eleva-
tion was controlled  by overflow  cups,
which were maintained above the media.
Headloss across the media was  mea-
sured by  a  mercury  manometer. Air
scrubbing and backwash were provided.
The experiments used both a single sand
medium (76 cm deep) and dual media of
anthracite and sand (45 cm and 30 cm
deep, respectively).
  The field-scale pilot plant was a 1.3-
L/sec (20-gpm) trailer-mounted package
water treatment plant. During periods of
low-turbidity water, this unit was located
adjacent to the Cache La Poudre River at
Fort Collins Water Treatment Plant No. 1.
  Both the laboratory-scale and the field-
scale pilot plants could be operated in
three modes of filtration: conventional
(rapid mix, flocculation, sedimentation,
filtration), direct (rapid mix, flocculation,
filtration), and in-line (rapid mix, filtra-
tion). The in-line mode was used for the
research,  except  for  the  beginning
exploratory work to ascertain the effect of
the filtration mode.


Experimental Design
  The  purpose of the laboratory-scale
pilot plant was to ascertain the effect of
selected process variables on a group of
dependent variables. The purpose of the
field-scale pilot plant was to conduct con-
firming tests for several of these varia-
bles.  The process  variables examined
included the  effects  of  temperature,
coagulant types and dosages, hydraulic
loading rate, and  media on removals of
Giardia cysts, total coliform bacteria, and
turbidity, with some measurements of
removals of standard plate count bacteria
and particles. Field-scale tests were con-
ducted  under   low-turbidity  low-
temperature river conditions to verify the
effects of no coagulation, nonoptimum
coagulation, and  optimum coagulation
on  removals of turbidity, total coliform
bacteria, and Giardia cysts.

Giardia Cyst Sampling and
Analysis
  Giardia cysts  in  the filter effluents
were  concentrated  by  polycarbonate
membrane filters with 5: m pore sizes.
The entire flow from the laboratory-scale,
rapid-rate filter column  was passed
through a 142-mm-diameter membrane
filter. To  sample the  field-scale  pilot
plant,  a portion  of the effluent  was
passed  through  a  293-mm-diameter
membrane filter.  The  total volume of
water passed through the membrane fil-
ter  was limited by the pressure increase
across the filter. Sampling was termi-
nated when the pressure became about 5
psi.  Usually  about  20  L was passed
through the 142-mm  membrane filter,
but 140 L was passed  in one test. When
the sampling was ended, the membrane
filter holder was removed from the filter
effluent line, and the membrane was

-------
removed and washed to remove cysts and
anything else that had collected on the
membrane. The sample was then sent to
a  laboratory, where the  cysts  were
counted using a micropipette technique.
  To determine removals of Giardia cysts
under a given set of filtration conditions,
the milk cooler (used as a reservoir for the
laboratory-scale  pilot  plant) was spiked
with several million cysts (determined to
be  Giardia lamblia) obtained from the
feces of infected dogs. To spike the field-
scale pilot plant, a cyst concentrate was
metered into the influent pipe where it
was mixed as the result of the turbulent
flow through four elbows.
  To  determine the  concentration  of
cysts in the milk cooler, a sample stream
was pumped through the 5-yum, 142-mm
polycarbonate filter. For the field-scale
pilot plant, the  influent cyst concentra-
tion was sampled by pumping a portion of
the flow (after the four mixing  elbows)
through the 293-mm membrane filter.
Sampling and Analysis of
Turbidity, Bacteria, and Particles
  Sampling  for  turbidity, bacteria, and
particles was done with grab samples.
For the laboratory-scale pilot plant,  these
samples were taken from water in the
milk cooler, after spiking, and from the
filter  effluent stream  after 30 min, 60
min, and (for some tests) 2 hr of opera-
tion. Cuvettes were used for the turbidity
samples, and sterile 250-mL plastic bot-
tles were used for bacteria sampling. For
particle sampling, 500-mL bottles were
used. They were cleaned and rinsed with
distilled water passed through  0.2-fjm
filters. For the field-scale pilot plant, sam-
ples were taken  after the mixing elbows
for the influent stream and from  the dis-
charge line for the filtered water.

Results
Removals
  Table 1  summarizes the removals of
turbidity, standard plate count bacteria,
total  coliform bacteria, particles, and
Giardia cysts for low-turbidity water at
low temperatures (2° to 4°C). Results of
21 test runs are shown for three catego-
ries of chemical  pretreatment: (1)  none,
in which no chemicals were used, (2)
nonoptimum, in which a  nonoptimum
chemical dose was used as determined
by turbidity removal, and (3) optimum, in
which an  optimum  chemical dose was
used as measured by  turbidity removal.
Raw water turbidity levels were0.4to 0.7
NTU for water hauled from the Cache La
Poudre River for the pilot plant tests, and
0.2 to 0.6 NTU for the water obtained by
diatomaceous earth filtration of Horse-
tooth Reservoir water (referred to in this
report as artificial low turbidity water).
  The effects of chemical pretreatment
can be seen in Table 1. For the eight tests
with no coagulant dosage, removals of all
parameters except particles were uni-
formly low. No explanation exists for the
higher percent removals of particles. For
the nonoptimum chemical dose, the per-
cent removals were generally higher but
not  uniformly  high.  For the  optimum
chemical dosage, percent removals of all
parameters were uniformly high, ranging
from about 80 to greater than 99.9. The
exception.  Run 106,  was for a  polymer
used  commonly as a filter aid for low-
turbidity waters. The result is  added  to
the table to illustrate that high removals
cannot be expected for some polymers.
For the optimum chemical dosage, fil-
tered water turbidity was generally about
0.05 NTU: the percent removals ranged
from 82 to 93.
  The field-scale pilot plant runs (117
and  129)  in Table  2  shows  results
obtained for tests with no coagulant dos-
age in which Giardia cysts and coliform
bacteria were injected into low-turbidity,
low-temperature raw water. Without a
coagulant,  coliform  bacteria removals
were 20 and 15 percent, respectively.
The Giardia cyst removal of Run 117 was
only 30 percent. No Giardia removal data
are reported  for Run 129 because the
cysts had  questionable identities for
analysis.  The  effluent turbidity  was
greater than  the influent turbidity for
each of these runs without chemicals.
  Runs  123  to 128 were classified as
nonoptimum.  Results for removals of tur-
bidity, total coliform bacteria, and Giardia
cysts were not significantly different than
those for tests without coagulant dosage.
For example, in Runs 123 and 124, remo-
vals of Giardia cysts were 45 and 40 per-
cent,  respectively;  coliform  removals
were  20  and 50 percent,  and  turbidity
removals were  less than 1 percent. The
coagulant aid commonly  used  in the
Rocky Mountain Region was simply not
appropriate for the low-turbidity waters.
  Run 138 was classified as optimum.
Removals in  all three categories were
high (i.e., 95 percent for Giardia cysts, 98
percent for coliform bacteria, and 42 per-
cent for turbidity). Coagulant chemicals
used for Run 138 were 7.0 mg/L of alum
as AI2(SO4)314H20 and 2 mg/Lof Mag-
nifloc 572C®.*  These  coagulant
* Mention of trade names or commercial products
  does not constitute endorsement or recommen-
  dation for use.
 dosages and chemicals were found to be
 effective in bench-scale and laboratory-
 scale testing. For this test, the raw water
 turbidity was 0.7 NTU, the effluent tur-
 bidity was 0.4  NTU, and the  water
 temperature was<1°C. Most important
 was the selection of the polymer used as
 a coagulant aid with alum.
  The data in Tables 1 and 2 show that
 with proper chemical pretreatment, rem-
 ovals  of turbidity, standard plate count
 bacteria, total coliform bacteria, parti-
 cles,  and Giardia  cysts were uniformly
 high—generally greater than 80 percent
 for turbidity and  98 percent for other
 parameters. Run 138 in Table 2 for field-
 scale testing showed a nominal deviation
 from this general finding, having only 42
 percent turbidity removal but 95 percent
 removal of Giardia cysts and 98 percent
 removal of total coliforms.  For  nonopti-
 mum  chemical  dosages, results were
 more variable, with both high  and low
 removals.  With  no chemical  pretreat-
 ment, removals were markedly lower for
 all  parameters except particles, which
 ranges from 81 to 99 percent. Using an
 effective polymer  is important  also,  as
 demonstrated by the results of Runs 127
 and 128, which  used some of the poly-
 mers found to be inefficient in filtration.
 These results show that  with proper
 chemical pretreatment, rapid-rate filtra-
 tion will generally remove 95 to 99.9 per-
 cent of Giardia cysts.
  The data in Tables 1 and 2 show that
 rapid-rate filtration will work only as a
 simple strainer when no chemicals are
 used  and will  pass appreciable percen-
 tages of turbidity,  bacteria, and Giardia
 cysts.  The  critical  importance of proper
 chemical coagulation is  demonstrated.
 From these results, little doubt exists that
 the rapid-rate  filtration process can be
 effective if the  proper chemicals are
 selected and if they are used at proper
 dosages.

 Effects of Process Variables
  The process  variables  investigated
 were chemical pretreatment (coagulant
 selection, coagulant dosages, and mode
 of filtration), comparison of single and
 dual media, hydraulic loading rate, and
 temperature. The effects of these varia-
 bles on turbidity removal was the main
 focus because it  indicated  removals  of
 both bacteria and Giardia cysts.

 Coagulant Selection
  Alum alone was not effective as a
chemical  coagulant  for  low-turbidity
waters unless a  high dosage was used
(e.g., 15to50mg/LasAI2(S04)314H20).

-------
 Table 1.    Effect of Chemical Pretreatment on Removal of Turbidity, Standard Plate Count Bacteria, Total Coliform Bacteria, Particles, and Giardia
            Cysts from Water with Artificial Low-Turbidity, and Low-Temperature Cache La Poudre River Water. Subject to In-Line, Laboratory-
            Scale Rapid-Rate Filtration*'^
Conditions
Filter Raw Water
Run
No..
46
47
49
48
119
120
121
122
69
82
114
50
51
52
53
70
81
104b
106
107b
118
V
Icm/minft
8.46
22.59
22.20
8.26
41.40
32.00
20.70
9.60
22.69
22.45
7.8
8.20
23.48
8.45
23.19
22.20
8.35
8.26
8.47
8.38
9.37
Media§
SandIL)
DuaKLI
Sand(L)
DuaKLi
DualfF)
Dual(F)
DuaKF)
DuaKF)
Dual(L)
Dual(L)
Dual(F)
SandIL)
DuaKU
DuallL)
SandIL)
DuallL)
DuallL)
DuaKF)
DuaKF)
DuaKF)
DuaKF)
Source"
HDE
HDE
HDE
HDE
CLP
CLP
CLP
CLP
HDE
HDE
CLP
HDE
HDE
HDE
HDE
HDE
HDE
CLP
CLP
CLP
CLP
Temp.
ret
3.0
3.0
2.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
4.0
3.0
4.0
4.0
3.0
3.0
3.5
3.5
3.0
3.0
Pretreatment
(Chemicals Used)
Dosage
Category^
None
None
None
None
None
None
None
None
Nonoptimum
Nonoptimum
Nonoptimum
Optimum
Optimum
Optimum
Optimum
Optimum
Optimum
Optimum
Optimum
Optimum
Optimum
Species
None
None
None
None
None
None
None
None
alum/573c
alum/572c
alum/572c
alum/572c
alum/572c
alum/572c
alum/572c
alum/573c
alum/572c
alum/572c
8102N
alum/572c
alum/572c
Dosage
99.9
83.0
>99.9
>99.9
99.5
99.9
79.8
>99.9
90.0
99.4
Particle
Count
86
99
99
94
90
86
82
82
-142.4
99.6
58.9
98.6
98.9
99.2
93.8
81.9
98.3
98.6
87.0
95.4
tt
Giardia
Cysts
7.6
96.3
>99.9
99.9
41.9
36.4
36.3
68.3
99.2
tt
95.3
97.8
99.1
99.7
99.5
99.4
tt
98.7
39.5
>99.9
97.6
 "Artificial water was obtained by diatomaceous earth filtration of Horsetooth Reservoir water; filtered water turbidity ranged from 0.2 to 0.6 NTU.

 f The term "in-line" filtration is the designation for treatment train comprising rapid mix and filtration (no flocculation or sedimentation).

 $ The term "V" designates a hydraulic loading rate, which equals flow divided by area of filter.

 §Sand /L) means the media was all sand and was obtained from Love/and Treatment Plant at Big Thompson Canyon. Bed depth was 76 cm. Dual (L) means the bed
  comprised 30 cm sand from Loveland and 45 cm anthracite having the trade name Philterkal Special No. 1 (produced by Reading Anthracite Coal Company. Pottsville,
  PA 17901). Dual (F) means that the bed contained 30 cm of sand obtained from Port Collins Treatment Plant No. 2 and 45 cm of Philterkal Special No. 1 (R) anthracite.

* * HDE is water obtained from Horsetooth Reservoir, filtered by diatomaceous earth to give low turbidity (0.2 to 0.6 N TU). CLP is low turbidity raw water obtained from the
  Cache La Poudre River during the period September to April when raw water turbidity  was generally 0.4 to 0.7 NTU.

ft "Optimum" and "none optimum" are  designations of coagulant dosages producing turbidities of filtered water that are minimum and greater than minimum.
  respectively.

tt No sample taken.
Furthermore, the polymers tested were
not  effective when  used  alone.  Thus
attention was focused on selection of a
polymer that could be an effective coagu-
lant aid when used with alum. To screen
polymers and determine dosages, turbid-
ity reduction was used as the measure of
effectiveness.   Turbidity  should  be
reduced from about 0.5 NTU in raw water
to 0.1 or 0.05 NTU in filtered  water. The
search for an effective coagulant aid was
wholly trial and error. The idea was to test
different polymers as coagulant aids and
then to stop when an effective one was
found. Nine polymers were tested using
the  laboratory-scale pilot plant. Two of
these, Magnifloc  572C® and Magnifloc
573C®, were determined to be effective
as measured by turbidity removal. Some
recommended polymers found effective
elsewhere by others were  not effective
for   the  low-turbidity raw waters.  All
chemicals were added to mixing basins
except 8102®, which was  injected into
the  pipeline  ahead  of the filter.  This
procedure simulated the practice of using
the polymer as a filter aid. The effect of no
chemical addition (i.e., using the filter as
a strainer) should be noted. Without the
use of coagulant chemicals, or with the
use of unsuitable coagulant chemicals,
removals of turbidity are erratic.
  The data showed that a combination of
alum and 572C® or 573C® will remove
85  percent or more of the Giardia cysts.
Many of the data show removals greater
than 99.9 percent. Without chemicals,
removals are  likely to be in the 0 to 50
percent  range, which  corroborates  the
turbidity removal results for runs with no
chemicals.

Dosages of Coagulants
   Data obtained showing filtered water
turbidity as a function of alum and 572C®
dosages showed that when alum is used
with Magnifloc 572C® as a coagulant aid,
filtered  water turbidity  levels  can be
reduced to 0.05 NTU nominally (as com-
pared with nominal raw water turbidity of
about  0.5 NTU). The response  surface
does not seem to be strongly sensitive to
either  alum  or  polymer  dosage, but  it
does show that either alum alone or poly-
mer alone is not effective.  The polymer
dosage used most often in  the research
was 1  to 2 mg/L, with alum dosages of 3
to 7 mg/L as AI2(SO4)3  14 HjO. When
the dose of an effective type of polymer
exceeded  1  mg/L and the alum dose
ranged between 3 and  15 mg/L, Giardia
removals were generally  high.

Mode of Filtration
  Coagulation  and  flocculation of the
low-turbidity waters did not produce any

-------
 Table 2.    Turbidity, Giardia. and Coliform Results of Using Low-Turbidity Raw Water* for Field-Scale. Rapid-Rate Filtration Pilot Plant
                       Coagulants Used
             Turbidity
Giardia Cysf+V
Run
Ho.
117
129
123
124
125
126
127
128
138
Coagulant
Dosage
Category
None
None
Nonoptimum
Nonoptimum
Nonoptimum
Nonoptimum
Nonoptimum
Nonoptimum
Optimum
Chemical
Species!
None
None
8102
8102
Alum
Alum
Alum/8102
Alum/8102
Alum/572-C
Chemical
Dose
0
0
0.1
0.4
0.4
5.0
3.0/0.2
3.0/0.4
7.0/2.0
Water
Temp.
2
1
1
1
1
1
1
1
<1
Influent" l
(NTU)
0.4
0.6
0.6
0.6
0.55
0.55
0.9
0.9
0.7
•ffluent*
ITU)
0.6
0.7
1.1
0.85
1.0
1.0
1.1
0.9
0.4
^ Percent Influent®
Removal cysts/L
<1 260
<1 Q"*
<1 325
<1 325
98
   * Cache La Poudre River water having raw water turbidities less than 1 NTU.
   + Nalco 8102, Magnifloc S72-C.
   J Alum doses are mg/L as AI2 (S04)3 14H, O.
  " Influent turbidity before contaminant injection.
  ++ Effluent turbidity after 1 hr of filtration.
  §§ Detected cyst concentrations, sampling influent stream after mixing by four elbows and before injection of coagulants. Membrane filters used were Nucleopore
     polycarbonate 5-micrometer pore size, 293-mm diameter. Samples were analyzed by micropipette technique.
  *** Procedures were the same as used for influent sampling and analysis.
  +++Q indicates cysts were of questionable viability; ND indicates no data, missed dilution range.
 visible floe unless  high  alum dosages
 were used. Therefore conventional filtra-
 tion using sedimentation was not used.
 The experimental work began by using
 direct filtration (rapid mix, flocculation,
 and filtration). Shortly thereafter, the in-
 line filtration mode, which included rapid
 mix followed by filtration, was also tried.
 Two comparisons of in-line and direct fil-
 tration, with all conditions the same for
 each,  resulted in filtered water turbidity
 levels of about 0.1 NTU for each. Based
 on these data, all further test runs were
 performed  using the in-line mode of
 filtration.

 Media

  Three test  runs  were conducted to
 compare filtered water turbidity levels for
 the same conditions with single medium
 (76 cm sand) and dual media (30 cm sand
 and 45 cm anthracite). The first compari-
 son was for runs with no chemical pre-
 treatment. With raw water turbidities of
 0.5 NTU, filtered water turbidities were
 0.4 NTU for both the single and dual
 media. Headloss was 92 cm of water for
 the single media and 54 cm of water for
 the dual media after 50 min of operation.
 Water temperatures were 2° to 4°C. The
 second comparison was conducted using
 alum and 573C® chemical pretreatment
 at optimum dosages with respect to tur-
 bidity removal. Effluent turbidities were
 0.04 NTU for both, and again, headless
 was higher for the single media. Based
on these results, the dual media was pre-
ferred  because of lower headloss.
 Hydraulic Loading Rate

   Tests  conducted  to  determine  the
 effect of hydraulic loading rate on remo-
 vals  of turbidity, standard plate count
 bacteria,  total  coliform  bacteria,  and
 Giardia cysts at optimum chemical dos-
 ages showed little influence on removals
 of  total  coliform  bacteria. Even at 25
 m/hr (10 gpm/ft2), the removal is 99 per-
 cent. Similar influences are seen for re-
 movals of standard plate count bacteria
 and  Giardia  cysts.  The  influence  is
 stronger, however, for removals of tur-
 bidity. Another series of tests  was con-
 ducted under nearly the same conditions,
 except the sand  in the dual media was
 obtained from the Loveland, Colorado,
 Water Treatment Plant. The trends were
 virtually the same. These results indicate
 that the hydraulic loading rate has little
 influence on percent removals of Giardia
 cysts and bacteria in  the range of 4.9 to
 19 m/hr (2 to 8 gpm/ft2). An influence
 begins to become discernible, however,
 at  25 m/hr (10 gpm/ft2). Pilot testing
 should be conducted to ascertain  the
 influence of hydraulic loading rate for the
 conditions  at  hand  in  a  particular
 situation.

Temperature

  The removals  of  turbidity,  standard
 plate count bacteria,  and total coliform
 bacteria measured at operating tempera-
tures of 5° and 18°C,  for four conditions
of chemical pretreatment showed either
no difference for the two temperatures or
conflicting trends. If there is an influence,
        it does not seem to be great. The data
        showed no conclusive trends.

        Surrogate Indicators for Giardia
        Cyst Removal

          Sampling of raw or filtered water to
        recover Giardia cysts requires passing a
        large volume of water through a mem-
        brane filter or through a fiberwound filter
        with a  pore size small  enough to strain
        the cysts. Whatever cysts were in the
        sample will be retained by the filter if its
        pores are smaller than the cysts. Once
        the sample is obtained, measurement of
        Giardia  cyst  concentration  requires
        skilled technique to process, identify, and
        count the cysts. Routine measurement of
        Giardia cysts is not likely to be incorpo-
        rated into water treatment practice. Thus
        a  surrogate indicator  for  removals  of
        Giardia cysts is desirable.  Several were
        investigated, including  turbidity,  parti-
        cles, standard plate count  bacteria, and
        total coliform bacteria. Histogram plots
        were developed to relate  removals  of
        Giardia cysts with  removals of turbidity
        and also  removals  of Giardia  cysts with
        removals of total coliform bacteria. Sim-
        ilar  histograms could have been  deve-
        loped using particles or standard plate
        count bacteria, but for the sake of brevity,
        this was not done. Also the use of particle
        counting was not continued throughout
        the project since considerable effort was
        required. Considering the effort and qual-
        ity of the relationships obtained, turbidity
        and total coliform bacteria were the most
        useful surrogate parameters.

-------
  For example, one of the  histograms
showed that if turbidity removal is high,
removal of Giardia cysts wilJ be high also.
Specifically, the plot showed 44 observa-
tions when turbidity removal was greater
than 70 percent  for low turbidity raw
water. Of  these  44 observations, 37
show removals of Giardia cysts exceed-
ing 99 percent. In other words, if turbidity
removal exceeds  70 percent and if fil-
tered water turbidity is lower than 0.10
NTU, the probability is 0.85 (37/44) that
removals of Giardia cysts would equal or
exceed 99 percent.
  A similar histogram  was constructed
using coliform bacteria as the surrogate.
The histogram showed that if high remo-
vals of total coliform bacteria occur by the
filtration process,  then  high removals of
Giardia  cysts can  also  be  expected.
Though  the  histogram  indicated  that
removals of total coliform bacteria would
be a good indicator for removals of Giar-
dia cysts, this use may not be practical for
water treatment plants using mountain
streams as a source of supply. In such
streams, concentrations of total coliform
bacteria are usually  less than 100 organ-
isms/100 ml. To evaluate filtration per-
formance, a pilot filter should be operated
alongside the full-scale filter and spiked
with raw sewage. Turbidity, on the other
hand, is easy to measure.
  To summarize,  removals of turbidity
could be used to monitor plant perfor-
mance  routinely.  Periodic evaluations
could be done by spiking a pilot column
with coliform bacteria. Both  are recom-
mended  when  filtering low-turbidity
waters.

Conclusions
  This research shows that proper chem-
ical  pretreatment is  imperative if  the
rapid-rate filtration process is to be effec-
tive  when using low-turbidity waters.
Most important is  selection of proper
coagulant polymers to use with the pri-
mary coagulant, such as alum. The range
of dosages must also be proper to achieve
high reductions of turbidity. With proper
chemical  pretreatment,  removal of  all
parameters can be expected to exceed 70
percent for turbidity, 99 percent for bac-
teria, and 95 percent for Giardia cysts.
With no chemical pretreatment, removal
of Giardia  cysts,  bacteria, and  turbidity
can be  expected  to range from 0 to 50
percent. The turbidity rule of thumb of 70
percent removal pertains to low turbidity
waters, nominally about 0.5 NTU for the
raw water. If raw water turbidity levels
are greater than  10 NTU, the rule does
not apply since the turbidity reductions
should be sufficient to meet standards. If
the raw water turbidity is about 0.1 NTU,
it would be very difficult to use this rule of
thumb as the turbidity reduction may not
be easily detectable.
  The  roles of other process  variables
were not as important as chemical pre-
treatment. In-line filtration was as effec-
tive  as direct filtration.  Singlemedium
(sand)  and dual media (anthracite and
sand) both have the same efficiencies in
reducing turbidity and bacteria.  Hydraulic
loading rate has very little effect on remo-
vals when it ranges between  5 and 19
m/hr (2 and 8 gpm/ft2). At 25  m/hr (10
gpm/ft2), a moderate effect is  indicated.
Investigation of temperature  influence
showed no trend in removals of turbidity,
bacteria, and Giardia cysts at 5°C com-
pared with  removals at  18°C. Further
work is recommended in this area.
  Analysis of data  by means  of  histo-
grams showed that both removals  of tur-
bidity and total coliform  bacteria could
serve as surrogate indicators of removals
of Giardia cysts. If percent removal of tur-
bidity is 70 or greater, for example, reduc-
ing turbidity from say  0.5 NTU to 0.15
NTU, the probability is 0.85 that removal
of Giardia cysts will exceed  99 percent.
Pilot filter columns spiked with coliform
bacteria are recommended for use along-
side the full-scale filters to evaluate prop-
erly filtration of low-turbidity waters.
  The full report was submitted in  fulfill-
ment of EPA Cooperative Agreement No.
CR808650-02 by Colorado State Univer-
sity under sponsorship of the U.S. Envi-
ronmental Protection Agency.
                                                                           U. S. GOVERNMENT PRINTING OFFICE:1985/559-l 11/20685

-------
     Mohammed A I-A ni is with the Water Board of the Scientific Research Council,
       Bagdad. Iraq; John M. McElroy is with CH2M-HHI Consulting Engineers,
       Bellevue, WA 98009-2050; David W. Hendricks and Charles P. Hibler are with
       Colorado State University, Fort Collins, CO 80523.
     Gary S. Logsdon is the EPA Project Officer (see below).
     The complete report, entitled "Filtration of Giardia Cysts and Other Substances:
       Volume 3. Rapid-Rate Filtration," (Order No. PB85-194 645/A S; Cost: $25.00,
       subject to change) will be available only from:
             National Technical Information Service
             5285 Port Royal Road
             Springfield, VA 22161
             Telephone: 703-487-4650
     The EPA Project Officer can be contacted at:
             Water Engineering Research Laboratory
             U.S. Environmental Protection Agency
             Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300

EPA/600/S2-85/027
            0000329    PS
                                            AGENCy
                                            60604

-------