United States
Environmental Protection
Agency
Municipal Environmental
Research Laboratory
Cincinnati OH 45268
Research and Development
EPA-600/2-84-068 May 1984
SEPA Project Summary
Removal of Trihalomethane
Precursors by Direct Filtration and
Conventional Treatment
James K. Edzwald
Removal of trihalomethane (THM)
precursors from drinking water was
studied in direct filtration and conven-
tional treatment plants. The research
had two objectives: (1) to investigate
direct filtration as a water treatment
process using cationic polymers as the
sole coagulant, and (2) to evaluate two
existing conventional water treatment
plants for THM formation through the
plants and for removals of THM precur-
sors. Direct filtration was evaluated by
means of traditional measures of filter
performance (head loss development
and filtered water turbidity and color)
and by removals of nonpurgeable total
organic carbon (NPTOC) and THM pre-
cursors. Conventional treatment was
studied in two full-scale plants through
evaluation of NPTOC and THM precur-
sor removals.
Two different water sources were
studied: the Grasse River, a highly col-
ored supply, and the Glenmore Reser-
voir, a low turbidity, protected upland
supply. Pilot plant studies showed that
direct filtration with cationic polymers
is an effective treatment process. Ex-
cellent performance was generally
achieved with filtered water turbidities
less than 0.3 nephelometric turbidity
unit (ntu) and apparent color less than
15 Pt-Co units. Though the two water
supplies are different in type and in raw
water NPTOC and THM precursor con-
centrations, similar removals were
achieved by direct filtration: 37 to 39 per-
cent for NPTOC and 41 to 42 percent for
THM precursors.
The two full-scale conventional water
treatment plants both use alum as a
primary coagulant and achieve greater
removals of organic matter and THM
precursors than direct filtration. The
removals were similar for both plants,
with 69 to 71 percent for NPTOC and 68
to 72 percent for THM precursors.
This study also demonstrated that UV
(254 nm) absorbance may be used as a
surrogate parameter for predicting raw
water NPTOC and THM precursors and
for monitoring treatment performance.
This Project Summary was developed
by EPA's Municipal Environmental Re-
search Laboratory, Cincinnati, OH, to
announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
Interest in direct filtration was stimulated
following passage of the Safe Drinking
Water Act in 1974 and the issuance of the
National Interim Primary Drinking Water
Regulations (NIPDWR). A turbidity standard
of 1 ntu was set. Many communities have
used low turbidity supplies and have pro-
vided no treatment other than disinfection
and pH control. They are now faced with
providing additional treatment to meet the
turbidity standard. In addition to concerns
about turbidity, it was found in the 1970's
that trihalomethanes (THM's) are produced
within water treatment plants as byproducts
of chlorination. A variety of organic com-
pounds (THM precursors) react with chlorine
to form trihalomethanes. The precursors
may exist in dissolved, colloidal, or par-
ticulate forms in water supplies. Humic
substances (natural color) constitute the
major fraction of organic matter and THM
precursors in most water supplies. In 1979,
-------
EPA amended the NIPDWR by establishing
a maximum contaminant level (MCL) of 0.10
mg/L for total trihalomethanes (TTHM's).
Several methods exist for reducing THM
formation and removing THM precursors
from water supplies. For example, low tur-
bidity humic (colored) waters are normally
treated by coagulation with alum in conven-
tional treatment plants employing coagu-
lation-flocculation, sedimentation, and
filtration processes. One method of control-
ling THM's is improved coagulation and
operation of the conventional plant. Direct
filtration, however, can offer savings in
capital costs and operating costs when com-
pared with conventional treatment. Since
most direct filtration plants have been
designed and operated for turbidity removal,
direct filtration must be evaluated for the
removals of organic matter and THM precur-
sors when used in the treatment of low tur-
bidity humic waters.
This research project had two major ob-
jectives. First, direct filtration was in-
vestigated as a water treatment process
using cationic polymers as the sole coagu-
lant. In some experiments, alum and polymer
filter aids were also examined. Direct filtra-
tion was evaluated for the removal of turbidi-
ty and color (traditional parameters in
assessing filter performance) and for
removals of TOC and THM precursors. The
second objective was to evaluate two ex-
isting conventional water treatment plants
for THM formation through the plants and
for removals of THM precursors.
Procedures
Water Supplies
Two water supplies were used in this
research — the Grasse River at Canton, New
York, and the Glenmore Reservoir at Oneida,
New York. The Grasse River was selected
for this research project because it is a low
turbidity, highly colored water typical of
many raw water supplies found in certain
regions of the United States (e.g.. Central
and Northern New York, New England, and
the Pacific Northwest). Another important
feature of the Grasse River site is that no
municipal or industrial wastewaters are being
discharged to the river upstream. Conse-
quently, the organic material in the river is
naturally occurring and high in humic mat-
ter, giving it the characteristic yellow-brown
color of colored waters.
The Glenmore Reservoir was chosen for
this study for the following reasons. First,
it is a protected upland water supply of low
turbidity. This supply therefore represents
many water supplies in the United States and
provides a good model for this research.
Second, the nature and concentration of
organic matter in the reservoir would differ
from that in the Grasse River and provide a
contrasting system for study. Finally, the
concentration of organic matter in Glenmore
Reservoir is much lower than that in the
Grasse River. In this research, then, two
water supplies were used in which the or-
ganic matter and THM precursors are natu-
rally occurring.
Raw water samples were collected peri-
odically from the Grasse River and Glenmore
Reservoir. The samples were characterized
as to their concentrations of organic matter
as measured by NPTOC, total trihalo-
methane formation potential (TTHMFP), and
UV (254 nm) absorbance; seasonal changes
in these parameters were also identified.
THM precursor concentrations were mea-
sured by determining the 7-day TTHMFP at
pH 7.5 and 20°C. For the Grasse River,
samples were collected on 57 dates between
February 1980 and March 1982. The Glen-
more Reservoir was sampled on 44 dates
between February and June 1982.
Conventional Water Treatment
Evaluation
Existing full-scale plants at Canton and
Oneida were monitored during this study to
evaluate removals of NPTOC and THM
precursors and to investigate THM forma-
tion through the plants. Monitoring of the
Canton plant took place 16 times between
February 1980 and March 1982. The Canton
plant normally operates at 11.4 million L/day
(3 mgd), and is a high rate conventional
water plant employing the processes of
coagulation-flocculation, sedimentation
(tube settlers), and filtration (mixed media
filters). Alum is used for coagulation, and a
nonionic polymer is used as a coagulant aid.
Prechlorination is normally practiced but at
low doses of 1 to 2 mg/L. For postchlorina-
tion, the chlorine is applied at a point be-
tween the filters and the clean/veil.
The Oneida plant has a rated process
capacity of 15.1 million L/day (4 mgd). This
plant was monitored on 16 occasions be-
tween April 1980 and May 1982. Oneida has
a conventional plant using coagulation-
flocculation, sedimentation (rectangular set-
tling basins), and filtration (dual media
filters). Alum is used for coagulation, and a
nonionic polymer is used as a filter aid.
Chlorine is added to the top of the filters and
again after filtration. During the course of
this study, Fe and Mn problems associated
with the raw water were treated by the ad-
dition of chlorine (prechlorination), chlorine
dioxide, and potassium permanganate.
Beginning in the spring of 1981 and for the
remainder of the study, potassium per-
manganate was used when needed.
Samples were collected at several loca-
tions throughout the plants — raw, settled,
filtered, and finished water (pumped water
from the clean/veil to the distribution
system). The analyses included temperature,
pH, turbidity, apparent color, true color, total
UV (254 nm) absorbance (unfiltered sam-
ples), soluble UV (254 nm) absorbance
(filtered samples), NPTOC, instantaneous
TTHM (Inst TTHM), TTHMFP, terminal
TTHM (Term TTHM), and free residual CI2.
For the raw waters, the 7-day TTHMFP's
(pH 7.5, 20°C) were used as the measure of
raw water THM precursor concentration.
Within the water plants. Term TTHM sam-
ples were buffered at pH 7.5 and spiked with
chlorine to maintain a residual for 7 days —
that is, Term TTHM's (7-day, pH 7.5, 20°C).
THM precursor removal through the plants
is the difference between the raw water
7-day TTHMFP and the finished water 7-day
Term TTHM.
Direct Filtration
Small-scale and large-scale direct filtration
pilot plant studies were conducted at the
Grasse River in Canton and at the Glenmore
Reservoir near Oneida. The small-scale
studies were conducted using columns of
4.45 cm ID. The large-scale studies used a
Waterboy* pilot plant with a filter area of
0.372 m2 (4 ft2) and was operated at flows
of 0.5 to 1.51 L/s (8 to 24 gpm).
The pilot plant studies were designed to
evaluate the effects of chemical variables
(pH, various polymers, polymer dosages,
and alum) and physical variables (filtration
rate, in-line direct filtration versus direct filtra-
tion with flocculation, and water tempera-
ture) on direct filtration performance. The
studies evaluated traditional measures of
filter performance such as filtered water tur-
bidity, color, and head loss development.
Reductions in UV (254 nm) absorbance,
NPTOC, and THM precursors were also in-
vestigated. THM precursor removal by the
pilot plants was calculated as the difference
between the 7-day TTHMFP's (pH 7.5,
20°C) of the raw and filtered waters.
Results and Conclusions
Characterization of Raw Waters
The Grasse River contains high concen-
trations of organic matter and THM precur-
sors as indicated by the data in Table 1. The
highest concentrations of organic matter and
THM precursors for the Grasse River occur
in the summer, particularly late in the season.
Summer mean values are 0.48 cm"1 UV (254
nm), 9.8 mg/L NPTOC, and 783
Mention of trade names or commercial products does
not constitute endorsement or recommendation for
use.
-------
Table 1. Summary of flaw Water Characteristics (February 1980 - June 19821
Grasse River Glenmore Reservoir
Parameter
Turbidity (ntul
Apparent Color IPt-Co Units)
Total UV (cm-1!
NPTOC (mg/L)
TTHMFP* (»9/L)
Mean
2.7
114
0.35
7.5
576
flange (n)
1.2-6.5 (56)
45-228 (57)
0.113-0.666 (57)
2.75-14.7 (53)
232-1108(54)
Mean
1.3
59
0.16
4.2
304
Range (n)
0.54-3.5 (44)
22-129(43)
0.084-0.309 (44)
2.0-10.6(40)
152-597 (42)
" 7-day, pH 7.5, 20°C.
TTHMFP. Lowest concentrations occur in
the winter when the Grasse River is ice-
covered - 0.18 cm-1 UV (254 nm), 4.2 mg/L
NPTOC, and 309 u.g/L TTHMFP.
This seasonal variation in organic matter
and THM precursors has practical applica-
tions for water supply utilities, consulting
engineering firms, and regulatory agencies
concerned with pilot plant studies or evalua-
tions of existing water treatment plants. For
rivers such as the Grasse, the worst water
quality in terms of organic matter and THM
precursors occurs in the summer, not in the
fall during leaf fall or in the spring during
snowmelt and runoff. Pilot plant studies
should be conducted during the summer,
particularly late in the season, to assess per-
formance under the highest concentrations
of organic matter and to determine co-
agulant requirements.
The Glenmore Reservoir contains lower
concentrations of organic matter and THM
precursors than the Grasse River, as shown
by the data in Table 1 . Like the Grasse River,
the Glenmore Reservoir undergoes a well-
defined seasonal variation. Lowest values
occur in the winter when the Glenmore
Reservoir is ice-covered. Mean winter values
are 0.10 cnrr1 UV (254 nm), 2.9 mg/L
NPTOC, and 200 u.g/1 TTHMFP. Highest
values occur in the summer with mean val-
ues for UV (254 nm), NPTOC, and TTHMFP
of 0.21 cm-1, 6.0 mg/L, and 412 Mg/L,
respectively.
Conventional Water Treatment
Plant Performance
Given the high concentrations of NPTOC
and THM precursors present in the Grasse
River, the Canton Water Filtration Plant
achieves excellent removals of NPTOC and
THM precursors. Average percent removals
for NPTOC, Term TTHM's (THM precur-
sors), and UV (254 nm) absorbance appear
in Table 2. In spite of the relatively high THM
precursor level of the raw water, which
ranged from 232 to 1108 u.g/L, the average
Inst TTHM concentration of the finished
water (16 plant monitorings) was only 37
operated between pH 6 and 7, and occa-
sionally the pH after alum addition is as low
as 5. The excellent performance of the Can-
ton plant in removing THM precursors and
controlling THM formation through the
water plant is due to the effectiveness of the
coagulation process, pH conditions during
treatment, chlorination practice, the short
hydraulic detention time of the plant, and
relatively low water temperatures. Though
prechlorination is practiced, the dosages are
low at 1 to 2 mg/L. The plant has tube set-
tlers so that the total hydraulic detention time
through the plant (excluding the clearwell)
is approximately 1 hr. The prechlorination
practice and the short hydraulic detention
time allow removal of THM precursors
before large concentrations of THM's are
formed.
The Oneida Water Treatment Plant also
achieves excellent removals of NPTOC and
THM precursors, as shown by the data in
Table 2. The Oneida plant uses alum
coagulation (measured pH after sedimenta-
tion is typically 6 to 6.5) with addition of a
nonionic polymer ahead of the filters to im-
prove filtration. Sedimentation is achieved
with conventional rectangular settling basins
so that the total hydraulic time through the
plant (excluding the clearwell) is approxi-
mately 6 hr.
This study showed that although the Glen-
more Reservoir contains lower concentra-
tions of organic matter and THM precursors,
the percent removals (Table 2) are quite
similar to those achieved at Canton. During
the summer stratification period in the Glen-
more Reservoir, reduced iron and manga-
nese in the raw water requires oxidation and
removal by the treatment plant. On two oc-
casions when the plant was using chlorine
at the front end of the facility for oxidation
purposes, the Inst TTHM concentrations of
the finished water were 134 and 148 /^g/L.
The plant later substituted KMn04 for CI2 at
the front end of the plant (a small CI2 dosage
of 1 mg/L or less is added to the top of the
filters), and the average finished water Inst
TTHM concentration for 12 monitorings
dropped to 27.5 pig/L (Table 2).
Direct Filtration
General
For the Grasse River pilot plant studies,
the raw water turbidity and apparent color
ranged from 1 to 6 ntu and 50 to 200 Pt-Co
units, respectively. Good filtration perfor-
mance was achieved using cationic polymers
as the sole coagulant with filtered water tur-
bidity and apparent color of 0.2 ntu and 5
to 10 Pt-Co units, respectively. The pH (from
5.5 to 7.5), filtration rate (4.88 to 14.6 m/hr),
type of filtration (in-line direct versus direct
with flocculation), and water temperature
had little or no effect on filter performance
based on removals of turbidity, color, and
organic matter. Figure 1 shows typical results
for two experiments comparing in-line direct
filtration with direct filtration preceeded by
flocculation. The experiments were con-
ducted at pH 7.4 (no pH adjustment) and a
filtration rate of 4.88 m/hr using Grasse River
water of similar raw water quality. Excellent
filter performance was achieved with low
filtered water apparent color and turbidity.
A major difference exists in head loss
development between in-line direct filtration
and direct filtration with flocculation, as
shown in the bottom plots of Figure 1. This
result will be discussed later.
The pilot plant studies on the Glenmore
Reservoir showed that cationic polymers can
be used successfully to treat a low turbidi-
ty, moderately colored water. With raw
water turbidities of 0.75 to 1.5 ntu and ap-
parent color values of 40 to 75 Pt-Co units,
filtered water turbidities of less than 0.3 ntu
and apparent color of 5 to 15 Pt-Co units
were obtained. Little difference was noted
between the performance of in-line direct
filtration and direct filtration with floccula-
Table 2. Performance Summaries for the Conventional Water Plants
Item
Canton Plant
Oneida Plant
The Canton plant uses alum coagulation
with the addition of a nonionic polymer to
aid floe formation. Normally the plant is
Percent removals
Total UV
NPTOC
TERM TTHM +
Mean Inst TTHM of
finished water tug/L)
80
69
72
37
82
71
68
27.5
For Canton, 16 monitorings between April 1980 and March 1982;
for Oneida, 12 monitorings between December 1980 and May 1982.
+7-day, free C/2 present, pH 7.5, 20°C.
-------
c
S
10
I
• In- Line
Polymer: 573C. 10.5 mg/L
pH: 7.4
Filt. Rate: 4.88 m/hr
C/OP
Period: 18.4 mm.
Polymer: 573C, 9.0 mg/L
pH: 7.4
Filt, Rate: 4.88 m/hr.
CO9P
?
ao
3 =5
"S °
I 5
0 2 4 6 8 10 12 14 16 18 20 22 24
Figure 1. Direct filtration by in-line treatment versus a flocculation period before filtration (raw
water quality conditions: 2.5 ntu turbidity. 108 to 115 Pt-Co units of apparent color,
0.37 to 0.40 cm^ UV(254nm). 7.3to8mg/LNPTOC. and657to674ljig/L TTHMFP).
tion, based on removals of turbidity, color,
and organic matter.
Pilot plant studies of direct filtration were
conducted on both a small and a large scale.
An important practical finding was that the
small-scale pilot plant gave similar results to
the large-scale plant in evaluating various
coagulants and coagulant dosages, filtered
water turbidity and color, and removals of
organic matter and THM precursors.
Polymer Dosage Selection
This study demonstrated that jar tests can
be used to select cationic polymer dosages
for direct filtration. Figure 2 illustrates this
procedure for an experiment with Grasse
River water. The jar test results plotted in the
4
top figure show a sharp drop in true color
at cationic polymer dosages of 2 to 4 mg / L;
above 4 mg/L, little change occurs in the
true color readings. These results indicate
that the required polymer dose for coagula-
tion is 4 mg/L. The bottom figure shows
direct filtration results where the polymer
dose was varied during the course of the ex-
periment. Best results were obtained at 4
mg/L, as the jar tests predicted. Underdos-
ing and overdosing occur in direct filtration
and can be predicted from the jar test data.
For the Grasse River, regression equations
were developed between the required cat-
ionic polymer (Magnifloc 573C) dosage for
direct filtration and either raw water apparent
color or total UV (254 nm) absorbance.
Figure 3 shows the data and regression equa-
tion for polymer dose versus total UV. Since
the UV absorbance is proportional to the
NPTOC concentration, these data indicate
a stoichiometric relation between polymer
dose and concentration of organic matter in
the raw water. These results are highly sig-
nificant and practical. The predictive equa-
tion developed in this study is specific to the
Grasse River, but the principle is generally
applicable. In practice, a water utility could
make in-line spectrophotometric UV mea-
surements of the raw water quality. From a
predictive equation relating polymer dose to
raw water UV (254 nm), the utility could then
select the polymer dosage.
In-Line versus Flocculation
As previously mentioned, the removals of
organic matter, filtered water turbidity, and
color did not differ between in-line direct
filtration and direct filtration with a floccula-
tion period provided by a flocculation basin.
For the Grasse River in particular, substan-
tially less head loss developed when treating
the same water by direct filtration with floc-
culation compared with in-line direct filtra-
tion (see Figure 1). These results are
significant and have design ramifications. In
the treatment of a water like the Grasse
River, direct filtration with flocculation will
give longer filter runs. This fact is explained
by considering that a water with large
amounts of humic matter contains high con-
centrations of submicron particles (the humic
particles). By providing a flocculation period,
through the use of a flocculation basin or
otherwise, the humic particles can aggregate
to larger sizes. The flocculation period
reduces the number and surface area of the
particles retained in the filter by increasing
their size. This process results in less head
loss development and longer filter runs.
A flocculation period is recommended in
direct filtration, since it will lengthen filter
runs, especially for waters containing sub-
micron particles (humics, viruses, asbestos,
etc.). The flocculation tank also provides the
water plant operator with some time to ad-
just chemical (coagulant) dosages for direct
filtration.
Removals of THM Precursors
Average percent removals of organic mat-
ter and THM precursors (TTHMFP) for direct
filtration with a cationic polymer (Magnifloc
573C) and with alum are summarized in
Table 3. For each method of treatment, the
removals of NPTOC and TTHMFP are ap-
proximately the same for both the Grasse
River (a highly colored river) and the Glen-
more Reservoir (a low turbidity, protected
upland reservoir). A comparison of the direct
filtration results with the performance results
-------
o _^
o o
O
-------
I
8
o
Q
20
16
72
- Polymer Dose = -1.2 + 26.7 (Toll UV)
r2 = 0.94
0.2 0.4
UVtcm"1)
0.6
Figure 3. Cationic polymer dose for direct filtration versus total UV absorbance for raw water
from the Grasse River.
Table 3. Average Percent Removals by Direct Filtration
Cationic Polymer
Percent Removals
Alum
Grasse River Glenmore Res. Grasse River G/enmore Res.
Total UV
NPTOC
TTHMFP*
58
37
42
51
39
41
76
57
59
71
56
59
7-day, pH 7.5, 20°C.
UV can be used to monitor direct filtration
pilot plant performance for removals of
NPTOC and TTHMFP. For pilot plant studies
on the Grasse River, r2 values of 0.86 for
NPTOC and 0.90 for TTHMFP were obtained
for the respective predictive equations. For
the Glenmore Reservoir, r2 values were 0.72
for NPTOC and 0.89 for TTHMFP.
Apparent color, total UV (unfiltered
samples), and soluble UV (filtered samples)
are all good surrogates for monitoring the
removals of NPTOC and TTHMFP by the
Canton and Oneida water treatment plants.
The full report was submitted in fulfillment
of Cooperative Agreement No. CR807034 by
Clarkson College of Technology under the
sponsorship of the U.S. Environmental Pro-
tection Agency.
-------
-J
1250-n
1000 —
750 —
ex.
fc 500-
250 —
TTHMFP = 38.2 + 1573.2 (Totl-UVj
r2 = 0.94
n = 54
Grasse R.
\ 1
0.0
1 | 1 1 1 1 | (
0. 1 0.2
' ' I
0.3
1
1 1
1 ' '
0.4
. 1 | 1
0.5
1 1 1 | 1 1
0.6
1 1
0
UV (cm'')
Figure 4. Correlation between TTHMFP (7-day, pH 7.5, 20°C) and total UV (254 nm)
absorbance for Grasse River raw water.
James K. Edzwald is with Clarkson College of Technology, Potsdam, NY 13676.
Gary S. Logsdon is the EPA Project Officer (see below).
The complete report, entitled "Removal of Trihalomethane Precursors by Direct
Filtration and Conventional Treatment." (Order No. PB 84-163 278; Cost:
$20.50, 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:
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
-------
------- |