United States
                    Environmental Protection
                    Agency
Water Engineering Research
Laboratory
Cincinnati OH 45268
                    Research and Development
EPA/600/S2-85/138  Jan. 1986
&ER&          Project  Summary

                    Cost  and  Performance
                    Evaluation  of In-Plant
                    Trihalomethane  Control
                    Techniques

                    J. S. Taylor, D. Thompson. B. R. Snyder, J. Less, and L. Mulford
                     A study was conducted to evaluate
                    the costs and performance of new
                    technology for reducing  trihalometh-
                    anes (THM)  in drinking  water on  a
                    bench-, pilot-, and plant-scale. The four
                    Florida plant sites that were selected for
                    study used highly organic surface or
                    ground water supplies and served popu-
                    lations of fewer than 30,000 or fewer
                    than 10,000. Low-pressure membrane
                    processes (ultrafiltration), polyvalent
                    aluminum chloride (PACI) coagulation,
                    flotation, lime softening succeeded by
                    alum  coagulation,  and conventional
                    lime softening and alum coagulation
                    were investigated for THM reduction at
                    these sites.
                     This Project Summary was developed
                   by EPA's Water Engineering Research
                   Laboratory, Cincinnati. OH.  to  an-
                   nounce key findings of the research
                   project that is fully documented in a
                   separate report of the same title fsee
                   Project Report ordering information at
                   back).

                   Introduction
                     The purpose of this cooperative study
                   was to demonstrate the performance and
                   costs of possible new technologies rel-
                   ative to  conventional technologies for
                   trihalomethane (THM) control in drinking
                   water. The scope of work was to identify
                   new technology for THM control, identify
                   sites  where THM control was needed,
                   and execute in-plant studies with the
                   necessary cost documentation.
  The technologies selected for this proj-
ect were polyvalent aluminum chloride
(PACI) coagulation, sequential treatment
of alum  coagulation preceded by lime
softening, membrane  processes, and
flotation. The selection  of sites was
coordinated with EPA Region  IV and
limited to Florida because the  varying
water supplies and THM problems in that
state allowed more technologies to be
investigated within the project  budget.
Coordination with the Florida  Depart-
ments of Environmental Regulation and
Public Health  resulted in the survey of
more than 50 potential sites. Following
plant visits, four sites were selected: (1)
The Village of Golf, Florida (VOG), a small
lime-softening plant  serving 2000 sea-
sonal residents with water containing
THM's averaging more than 700yug/L; (2)
the Acme Improvement District (AID), a
lime-softening plant serving slightly few-
er than 10,000 residents and averaging
THM's of 380 //g/L (both AID and VOG
are near West Palm Beach, Florida, and
use ground waters);  (3) the Olga water
treatment plant (near Ft. Myers, Florida),
an alum plant that uses Caloosahatchee
River water, has THM's of 700 yug/L, and
serves about  18,000 people; and (4)
Venice, Florida (the  alternative  site), a
water treatment plant that serves 13,000
with a blended water from lime softening
and reverse osmosis (RO) plants.

Plant Optimization
  Optimization of existing plant processes
for THM reduction was conducted at all
four sites to develop comparable cost data

-------
and for the benefit of the plants. Bench-
scale and pilot-plant membrane process-
es were investigated at VOG, AID, and
Olga.  Flotation experiments were con-
ducted at Olga.  Pilot-plant sequential
treatment was done at  AID. Seasonal
PACI/alum coagulation was studied on a
plant-scale at Olga.
  Bench- and plant-scale investigations
for plant optimization were conducted at
three  lime softening plants located  in
VOG, AID, and Venice, Florida. The bench
tests  clearly showed that  the  organic
parameters studied-color, dissolved or-
ganic  carbon (DOC), trihalomethane for-
mation potential (THMFP), and total or-
ganic  halogen  formation  potential
(TOXFP) could be reduced by increasing
softening pH. These plants typically soft-
ened at pH 10.3 to a total hardness (TH) of
100 ±50 mg/L as CaCOs and removed
20% to 30% of the raw  color,  DOC,
THMFP, and TOXFP. The jar tests indi-
cated  an  additional 1% to 2% DOC
removal for every 0.1 pH unit increase in
the reaction  pH from 9.5 to 11.5. Gen-
erally, the THMFP reduction was greater
than  the  DOC  reduction because  CI2
demand was decreased as a result of the
lower DOC.  The jar test studies  also
indicated that 20 mg/L or less of alum
would increase organic  removal 5%  to
10% during softening and would not be
present in the finished water.
  As a result of the bench-scale studies,
plant-scale investigations were  imple-
mented at each of  the lime-softening
plants. Similar  process  changes were
made at the three  sites, including (1)
raising the reaction pH during softening
from 10.3 to 11.0-11.3, (2) reducing the
chlorine dose by eliminating chlorine as a
means of color removal and pH reduction,
and (3)  adding H2SO4 to offset the in-
creased softening pH. At all three sites,
THM's were reduced by nearly half,  as
were  TOX and DOC relative to normal
operations. Generally, 80% of the color
was removed during softening, and hard-
ness was increased 50 to 80 mg/L. THM's
at AID and VOG averaged 240 and 330
fjg/L, respectively, during the plant test,
and were therefore above the maximum
contaminant level (MCL). Venice reduced
their THM's to 85 fjg/L during the plant
test and has adopted a high-pH softening
process for THM control. The additional
cost of  high-pH softening ranged from
$0.02/1000 gal atVenice to $0.22/1000
gal at VOG. The percentage of unit cost
increase for high-pH softening was 2% at
Venice, 19% at AID, and 20% at VOG,
which corresponded  with 32%, 50%, and
45% THM reduction, respectively.
PACI and Alum Treatment
  The PACI and alum investigations were
made on both a bench- and a plant-scale
at the Olga water treatment plant in Lee
County, Florida, for both the wet and dry
seasons. The organic  parameters inves-
tigated were DOC, THM, TOX, and color
as a result of  coagulant dose and pH.
Aluminum  residuals, turbidity, sludge
volumes, and dry-season softening were
also investigated.
  The raw water color averaged about 57
chloroplatinate units (cpu) during the dry
season tests and increased 228% to 130
cpu during the wet season tests. Raw
water DOC averaged  19.6 mg/L during
the dry season  plant tests and increased
9% to 21.4 mg/L during the wet season
tests. The  raw turbidity  averaged 6 NTU
during the wet  season and 4 NTU during
the dry season. The alkalinity and hard-
ness during the wet season were ap-
proximately 140 and 110 mg/L as CaC03,
and they increased to 250 and 160 mg/L
as CaC03 during the dry season. The raw
water used during both plant tests was a
low-turbidity, high-color water that would
normally be treated by sweep coagulation
for color removal  only. Two PACI coag-
ulants—1 DOS,* a  sulfate chloride base,
and 190,  a chloride-based coagulant-
plus alum were compared  for organic
removal.
  Jar testing during  the wet and dry
seasons demonstrated that the optimum
coagulation pH was 5.0 to 6.0 for DOC
and color removal for all coagulants. PACI
100S was superior to PACI  190 for the
removal of DOC and color in jar tests; thus
it was selected for plant testing. DOC
removal increased with  increasing coag-
ulant dose between pH 5.0 and 6.0, but it
appeared to approach a limiting value.
Coagulant doses of 117 mg/L as alum
removed 60% of the DOC and 90% of the
color in the wet and dry season jar test.
Increasing the  coagulant dose to 175
mg/L increased DOC removal 5% to 10%
and had less effect on color removal. The
percentages of color  and DOC  removal
during the wet season were slightly
higher (<5%) than during the dry season
for all coagulants. However, the actual
DOC concentration and, to a less extent,
the color remaining in each season were
approximately  the same at equivalent
conditions for each coagulant. Wet-sea-
son THMFP reduction  was very similar to
DOC removal. The optimum pH range for
 •Mention of trade names or commercial products
 does not constitute endorsement or recommenda-
 tion for use.
THMFP reduction was 5.0 to 6.0 for all
coagulants, and increasing the coagulant
dose increased the THMFP reduction in
all jar tests. Doses of 117 to 175 mg/L as
Alz(S04)3-14H20 realized 90% of the max-
imum THMFP reduction during coagula-
tion.
  In all the jar testing, alum generally
removed slightly more «5%) DOC, color,
and THMFP than PACI. The seasonal jar
tests did not indicate a major difference in
coagulant demand  for organic removal,
possibly because of the small DOC in-
crease during the wet season. Aluminum
residuals after coagulation  with  either
PACI-100S or alum were typically less
than  0.4  mg/L aluminum  and were
controlled by coagulation pH. These levels
would not cause post-precipitation prob-
lems. Removal of DOC and THMFP during
coagulation should exceed 60% and be
accompanied by more  than  90% color
removal. Although DOC, THMFP,  and
color removal vary  directly with equiva-
lent coagulation dose and pH, DOC is the
better indicator of THMFP than color.
  Plant tests using alum and PACI were
conducted in the dry and wet season for
approximately 30 days each. The alum-
inum residuals in the distribution system
were generally less than 0.2 mg/L. The
aluminum residuals of 5 mg/L occurred
in  the distribution system  under  the
normal plant process of using alum in
conjunction with softening. The seasonal
plant data demonstrated that finished
aluminum  residuals  are  controlled by
coagulation pH, which should be 5.5 to
6.5 for minimum aluminum residuals. No
significant difference existed between the
sludge volumes produced during alum or
PACI coagulation during the plant tests
for equal operating conditions. The tur-
bidity carryover from the settled water
using either alum  or PACI  coagulation
was typically less than 2 NTU and ex-
hibited no differences between the  two
coagulants. A PACI-100S dose  of 80
mg/L was not sufficient to produce  a
settleable floe during the plant test. An
equal alum dose produced a settleable
floe.
  Softening preceded coagulation during
the dry-season plant test. Although soft-
ening removed 20% of the DOC and 43%
of the color, the succeeding alum or PACI
coagulation produced no  more color or
DOC  removal  than if coagulation  had
been  used alone.  The addition of  1.2
meq/L  of  alkalinity to the  softening
process reduced the calcium hardness to
only 100 mg/L as  CaC03 although  the
alkalinity and total hardness were bal-
anced. The softening  pH varied from 9.5

-------
to 10.6, with 10.0 achieving the lowest
hardness.
  During the wet season, the DOC aver-
age reduction was 57% to an 8.6 mg/L
residual by PACI and 58% to a 7.0 mg/L
residual by alum. The  dry season DOC
removals averaged 48% (to a 10.1 mg/L
residual) for alum and 39% (to a 12.1
mg/L residual) for PACI. This lower reduc-
tion during the dry season was due to a
higher coagulation pH caused by a hard
CaC03 scale in the reactor. The THM's at
the plant tap  during  the wet  season
averaged 235 jug/L  for PACI and 248
fjg/L for alum. The dry-season THM's at
the plant tap averaged 204 jug/L for PACI
and 175pg/Lforalum, which represented
a decrease of  13%  to 30% in the dry
season.  However, TOX for alum or PACI
were 710 and 714 fjg/L during the wet
season,  and they decreased to 421 and
380 /ug/L during the dry season. Color
removal was similar to  DOC removal,
reaching levels of 91% to 94% for PACI
and alum during the wet season and 78%
to 89% for PACI and alum during the dry
season.
  The DOC:THM:TOX  ratios averaged
1:32:94 during the wet season and
1:16:39 during the dry season. The
THM:TOX ratio did not vary as much as
the DOC:THM:TOX ratio, since both THM
and TOX are directly affected by CI2 dose.
The THM:TOX ratio was 1:2.9 and 1:2.4
during the wet  and dry seasons, respec-
tively.
  The operation cost increased  from
$1.07/1000 gal to $1.16/1000 gal when
the maximum alum dose was used during
the wet season. This 8% cost  increase
decreased  plant THM's by 72%. The
operational cost during the wet season
for PACI was $1.62/1000 gal, and similar
THM reductions were obtained. The cost
of producing water was less during the
dry season because of relativley constant
labor and power  costs  and increased
production. The plant costs were  in-
creased from $0.60/1000 gal to $0.787
1000 gal using the maximum alum dose.
This cost increase of 23% reduced THM's
by 39%. Dry-season PACI costs were
$1.02/1000 gal and decreased THM's by
29%.


Sequential Treatment
  Lime softening followed by alum coag-
ulation was investigated on a bench- and
pilot-plant scale at the AID water treat-
ment plant. Jar tests were conducted to
determine the best sequence of softening
and coagulation for THM precursor  re-
moval. Pilot plant testing was executed in
two unused reactors that were serially
connected for the project. The raw water
contained no turbidity. The hardness and
alkalinity were typically 350 mg/L as
CaC03,  with  color  and DOC values of
approximately 40 cpu  and 15  mg/L,
respectively.
  The initial bench-scale sequential treat-
ment work indicated that 60% of the DOC
could be removed by sequentially treating
AID raw water with softening and coag-
ulation, regardless of order. However, if
pH was adjusted to 5.5 during coagula-
tion, this DOC  removal was increased
25%. Color removal was 85% to 90% and
varied directly with removals of DOC and
THM precursors. THMFP was reduced to
a range from 167 to 217 fjg/L by softening
at pH 11.0 and coagulation at pH  5.5,
which represented an increased average
reduction of 45% relative to softening
only at pH  10.3. TOX formation was
generally three times THM formation.
  The  sequential treatment isopleths
showed that softening at pH 11.0 aver-
aged41 % DOC, 67%THMFP, 65%TOXFP,
and  63%  color removal. These were
increased removals relative to softening
at pH 10.3 (+11 % DOC, +3% THMFP, +8%
TOXFP, +3% color). Following lime soften-
ing with alum coagulation also increased
organic  removal. The optimum pH  and
dose were 5.0 to 6.0 and 117 mg/L alum.
The DOC, THMFP, TOXFP, and color were
reduced an additional 30%,  13%, 23%,
and 30%, respectively. Alum coagulation
following softening at pH 10.3 increased
the percentage of  removal  for all the
organic parameters more than alum
coagulation following softening at pH 11.
However, the lowest concentrations of
DOC,  THMFP, TOXFP, and  color were
from samples softened at pH 11.0  and
coagulated  at pH  5.0 to 6.0, although
these differences were typically less than
5%.
  Plant testing of sequential treatment
was conducted by softening  at pH 10.3
and 11.0 followed by alum coagulation
from pH 7.0 to 5.0. The alum dose was
varied from 60 to 109 mg/L. The results
demonstrated, as did the jar tests, that
alum coagulation following softening will
decrease the DOC, THMFP, TOXFP, and
color relative to softening alone. The
maximum DOC  and  color removals ob-
tained in the plant tests were approx-
imately 70% for DOC and 93% for color.
The THMFP  and TOXFP  were reduced
from 20% to 25%. Slightly better removal
occurred for all parameters when pH 11
rather than pH 10.3 softening preceded
coagulation.  The  organic removal
achieved in  sequential treatment indi-
cates that DOC, THMFP, and TOXFP can
be reduced an additional 20% to  25%
when alum coagulation is used in series
with  softening. The chemical cost in-
crease would vary from $0.10 to $0.157
1000 gal, which would increase AID O&M
cost  ($0.35/1000 gal) by 30% to 43%.
These data do indicate that, relative to the
coagulation testing at other sites, adding
softening to a coagulation process could
not increase organic removal.

Membrane Processes
  Bench-scale (1000 gpd) and pilot-scale
(25,000 gpd) investigations of membrane
processes were conducted at two sites
that  used ground water supplies—VOG
and AID. Bench-scale investigations were
also  conducted at one site  (Olga)  that
uses a surface supply. Initially, one RO
and six ultrafiltration (UF) low-pressure
membranes were purchased and tested
for product (permeate) water quality  on a
bench-scale level. The UF membranes
are designed for operation up to 100 psi,
whereas the RO membrane is intended to
operate at 200 to 250 psi. Bench results
from all three  sites demonstrated  that
only  two  membranes  could produce a
water from these highly organic sources
that  would meet  the  THM MCL  and
maintain a CI2  residual—the Filmtec UF
membrane (N-50)  and the Filmtec RO
membrane (BW 3030).
  The  nominal molecular weights  re-
jected by each membrane were supplied
by each manufacturer  and ranged from
40,000 to 100. Bench-scale testing indi-
cated that a molecular weight rejection of
2,000 would typically  pass 50% of the
raw DOC but only 20% of the color.  The
resulting product THMFP was generally
800  A/g/L for the surface water source
and  400 jug/L for either ground water
source. The ultrafilter with a molecular
weight rejection of 400 passed less than
10%  of the DOC and 3% of the color at any
site, and it typically produced a THMFP of
less  than  50 /ug/L at the ground water
sites and approximately 100 jug/L at the
surface water  sites. The product  was
essentially colorless, and the DOC was 2
mg/L or less at all three sites. The color,
DOC, and THMFP of the  RO  membrane
were approximately equal to the color,
DOC, and THMFP of the  best UF mem-
brane. However, the RO membrane oper-
ated  at 200 psi and rejected species with
a molecular weight of 100 or greater as
opposed  to the UF membrane, which
rejected species with a molecular weight
of 400 or greater at a pressure of 100 psi.
The  RO  membrane rejected a  much

-------
higher inorganic fraction than did the UF
membrane. The RO membrane rejected
more  than 90% of the total dissolved
solids (IDS), total hardness (TH), chloride
(CO, and sodium (Na+) at all sites, wher-
eas the UF membrane inorganic rejection
varied from 50% to 70% for the various
parameters.  Since the N-50 was the
membrane that operated at the lowest
pressure and still produced a water that
met the THM MCL, it was selected for
extended operation.
  The N-50  ultrafilter was installed at
Olga  in a bench-scale unit capable of
producing 1000 gpd, and it operated for
740 hr over a 45-day period. An opera-
tional percentage of recovery and feed
pressure matrix was developed to deter-
mine the extended study operating condi-
tions. Over matrix conditions of 60 to 120
psi feed pressure and 60% to 90% recov-
ery, product water quality improved at
high pressure (105+ psi) and lower recov-
ery (60%). The matrix results indicated
the THM MCL could be met if the opera-
tional conditions were 105 psi with 60%
recovery.  At these conditions,  product
water quality  and the  percentage  of
rejections were 1.6 mg/L DOC (92%), 3
cpu color (93%), 172 mg/L TDS (64%), 78
mg/L as CaCO3 TH (65%), 60 mg/L CI"
(40%), and 68 mg/L as CaC03 alkalinity
(59%) with pH 7.8. During the extended
operation, pressure was varied from 60 to
100 psi with recoveries of 60% to 90%.
The THM MCL was met for 105 psi and
60% recovery,  and for 75 psi and 60%
recovery immediately after the membrane
was chemically cleaned. The flux  de-
creased with time during the extended
study from 18 to 14 gpd/ft2 over 150 hr of
operation. After cleaning with the pres-
sure at 95 psi, the flux declined from 22 to
16 gpd/ft2, but the product quality re-
mained  constant. Flux was independent
of recovery during the extended study.
The  water quality and  percentage  of
rejection for the conditions meeting the
THM  MCL were 2.7 mg/L DOC (88%),
156 fjg/L TOXFP (80%), and 3 cpu color
(98%). The inorganic water quality was
145 mg/LTDS (65%), 68 mg/L Cf(32%),
85 mg/L as CaC03TH(64%), and 7 mg/L
as CaC03 alkalinity (58%). The pH was
7.8, and the water was stable.
  A 25,000-gpd mobile  UF pilot plant
using the N-50 membrane was built in a
30-ft trailer. The UF plant was housed in
the 20-  by 8-ft rear section of the trailer
and equipped for  antiscalant feed, acid
feed, chlorination, stabilization, prefiltra-
tion,  and  storage as well as UF  with
variable recovery (50% to 90%) and feed
pressure (80 to 120 psi). This plant was
installed and operated at VOG for 365 hr
from January 2 to March 3,1985. Initially,
an operational test matrix was developed
for water quality and flux from varying
percentages  of  recovery and pressure.
Product water DOC, color, and THMFP
were independent of recovery and pres-
sure over the test conditions and averaged
less than 2 mg/L, 1 cpu,  and 50 fjg/L,
respectively. Product water TDS and TH
increased with increasing recovery, were
independent of pressure, and varied from
25% to 75%  of the raw  water value.
Operating conditions at VOG were set at
90 to 105 psi and 75% recovery to produce
a water with a TH of 150 mg/L as CaC03,
essentially no color, and THMFP of 50
/jg/L or less. During the VOG operation,
the product water quality and the per-
centage of rejection from the raw water
were 1.9  mg/L  DOC (88%), 3 cpu color
(97%),  27  yug/L THMFP,  and  47 /ug/L
TOXFP. The raw  water TDS,  TH, and
alkalinity  were  reduced to 195 mg/L
(60%),  142  mg/L  as CaCO3 (62%), and
135 mg/L as CaC03(60%), respectively.
The final pH was 7.5, and the water was
stable. The product water flux declined
32% during the VOG operation from 20 to
13.4 gpd/ft2. The water temperature was
approximately 25°C and essentially did
not vary during the operation.
  On March 3,  1985, the  UF pilot plant
was moved from VOG to AID and oper-
ated until May 31,1985, with an elapsed
time of operation of 1020  hrs. A second
operational test matrix was developed
and showed that color, DOC, and THMFP
removal were independent of product
recovery and feed pressure over the test
conditions (50% to 90% product recovery
and 80 to 120 psi). Product water color,
DOC, and THMFP were typically less than
2 cpu, 2 mg/L, and 30/ug/L, respectively.
TH and TDS removals were independent
of pressure and dependent on product
recovery, and they varied from 25% to
75% of the raw water value. Long-term
operating conditions varied from  90 to
103 psi and 67%  to 83%  recovery. The
product water quality and percentage of
rejection  from the raw water were 1.5
cpu color (97%),  2.0 mg/L DOC (86%), 50
fjg/L THMFP, and 48 fjg/L TOXFP. The
product water TDS, TH,  and  alkalinity
values and percentage of rejection were
282 mg/L (43%), 187 mg/L as CaC03
(40%), and 180 mg/L as  CaC03 (37%),
respectively. The pH averaged 7.5, and
the product was stable.
  The  UF pilot plant was designed with
the membranes in four pressure vessels
connected two each in series. The aver-
age pressure drop in the first pressure
vessel was 32 psi (11 psi/membrane)and
41 psi (13 psi/membrane) in the second
pressure vessel. The flux at AID rose
slowly from the range of 14 to 15 gpd/ft2
during the first 150 hr of operation. After
a chemical cleaning, it rose to slightly
more than 20 gpd/ft and remained there
for the duration of the study.
  Cost of construction of a UF  plant
should be slightly less than an equivalent
RO  plant because of  less expensive
membranes and lower pressure require-
ments.  Power costs  should generally
decrease from 75% (100 psi/400 psi) to
50% (100 psi/200 psi). The construction
and O&M costs were estimated at $0.29
and $0.47/1000 gal  for  a  1-MGD  UF
plant.

Flotation
  Bench-scale investigations of dissolved
air flotation (DAF) in conjunction with and
after alum  coagulation  of a high-color,
low-turbidity water were conducted using
a small  DAF pilot  plant supplied by
Komline Sanderson. After contact with
DAF,  alum  floe was  found to  shear,
solubilize,  and  post-precipitate  after
sampling was complete. DOC and color
removal following DAF of alum-coagulat-
ed waters were less than those having
only alum  coagulation. Foam flotation
using alcohol-based surfactants common
to the mining industry did not remove
DOC from the same high-color, low-tur-
bidity potable water source.


Summary of  Results
  The DOC removal from each  of the
processes  investigated was different.
Lime softening typically removed 10% to
30% of the raw  DOC at normal reaction
pH's of  9.0 to 10.3. The  DOC removal
during softening could be generally in-
creased to  40% to 50% by  raising the
reaction  pH. Alum or PACI coagulation
should remove 60% to 70% of the raw
DOC if the  reaction pH is 5.0 to 6.5 and
the coagulant dose is  adequate—80 to
180 mg/L as AlzfSO^a'IAH^  in the
waters investigated. Sequentially treating
water with lime softening and  alum
coagulation would remove 60% to 70%
DOC  and offers essentially the  same
removal as alum coagulation. RO at 200
psi  or  UF  at  100  psi  with  selected
membranes will generally remove 90% or
more of  the initial DOC. No process other
than membranes can achieve this high
removal  for a prolonged  period. These
general  statements concerning DOC re-
movals are supported by the test results
of investigations conducted on a low-

-------
turbidity, high-color surface water and on
a low-turbidity, high-color ground water.
  Although DOC  is  a better surrogate
than color for THM  or TOX formation,
DOC cannot be used indiscriminately as a
surrogate for chlorinated organic forma-
tion. These results and others have
suggested that DOC's of less than 4 mg/L
following conventional coagulation or
softening are required to meet the THM
MCL. Indeed, when the DOC was reduced
to a range of 5 to 6 mg/L by coagulation,
THMFP's of  less  than 200  yug/L were
recorded. The DOC could not be reduced
any lower by coagulation or softening of
these waters. However, DOC's below 2
mg/L achieved by UF hadTHMFP's above
100/ug/L. The results indicated that DOC
is the best THM surrogate on a process
basis. DOC from membrane processes is
generally more reactive than DOC from
precipitative processes. The THM reduc-
tion is significantly greater than the DOC
removal because of the reduced chlorine
demand, which also  reduces THM's. The
plant experiences indicate that a 30%
reduction inTHM is possible if the process
is tuned  for increasing organic removal
and minimizing chlorine dose.
Conclusions

Plant Optimization

• DOC, THMFP, TOXFP, and color are
   reduced as the reaction pH  during
   softening is increased.
• If the THM's are 150//g/L or less, the
   THM and CI2 residual MCL can prob-
   ably be met in a lime-softening plant
   by  increasing the  reaction pH and
   lowering the CI2 dose.
• Optimizing a lime-softening plant for
   THM control will increase product
   hardness by 75 mg/L as CaCO3, or
   more.

PAC I/Alum Coagulation
• PACI has less  base-neutralizing ca-
   pacity than alum, which would gener-
   ally be a  disadvantage  for organic
   removal from a low-turbidity, high-
   color water because of the importance
   of a low-coagulation pH.
• Color,  DOC, THMFP, and TOXFP re-
   duction by  either PACI or alum coag-
   ulation was maximized near pH 5.5.
• Aluminum residuals after PACI or alum
   coagulation were dependent  on co-
   agulation pH and minimized from pH
   5.0 to 6.0.
•  DOC was a better surrogate parameter
   for THM and TOX than was color.
•  Unit cost depended more on seasonal
   water demand than  on chemicals
   required for optimum organic removal.
•  Preceding either PACI or alum coag-
   ulation with lime softening does not
   increase the organic removal during
   coagualtion.
•  THM:TOX ratios were less  variable
   than DOC:THM:TOX ratios because of
   the effect of chlorine dose on THM and
   TOX.
•  Alum floe was more settleable at low
   coagulant doses (80 mg/L) than was
   PACI floe.
•  Alum was slightly superior to  PACI for
   color, DOC, and THMFP reduction.

Sequential Treatment

•  The order of lime softening and alum
   coagulation  was  not  significant  for
   maximizing organic removal  if alum
   coagulation  is added  to an  existing
   lime-softening plant.
•  Alum coagulation added to an existing
   lime-softening  process  significantly
   increases the organic removal  and
   reduces the THMFP;  however, lime
   softening will not increase the organic
   removal when  added  to an  existing
   alum coagulation process.

Membranes

•  A 150-day pilot-plant UF study of two
   highly organic ground waters (both of
   which produced more than 400 /ag/L
   THM's when conventionally  treated)
   produced  a finished water that easily
   met the MCL's for CI2 residual and
   THM's.
•  Compared with membranes,  no other
   process has the same practical capac-
   ity for the removal of THM precursors.
•  The  vast majority of THM precursors
   were rejected by a membrane that has
   a molecular weight cutoff of less than
   2000 but greater than  400.
•  The effect of recovery percentage and
   feed pressure on product water quality
   was site-specific.
•  Flux was site-specific, ranging from
   approximately 20 to 10 gpd/ft2, and it
   was generally independent of the
   recovery percentage.
•  Caloosahatchee River water needs
   more  elaborate pretreatment than
   single-pass sand filtration to maximize
   N-50 permeate flux. Flux values during
   the extended study were 12  to  18
   gpd/ft2, and they decreased contin-
   uously with time of operation.
 • N-50 permeate quality is a function of
   feed pressure, percentage of recovery,
   and membrane surface condition. The
   best inorganic and organic rejection
   occurred at the highest pressures, the
   lowest recoveries, and the cleanest
   membrane condition. With the Caloo-
   sahatchee water, the THM MCL was
   met only at the 105 psi, 60% recovery
   and the 75 psi, 60% recovery condi-
   tions for a cleaned membrane.  Mem-
   brane surface conditions appear to be
   as important as the pressure-recovery
   setting  for organic rejection.  Poor
   permeate quality during extended op-
   eration was primarily  due to  mem-
   brane fouling.
 • Inorganic parameters such asTDS and
   TH were reduced approximately 40%
   by UF at approximately 100 psi and
   75% recovery.
 • The product water at VOG, AID, and
   Olga exhibited low  DOC and color
   values that would not require addi-
   tional  color removal through chlorine
   bleaching.
 • The organic quality produced during
   the VOG and AID UF pilot-plant tests
   were independent of the percentage of
   recovery and feed pressure for the test
   conditions.
 • The inorganic quality produced during
   the UF pilot-plant tests at VOG and AID
   were  independent of the feed pres-
   sures and improved as the percentage
   of recovery decreased.
 • UF with the N-50 membrane produced
   a stable, noncorrosive water.
 • The operation  of the UF membrane
   pilot plant required less effort and skill
   than the operation of the other pro-
   cesses investigated, and it produced a
   vastly  superior organic water quality.

Flotation

 • DAF of alum sludge caused floe shear
   and post-precipitation. It removed  no
   more organics than did conventional
   alum coagulation.
 • Foam  flotation  using  conventional
   alcohols common to the mining indus-
   try removed no DOC from the raw
   water.

Recommendations
 • PACI studies should  be expanded  to
   include high-turbidity and/or low-alk-
   alinity waters.

-------
• DOC could be used as a surrogate for
  THM formation on a site-specific basis.
• Specific organic rejection by  mem-
  brane processes should be investigat-
  ed to include priority pollutant remov-
  als and  other organics  capable of
  adverse health effects.
• Cost and performance information on
  membrane processes should be made
  available to the water utility  industry
  as soon as possible so that they can be
  considered in  planning for plant ex-
  pansion or new plant construction.
• A national survey of membrane plants
  should be conducted to determine (1)
  the cost of construction, operation,
  maintenance, (2) operating conditions
  and problems, (3) raw and  finished
  water quality, and (4) brine  disposal
  and membrane life.
• Further bench and pilot-plant testing
  using UF and  RO membranes should
  be conducted  on surface and ground
  waters of varying organic quality to
  ascertain the  relationships among
  water q ual ity, recovery, pressu re, flux,
  membrane cleaning, membrane mater-
  ials, and membrane life.
• Newly constructed membrane plants
  using raw waters normally treated by
  lime softening or coagulation should
  be monitored  for cost, water quality,
  and operating conditions,  and  that
  information should be published for
  the water utility industry.
• A permanent  membrane plant  that
  uses UF and RO should be constructed,
  operated, and monitored for 2 to 5
  years to provide for a small community
  (fewer than  3,000 residents) water
  that does not exceed the THM MCL.
  Such a project could (1) establish mem-
  brane processes as a practical means
  of organic control, (2) provide the best
  possible data  on cost, water quality,
  and operation, and (3) demonstrate the
  ease of operation, modular expand-
  ability,  operator  skill  required, and
  consistency of membrane processes.

• A mobile, pilot-plant-scale, hybrid sys-
  tem of low-pressure RO and UF mem-
  branes should be constructed  and
  operated to assess organic control and
  production of  waters with  suitable
  hardness values. Also,  saltwater-in-
  truded, high-organic waters could be
  studied, particularly the coastal wat-
  ers.
• Research and development should
  continue  on molecular-weight rejec-
  tion by membranes.  This work has
  indicated  that significant THM pre-
  cursor rejection  is accomplished be-
  tween molecular-weight rejections of
  400 and 2000. A membrane that has a
  molecular  weight rejection  greater
  than 400 may have as good a THM
  precursor rejection and  be  less ex-
  pensive to operate.

  The full report was submitted in fulfill-
ment of Cooperative Agreement No. CR-
811039-01-0 by the University of Central
Florida under the sponsorship of the U.S.
Environmental Protection Agency.
                                                                           . S. GOVERNMENT PRINTING OFFICE:1986/646-l 16/20752

-------
     J. S.  Taylor, D.  Thompson, B. R.  Snyder, J. Less, and L. Mulford are with the
       University of Central Florida, Orlando. FL 32816.
     J. Keith Carswell is the EPA Project Officer (see below).
     The complete report, entitled "Cost and Performance Evaluation of In-Plant
       Trihalomethane Control Techniques." (Order No.  PB 86-130 515/AS; Cost:
       $34.95, 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
     BULK RATE
POSTAGE & FEES PAID
        EPA
  PERMIT No. G-35
Official Business
Penalty for Private Use $300

EPA/600/S2-85/138
                          0063240    WERL
                          LOU  w  TILL!?
                          REGION V  EPfl
                          LIBRARIAN
                          |3?rS  OiARSCRN  $T
                          CHICAGO
                                                         60604

-------