Professional Documents
Culture Documents
Arsenic
removal
by coagulation
This utility’s goal of 90 percent arsenic removal
was attained through treatment with 6.5 mg/L
ferric chloride.
Test Scale
Total
Dose Filtration pH Alkalinity
µM/L Polymer Module Rate of Tem- mg/L
Dose Fe(III) Dose Flow gpm/ Coagu- Turbidity Arsenic perature TDS TOC as
Coagulant mg/L or Al(III) mg/L mgd sq ft lation ntu µg/L pH oC mg/L mg/L CaCO3
Finished-
water
reservoir
Module 1 Module 2
Control
room
Sampling
location
A
added to maintain a free-
chlorine residual of approx-
lum was less effective than ferric imately 1.3 mg/L in the
plant effluent during the
chloride at removing arsenic. alum testing. Sedimentation
basin sludge and filter-back-
wash water are gravity-fed
to one of seven retention
centration and thereby comply with the lowest basins for drying prior to disposal. Separate basins
anticipated arsenic MCL. were used for the ferric chloride and alum sludge.
• Validate the results of bench-, pilot-, and demon-
stration-scale studies and ensure that their results Experimental setup
can be applied to a full-scale plant. Ferric chloride doses of 3, 4, 5, 6.5, and 10 mg/L as
• Ensure that, in the process of complying with FeCl3 [or 18.5, 24.7, 30.8, 40.1, and 61.7 µM Fe(III)/L]
the arsenic regulations for drinking water, MWDSC and alum doses of 6, 10, and 20 mg/L as Al2(SO4)3 ·
is not compromising other water quality parameters 14H2O [or 20.2, 33.6, and 67.3 µM Al(III)/L] were
or creating hazardous wastes that cannot be disposed applied for one-week periods (Table 2). Dosages started
of in the landfills being used for sludge disposal. at levels normally used to maintain effluent turbidity
levels of <0.1 ntu and were gradually adjusted, based
Test site on Cheng and co-workers’ results,9 in an attempt to
This study was conducted at the Henry J. Mills Fil- attain 90 percent arsenic removal. Test conditions were
tration Plant, a 106-mgd conventional drinking water set on Mondays, and samples were typically collected
treatment plant in Riverside, Calif. The Mills plant is on Thursdays and the following Mondays. Optimal
composed of two nearly identical modules, each of polymer* doses of 2–3 mg/L were determined by jar
which consists of one rapid-mix basin; two parallel,
rectangular flocculation basins; and two parallel, rec- *Cationic polyDADMAC, Magnifloc 587C, Cytec Industries, West Pater-
tangular sedimentation basins (Figure 1). The con- son, N.J.
Metal Concentration
Metal TTLC STLC Total Total Total Soluble Total Soluble Total Soluble
mg/kg mg/L mg/kg mg/kg mg/kg mg/L mg/kg mg/L mg/kg mg/L
testing, and no acid was applied for adjusting the coag- Trace metals in sludge. Sludge samples were
ulation pH. Filter aid* was applied to the settled-water collected from the sedimentation basins for each
conduit at 0.01 mg/L throughout the study. Because the test condition and from the alum and ferric chloride
plant was in service, flows in module 2 fluctuated from sludge lagoons after the sludge had accumulated.
16.1 to 48.4 mgd, and the filtration rates ranged from To simulate sludge that is disposed of in landfills,
2.87 to 6.48 gpm/sq ft (Table 2). The filters were back- the sludge samples were settled, decanted, and dried
washed when the filter head loss reached approxi- in an oven at 104oF. The resultant dried sludge rep-
mately 6.0 ft, the filter effluent turbidity reached resented the driest—and therefore the most con-
approximately 0.25 ntu, or the filter length of run centrated (85–100 percent solids)—sludge that
(LOR) approached 30 h, whichever came first. MWDSC sends to the landfill. The sludge samples
Silverwood Lake state project water (SPW) was were analyzed by the laboratory for the inorganics
used for all the tests except the 4-mg/L ferric chloride regulated in California as hazardous waste14 (Tables
test, which used SPW from Lake Perris. The perti- 3 and 4) to ensure that the optimization of arsenic
nent raw-water quality parameters of the water removal would not create waste disposal problems.
treated in this study are listed in Table 2. Analyses The total metals concentrations were determined
for the following constituents were conducted using an ICPMS method (USEPA 6020).15 Those
throughout this study: arsenic, trace metals in sludge samples that exhibited total metal concentrations
and in the water, turbidity, total suspended solids exceeding 10 times the soluble threshold limit con-
(TSS), total organic carbon (TOC), and total alkalin- centration (STLC; a California MCL) were analyzed
ity. In addition, particle counts, pH, temperature, fil- further to determine the soluble metal concentra-
ter head loss, and the LOR for each filter were mon- tions. The waste extract test procedure and an ICPMS
itored; jar tests were routinely conducted prior to method (USEPA 6020)15 were used to determine
changes in the coagulant dosage at the plant. the soluble metal concentrations.
Trace metals in finished water. Water samples
Sample analyses were collected for analysis of the inorganics regu-
Arsenic. Arsenic samples were collected at the lated in California’s drinking water14 (Table 5) at the
plant influent and combined filter effluent on each plant influent and combined filter effluent. The pur-
sampling date and at the filter influent for selected pose of these analyses was to ensure that water qual-
ferric chloride tests. The arsenic analyses were con- ity would not deteriorate when elevated dosages of
ducted by an outside laboratory† using a hydride gen-
*Nonionic polyacrylamide, Magnifloc 985N, Cytec Industries, West
eration–inductively coupled plasma–mass spectrom- Paterson, N.J.
etry (ICPMS) method (detection limit 0.02 µg/L).13 †West Coast Analytical Service Inc., Santa Fe Springs, Calif.
Metal Concentration
Lagoon Samples
California FeCl3 Dose (mixed dosages)
Regulatory
Limit (Title 22)† 6.5 mg/L 10 mg/L 2/14/94 7/11/94 7/18/94
TTLC STLC Total Soluble Total Soluble Total Soluble Total Total Soluble
Metal mg/kg mg/L mg/kg mg/L mg/kg mg/L mg/kg mg/L mg/kg mg/kg mg/L
Metal Concentration
Title 22
Limit† 6 mg/L Alum 10 mg/L Alum 20 mg/L Alum Sludge Lagoon
TTLC STLC Total Soluble Total Soluble Total Soluble Total Soluble
Metal mg/kg mg/L mg/kg mg/L mg/kg mg/L mg/kg mg/L mg/kg mg/L
ferric chloride and alum were used. The analyses combined filter effluent. These samples allowed
were conducted by the outside lab, using an ICPMS comparison of the relative efficacy of the various
method (USEPA 200.8).16 dosages of ferric chloride and alum for turbidity
Turbidity. Turbidity samples (grab samples) were removal and served as indicators of water quality
collected at the plant and filter influents and at the during the testing. The samples were measured
Aluminum 1,000 ND* <20 ND <20 39.0 2.5 93.6 33.7 4.6 86.4
Arsenic 50 (2.0–20) 1.60 0.27 83.10 1.30 0.14 89.23 1.60 0.07 95.63
Barium 1,000 27.7 25.3 8.7 24.4 20.6 15.6 27.2 25.7 5.5
Cadmium 10 (5) ND <0.3 ND <0.3 ND <0.1 ND <0.1 0.5 ND <0.3 NC
Chromium 50 ND <1 ND <1 ND <0.4 ND <0.4 ND <0.8 ND <0.8
Copper 1,000 2.3 1.5 34.8 2.6 1.8 30.8 2.9 2.0 31.0
Iron 300 ND <100 ND <100 105.0 64.0 39.0 102.0 58.0 43.1
Lead 50 (BAT) ND <0.3 ND <0.3 0.2 0.1 NC 7.7 0.5 93.5
Manganese 50 5.0 ND <2 > 60 17.5 4.1 76.6 11.3 10.4 8.0
Mercury 2 ND <0.1 ND <0.1 ND <0.2 ND <0.2 ND <0.4 ND <0.4
Selenium 10 (50) ND <4 ND <4 ND <9 ND <9 ND <40 ND <40
Silver 50 (no MCL) ND <0.1 ND <0.1 ND <0.1 ND <0.1 ND <0.2 ND <0.2
Zinc 5,000 4.0 2.0 50.0 5.4 5.5 NC 3.0 2.0 33.3
*ND—not detected
†Negative numbers denote an increase of metals in the treatment train; however, with the exception of the aluminum, these apparent increases are insignificant.
‡NC—no change; the influent and effluent values were the same within experimental error.
§Samples were taken from an alternate location, which was inappropriate for this test. Copper tubing, leaded solder, and brass fittings came in contact with the water
sample.
using a nephelometric turbidimeter,* which was ent to determine relative particle removal on each
calibrated daily. sampling date. The particle counter‡ was set up in
TSS. TSS samples were analyzed by standard batch mode, and each grab sample was measured
laboratory procedures.17 The samples were collected three times. Total and size differential particle counts
at the filter influent for each test condition. This were recorded for each sample. Six channels were
parameter was used as a measure of filter solids programmed to measure the following size ranges:
loading. 2–4, >4–6, >6–8, >8–10, >10–15, and >15–200 µm.
TOC and total alkalinity. The TOC and total alka- pH. The pH was measured at the plant influent,
linity measurements provide raw-water quality infor- after coagulant addition (in the flocculation basin),
mation (Table 2). The TOC was measured with a car- and at the reservoir inlet on each sampling date. A pH
bon analyzer,† and the total alkalinity tests were meter§ was used for these analyses.
conducted by standard titration methods.17 Filter head loss and LOR. Terminal headloss, as
Particle counts. Particle counts were measured measured by differential pressure cells, and LOR for
at the plant influent and the combined filter efflu- each filter were monitored continuously on the con-
trol system computer. De-
creases in these parameters
FIGURE 2 Average arsenic removal in SPW with alum and ferric chloride indicate stress on the filters
coagulation and deterioration of plant
performance.
FeCl
3
Al (SO ) . 14H O
2 4 3 2
Jar testing. Jar testing
3 82
was completed each week,
prior to changes in test con-
4 Lake Perris water 89 ditions, to ensure that the
Coagulant Dose—mg/L
86
coagulant dosages of the next
5
test condition would not com-
6 23 promise the Mills plant’s
effluent water quality.
6.5 92
10 95
*Model 2100A, Hach Co., Loveland,
Colo.
10 45 †Dohrmann model DC-180, Rose-
mount Analytical Corp., Santa Clara, Calif.
20 69 ‡Model 250 batch sampler, model
214 single-sensor counter, or model LB100
light-blocking laser sensor, Met One Inc.,
0 10 20 30 40 50 60 70 80 90 100 Grants Pass, Ore.
A i R l t §Model 130, Corning Inc., Corning,
N.Y.
*Negative numbers denote an increase of metals in the treatment train; however, with the exception of the aluminum, these apparent increases are insignificant.
†NC—no change; the influent and effluent values were the same within experimental error.
‡ND—not detected
§Samples were taken from an alternate location, which was inappropriate for this test. Copper tubing, leaded solder, and brass fittings came in contact with the water
sample.
Results and discussion sedimentation basin (Table 6). Alum was less effective
Arsenic removal. Average arsenic removals of than ferric chloride at removing arsenic. Alum doses
82, 89, 86, 94, and 96 percent were achieved when of 6, 10, and 20 mg/L [or 20.2, 33.6, and 67.3 µM
3, 4, 5, 6.5, and 10 mg/L ferric chloride [or 18.5, Al(III)/L] reduced the SPW arsenic concentrations
24.7, 30.8, 40.1, and 61.7 µM Fe(III)/L] were used by 23, 45, and 69 percent, respectively. These per-
(Table 6 and Figure 2). This treatment reduced the centages correspond to reductions in arsenic con-
arsenic level from approximately 1.6 µg/L in the plant centrations from approximately 2.15 µg/L to 1.65,
influent to 0.07–0.32 µg/L in the filter effluent when 1.15, and 0.66 µg/L, respectively.
SPW and Lake Perris water were used. Most of the Previous studies have shown that the percentage
arsenic removal (48–88 percent) was achieved in the of arsenic removed by the use of alum and ferric
Turbidity—ntu
Total
Conditions Removal TSS— Particle Count (2–200 µm)
(at mg/L number/mL
Coagulant Filtration Combined Combined
Dose Rate Plant Filter Filter Filter Filter Plant Filter Log
Coagulant mg/L gpm/sq ft Influent Influent Effluent Effluent—% Influent Influent Effluent Removal
Conditions
*Detection limit = 0.02 mg/L, except for bench scale (0.5 mg/L) and where noted; all removals are based on samples collected from the raw water and filter effluents.
†BS—bench scale, PS—pilot scale, DS—demonstration scale, FS—full scale
‡No arsenic detected in filter effluent (detection limit = 0.5 mg/L)
§No arsenic detected in filter effluent (detection limit = 1.0 mg/L)
cent (average 93.9 ± 1.8) in the SPW ferric chloride 2.76, 2.49, and 2.46 for the 6, 10, and 20 mg/L alum
tests to achieve an average filter effluent turbidity of tests, respectively. In addition to the greater log
0.09 ntu. The alum tests achieved slightly better turbidity removals of particles during the alum tests, the filter
removal (96–98.1 percent; average 97.1 ± 0.8) and effluent particle counts were also lower throughout
averaged 0.07 ntu in the filter effluent. Most of the dif- the alum tests as compared with the ferric chloride
ference in percent turbidity removal can probably be tests (averages of 8.8–14.5 particles/mL versus aver-
attributed to the slightly higher influent turbidities dur- ages of 47.3–57.5 particles/mL in SPW). These results
ing the alum tests. The plant influent turbidities ranged are consistent with those of previous studies,22,23
from 0.92 to 2.8 ntu (average 1.49 ± 0.48) for the fer- which, at ambient pH, showed better particle removal
ric chloride SPW tests and from 1.9 to 3.4 ntu (average by alum than by ferric chloride. However, it should be
2.37 ± 0.66) in the alum tests. The warmer water tem- noted that the superior particle removal by alum dur-
peratures experienced during the alum tests (average ing the study may have resulted from water quality
23.4oC versus 12.6oC) may have also enhanced the differences such as increased temperature and influ-
effectiveness of the alum coagulation. ent particle counts during the alum testing.
TSS. The TSS measured at the filter influent fluc- Comparing arsenic, turbidity, and particle
tuated between 1.4 and 4.7 mg/L (average 3.8 mg/L) removal in bench-, pilot-, and demonstration-scale
for the ferric chloride tests, with no clear trends emerg- studies. Bench-, pilot-, and demonstration-scale
ing from the data (Table 7). TSS during the alum tests arsenic removal studies were performed at MWDSC by
ranged from 0.7 to 2.8 mg/L (average 2.0 mg/L). The Cheng and co-workers from 1992 through 1994.9,10,23
sudden drop in the TSS levels from 2.8 mg/L at 10 Key design and water quality parameters of each of the
mg/L alum to 0.7 mg/L at 20 mg/L alum indicated tests are listed in Table 1. Despite design and water
that sweep coagulation had been attained. This find- quality differences and the inherent variability in full-
ing supports previous research, which shows alum scale studies (e.g., changing flow rates), the studies
doses of 20 mg/L at pH 7.2 to be in the sweep coagu- conducted at the different scales demonstrated trends
lation zone of the alum coagulation diagram.20,21 The in arsenic, turbidity, and particle removal that were
decrease in TSS during the alum tests as compared similar to those found in this full-scale study (Table 8).
with the ferric chloride tests may also have resulted Excellent turbidity removal was achieved at all scales
from better coagulation due to warmer water tem- and dosages when either ferric chloride or alum was
peratures and slightly higher raw-water turbidity. used; however, correspondingly good arsenic removal
Particle counts. Improvement in arsenic removal was achieved at all scales and dosages only with fer-
did not correlate with increased particle removal. The ric chloride treatment. Particle removal did not cor-
particle count data presented in Table 7 and Figure 3 relate with arsenic removal.
indicate that alum may be more effective than ferric Figure 4 shows that the bench-, pilot-, and demon-
chloride at removing particles. Average log removals stration-scale tests provided reasonable predictions of
of 1.76, 1.88, 2.09, and 1.98 were achieved when arsenic removal at a full-scale plant with ferric chloride
SPW was treated with 3, 5, 6.5, and 10 mg/L ferric treatment. The 10-degree increase in water tempera-
chloride, respectively. Average log removals were ture during the alum tests in the full-scale study appears
100
Arsenic Removal—percent
100
90 90 Trace metals in sludge.
80 80
70 95 92 70 The results of the sludge analy-
60 92 60 ses are presented in Tables 3
50 50
40
86
89 83 40 and 4. The sludge samples
30 82
30 evaluated throughout the
20 20
10 10
study were all well below the
Lake
Perris current California (Title 22)
30 Bench Scale
20
water MCLs and were suitable for
Pilot Scale
Fer 10 6.5 nonhazardous waste landfills.
Dorsic Ch
e— lorid
mg/ e
5
4
Demonstration Scale Treatment of SPW with 3–10
L Full Scale mg/L ferric chloride or 6–20
3
mg/L alum should not create
hazardous waste or present a
FIGURE 5 Arsenic removal in SPW with bench-, pilot-, demonstration-, and full-
disposal problem. However, if
scale alum coagulation
current MCLs for hazardous
waste are lowered, or if poten-
96 tial new sources of water con-
89 taining high levels of arsenic
100
are developed by MWDSC,
Arsenic Removal—percent
Arsenic Removal—percent