Jar Tests For Evaluation of Atrazine Remova
Jar Tests For Evaluation of Atrazine Remova
Jar Tests For Evaluation of Atrazine Remova
ABSTRACT
The objective of this study was to evaluate the effects of major factors such as PAC dosage, pH,
contact time,mixing energy, alum dosage, and enhanced coagulation on the effectiveness of atrazine
removal. Jar tests and response surface methodology were used to simulate conditions found in dif-
ferent treatment facilities. The time course of atrazine concentration with an initial atrazine con-
centration of 12 /JLgfL and initial (Acticarb) PAC of 16 mg/L indicated that it took approximately
5 days to reach equilibrium with the maximum atrazine removal of about 73%. Therefore, in treat-
ment facilities, the adsorption of atrazine with this kind of PAC will be less than the removal achieved
at equilibrium, due to the short retention time in a dynamic process. Mixing energy is a major fac-
tor affecting atrazine absorption. With jar test times ranging from 30 to 60 min, increasing rpms
from 5 to 100 (G = 4 to 321 s_1) resulted in atrazine removals ranging from 34 to 59%. Without
addition of PAC, neither lime softening nor alum coagulation (conventional or enhanced dosages
ranging from 6 to 18 mg/L) demonstrated atrazine removal. A synergistic relationship appears to
exist between PAC dosage and enhanced coagulation (with pH adjusted to about 5.8); neither PAC
nor enhanced coagulation resulted in as high a removal rate of atrazine as the two did together
(greater than 60%). The results of this study are useful for evaluation of different PAC application
points in conventional drinking water treatment plants.
Key words: Atrazine; powdered activated carbon (PAC); jar tests
417
418 ZHANG AND EMARY
the application of powdered activated carbon (PAC) to as pH, or PAC dose are the major factors that can be eval-
the treatment facilities (Miltner et al, 1989; Qi et al, uated. In this preliminary study, jar tests were introduced
1994; Adams and Watson, 1996). to study the effects of PAC on atrazine removal in drink-
Conventionally, PAC is often used in water-supply ing water treatment plants. Using jar tests to evaluate per-
utilities worldwide for taste and odor control (U.S. EPA, formance of PAC adsorption on organic compounds may
1976; AWWA, 1977, 1986; Lettinga et al, 1978; have several advantages. (1) The jar test is well known
Lalezary-Craig et al, 1986, 1988; Miltner et al, 1989). by workers in drinking-water treatment plants. Therefore,
Since the late 1980s, considerable research has been con- once the evaluation procedure is established, the tech-
centrated on using PAC to control organic contaminants nique transfer is easier than other methods, such as the
in drinking water (Najm et al, 1991a, 1993; Qi et al, HSDM method, which needs a full understanding of
1994; Adams and Watson, 1996), with two major direc- mathematical modeling and adsorption theory. (2) Mix-
tions: theoretical- and practical-oriented research. For ing energy (e.g., the velocity gradient, G) can be intro-
theoretical-oriented research, most studies have been fo- duced into PAC-adsorption jar tests. This is important
cused on the predictive methods (e.g., equilibrium mod- when optimization of PAC application points is needed.
els and mass transfer models) for evaluating the perfor- The objectives of this study were to: (1) evaluate the
mance of a PAC adsorption system. The homogeneous effects of major factors, such as PAC dosage, pH, con-
surface diffusion model (HSDM) is commonly used to tact time, mixing energy, alum coagulant, enhanced co-
evaluate adsorption kinetics (Crittenden and Weber, agulation, on the effectiveness of atrazine removal; and
1978a, 1978b; Hand et al, 1983; Traegner and Suidan, (2) investigate the interrelationship(s) among these fac-
1989, 1991). The procedure for using the equivalent tors through conducting jar tests. The experimental re-
background compound coupled with the ideal adsorbed sults were then used to evaluate the best PAC applica-
solution can be used as a method to qualify the compe- tion location(s) of conventional (surface) drinking-water
tition between trace organics and background organic treatment plants, with a drinking-water treatment plant
matter in water (Najm et al, 1990; 1991b; Qi, era/., 1994; being used for a case study.
Hopman et al, 1998).
For practical-oriented research, PAC has been used in
the Robert-Haberer process (Stukenberg and Hesby, MATERIALS AND METHODS
1991; Harberer and Norman-Schmidt, 1991), the upflow Chemicals and Waters
floc-blanket reactor (Najm et al, 1991c, 1993), and the
solids contact slurry recirculating clarifier (Kassam et al, The atrazine (2-chloro-4-ethyamino-6-isopropylamino-
1991), jet flocculation systems (Sobrinho et al, 1997), 1,3,5-triazine) used in this study was obtained from Chem
and other reactors (Najm et al, 1991b; Qi et al, 1994). Service (West Chester, PA). Purity and lot number were
Najm et al. (1991a) summarized the advantages and dis- 99.7% and 25-116A, respectively. The adsorbent used in
advantages of different points for PAC application and this study was Acticarb PAC (Calgon Carbon Corpora-
the important criteria for selecting these points. However, tion, Pittsburgh, PA). The characteristics of this carbon
information still is not sufficient on how (1) to evaluate are as follows: iodine number, 1199 mg/g; moisture as
the effects of different operating factors on PAC perfor- packed, 3%; apparent density, 0.54 g/cm3; ash content,
mance, and (2) to find the best location(s) for PAC ap- 6%; and 98% of the carbon passing through #325 mesh.
plication for the optimum removal of atrazine. Because Aluminum sulfate obtained from General Chemical Co.
the drinking water treatment train changes with different (St. Louis, MO) was used as a coagulant in the corre-
water supply sources, it is imperative to study the appli- sponding jar tests, and is also used in the upflow floc-
cation of PAC for atrazine removal and to optimize the blanket reactors (primaries) for clarification in FTP (see
locations of PAC application in treatment trains. Case Study). NaOH (8 N) was used in jar tests to demon-
Traditionally, conducting jar tests is one of the most strate the impact of pH adjustment on PAC adsorptivity
common methods to evaluate coagulation efficiency. Re- in lime-softening basins.
cently, jar tests were used to evaluate criteria for en- Three different kinds of water were used in this study.
hanced coagulation compliance (U.S. EPA, 1993, 1994; Organic-free water was distilled water followed by treat-
Krasner and Amy, 1995; White et al, 1997; Vrijenhoek ment of GAC adsorption, ion exchange, and microfiltra-
et al, 1998). Jar tests are often used to estimate the PAC tion; (2) influent of pre-sedimentation tank; and (3) influ-
dose that is required to achieve the desired removal in ent of upflow flow-blanket reactors (primary basins in Fig.
conventional plants for test and odor control. In ajar test, 8), originally sampled from Florence Treatment Plant (FTP)
contact time, mixing energy, operational conditions such located on the Missouri River, Omaha, NE (see Table 1).
JAR TESTS FOR ATRAZINE REMOVAL 419
Equilibrium Isotherm Test ter, spiked with atrazine of 12 pg/L and PAC of 16 mg/L,
was used for these tests. For two different contact times,
To determine the equilibrium time for PAC to adsorb
the atrazine residual in the test bottle was analyzed un-
atrazine, 12 4.5-L amber glass bottles were used. Vol- der different velocity gradients (G values, calculated as
umes (1,000 mL) of organic-free water were spiked with
12 pg/L atrazine and treated with 16 mg/L PAC. These follows):
bottles were mixed at a constant 100 rpm using a Labline G =
VCdAu3/2vV (1)
model 208 magnetic stirrer and a 3" Teflon-coated stir-
where C<¡ 1.8, the drag coefficient of the paddle; A
= =
ring bar. The mixing was then stopped at different times 0.001875 m2 (=0.025 X 0.075 m), the area of the pad-
(ranging from 15 min to 20 days) for sampling and analy- dle; u rpm*0.236 m/60 s, the velocity of the paddles
=
in the beaker.
The second step was to conduct isotherm testing, us-
ing either 4.5-L glass bottles containing 1,000-mL vol-
umes of organic-free water. Four of these bottles were Chlorine Demand
spiked with 16 pg/L atrazine and treated with 2, 4, 8, The effect of PAC on chlorine demand was evaluated
and 16 mg/L PAC, respectively. The other four bottles
were spiked with 3 pg/L atrazine and treated with 2,
by adding PAC to some samples and analyzed total chlo-
rine using the Standard Methods, procedure 2350 B
4, 8, and 16 mg/L PAC, respectively. The bottles were (American Public Health Association, 1995). One-liter
sealed with Teflon-lined caps and stirred at 100 rpm
and at room temperature (25°C). After the 5-day equi- samples of the influent of upflow floc-blanket reactors
were placed in a water bath held at the temperature of
librium time was reached, samples were withdrawn, 16°C (the water temperature at FTP's intakes). Each sam-
centrifuged, and filtered through 0.22-¿im nylon mem- ple was dosed with a known amount of chlorine and PAC
brane filters (Millipore, Bedford, MA) for atrazine and placed in the dark for 1 h. After 1 h each sample was
analyses. titrated with an amperemetric titrimeter (Wallace & Tier-
nan model 311 ) to determine free and combined chlorine
Mixing Energy concentration. When added together, these parameters are
In this study, two series of tests—one with a specific referred to as the total chlorine in the solution. When the
contact time of 30 min and the other 60 min—were con- total chlorine value is subtracted from the original dos-
ducted with ajar-test apparatus (model #3461 made by ing amount of chlorine, the difference is the chlorine de-
Phipps & Bird, Inc., Richmond, VA). Organic-free wa- mand of the solution.
Jar Tests and Experimental Designs 507 (U.S. EPA, 1990). All of the jar tests were performed
in random order to eliminate errors arising from a pre-
Jar tests were to evaluate atrazine removal
designed determined sequence.
corresponding different
to PAC application points and
operating conditions. The following factors were tested: Jar tests individual factors. Two sets of
alum dose, PAC dose, and pH. Alum was 6 to 18 mg/L investigating
jar tests conducted
were to investigate independent im-
(midpoint 12 mg/L); PAC, 8 to 16 mg/L (midpoint 12
pact of each individual factor alone such as alum, PAC,
mg/L); pH was adjusted to 11.20 to 11.60 (midpoint and pH on atrazine removal (Table 2). Influent of pre-
11.40) for lime softening and 5.80 to 6.60 (midpoint 6.2)
for enhanced coagulation in the upflow floc-blanket re- sedimentation tanks and upflow floc-blanket reactors,
actors. These ranges and midpoints were selected based spiked with atrazine of 12 pg/L, were used in these jar
on conventional drinking-water treatment plants and in- tests, respectively.
corporated into the following experimental designs for
jar tests. Jar
investigating interrelationships among dif-
tests
ferent factors. In this study, a rotatable central compos-
Jar test procedures. The procedures are: (1) rapid mix ite design based on the response surface methodology
at 100 rpm for 3 min; (2) slow mix at 30 rpm for 12 min; (Montgomery, 1991) was used to design the jar tests to
(3) allow to settle for 15 min, decant, and filter; and (4) investigate the interrelationships among different factors
collect eluent for analysis of atrazine with EPA method (Tables 2 to 4). After accomplishment of each set of those
Table 2. Rotatable Central Composite Design and Experimental Results of Jar Tests for
Evaluation of Alum and PAC on Atrazine Removal in Pre-sedimentation Reactor"
aInitial atrazine concentration was 12 /¿g/L. Two sets of jar tests were conducted using influent of pre-sedimentation tank and
upflow floc-blanket reactors, respectively. In this experimental design, 0 means central point; "—1" can be obtained using semi-
range central point (6-12 —6); "+1" can be obtained using central point + semirange (12 + 6 18); "—1.414" can be ob-
= =
tained using (semirange) * 1.414 central point (6*1.414 —3.5); and "+1.414" can be obtained using central point +
—
12 =
bRun number is the design number, while the random number is the actual experimental number.
JAR TESTS FOR ATRAZINE REMOVAL 421
Table 3. Rotatable Central Composite Design and Experimental Results of Jar Tests for Evaluation
of pH and Alum on Atrazine Removal in the Upflow Floc-Blanket Reactors3
"Initial atrazine concentration was 12 pg/L, and influent of upflow floc-blanket reactors was used.
jar tests, Statgraphics (STSC, Inc., 1988) was used to (1) lutions were refrigerated. At least three standard solutions
conduct multiple regression using least squares to esti- were used each day in the operation of the gas Chro-
mate the regression models, which were approximated matograph (GC) to confirm the standard curve of five
using second-order polynomial functions, (2) produce standard points. If the daily variation in the slope of the
multiple X-Y-Z plots, and (3) produce response surface standard curve was higher than 10%, all standard solu-
plots (Emary, 1996). tions were injected into the GC, and a new standard curve
was determined. All of the data points presented in this
Analytical Methods and Quality paper are the results of duplicate samples. Precision was
Assurance/Quality Control assessed by analyzing duplicate samples taken from each
and comparing the results. The objective was
Atrazine was analyzed using EPA method 507 (U.S. experiment
a relative error of less than 10% between duplicate sam-
EPA, 1990). A 1-L sample was fortified with 50 pL of
If the difference was greater than 10%, the experi-
surrogate standard solution and extracted in a 2-L sepa- ples.
ment was repeated, and the cause of the discrepancy was
rately funnel. Samples were adjusted to pH 7 with a phos-
phate buffer. NaCl was added to improve the partition investigated.
coefficient, and the sample was extracted three times with
60 mL méthylène chloride. The méthylène chloride was
then concentrated and exchanged with methyl ierr-butyl RESULTS AND DISCUSSION
ether using a Kuderna-Danish apparatus. The sample vol-
Isotherm Data
ume was then adjusted to 1-mL and analyzed using a Tra-
cor 540 gas Chromatograph (Tremetric, Austin, TX) hav- It took approximately 5 days to reach the equilibrium,
ing a nitrogen-phosphorous detector with a DB-5 and the corresponding maximum atrazine removal was
capillary column. about 73% (Fig. 1). During the tests, there was no pH
Standard solutions of atrazine covering a range of 1 to change in the batch reactors, which may be due to the
20 pg/L were prepared in the stock solutions being tested low concentration of atrazine. Based on 5 days as the
and extracted using EPA method 507. The standard so- equilibrium time, the isotherm of atrazine adsorption on
"Initial atrazine concentration was 12 pg/L, and influent of upflow floc-blanket reactors was used.
100
14
Atrazine adsorbed
12
Atrazine removal 10
60 TO
O
8 E
<u
6 <D
c
Atrazine remaining 4
420
2
-10
100 150 200 250
Time (hr)
FIG. 1. Equilibrium test results of atrazine adsorption on PAC. The initial atrazine concentration was 12 pg/L, while PAC con-
centration was 16 mg/L. Experiment was conducted using 12 batch reactors.
JAR TESTS FOR ATRAZINE REMOVAL 423
PAC was accomplished. The Freundlich isotherm para- 5 to 100) resulted in an increase in atrazine removals from
meters, Un and K, were 0.852 and 162.39 (mg/g) 37 to 59%. It seems that, if the G value continuously in-
isotherm also creases (in this study, this was beyond the capability of
(L/mg)1/n, respectively. A Langmuir was
tried, but no reasonable parameters were found. the jar test apparatus), the atrazine removal will be reach-
Adams and Watson (1996) used Calgon Water Pow- ing its maximum removal. Therefore, mixing energy (the
dered Low Activity, a bituminous coal carbon similar to G value), instead of the hydraulic retention time (HRT),
pulverized Calgon F-200, and found that Un 0.44 to = is more important when the supplied HRT in a PAC con-
0.56, K 140 to 572 (mg/g) (L/mg)l/n, depending on the
= tactor is not sufficient for equilibrium to be reached.
initial pH in the batch reactor. Qi et al (1994) used WPH Qi et al. (1994) reported that the atrazine removal ef-
powdered activated carbon, and found Un 0.335, K
= =
ficiencies were between 30 to 65% for a reaction time
287.33 (mg/g) (L/mg)1/n (796.9 (/¿m/g) (L/pm)l,n). Najm between 15 and 100 min in a batch reactor with an ini-
et al. (1991a) reported that Un =
0.29, K 283.18 for
= tial atrazine concentration of 22.5 pg/L and a PAC dose
DDW water and Un 0.36, K 293.35 (mg/g)
= =
of 3.325 mg/L. Similar results were obtained in a batch
(L/mg)1/n for groundwater. reactor with an initial atrazine concentration of 252.4
The K value found in this study is comparable to the /xg/L and a PAC dose of 2.63 mg/L. The results of this
K values found in previous studies, but the Un value study are consistent with these previous results. How-
seems higher. The reasons for these differences may be ever, Qi et al (1994) also reported that the atrazine re-
due to the different PAC and the very low initial atrazine moval efficiencies were between 60 to 93%, for a reac-
concentrations (3 to 12 pg/L) used in this study. The tion time of 15 to 100 min in a batch reactor with an
lower initial atrazine concentration provides a lower sur- initial atrazine concentration of 22.5 pg/L and a PAC
face loading of atrazine (Qi et al, 1994), resulting in a dose of 8.4 mg/L, or 222 pg atrazine/L and a PAC dose
lower isotherm, which is the case in this study. There- of 12.88 mg/L. As one can see from Fig. 1, the PAC used
fore, the atrazine isotherm at a given initial concentration in this study can never achieve an atrazine removal effi-
in water cannot be represented accurately by one Freund- ciency higher than 73%. Therefore, in some cases, the re-
lich equation (Qi et al, 1994). The isotherm data obtained sults are not consistent with what Qi et al (1994) re-
in this study suggests that, in treatment facilities, atrazine ported.
removal by PAC be less than 73% (the removal achieved Traditionally, the dynamic processes of PAC adsorp-
at equilibrium) if Acticarb PAC is used, due to the short tion of organic compounds are studied using completely
retention time of treatment facilities. mixed batch reactors or CSTRs. Most of these previous
studies did not pay enough attention to mixing energy be-
Mixing Energy cause G values were not mentioned in these studies. For
Increasing the contact time from 30 to 60 min only re- a conventional flocculation basin with its G values be-
sulted in approximately 5% additional atrazine adsorp- tween 70 s"1 and 10 s_1, the atrazine removal of PAC
tion (Fig. 2). However, increasing the G value from ap- application should range from 34 to 44%. Because, based
proximately 4 to 321 s"1 (the corresponding rpms from on tracer studies, sludge residence times in the floccula-
Jar Tests So 36
E
To evaluate the best locations for PAC addition points,
two groups of jar tests were tested, and the results are re- 24
ported as follows.
12
Jar tests investigating individual factors. Virtually no
removal was seen with the alum coagulant alone using
influent of the pre-sedimentation and upflow floc-blan-
PAC dosage (mg/L)
ket reactors (Fig. 3a). When PAC alone was used,
atrazine removal ranged from 40.8 to 49.2% for pre-sed-
imentation water and 40.0 to 46.7% for upflow floc-blan-
ket reactors influent water and increased as PAC dosages
increased (Fig. 3b). When pH effect was evaluated, pH
adjustment was accomplished through the addition of 8
N NaOH during the rapid-mix segment of the jar test se-
quence. Elevated pH adjustment (pH = 11.2 to 11.6), as
seen in the lime-softening process, resulted in negligible
atrazine removal ranging from 0.5 to 2.6% (Fig. 3c). Be-
cause of the poor relationship between elevated pH and
atrazine adsorptivity, it was deemed unnecessary to ex-
amine the process using slaked lime. The softening
process would further inhibit the atrazine adsorption by
possibly encapsulating the PAC and thus, reducing the
sites available for adsorption.
FIG. 3. Effects of each individual factor on atrazine removal.
Jar tests investigating interrelationships among dif- (a) Alum coagulation only without addition of PAC; (b) PAC
ferentfactors. Four additional series of jar tests were con- only; and (c) pH adjustment only without PAC or alum.
ducted. One series investigated the effect of alum and
PAC together on atrazine removal using the influent of under conditions of enhanced coagulation with alum and
the pre-sedimentation tanks. The remaining three series pH adjustment (but without PAC addition); and (3) ex-
were conducted using the influent of upflow floc-blanket amine the impact of PAC, alum, and pH adjustment used
reactors. The objectives of these three series jar tests were together on atrazine adsorption efficiency.
to: (1) determine the impact of PAC and alum used to- Table 2 shows the rotatable central composite design
gether on atrazine removal; (2) evaluate atrazine removal and results of jar tests for evaluation of the effects of
JAR TESTS FOR ATRAZINE REMOVAL 425
alum and PAC together on atrazine adsorption. Based on conducted to examine the effect of alum, pH, and PAC
the results shown in Table 2, multiple regression using on atrazine adsorptivity (Table 4). Jar test procedures
Statgraphics was conducted. Two polynomial equations were the same as those described before. Based on the
were obtained to represent the response surface of results in Table 4 and the same procedure aforementioned
atrazine concentration remained (pg/L) after jar tests vs. for multiple regression using Statgraphics, a polynomial
alum concentration (mg/L) and PAC dosage (mg/L). equation to represent the response surface of atrazine con-
Equations (2) and (3) are for influent of pre-sedimenta- centration remained after jar tests (pm/L) vs. alum con-
tion tanks and upflow floc-blanket reactors, respectively. centration (mg/L), PAC dosage (mg/L), and pH (units)
can be obtained:
Atz. cone, remained =
8.56 0.20 *(PAC)
0.0030*(Alum)2 Atz. remained 1.03 0.027*(Alum)
- -
0.007*(Alum)2
-
0.21*(PAC)
-
0.00056*(Alum)2 (3)
- -
- -
0.0012*(Alum)*(PAC) +
The coefficient of determination, R2 is 0.87 for Equa- 0.034*(Alum)*(pH) + 0.064*(PAC)*(pH) (4)
tion (2) and 0.93 for Equation (3). The response surfaces The coefficient of determination, R2 is 0.95. The inabil-
are shown in Fig. 4. The statistical model fitting results, ity to plot a four-dimensional graph necessitates the use
ANOVA analysis, and other detailed information on mul- of box graphs denoting rotatable design representing the
tiple regression can be found in Emary's thesis (1996). three independent variables examined in this jar test se-
As PAC dosage increases, the remaining atrazine con- ries (Fig. 5).
centration after the test decreases (Fig. 4). Application of Lower pH (with H2S04) in conjunction with elevated
alum to pre-sedimentation tanks has almost no effect on PAC application dosages results in higher atrazine re-
atrazine removal (Fig. 4a). However, increase in alum moval (adsorptivity) (Fig. 5a and b). The lower pH val-
concentration in upflow floc-blanket water will decrease ues seem to increase the hydrophilic characteristics of the
atrazine removal (if pH is not a variable), which may be soluble atrazine (U.S. EPA, 1991), resulting in higher ad-
due to the encapsulation of PAC by coagulation and floc- sorption on the PAC. The higher PAC dosages make
culation (Fig. 4b). The difference in water quality para- more adsorption sties available. There is a net effect of
meters between the two locations does not impact greatly increase in atrazine removal (adsorptivity) with enhanced
the relationship among alum, PAC dosages, and atrazine coagulation (adjusting pH with H2SO4) (Fig. 5c).
adsorptivity. The best atrazine removal is about 50%, Atrazine removal can be higher than 60%, with a PAC
with the PAC dosage of about 16 mg/L. dosage of 16 mg/L and a pH adjustment to 5.8 (Fig. 5c).
The impact of enhanced coagulation alone on atrazine This synergistic relationship between the PAC dosage
removal was evaluated by adjusting the pH and varying and enhanced coagulation can be explained according to
the alum dosage. This series of jar tests did not include the the previous studies. The presence of natural organic mat-
addition of PAC. Influent of upflow floc-blanket reactors ter (NOM) significantly decreases the PAC capacity and
was spiked with 12 pg/L atrazine, and jar test procedures rate of adsorption of synthetic organic compounds (SOC)
were the same as those described before. Table 3 shows due to the competition between NOM and SOC (Najm
the rotatable central composite design and results of jar et al, 1990). Enhanced coagulation at pH about 5 to 6
tests. Based on Statgraphics analysis, the best coefficient significantly increases the removal of NOM compared
of determination, R2 found is 0.32. Therefore, there is no with coagulation at ambient pH (U.S. EPA, 1993; Kras-
good relationship among atrazine removal, alum, and pH. ner and Amy, 1995; Vrijenhoek et al, 1998). Generally,
As pH is adjusted and alum dosage is varied using upflow aluminum (or iron) coagulation can remove NOM ac-
floc-blanket reactors influent water, atrazine removal is not cording to two mechanisms: (1) adsorption onto alu-
enhanced. The highest atrazine removal value is 5.0%. This minum hydroxide or ferric hydroxide floe, and (2) for-
occurred at a pH of 5.8 and an alum dosage of 6 mg/L mation of insoluble complexes (aluminum or iron
(Table 3). Therefore, atrazine removal is negligible using humâtes or fulvates) in a manner analogous to charge
enhanced coagulation without PAC. neutralization (Krasner and Amy, 1995). Lower pH re-
Although coagulation did not demonstrate any appre- duces the charge density of humic and fulvic acids, mak-
ciable atrazine removal, literature (U.S. EPA, 1991) in- ing them more hydrophocic and adsorbable (Krasner and
dicates the stability of atrazine will be reduced at lower Amy, 1995; Kranser et al, 1996). The second mecha-
pH values. Miltner et al (1989) indicates that a syner- nism is relatively more dominant at lower dosages and
getic relationship exists between PAC and enhanced co- pH conditions. Therefore, the synergy effect of the PAC
agulation. Therefore, the fourth series of jar tests were dosage and enhanced coagulation is attributable to the in-
ENVIRON ENG SCI, VOL. 16, NO. 6, 1999
426 ZHANG AND EMARY
PflC (ng/L)
fllun (ng/L)
PAC (ng/L)
12 tó
AI un (ng/L)
FIG. 4. Response surface for evaluation of the interrelationship among PAC, alum, and atrazine removal using influent of (a)
pre-sedimentation tanks; and (b) upflow floc-blanket reactors.
hibition of NOM competition with atrazine for adsorp- flow floc-blanket reactors could be a much better PAC
tion sites. This inhibition is presumably due to the for- application point than pre-sedimentation basins if en-
mation of insoluble alum-humate complexes (pH at 5.8 hanced coagulation (with pH adjustment) is coupled with
and low alum dosage) (Hubel and Edzwald, 1987; Vri- PAC application. These results are consistent with the
jenhoek et al, 1998). These results indicate that the up- previous modeling prediction reported by Qi et al (1994).
JAR TESTS FOR ATRAZINE REMOVAL 427
nhiM (ag/L)
FIG. 5. Response surface for evaluation of the interrelationship among PAC, enhanced coagulation (alum and pH), and atrazine
removal in upflow floc-blanket reactors. Atrazine concentration remained in the test solution vs. PAC and pH at alum dosage of
(a) 12 mg/L and (b) 3 mg/L; and (c) atrazine removal vs. PAC and alum at pH 6.2.
=
PAC Application
PAC Application
Point # 4
Point Ü 2 „.„ CI, Application
PAC Application ,.
Point ft 3
.
Point H 2
Selection of PAC Application Points ganic compounds may be adsorbed that would otherwise
probably be removed by the treatment processes followed
Najm al (1991a) reported that, in a conventional
et up (Najm et al, 1991a), and thus the PAC usage rate in-
treatment plant, the common PAC application points are creases. In addition, some treatment plants apply a poly-
the plant intake, rapid mix, and filter. They further rec- mer at the river station at the
same time PAC would be
ommended that a continuous-flow slurry contractor that applied. Addition of the polymer would cause the PAC
precedes the rapid mix should also be considered. Qi et to drop out of solution on the way to the rapid mix basin.
al (1994) suggested that efficient use of PAC can be PAC application point before the filter is not recom-
achieved using an upflow floc-blanket reactor or a mended, because the proximity to the filtration process
PAC/UF process. To discuss the best point(s) for PAC would likely lead to breakthrough of the filter beds. In
addition, a hypothetical treatment train with four poten- addition, PAC application will increase CI2 demand. Fig-
tial PAC application points (Fig. 6) is used for the fol- ure 7 shows chlorine demand resulting from PAC addi-
lowing discussions. Note that a conventional drinking- tion, indicating that the impact of PAC on chlorine de-
water treatment plant may not contain some of the mand is very obvious, and depends on PAC dosages.
reactors. The advantage of PAC application point #1 is the com-
Application of PAC at the river intake may provide bined retention time (about 60 min) resulting from the
long contact time and good mixing. However, some or- CSTR in series with the flocculation basin. A CSTR typ-
-o 3
<D 2
O
1 1
8 12 16
PAC dosage (mg/L)
FIG. 7. Chlorine demand resulting from PAC addition.
JAR TESTS FOR ATRAZINE REMOVAL 429
ically has a 30-min retention time, while a rapid mixing tional contact time, and the rapid mix process would en-
tank (point #2) may have a contact time less than a few courage more contact between the atrazine and the PAC.
minutes. Higher mixing energies could be achieved Because coagulant is added in the conventional treatment
through this arrangement, resulting in higher atrazine re- scheme at the rapid mix chamber, the opportunity for en-
moval (Fig. 2). The carbon is completely mixed with the hanced coagulation (adjustment of the pH to 5.6 to 6.8)
water in the CSTR before the water flows to the rapid may exist.
mixing basin, where it is coagulated, and then to the floc-
culation and sedimentation basins.
Another advantage of the PAC application point #1 is Case Study
the distance in the treatment train from the first CI2 ap- The results obtained from the aforementioned tests can
plication point. The discovery that trihalomethanes be applied to analyze the best PAC application point(s)
(THMs) can be formed during drinking-water treatment at FTP. The FTP is a split treatment lime-softening/alum
(Rook, 1974) has caused major concern due to the po- coagulation facility (Fig. 8). Cationic polymer coagulant
tential health hazard associated with these compounds. is adding to the river intake. Water then enters pre-sedi-
An increase in chlorine demand results in higher THM mentation tanks with an HRT of 18 to 24 h. The mixing
production. For this reason the proximity of the PAC ap- basins are baffled basins without rapid mixing function.
plication to the chlorination application is very impor- The basin #6 is for contact and settling. The mixing basins
tant. To minimize chlorine demand as a result of PAC and basin #6 were built for the old treatment plant be-
application, the PAC should be maintained as far from fore the primaries were built, and do not function specif-
the CI2 application point #1 as possible. Therefore, PAC ically. Water leaves basins #6 and is divided into the four
application point #1 is advantageous over PAC point #2. primaries (four upflow floc-blanket reactors). Primaries
However, if the treatment plant does not have a CSTR, #1 and #4 are the alum coagulation basins, while pri-
PAC application point #2 can also be used. maries #2 and #3 are lime softening. The water is then
If a treatment plant does not have a CSTR prior to the recombined in a center flume and enters settling basin #3
rapid mixing basin but has upflow floc-blanket reactors, for further blending. The "D" basin, a baffled tank, is
as FTP (see Case Study), point #4 may be the best loca- used for adding anionic polymer for the second stage of
tion for PAC application due to the following reasons. coagulation during runoff time. Basins #2 and #1 are for
(1) Based on the results obtained in this study, a syner- secondary sedimentation and stabilization of alkalinity.
gistic relationship was demonstrated between PAC and The filters and clearwells are conventional.
enhanced coagulation (with a pH adjusting to around 5.8), Four potential PAC application points exist: (1) mix-
and the best atrazine removal (greater than 60%, Fig. 5) ing basin, (2) "D" basin, (3) the pre-sedimentation basins,
could be achieved. (2) The maximum adsorptive capac- and (4) the upflow floc-blanket reactors upflow basins
ity of the carbon in the floc-blanket reactor would be (primaries). The current application point for PAC is the
achieved due to a longer carbon resident time (e.g., 9 h pre-sedimentation basins.
to several days, Najm et al, 1991a; Qi et al, 1994) com- Based on results of this study, the mixing basin is not
pared to 30 to 60 min carbon residence times in the pre- recommended as a PAC application point due to the mix-
sedimentation basins. Mixing energy should be as high ing energy not being high enough to keep the PAC sus-
as the floc-blanket will tolerate (Fig. 2) to obtain the pended. The "D" basin is not recommended for a PAC
highest atrazine removal. (3) PAC adsorption of atrazine addition because carbon applied at this location eventu-
or other organic compounds can be enhanced through a ally breaks through the filter beds and can be detected in
pH adjustment. (4) The need for a major capital expen- the clear well (Fig. 8).
diture required to build a PAC contact facility (e.g., Comparison of jar tests results shows that the best
rapid mix contactor, CSTR, etc.) can be avoided. (5) atrazine removal at pre-sedimentation tanks can be about
There is a potential to regulate the amount of water in 50%, while that at upflow floc-blanket reactors can be
contact with PAC by adjusting the split in the upflow higher than 60%, with a PAC dosage of 16 mg/L and a
floc-blanket reactors (see Case Study). However, more pH adjustment to 5.8. Therefore, for FTP, the upflow
studies are needed to evaluate whether point #1 is bet- floc-blanket reactor would be the best PAC application
ter than point #4. point.
For a treatment plant that has no CSTR reactor and no More studies are needed if the upflow floc-blanket re-
upflow floc-blanket reactor followed by a settling basin, actorwith chlorine addition at its effluent is used as a
points #1 and #4 do not exist, and point #2 will be the PAC application point. Currently, it is not very clear how
best PAC application point. Application of PAC prior to much PAC will escape for the effluent of the upflow floc-
the rapid mix basin would have the advantage of addi- blanket reactor. In FTP, additional jar tests and a field-
CI. Addition
Alum
Polymer
Low Basin # 6
River Intake Mixing
Service Contact &
Water Screens Basins
Pumps Settling
PAC PAC
CI. Addition
Pump
Station
Basin tt 3 D' Basin Basin # 2 Basin # I Filters
H Blending TTÎ1
Mixing Settling Settling
Basin # 1
rn
PAC Storage Pump
Fluoride (Clearwell) Station
To Distribution
Svstem
FIG. 8. Schematic of the Florence Treatment Plant. Three potential PAC application points shown in the figure. In this study,
it was hypothesized that primary basins 1 and 4 (upflow floc-blanket reactors) would be a better PAC application point.
scale test conducted in the upflow floc-blanket reactors of organiccompounds using batch or CSTRs did not fo-
showed that some PAC did escape from the reactors. Ox- cus on mixing energy.
idizing agents (such as CI2) can greatly reduce the ad- Based on jar tests and response surface technology, it
sorbing potential of PAC and increase the disinfectant was found that (1) lime-softening processes alone will
demand. If the oxidizing disinfectant is chlorine, tri- not enhance atrazine removal; (2) alum coagulant alone
halomethane production may be increased, risking the or enhanced alum coagulation without the addition of
noncompliance of another U.S. EPA regulation (Federal PAC will result in a minimal increase of atrazine removal
Register, 1979). Therefore, the location of chlorine ap- (less than 5%); and (3) a synergistic relationship appears
plication point after the upflow floc-blanket reactor at to exist between PAC dosage and pH adjustment occur-
FTP may need to be moved. ring with enhanced coagulation; neither PAC nor en-
hanced coagulation (with a pH adjusted to 5.8) resulted
in as high a removal rate of atrazine as the two did to-
SUMMARY gether (greater than 60%). Therefore, the two most im-
portant variables impacting atrazine removal are PAC
The results obtained in this study demonstrate that jar and enhanced coagulation (with a pH around 5.8).
tests can be used to evaluate PAC performance for Selection of the best point for PAC application depends
atrazine removal. The efficiency of PAC application for on treatment trains and chlorination points. For a con-
atrazine removal depends on many factors. In treatment ventional treatment plant with a CSTR or rapid mixing
facilities, the adsorption of atrazine with PAC is usually tank before flocculation and sedimentation basins, appli-
a dynamic process, with an atrazine removal less than the cation of PAC to the CSTR or rapid mixing would be a
removal achieved at equilibrium, due to the short reten- good selection. For a treatment plant with upflow floc-
tion time. Specifically, this study demonstrates that mix- blanket reactors, as in the Florence Treatment Plant, the
ing energy is a major factor affecting atrazine adsorption. best point for PAC application could be at the upflow
Traditional studies on PAC performance for adsorption floc-blanket reactors. In addition, the distance that a PAC
JAR TESTS FOR ATRAZINE REMOVAL 431
application point is upstream from a chlorine application HABERER, K., and NORMANN-SCHMIDT, S. (1991). The
point should be maximized to avoid an increase in chlo- Haberer process: Combining contact flocculation. Filtration,
rine dosage. and PAC adsorption. J. AWWA 83, 82-89.
HAND, D.W., CRITTENDEN, J.C., and THACKER, W.E.
(1983). User-oriented batch reactor solutions to the homo-
ACKNOWLEDGMENT geneous surface diffusion model. J. Environ. Eng (ASCE)
109,82-101.
The authors acknowledge the following people for
HOPMAN, R., VAN DER HOEK, J.P., VAN PAASSEN, J.M.,
their assistance in this study: Li Li, and Jim and Patty and KRUITHOF, J.C. (1998). Impact of NOM presence on
Trebbien for providing computer assistance and statis- removal Problems and solutions.
pesticide by adsorption:
tical calculation; Drs. J. Stansbury, M. Moussavi, and Water Supply 16, 497-501.
Ms. A. Blakey of the University of Nebraska-Lincoln
HUBEL, R.E., and EDZWALD, J.K. (1987). Removing tri-
for their useful suggestions and comments about the halomethane precursors 79,
by coagulation. J. AWWA
project and the manuscript. The authors appreciate the 98-107.
Florence Treatment Plant for providing laboratory
KASSAM, K., HUCK, P.M., VAN ROODSELAAR, A., and
equipment and chemicals, and the National Science SHARIFF, R. (1991). Accumulation and adsorption capac-
Foundation/Experimental Program to Stimulate Com- ity of PAC in a slurry recirculating clarifier. / AlVTVA 83,
petitive Research (NSF/EPSCoR) for partially funding 69-78.
the project.
KRASNER, S.W., CROUE, J.-P., BUFFLE, J., and PERFUE,
E.M. (1996). Three approaches for characterizing NOM. J.
REFERENCES AWWA 88, 66-79.
Effect of initial concentration of a SOC in natural water on U.S. EPA. (1994). National primary drinking water regulations;
its adsorption by activated carbon. J. AWWA 83, 57-63. Disinfectants and disinfection by products; Proposed rule.
Fed. Reg. 59, 145:38668.
NAJM, I.N., SNOEYINK, V.L., and GALVIN, T.L. (1991c).
Control of Organic Compounds with Powdered Activated U.S. EPA. (1993). Draft Guidance Manual for Enhanced Co-
Carbon. Report to AWWA Res. Found. agulation and Enhanced Precipitative Softening. Washing-
ton, DC: U.S. EPA.
NAJM, I.N., SNOEYINK, V.L., SUIDAN, S.T., LEE, C.H.,
and RICHARD, Y. (1990). Effect of particle size and back- U.S. EPA. (1992). Another Look: National Survey of Pesticides
ground natural organics on the adsorption efficiency of PAC. in Drinking Water Wells, Phase II Report. USEPA, Office
J. AWWA 82, 65-76. of Water and Office of Pesticides and Toxic Substances, EPA
and Treatment. New
579/09-91-020, 1992.
PONTIUS, F.W. (1990). Water Quality
York: McGraw-Hill. U.S. EPA. (1991). Pesticide Fact Sheet. U.S. EPA Office of
Pesticide Programs.
QI, S., ADHAM, S.S., SNOEYINK, V.l., and LYKINS, B.W.,
Jr. (1994). Prediction and verification of atrazine adsorption U.S. EPA. (1990). Methods for the Determination of Organic
by PAC. J. Environ. Eng. (ASCE) 120, 202-218. Compounds in Drinking Water, Supplement I.
RICHARDS, R.P., BAKER, D., CHRISTENSEN, B.R., and U.S. EPA. (1976). Powdered Activated Carbon Treatment for
TIERNEY, D.P. (1995). Atrazine exposures through drink- Removal of Trace Organic Compounds From the Now Or-
leans Water Supply. USEPA Grant R-804404.
ing water: Exposure assessments for Ohio, Illinois, and Iowa.
Environ. Sei. Technol. 29, 406-421. U.S.G.S. (1991). Distribution of selected herbicides and nitrate
ROOK, J.J. (1974). Formation of haloforms during chlorina- in the Mississippi River and its major tributaries, April
tion of natural waters. Proc. Water Treatment Exam 23, through June 1991. U.S.G.S. Water-Resources Investigation
234-243. Report 91-4163, p. 1035.
SCHOTTLER, S.P., EISENREICH, S.J., and CAPEL, P.D. VRIJENHOEK, E.M., CHILDRESS, A.E., ELIMELECH, M.,
(1994). Atrazine, alachlor, and cyanazine in a large agricul- TANAKA, T.S., and BEUHLER, M.D. (1998). Removing
tural river system. Environ. Sei. Technol. 28, 1079-1089. particles and THM precursors by enhanced coagulation. J.
AWWA 90, 139-150.
SOBRINHO, J.A.H., THIEM, L.T., and ALKHATIB, E.A.
(1997). Benzene removal by PAC in jet flocculation system. WHITE, M.C., THOMPSON, J.D., HARRINGTON, G.W., and
J. Environ. Eng. (ASCE) 123, 1011-1018. SINGER, P.C. (1997). Evaluating criteria for enhanced co-
agulation compliance. J. AWWA 89, 64-77.
STSC, Inc. ( 1988). Statgraphics Manual. Englewood Cliffs, NJ:
Statistical Graphics Co. Corresponding author:
Tian Cheng Zhang
STUKENBERG, J.R., and HESBY, J.C. (1991). Haberer
process in the United States. J. AWWA 83, 90-96. Department of Civil Engineering
205D PKI
TRAEGNER, U.K., and SUIDAN, M.T. (1991). Steady state
University of Nebraska-Lincoln Omaha Campus
at
performance of activated carbon contactors. Water Sei. Tech- Omaha, NE 68182-0176
nol. 23, 1677-1686.
TRAEGNER, U.K., and SUIDAN, M.T. (1989). Parameter phone: 402-554-3784
evaluation for carbon adsorption. J. Environ. Eng. (ASCE) fax: 402-554-3288
115, 109-128. E-mail: tzhang@unomaha.edu