- Email: [email protected]
The impact of ferrous ion reduction of chlorite ion on drinking water process performance
PII: S0043-1354(01)00172-5
Wat. Res. Vol. 35, No. 18, pp. 4464–4473, 2001 r 2001 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/01/$ - see front matter
RESEARCH NOTE THE IMPACT OF FERROUS ION REDUCTION OF CHLORITE ION ON DRINKING WATER PROCESS PERFORMANCE RICHARD HENDERSON, KENNETH CARLSON* and DEAN GREGORY Department of Civil Engineering, Colorado State University, Fort Collins, CO 80521, USA (First received 1 April 2000; accepted in revised form 1 March 2001) AbstractFThe use of chlorine dioxide (ClO2) as a primary disinfectant and pre-oxidant in drinking water treatment is being explored as an alternative to chlorine for reducing disinfection by-product formation and to assure compliance with United States Environmental Protection Agency’s Stage 1 Disinfection/ Disinfection By-Products Rule. However, the ClO2 by-product chlorite ion (ClO@ 2 ) is also regulated by the same regulation. Ferrous iron (Fe(II)) has been shown to effectively reduce chlorite ion to chloride ion (Cl@) and this study was conducted to evaluate the impact on overall treatment process performance due to the ferric hydroxide solids that form from the reaction. Ferrous iron application was explored at three different points in a pilot-scale water treatment system: pre-rapid mix, pre-settling and pre-filter. Chlorite ion concentrations were effectively reduced from 2 mg/L to less than 0.3 mg/L using an Fe(II) dose of approximately 6 mg/L for all trials. Fe(II) addition at the rapid mix caused no adverse effects and, in fact, allowed for reduction of the alum dose due to the newly formed ferric hydroxide acting as a supplemental coagulant. An increase of 241 and 247% of total suspended solids influent to the filter process was observed when Fe(II) was applied at the pre-settling and pre-filter locations. Pilot-scale filter runs during these trials were less than 2 h and never obtained true steady state conditions. Jar testing was performed to better understand the nature of the ferric hydroxide solids that are formed when Fe(II) was oxidized to Fe(III) and to explore the effectiveness of Fe(II) addition at intermediate stages in the flocculation process. r 2001 Elsevier Science Ltd. All rights reserved Key wordsFwater treatment, chlorine dioxide, chlorite ion, reduction, ferrous ion
INTRODUCTION
The first application of chlorine dioxide (ClO2) in a United States drinking water system was in Niagara Falls, NY in 1944 for taste and odor control (USEPA, 1978). Presently, ClO2 is used for disinfection (CT credit), taste and odor control, iron and manganese removal, and control of hydrogen sulfide (USEPA, 1999). The use of ClO2 as a primary disinfectant is of particular interest because it does not produce the organic disinfection by-products common to conventional chlorination practices. In particular, Werdehoff and Singer (1987) showed that the application of ClO2 does not produce trihalomethanes (THMs) and produces only a small amount of total organic halide (TOX). ClO2 is primarily an oxidizing agent, not a chlorinating agent. Gates (1998) described the chemistry as being ‘‘free radical electrophilic (i.e. electron attracting) abstraction’’ and not ‘‘oxidative substitution or addition (as in chlorinating agents)’’. Chlorine dioxide has also been proven to be effective for inactivation of chlorine-
resistant pathogens such as Giardia and Cryptosporidium (White, 1999; Finch et al., 1999). The United States Environmental Protection Agency (USEPA) promulgated Stage 1 of the Disinfection/Disinfection By-Products Rule (D/DBP) Federal Register, 1998). This regulation established the maximum contaminant levels (MCL) of 0.080 and 0.060 mg/L for THMs and haloacetic acids (HAAs), respectively. This has prompted some water utilities to study ClO2 as a primary disinfectant to reduce the formation of THMs and HAAs during the primary disinfection process. Although ClO2 does not produce THMs, it does produce chlorite ion (ClO@ 2 ). The Stage 1 D/DBP Rule sets the MCL for ClO@ at 2 1.0 mg/L limiting the dose of ClO2 and, therefore, the applicability of ClO2 as a disinfectant. United States water utilities serving more than 10,000 people will need to be in compliance with this regulation by January 1, 2002, while compliance for smaller utilities is required by January 1, 2004.
BACKGROUND
*Author to whom all correspondence should be addressed. E-mail: [email protected]
Three approaches for removing ClO@ have been 2 explored in recent years. Two of these involve the
4464
Ferrous reduction of chlorite
chemical reduction of ClO@ 2 , one through the use of sulfite ion (SO2@ ) and one using ferrous ion (Fe2+ 3 or Fe(II)). The third technique that has been explored is the use of granular activated carbon (GAC) as a reductant. The use of sulfite ion has been proven to be somewhat limited due to the reaction’s dependency on pH. The reaction is efficient over the pH range of 5–6.5, but above pH=7 it slows considerably (White, 1999; Gregory et al., 1997). Another disadvantage of SO2@ is the formation of chlorate ion (ClO@ 3 3 ) from the reaction. Chlorate ion has been shown to form in pH ranges of 4–8.5 by Griese et al. (1991) and 8.5–11 by Suzuki and Gordon (1978). GAC has also been shown to be extremely limited because of the short life of the carbon when used for this purpose, especially if the carbon is preloaded (Dixon and Lee, 1991; Ellenberger et al., 1998). The use of Fe(II) appears to be the most promising of the ClO@ removal techniques and has been 2 successfully used in laboratory and, to some extent, in pilot and full-scale studies (Griese et al., 1991, 1992; Iatrou and Knocke, 1992; Hurst and Knocke, 1997; Rittman, 1997). The reaction stoichiometry for the reduction of ClO@ 2 by Fe(II) is given in equations (1)–(3). 12H2 O þ 4Fe2þ "4FeðOHÞ3 þ 4e@ þ 12Hþ
ð1Þ
4Fe2þ þ ClO@ 2 þ 10H2 O"4FeðOHÞ3ðsÞ þ Cl@ þ 8Hþ @ @ þ 2H2 O 4Hþ þ ClO@ 2 þ 4e "Cl
ð2Þ ð3Þ
Based on equation (3), 3.3 mg of Fe(II) is needed to reduce 1 mg of ClO@ 2 . In laboratory experiments, Iatrou and Knocke (1992) determined the ratio to be approximately 3.1 mg Fe(II)/mg ClO@ 2 . This result suggests that the majority of ClO@ is reduced to 2 chloride, a 4 electron transfer reaction. The study also concluded that the reaction was complete within 5–15 s (at pH=5–7) and that the presence of Fe(OH)3(s) did not adversely affect alum as a coagulant during jar test experiments. Lastly, it was concluded that any excess Fe(II) would be oxidized by dissolved oxygen when the pH was greater than 7. Later experiments by Hurst and Knocke (1997) explored the use of Fe(II) under alkaline conditions (pH 8–10) similar to those used in a typical softening process. Ferrous ion proved to be effective with dosages 10–20% above the stoichiometric requirement. The only adverse effect was that high dissolved oxygen concentrations at pH=10 inhibited the reaction. Also explored in these experiments was the influence of dissolved organic carbon (DOC) on the Fe(II)/ClO@ 2 reaction. No effect was observed at a pH of 8, but significant and substantial effects were seen at pH=9 and 10, respectively. Further experimentation involving nitrogen stripping of oxygen led to the hypothesis that the DOC may act to enhance
4465
the oxidation of ferrous by oxygen. The last consideration for Fe(II) reduction of ClO@ is that 2 of ClO@ 3 formation. Studies have shown that, in pH ranges of 5–10, ClO@ formation is minimal and 3 levels are approximately equal to those in the influent (Griese et al., 1991; Iatrou and Knocke, 1992; Hurst and Knocke, 1997).
RESEARCH PURPOSE AND OBJECTIVE
The use of existing infrastructure (flocculation and settling basins) for ClO2 disinfection contact time (and the associated CT credit) is an appealing situation. Increasing the amount of disinfection that can be gained in this manner would increase the number of treatment plants that could reduce the formation of THMs and HAAs by replacing chlorine with ClO2 as the primary disinfectant. For example, plants that are already using ClO2 for pre-oxidation of iron and manganese (which is normally done before the rapid mix) would be able to raise the oxidant dose and achieve adequate disinfection credit with their existing system. However, application of Fe(II) later in the treatment process may have its own consequences. The ferric hydroxide solids that are formed when Fe(II) is oxidized may impact settling and filtration. Ferrous ion has been applied successfully in full-scale operation at the rapid mix location in El Paso, Texas (Rittman, 1997), but studies exploring the application of Fe(II) after the rapid mix have been somewhat limited. Griese et al. (1992) applied Fe(II) post-settling in pilot plant studies, but the reaction occurred in a contact chamber that was designed for a 30-min hydraulic residence time. In addition, secondary flocculation and sedimentation basins were used to lessen the impact of the ferric hydroxide solids on the filtration process. The main objective of this research was to evaluate the impact of Fe(II) reduction of ClO@ on overall 2 process performance using a pilot-scale treatment plant. In particular, the study focused upon the settleability and filterability of the ferric hydroxide solids that are formed as a result of the oxidation/ reduction reactions. Pre-filter total suspended solid concentration (TSS) and turbidity measurements were used to evaluate settleability. Pilot-scale filter performance was evaluated with finished water particle counts, turbidity, and iron concentration measurements. The experimental plan involved the application of Fe(II) at the rapid mix and at presettling and pre-filter locations with no modifications (other than the Fe(II) feed) to a pilot-scale drinking water treatment plant. As mentioned earlier, Iatrou and Knocke (1992) concluded that the Fe(II)/ClO@ 2 reaction is complete within 15 s when the pH is between 5 and 7. Because the reaction pH during this study was within the range cited by Iatrou and Knocke, a separate reaction basin for chlorite
4466
Richard Henderson et al.
reduction was not required. Fe(II) was also applied slightly under the stoichiometric values to further ensure that unreacted Fe(II) was not influent to the filter. This was important because Fe(II) being oxidized in the filter media would have added an extra variable to the study.
METHODS AND MATERIALS
Pilot scale experiments All of the pilot plant research was conducted using a 10 gal/min (37.85 L/min) pilot plant located at the Engineering Research Center, Colorado State University, Fort Collins, Colorado. The plant consists of a rapid mix chamber, three-stage flocculation basin, lamella plate settling basin, and two dual media filters consisting of 1 ft (0.30 m) of sand covered with 2 ft (0.61 m) of anthracite (Fig. 1). Hydraulic loading rates (HLR) for the settling basin and filter were 0.4 gal/min/ft2 (16.3 L/min/m2) and 5 gal/min/ft2 (203.7 L/min/m2), respectively. On line flow meters (Signet, Model 3-8500) were used to monitor influent and filter flows throughout the course of the tests. Flow rates were manually checked at the start of each filter run and at least once during the run for each of the following chemical feeds: liquid alum (General Chemical), LT-22 cationic polymer (Ciba Specialty Chemical Water Treat@ (80% ments, Inc.), sodium bicarbonate (HCO@ 3 ), ClO2 technical grade sodium chlorite, Aldrich Chemical Company, Inc.), and Fe(II) (ferrous sulfate (FeSO4), 5.08% ferrous, specific gravity=1.2, Kemiron, Inc.). Fe(II) locations given in Fig. 1 indicate the different application points that were explored during this study. The source water for the plant came from Horsetooth Reservoir, Fort Collins, Colorado. Influent pH (Fischer Scientific, Accumet, pH meter 50), alkalinity (APHA, 1998, Standard Method 2320A), on-line turbidity (Hach 1720D Turbidimeter), on-line particle counts (Hach 1900 WPC Particle Counter), and TOC concentration (Rosemont Dohrman DC-80 Total Organic Carbon Analyzer) were measured during the study. The ranges of these parameters can be found in Table 1. Table 2 gives the experimental trial sequence along with the chemical additions for each experimental trial. This sequence was chosen so that
comparisons between control runs and experimental runs could be made with the assurance that the water quality was not changing significantly through the course of the study (January 22, 2000 through March 10, 2000). The protocol of alternating control runs with experimental runs was ended once three control runs had been completed and once it was realized (as will be discussed later) that the pre-settling and pre-filter FeSO4 application experiments could be completed within a short period of time. solution was pumped into the raw A 7000 mg/L ClO@ 2 water at a rate of 10.8 mL/min in order to obtain a 2 mg/L concentration. Chlorite ion was measured directly ClO@ 2 before Fe(II) addition and in the finished water. Amperometric titration Method II (APHA, 1998, Standard Method E) was used for ClO@ determination. The 4500-ClO@ 2 2 analyses were completed within 60 min following ClO@ 2 sample collection. Pre-filter samples were collected and analyzed for TSS concentration according to Standard Method 2540 D (APHA, 1998). Finished water iron levels (Fe(II) and total) were determined using spectrophotometric techniques (Hach Corporation 1,10-phenanthroline method, Hach DR/3000 Spectrophotometer). Ferric ion (Fe(III)) concentrations were determined by taking the difference between the Fe(II) and total iron concentrations. Iron analyses were completed within 30 min of sample collection. On-line particle counts (Hach 1900 WPC Particle Counter) were measured for the finished water. Pre-filter and finished water turbidities (Hach 1720D Turbidimeter) were also measured on-line. The filter-ripening time was defined as the time after filter startup when the number of particles greater than 2 mm exiting the filter decreased to 20 counts per mL or less. Filter breakthrough was defined as
Table 1. Horsetooth reservoir water qualitya Water quality parameter
Range
pH Alkalinity (mg/L as CaCO3) Turbidity (NTU) Total organic carbon (mg/L)
7.2–7.6 26.5–31 1.9–2.1 3.7–4.3
a
January 22, 2000–March 10, 2000.
Fig. 1. Configuration of pilot plant at Colorado State University, Fort Collins, Colorado.
Ferrous reduction of chlorite
4467
Table 2. Experimental sequence and chemical additionsa Trial sequence
Control #1 Fe(II) applied Control #2 Fe(II) applied Control #3 Fe(II) applied Fe(II) applied Fe(II) applied Fe(II) applied a
at rapid mix #1 at rapid mix #2 pre-settling #1 pre-settling #2 pre-filter #1 pre-filter #2
Alum (mg/L)
LT-22 Flocculant AID (mg/L)
Bicarbonate (mg/L as HCO@ 3 )
Chlorite (mg/L)
Ferrous (mg/L as Fe2+)
22 10 22 10 22 22 22 22 22
0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
15 15 15 15 15 15 15 15 15
0 2 0 2 0 2 2 2 2
0 6.2 0 6.2 0 6 6 6 6
Bicarbonate was added to raise alkalinity and to provide some buffering for the addition of ferrous sulfate (pHo2).
the point where particle counts of particles greater than 2 mm increased above 20 counts per mL. Finished water pH (Fischer Scientific, Accumet, pH meter 50) and alkalinity (APHA, 1998, Standard Method 2320A) were also measured. All on-line measurements were taken at 1-min intervals. Headloss was monitored throughout the course of the study. An upper limit range of 6.5–7 ft was set for run termination if particle breakthrough did not occur. However, the particle count breakthrough limits mentioned earlier were the limiting factor in all trials.
Bench-scale experiments Standard jar tests (Phipps and Bird PB-900 Paddle Stirrer) were conducted as a means of evaluating the benefits of FeSO4 addition at different stages in the treatment process and to facilitate analyses of data collected during the pilot-scale experiments. The mixing and settling sequence for the jar tests was as follows: (1) 300 rpm for 40 s; (2) 48 rpm for 10 min; (3) 32 rpm for 10 min; (4) 18 rpm for 10 min; and (5) 15 min of settling. Settled water turbidities were then recorded (Hach 2100N Turbidimeter). Alum and ferric chloride were both evaluated as primary coagulants at 3 mg/L intervals ranging from 7 to 25 mg/L. Chlorite ion were introduced to the raw water using the and HCO@ 3 same stock solutions as those used in the pilot plant portion of the study. Likewise, the FeSO4 solution used for ClO@ 2 reduction was the same as that used in the pilot plant portion of the study. Lastly, the addition of iron as Fe(III) (6 mg/L) was compared to the addition of iron as Fe(II) (6 mg/L) at an alum dose of 22 mg/L. These experiments were conducted to explore whether the freshly formed Fe(III) from Fe(II) oxidation behaves similarly (in terms of settleability) to pre-formed Fe(III) from a ferric chloride solution.
Data analysis Samples for measurement of ClO@ 2 , TSS concentration, pH, and alkalinity were taken once during the steady state (or point of lowest particle counts if there was no true steady state) of each pilot plant trial. Effluent samples were collected for iron analyses during the steady state and particle breakthrough periods for the trials in which Fe(II) was applied pre-settling and pre-filter. All samples were tested in triplicate. Error bars that are given on the graphical representations of the data represent the minimum and maximum values of these analyses. Each bench-scale scenario was run in triplicate. Average values given for settled water turbidities represent the average of the three independent settled water samples.
Fig. 2. Chlorite removal during application of Fe(II) at the rapid mix and at pre-settling and pre-filter locations.
RESULTS AND DISCUSSION
Pilot scale experiments Chlorite ion was effectively reduced to concentrations lower than 0.3 mg/L for all trials (Fig. 2). Ferrous sulfate dosing was reduced slightly (from 6.2 mg/L as Fe(II) to 6.0 mg/L as Fe(II)) when applied at pre-settling and pre-filter locations to ensure that no unreacted Fe(II) was influent to the filters. As discussed earlier, this was done as a precautionary measure to ensure that the extra variable of Fe(II) being oxidized in the filter, possibly causing negative effects, was eliminated. The finished water pH values (less than 7 for all trials) were within the range cited by Iatrou and Knocke (1992) where Fe(II) oxidation is both rapid and complete, thus adding extra confirmation that no oxidation of Fe(II) should have taken place in the filter media. These data were also supported by effluent iron concentrations, which are discussed at a later point in this paper. Figs 3(A)–(D) give particle count data for selected examples of each type of trial (i.e. no Fe(II), Fe(II) applied at the rapid mix, Fe(II) applied pre-settling, and Fe(II) applied pre-filter). Based upon the 20 counts per mL particle count criteria set as lower and upper limits of ripening and breakthrough, respectively, the three control experiments averaged run
4468
Richard Henderson et al.
Fig. 3. Examples of raw and finished water particle counts (>2 mm) for each experimental condition.
times of 21.3 h (range=19.5–22.5 h) and yielded an average unit filter run volume (UFRV=gal/ft2 of filter area/trial) of 6400 gal/ft2 (263,304 L/m2). The two trials in which FeSO4 was applied at the rapid mix produced similar results (run times =22.3 and 21.9 h, average UFRV=6630 gal/ft2=272,767 L/m2) to those of the controls. The addition of FeSO4 at the rapid mix also allowed for reduction of the alum dose from 22 to 10 mg/L (Fig. 3(B)). This was due to the ferric hydroxide acting as a supplemental coagulant.
These results are supported by the full-scale operation in El Paso, Texas where ferrous chloride is added at the rapid mix and the resulting ferric hydroxide solids are used as the primary coagulant (Rittman, 1997). When FeSO4 was applied at the pre-settling (Fig. 3(C)) and pre-filter (Fig. 3(D)) locations, particle counts less than 20 counts per mL were barely obtained. Within 2 h of filter start up of these trials, particle counts increased to greater than 100 counts per mL.
Ferrous reduction of chlorite
4469
Fig. 4. Examples of raw, pre-filter, and finished water turbidity data for each experimental condition.
Selected examples of turbidity data for each type of experimental trial are given in Figs 4A–D. Finished water turbidities for the control (Fig. 3(A)) and for the trials when FeSO4 was applied at the rapid mix (Fig. 3(B)) were consistently below 0.07 NTU during the steady state portion of the filter run. Ferrous sulfate application at the pre-settling and pre-filter locations produced finished water turbidities that
were equal to or less than 0.1 NTU for only a brief time (Figs 3C, D). Post-settling turbidity measurements averaged 4.5 NTU (FeSO4 applied presettling) and 5.1 NTU (FeSO4 applied pre-filter). This was an observed increase of 157 and 191%, respectively, when compared to the average of 1.75 NTU for the control runs. The marginal difference between these two values indicates that the freshly
4470
Richard Henderson et al.
formed ferric hydroxide solids were not settleable in the time afforded by the settling basin when FeSO4 was applied post-flocculation (pre-settling). Figure 5 gives the results of the TSS analysis. The pre-settling and pre-filter applications of FeSO4 resulted in filter influent TSS levels that were nearly identical (11.6 and 11.8 mg/L, respectively). This corresponded to an observed increase of 241 and 247% of TSS influent to the filter when compared to those of the control runs. Again, this corresponds
Fig. 5. Total suspended solids measured immediately before filtration.
with the increase in the pre-filter turbidity measurements mentioned earlier providing further support that the freshly formed ferric hydroxide solids were not settleable. Thus, the majority of the solids that were produced from the reaction were being sent to the filter. Although these data indicate that the solids loading to the filters increased substantially, the authors speculated that these increases probably did not completely explain the degradation in filter run times when applying FeSO4 pre-settling or prefiltration. The significant decrease in filter run times suggested that not only did the additional solids loaded to the filter play a part in poor filter performance when applying FeSO4 pre-settling and pre-filter, but degradation of filter performance might have been related to the nature of the freshlyoxidized ferric solids. Although no definitive study of the charge characteristics or size of these particles was done, a qualitative bench-scale experiment was performed exploring this hypothesis and is discussed later. Finished water iron concentrations are given in Fig. 6(A). This graph gives total iron and Fe(III) concentrations for the control runs and experimental trials during the steady state (or point of lowest
Fig. 6. (A) Finished water and Fe(III) concentrations during the steady state of filter runs. (B) Fe(III) concentration during steady state and during breakthrough for pre-settling and pre-filter application of Fe(II) trials. Error bars equal maximum and minimum values of triplicate analyses.
Ferrous reduction of chlorite
4471
Fig. 7. 15 min settled water turbidities for jar tests exploring application of Fe(II) at different points in flocculation process. (A) Alum as coagulant, (B) ferric chloride as coagulant.
effluent particle counts for runs lasting less than 2 h) portion of the filter runs. The total iron and ferric ion concentrations were nearly equivalent during the trials when FeSO4 was applied pre-settling and prefilter. These results provide further support that the extremely short run times observed during these trials were not due to the fact that unreacted Fe(II) was being oxidized in the filter bed. Figure 6(A) also shows that during the application of FeSO4 at the pre-settling and pre-filter locations, total iron and Fe(III) concentrations were 73 and 61 mg/L, respectively. When compared to the total iron concentrations during the control runs and those where FeSO4 was applied at the rapid mix (5 and 7 mg/L, respectively), it appears that the filtration process is less efficient at removing the freshly formed iron solids that contributed to the increase in TSS previously discussed. Figure 6(B) gives the Fe(III) concentrations during the breakthrough period for the trials where ferrous was applied pre-settling and pre-filter. At this point in the runs, effluent concentrations of Fe(III) were above the USEPA secondary maximum contaminant level of 0.3 mg/L, thus indicating that the increase in particle counts was at
least partially due to the increase of ferric solids breaking through the filter.
Bench scale experiments The results of the jar tests can be found in Fig. 7. As FeSO4 applications were moved towards the end of the simulated flocculation basin, it can be seen that settled water turbidities increased accordingly. Ferrous sulfate addition at the rapid mix and directly after the rapid mix yielded settled water turbidities of 1.3 and 1.4 NTU, respectively, at an alum dose of 10 mg/L. These results indicate that FeSO4 could be applied directly after the rapid mix with minimal to no impact on process performance. When FeSO4 was applied after the first stage of flocculation, the average settled water turbidity was 2.3 NTU. These data also indicated that the mixing afforded by using the last two flocculation cells may be enough to condition the ferric solids so that they may be removed through settling, allowing use of the first cell in the flocculation basin for ClO2 disinfection contact time.
4472
Richard Henderson et al.
Fig. 8. 15 min settled water turbidities for jar tests exploring the application of iron added as Fe3+ (6 mg/ L) versus iron added as Fe2+ (6 mg/L) at different points in the flocculation process. Alum=22 mg/L, chlorite=2 mg/L.
As shown in Fig. 7, addition of FeSO4 after the second flocculation mixing step (simulating addition after the second cell in the flocculation basin) did not appear to give enough mixing time for coagulation and flocculation of the ferric hydroxide. The observed optimum settled water turbidities for these trials were 7.2 NTU (alum=22 mg/L) and 7.4 NTU (alum=22 mg/L). These high settled water turbidity values suggest that filter runs in which Fe(II) is applied between the 2nd and 3rd stage of flocculation will have similar results to the pilot plant trials when Fe(II) was applied post-flocculation (pre-settling). As mentioned earlier, a qualitative experiment was performed to explore the hypothesis that the chemical nature of the freshly formed ferric solids were such that they contributed to the poor filter performance observed during the pilot plant trials in which FeSO4 was applied at the pre-settling and prefilter locations. Many water treatment facilities use pre-formed ferric (either ferric sulfate or ferric chloride) as a coagulant. These bench-scale experiments were intended to compare the settleability of ferric hydroxide solids formed from the addition of ferric chloride as opposed to those Fe(III) solids that are formed when Fe(II) is oxidized. Figure 8 shows the results of jar tests that were performed using a 22 mg/L alum dose with 6 mg/L addition of Fe(III) (applied as ferric chloride) at the rapid mix, post rapid mix, after the 1st flocculation cell, after the 2nd flocculation cell, and pre-settling. These results are presented in conjunction with the results from the previously presented data in which a 6 mg/L Fe(II) dose was applied to water containing 2 mg/L of ClO@ (22 mg/L of alum as coagulant). These data 2 suggest that settleable ferric hydroxide solids form more readily from pre-formed Fe(III) than those formed when Fe(II) is oxidized to Fe(III). Specifically, Fe(III) addition at the pre-settling location yielded settled water turbidities of 4 NTU. This is considerably less than the 7.4 NTU value observed when iron was added as Fe(II). These results suggest that the filterability of the iron may be linked not
only to the increased solids loading of the filter, but also to the chemical nature of the ferric hydroxide solids that are formed in the ClO@ reduction 2 process.
SUMMARY AND CONCLUSIONS
The results from a study on the impact of Fe(II) reduction of ClO@ on the performance of a pilot2 scale filtration process have been presented. Main conclusions resulting from the study are as follows: *
*
*
*
*
Filter headloss rate and effluent turbidity should not be adversely affected when Fe(II) is added for ClO@ 2 reduction if the ferrous compound is added prior to flocculation. Fe(III) solids that are formed during ClO@ 2 reduction process are an effective supplement to the coagulant that is being used during normal process operation. Fe(II) reduction of ClO@ yielded significant 2 negative impacts to the filtration process in terms of both run time and steady-state particle counts when the ferrous compound was applied postflocculation or post-settling. The suspended solids influent to the filter increased significantly regardless of whether the Fe(II) was applied at pre-settling or pre-filter locations. Bench-scale results indicate that Fe(III) solids formed by oxidizing Fe(II) appear to be significantly less settleable than iron solids formed from ferric chloride.
It should be noted that only one set of conditions was used throughout the course of this study. These included a source water that had a near neutral pH, low alkalinity and low TOC, and a pilot plant configuration that remained constant throughout the course of this study. Additional research should be conducted with a wider range of raw waters and
Ferrous reduction of chlorite
further pilot-scale testing should consider lower Fe(II) doses, the use of polymer filter aids, lower hydraulic loading rates, and different filter media. REFERENCES
APHA (1998) Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, American Water Works Association and Water Pollution Control Federation, Washington, DC. Dixon K. and Lee R. (1991) The effect of sulfur-based reducing Agents and GAC filtration on chlorine dioxide by-products. J. Am. Water Works Assoc. 83(5), 48–55. Ellenberger C. S., Hoehn R. C., Gallagher D. L., Knocke W. R., Via C. E., Wiseman E. V., Benninger R. W. and Rosenblatt A. (1998) Water quality impact of pure chlorine dioxide pretreatment at the Roanoke County (Virginia) water treatment plant. Proceedings from the 1998 AWWA Symposium, pp. 131–143. Federal Register (1998) National Primary Drinking Water Regulations: Disinfectants and Disinfection Byproducts; Final Rule. 40 CFR Parts 9, 141 and 142; Vol. 63 (No. 241), pp. 69,389-69,476. December 16. Finch G. R., Li H. and Belosevic M. (1999) Chlorine dioxide inactivation of Cryptosporidium Parvum. Proceedings of the Second European Symposium on Chlorine Dioxide, Paris, France. Gates D. A. (1998) The Chlorine Dioxide Handbook. American Water Works Association, Denver, CO. Gregory D., Carlson K. H., Gordon G. and Bubnis B. (1997) Removal of chlorite ion using sulfite ion. Proceedings of the 1997 AWWA WQTC, Denver, CO.
4473
Griese M. H., Hauser K., Berkemeier M. and Gordon G. (1991) Using reducing agents to eliminate chlorine dioxide and chlorite ion residuals in drinking water. J. Am. Water Works Assoc. 83(5), 56–61. Griese M. H., Kaczur J. J. and Gordon G. (1992) Combining methods for the reduction of oxychlorine residuals in drinking water. J. Am. Water Works Association. 84(11), 68–77. Hurst G. H. and Knocke W. R. (1997) Evaluating ferrous iron for chlorite removal. J. Am. Water Works Assoc. 89(8), 98–105. Iatrou A. and Knocke W. R. (1992) Removing chlorite by addition of ferrous iron. J. Am. Water Works Assoc. 84(11), 63–68. Rittman D. D. (1997) ‘‘Can You have Your Cake and Eat it Too’’ with Chlorine Dioxide? Water/Eng Manage. April, 30–35. Suzuki K. and Gordon G. (1978) Stochiometry and kinetics of the reaction between chlorine dioxide and sulfur(IV) in solution. Inorg. Chem. 17, 3315. US Environmental Protection Agency (1978) An Assessment of Ozone and Chlorine Dioxide Technologies for Treatment of Municipal Water Supplies. EPA-600/8-78-018. US Environmental Protection Agency (1999) Alternative Disinfectants and Oxidants Guidance Manual. EPA 815-R99-014. Werdehoff K. S. and Singer P. C. (1987) Chlorine dioxide effects on THMFP, TOXFP, and the formation of inorganic by-products. J. Am. Water Works Assoc. 79(9), 107–113. White G. C. (1999) The Handbook of Chlorination and Alternative Disinfectants, pp. 1153–1202. Wiley, New York.
Wat. Res. Vol. 35, No. 18, pp. 4464–4473, 2001 r 2001 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/01/$ - see front matter
RESEARCH NOTE THE IMPACT OF FERROUS ION REDUCTION OF CHLORITE ION ON DRINKING WATER PROCESS PERFORMANCE RICHARD HENDERSON, KENNETH CARLSON* and DEAN GREGORY Department of Civil Engineering, Colorado State University, Fort Collins, CO 80521, USA (First received 1 April 2000; accepted in revised form 1 March 2001) AbstractFThe use of chlorine dioxide (ClO2) as a primary disinfectant and pre-oxidant in drinking water treatment is being explored as an alternative to chlorine for reducing disinfection by-product formation and to assure compliance with United States Environmental Protection Agency’s Stage 1 Disinfection/ Disinfection By-Products Rule. However, the ClO2 by-product chlorite ion (ClO@ 2 ) is also regulated by the same regulation. Ferrous iron (Fe(II)) has been shown to effectively reduce chlorite ion to chloride ion (Cl@) and this study was conducted to evaluate the impact on overall treatment process performance due to the ferric hydroxide solids that form from the reaction. Ferrous iron application was explored at three different points in a pilot-scale water treatment system: pre-rapid mix, pre-settling and pre-filter. Chlorite ion concentrations were effectively reduced from 2 mg/L to less than 0.3 mg/L using an Fe(II) dose of approximately 6 mg/L for all trials. Fe(II) addition at the rapid mix caused no adverse effects and, in fact, allowed for reduction of the alum dose due to the newly formed ferric hydroxide acting as a supplemental coagulant. An increase of 241 and 247% of total suspended solids influent to the filter process was observed when Fe(II) was applied at the pre-settling and pre-filter locations. Pilot-scale filter runs during these trials were less than 2 h and never obtained true steady state conditions. Jar testing was performed to better understand the nature of the ferric hydroxide solids that are formed when Fe(II) was oxidized to Fe(III) and to explore the effectiveness of Fe(II) addition at intermediate stages in the flocculation process. r 2001 Elsevier Science Ltd. All rights reserved Key wordsFwater treatment, chlorine dioxide, chlorite ion, reduction, ferrous ion
INTRODUCTION
The first application of chlorine dioxide (ClO2) in a United States drinking water system was in Niagara Falls, NY in 1944 for taste and odor control (USEPA, 1978). Presently, ClO2 is used for disinfection (CT credit), taste and odor control, iron and manganese removal, and control of hydrogen sulfide (USEPA, 1999). The use of ClO2 as a primary disinfectant is of particular interest because it does not produce the organic disinfection by-products common to conventional chlorination practices. In particular, Werdehoff and Singer (1987) showed that the application of ClO2 does not produce trihalomethanes (THMs) and produces only a small amount of total organic halide (TOX). ClO2 is primarily an oxidizing agent, not a chlorinating agent. Gates (1998) described the chemistry as being ‘‘free radical electrophilic (i.e. electron attracting) abstraction’’ and not ‘‘oxidative substitution or addition (as in chlorinating agents)’’. Chlorine dioxide has also been proven to be effective for inactivation of chlorine-
resistant pathogens such as Giardia and Cryptosporidium (White, 1999; Finch et al., 1999). The United States Environmental Protection Agency (USEPA) promulgated Stage 1 of the Disinfection/Disinfection By-Products Rule (D/DBP) Federal Register, 1998). This regulation established the maximum contaminant levels (MCL) of 0.080 and 0.060 mg/L for THMs and haloacetic acids (HAAs), respectively. This has prompted some water utilities to study ClO2 as a primary disinfectant to reduce the formation of THMs and HAAs during the primary disinfection process. Although ClO2 does not produce THMs, it does produce chlorite ion (ClO@ 2 ). The Stage 1 D/DBP Rule sets the MCL for ClO@ at 2 1.0 mg/L limiting the dose of ClO2 and, therefore, the applicability of ClO2 as a disinfectant. United States water utilities serving more than 10,000 people will need to be in compliance with this regulation by January 1, 2002, while compliance for smaller utilities is required by January 1, 2004.
BACKGROUND
*Author to whom all correspondence should be addressed. E-mail: [email protected]
Three approaches for removing ClO@ have been 2 explored in recent years. Two of these involve the
4464
Ferrous reduction of chlorite
chemical reduction of ClO@ 2 , one through the use of sulfite ion (SO2@ ) and one using ferrous ion (Fe2+ 3 or Fe(II)). The third technique that has been explored is the use of granular activated carbon (GAC) as a reductant. The use of sulfite ion has been proven to be somewhat limited due to the reaction’s dependency on pH. The reaction is efficient over the pH range of 5–6.5, but above pH=7 it slows considerably (White, 1999; Gregory et al., 1997). Another disadvantage of SO2@ is the formation of chlorate ion (ClO@ 3 3 ) from the reaction. Chlorate ion has been shown to form in pH ranges of 4–8.5 by Griese et al. (1991) and 8.5–11 by Suzuki and Gordon (1978). GAC has also been shown to be extremely limited because of the short life of the carbon when used for this purpose, especially if the carbon is preloaded (Dixon and Lee, 1991; Ellenberger et al., 1998). The use of Fe(II) appears to be the most promising of the ClO@ removal techniques and has been 2 successfully used in laboratory and, to some extent, in pilot and full-scale studies (Griese et al., 1991, 1992; Iatrou and Knocke, 1992; Hurst and Knocke, 1997; Rittman, 1997). The reaction stoichiometry for the reduction of ClO@ 2 by Fe(II) is given in equations (1)–(3). 12H2 O þ 4Fe2þ "4FeðOHÞ3 þ 4e@ þ 12Hþ
ð1Þ
4Fe2þ þ ClO@ 2 þ 10H2 O"4FeðOHÞ3ðsÞ þ Cl@ þ 8Hþ @ @ þ 2H2 O 4Hþ þ ClO@ 2 þ 4e "Cl
ð2Þ ð3Þ
Based on equation (3), 3.3 mg of Fe(II) is needed to reduce 1 mg of ClO@ 2 . In laboratory experiments, Iatrou and Knocke (1992) determined the ratio to be approximately 3.1 mg Fe(II)/mg ClO@ 2 . This result suggests that the majority of ClO@ is reduced to 2 chloride, a 4 electron transfer reaction. The study also concluded that the reaction was complete within 5–15 s (at pH=5–7) and that the presence of Fe(OH)3(s) did not adversely affect alum as a coagulant during jar test experiments. Lastly, it was concluded that any excess Fe(II) would be oxidized by dissolved oxygen when the pH was greater than 7. Later experiments by Hurst and Knocke (1997) explored the use of Fe(II) under alkaline conditions (pH 8–10) similar to those used in a typical softening process. Ferrous ion proved to be effective with dosages 10–20% above the stoichiometric requirement. The only adverse effect was that high dissolved oxygen concentrations at pH=10 inhibited the reaction. Also explored in these experiments was the influence of dissolved organic carbon (DOC) on the Fe(II)/ClO@ 2 reaction. No effect was observed at a pH of 8, but significant and substantial effects were seen at pH=9 and 10, respectively. Further experimentation involving nitrogen stripping of oxygen led to the hypothesis that the DOC may act to enhance
4465
the oxidation of ferrous by oxygen. The last consideration for Fe(II) reduction of ClO@ is that 2 of ClO@ 3 formation. Studies have shown that, in pH ranges of 5–10, ClO@ formation is minimal and 3 levels are approximately equal to those in the influent (Griese et al., 1991; Iatrou and Knocke, 1992; Hurst and Knocke, 1997).
RESEARCH PURPOSE AND OBJECTIVE
The use of existing infrastructure (flocculation and settling basins) for ClO2 disinfection contact time (and the associated CT credit) is an appealing situation. Increasing the amount of disinfection that can be gained in this manner would increase the number of treatment plants that could reduce the formation of THMs and HAAs by replacing chlorine with ClO2 as the primary disinfectant. For example, plants that are already using ClO2 for pre-oxidation of iron and manganese (which is normally done before the rapid mix) would be able to raise the oxidant dose and achieve adequate disinfection credit with their existing system. However, application of Fe(II) later in the treatment process may have its own consequences. The ferric hydroxide solids that are formed when Fe(II) is oxidized may impact settling and filtration. Ferrous ion has been applied successfully in full-scale operation at the rapid mix location in El Paso, Texas (Rittman, 1997), but studies exploring the application of Fe(II) after the rapid mix have been somewhat limited. Griese et al. (1992) applied Fe(II) post-settling in pilot plant studies, but the reaction occurred in a contact chamber that was designed for a 30-min hydraulic residence time. In addition, secondary flocculation and sedimentation basins were used to lessen the impact of the ferric hydroxide solids on the filtration process. The main objective of this research was to evaluate the impact of Fe(II) reduction of ClO@ on overall 2 process performance using a pilot-scale treatment plant. In particular, the study focused upon the settleability and filterability of the ferric hydroxide solids that are formed as a result of the oxidation/ reduction reactions. Pre-filter total suspended solid concentration (TSS) and turbidity measurements were used to evaluate settleability. Pilot-scale filter performance was evaluated with finished water particle counts, turbidity, and iron concentration measurements. The experimental plan involved the application of Fe(II) at the rapid mix and at presettling and pre-filter locations with no modifications (other than the Fe(II) feed) to a pilot-scale drinking water treatment plant. As mentioned earlier, Iatrou and Knocke (1992) concluded that the Fe(II)/ClO@ 2 reaction is complete within 15 s when the pH is between 5 and 7. Because the reaction pH during this study was within the range cited by Iatrou and Knocke, a separate reaction basin for chlorite
4466
Richard Henderson et al.
reduction was not required. Fe(II) was also applied slightly under the stoichiometric values to further ensure that unreacted Fe(II) was not influent to the filter. This was important because Fe(II) being oxidized in the filter media would have added an extra variable to the study.
METHODS AND MATERIALS
Pilot scale experiments All of the pilot plant research was conducted using a 10 gal/min (37.85 L/min) pilot plant located at the Engineering Research Center, Colorado State University, Fort Collins, Colorado. The plant consists of a rapid mix chamber, three-stage flocculation basin, lamella plate settling basin, and two dual media filters consisting of 1 ft (0.30 m) of sand covered with 2 ft (0.61 m) of anthracite (Fig. 1). Hydraulic loading rates (HLR) for the settling basin and filter were 0.4 gal/min/ft2 (16.3 L/min/m2) and 5 gal/min/ft2 (203.7 L/min/m2), respectively. On line flow meters (Signet, Model 3-8500) were used to monitor influent and filter flows throughout the course of the tests. Flow rates were manually checked at the start of each filter run and at least once during the run for each of the following chemical feeds: liquid alum (General Chemical), LT-22 cationic polymer (Ciba Specialty Chemical Water Treat@ (80% ments, Inc.), sodium bicarbonate (HCO@ 3 ), ClO2 technical grade sodium chlorite, Aldrich Chemical Company, Inc.), and Fe(II) (ferrous sulfate (FeSO4), 5.08% ferrous, specific gravity=1.2, Kemiron, Inc.). Fe(II) locations given in Fig. 1 indicate the different application points that were explored during this study. The source water for the plant came from Horsetooth Reservoir, Fort Collins, Colorado. Influent pH (Fischer Scientific, Accumet, pH meter 50), alkalinity (APHA, 1998, Standard Method 2320A), on-line turbidity (Hach 1720D Turbidimeter), on-line particle counts (Hach 1900 WPC Particle Counter), and TOC concentration (Rosemont Dohrman DC-80 Total Organic Carbon Analyzer) were measured during the study. The ranges of these parameters can be found in Table 1. Table 2 gives the experimental trial sequence along with the chemical additions for each experimental trial. This sequence was chosen so that
comparisons between control runs and experimental runs could be made with the assurance that the water quality was not changing significantly through the course of the study (January 22, 2000 through March 10, 2000). The protocol of alternating control runs with experimental runs was ended once three control runs had been completed and once it was realized (as will be discussed later) that the pre-settling and pre-filter FeSO4 application experiments could be completed within a short period of time. solution was pumped into the raw A 7000 mg/L ClO@ 2 water at a rate of 10.8 mL/min in order to obtain a 2 mg/L concentration. Chlorite ion was measured directly ClO@ 2 before Fe(II) addition and in the finished water. Amperometric titration Method II (APHA, 1998, Standard Method E) was used for ClO@ determination. The 4500-ClO@ 2 2 analyses were completed within 60 min following ClO@ 2 sample collection. Pre-filter samples were collected and analyzed for TSS concentration according to Standard Method 2540 D (APHA, 1998). Finished water iron levels (Fe(II) and total) were determined using spectrophotometric techniques (Hach Corporation 1,10-phenanthroline method, Hach DR/3000 Spectrophotometer). Ferric ion (Fe(III)) concentrations were determined by taking the difference between the Fe(II) and total iron concentrations. Iron analyses were completed within 30 min of sample collection. On-line particle counts (Hach 1900 WPC Particle Counter) were measured for the finished water. Pre-filter and finished water turbidities (Hach 1720D Turbidimeter) were also measured on-line. The filter-ripening time was defined as the time after filter startup when the number of particles greater than 2 mm exiting the filter decreased to 20 counts per mL or less. Filter breakthrough was defined as
Table 1. Horsetooth reservoir water qualitya Water quality parameter
Range
pH Alkalinity (mg/L as CaCO3) Turbidity (NTU) Total organic carbon (mg/L)
7.2–7.6 26.5–31 1.9–2.1 3.7–4.3
a
January 22, 2000–March 10, 2000.
Fig. 1. Configuration of pilot plant at Colorado State University, Fort Collins, Colorado.
Ferrous reduction of chlorite
4467
Table 2. Experimental sequence and chemical additionsa Trial sequence
Control #1 Fe(II) applied Control #2 Fe(II) applied Control #3 Fe(II) applied Fe(II) applied Fe(II) applied Fe(II) applied a
at rapid mix #1 at rapid mix #2 pre-settling #1 pre-settling #2 pre-filter #1 pre-filter #2
Alum (mg/L)
LT-22 Flocculant AID (mg/L)
Bicarbonate (mg/L as HCO@ 3 )
Chlorite (mg/L)
Ferrous (mg/L as Fe2+)
22 10 22 10 22 22 22 22 22
0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
15 15 15 15 15 15 15 15 15
0 2 0 2 0 2 2 2 2
0 6.2 0 6.2 0 6 6 6 6
Bicarbonate was added to raise alkalinity and to provide some buffering for the addition of ferrous sulfate (pHo2).
the point where particle counts of particles greater than 2 mm increased above 20 counts per mL. Finished water pH (Fischer Scientific, Accumet, pH meter 50) and alkalinity (APHA, 1998, Standard Method 2320A) were also measured. All on-line measurements were taken at 1-min intervals. Headloss was monitored throughout the course of the study. An upper limit range of 6.5–7 ft was set for run termination if particle breakthrough did not occur. However, the particle count breakthrough limits mentioned earlier were the limiting factor in all trials.
Bench-scale experiments Standard jar tests (Phipps and Bird PB-900 Paddle Stirrer) were conducted as a means of evaluating the benefits of FeSO4 addition at different stages in the treatment process and to facilitate analyses of data collected during the pilot-scale experiments. The mixing and settling sequence for the jar tests was as follows: (1) 300 rpm for 40 s; (2) 48 rpm for 10 min; (3) 32 rpm for 10 min; (4) 18 rpm for 10 min; and (5) 15 min of settling. Settled water turbidities were then recorded (Hach 2100N Turbidimeter). Alum and ferric chloride were both evaluated as primary coagulants at 3 mg/L intervals ranging from 7 to 25 mg/L. Chlorite ion were introduced to the raw water using the and HCO@ 3 same stock solutions as those used in the pilot plant portion of the study. Likewise, the FeSO4 solution used for ClO@ 2 reduction was the same as that used in the pilot plant portion of the study. Lastly, the addition of iron as Fe(III) (6 mg/L) was compared to the addition of iron as Fe(II) (6 mg/L) at an alum dose of 22 mg/L. These experiments were conducted to explore whether the freshly formed Fe(III) from Fe(II) oxidation behaves similarly (in terms of settleability) to pre-formed Fe(III) from a ferric chloride solution.
Data analysis Samples for measurement of ClO@ 2 , TSS concentration, pH, and alkalinity were taken once during the steady state (or point of lowest particle counts if there was no true steady state) of each pilot plant trial. Effluent samples were collected for iron analyses during the steady state and particle breakthrough periods for the trials in which Fe(II) was applied pre-settling and pre-filter. All samples were tested in triplicate. Error bars that are given on the graphical representations of the data represent the minimum and maximum values of these analyses. Each bench-scale scenario was run in triplicate. Average values given for settled water turbidities represent the average of the three independent settled water samples.
Fig. 2. Chlorite removal during application of Fe(II) at the rapid mix and at pre-settling and pre-filter locations.
RESULTS AND DISCUSSION
Pilot scale experiments Chlorite ion was effectively reduced to concentrations lower than 0.3 mg/L for all trials (Fig. 2). Ferrous sulfate dosing was reduced slightly (from 6.2 mg/L as Fe(II) to 6.0 mg/L as Fe(II)) when applied at pre-settling and pre-filter locations to ensure that no unreacted Fe(II) was influent to the filters. As discussed earlier, this was done as a precautionary measure to ensure that the extra variable of Fe(II) being oxidized in the filter, possibly causing negative effects, was eliminated. The finished water pH values (less than 7 for all trials) were within the range cited by Iatrou and Knocke (1992) where Fe(II) oxidation is both rapid and complete, thus adding extra confirmation that no oxidation of Fe(II) should have taken place in the filter media. These data were also supported by effluent iron concentrations, which are discussed at a later point in this paper. Figs 3(A)–(D) give particle count data for selected examples of each type of trial (i.e. no Fe(II), Fe(II) applied at the rapid mix, Fe(II) applied pre-settling, and Fe(II) applied pre-filter). Based upon the 20 counts per mL particle count criteria set as lower and upper limits of ripening and breakthrough, respectively, the three control experiments averaged run
4468
Richard Henderson et al.
Fig. 3. Examples of raw and finished water particle counts (>2 mm) for each experimental condition.
times of 21.3 h (range=19.5–22.5 h) and yielded an average unit filter run volume (UFRV=gal/ft2 of filter area/trial) of 6400 gal/ft2 (263,304 L/m2). The two trials in which FeSO4 was applied at the rapid mix produced similar results (run times =22.3 and 21.9 h, average UFRV=6630 gal/ft2=272,767 L/m2) to those of the controls. The addition of FeSO4 at the rapid mix also allowed for reduction of the alum dose from 22 to 10 mg/L (Fig. 3(B)). This was due to the ferric hydroxide acting as a supplemental coagulant.
These results are supported by the full-scale operation in El Paso, Texas where ferrous chloride is added at the rapid mix and the resulting ferric hydroxide solids are used as the primary coagulant (Rittman, 1997). When FeSO4 was applied at the pre-settling (Fig. 3(C)) and pre-filter (Fig. 3(D)) locations, particle counts less than 20 counts per mL were barely obtained. Within 2 h of filter start up of these trials, particle counts increased to greater than 100 counts per mL.
Ferrous reduction of chlorite
4469
Fig. 4. Examples of raw, pre-filter, and finished water turbidity data for each experimental condition.
Selected examples of turbidity data for each type of experimental trial are given in Figs 4A–D. Finished water turbidities for the control (Fig. 3(A)) and for the trials when FeSO4 was applied at the rapid mix (Fig. 3(B)) were consistently below 0.07 NTU during the steady state portion of the filter run. Ferrous sulfate application at the pre-settling and pre-filter locations produced finished water turbidities that
were equal to or less than 0.1 NTU for only a brief time (Figs 3C, D). Post-settling turbidity measurements averaged 4.5 NTU (FeSO4 applied presettling) and 5.1 NTU (FeSO4 applied pre-filter). This was an observed increase of 157 and 191%, respectively, when compared to the average of 1.75 NTU for the control runs. The marginal difference between these two values indicates that the freshly
4470
Richard Henderson et al.
formed ferric hydroxide solids were not settleable in the time afforded by the settling basin when FeSO4 was applied post-flocculation (pre-settling). Figure 5 gives the results of the TSS analysis. The pre-settling and pre-filter applications of FeSO4 resulted in filter influent TSS levels that were nearly identical (11.6 and 11.8 mg/L, respectively). This corresponded to an observed increase of 241 and 247% of TSS influent to the filter when compared to those of the control runs. Again, this corresponds
Fig. 5. Total suspended solids measured immediately before filtration.
with the increase in the pre-filter turbidity measurements mentioned earlier providing further support that the freshly formed ferric hydroxide solids were not settleable. Thus, the majority of the solids that were produced from the reaction were being sent to the filter. Although these data indicate that the solids loading to the filters increased substantially, the authors speculated that these increases probably did not completely explain the degradation in filter run times when applying FeSO4 pre-settling or prefiltration. The significant decrease in filter run times suggested that not only did the additional solids loaded to the filter play a part in poor filter performance when applying FeSO4 pre-settling and pre-filter, but degradation of filter performance might have been related to the nature of the freshlyoxidized ferric solids. Although no definitive study of the charge characteristics or size of these particles was done, a qualitative bench-scale experiment was performed exploring this hypothesis and is discussed later. Finished water iron concentrations are given in Fig. 6(A). This graph gives total iron and Fe(III) concentrations for the control runs and experimental trials during the steady state (or point of lowest
Fig. 6. (A) Finished water and Fe(III) concentrations during the steady state of filter runs. (B) Fe(III) concentration during steady state and during breakthrough for pre-settling and pre-filter application of Fe(II) trials. Error bars equal maximum and minimum values of triplicate analyses.
Ferrous reduction of chlorite
4471
Fig. 7. 15 min settled water turbidities for jar tests exploring application of Fe(II) at different points in flocculation process. (A) Alum as coagulant, (B) ferric chloride as coagulant.
effluent particle counts for runs lasting less than 2 h) portion of the filter runs. The total iron and ferric ion concentrations were nearly equivalent during the trials when FeSO4 was applied pre-settling and prefilter. These results provide further support that the extremely short run times observed during these trials were not due to the fact that unreacted Fe(II) was being oxidized in the filter bed. Figure 6(A) also shows that during the application of FeSO4 at the pre-settling and pre-filter locations, total iron and Fe(III) concentrations were 73 and 61 mg/L, respectively. When compared to the total iron concentrations during the control runs and those where FeSO4 was applied at the rapid mix (5 and 7 mg/L, respectively), it appears that the filtration process is less efficient at removing the freshly formed iron solids that contributed to the increase in TSS previously discussed. Figure 6(B) gives the Fe(III) concentrations during the breakthrough period for the trials where ferrous was applied pre-settling and pre-filter. At this point in the runs, effluent concentrations of Fe(III) were above the USEPA secondary maximum contaminant level of 0.3 mg/L, thus indicating that the increase in particle counts was at
least partially due to the increase of ferric solids breaking through the filter.
Bench scale experiments The results of the jar tests can be found in Fig. 7. As FeSO4 applications were moved towards the end of the simulated flocculation basin, it can be seen that settled water turbidities increased accordingly. Ferrous sulfate addition at the rapid mix and directly after the rapid mix yielded settled water turbidities of 1.3 and 1.4 NTU, respectively, at an alum dose of 10 mg/L. These results indicate that FeSO4 could be applied directly after the rapid mix with minimal to no impact on process performance. When FeSO4 was applied after the first stage of flocculation, the average settled water turbidity was 2.3 NTU. These data also indicated that the mixing afforded by using the last two flocculation cells may be enough to condition the ferric solids so that they may be removed through settling, allowing use of the first cell in the flocculation basin for ClO2 disinfection contact time.
4472
Richard Henderson et al.
Fig. 8. 15 min settled water turbidities for jar tests exploring the application of iron added as Fe3+ (6 mg/ L) versus iron added as Fe2+ (6 mg/L) at different points in the flocculation process. Alum=22 mg/L, chlorite=2 mg/L.
As shown in Fig. 7, addition of FeSO4 after the second flocculation mixing step (simulating addition after the second cell in the flocculation basin) did not appear to give enough mixing time for coagulation and flocculation of the ferric hydroxide. The observed optimum settled water turbidities for these trials were 7.2 NTU (alum=22 mg/L) and 7.4 NTU (alum=22 mg/L). These high settled water turbidity values suggest that filter runs in which Fe(II) is applied between the 2nd and 3rd stage of flocculation will have similar results to the pilot plant trials when Fe(II) was applied post-flocculation (pre-settling). As mentioned earlier, a qualitative experiment was performed to explore the hypothesis that the chemical nature of the freshly formed ferric solids were such that they contributed to the poor filter performance observed during the pilot plant trials in which FeSO4 was applied at the pre-settling and prefilter locations. Many water treatment facilities use pre-formed ferric (either ferric sulfate or ferric chloride) as a coagulant. These bench-scale experiments were intended to compare the settleability of ferric hydroxide solids formed from the addition of ferric chloride as opposed to those Fe(III) solids that are formed when Fe(II) is oxidized. Figure 8 shows the results of jar tests that were performed using a 22 mg/L alum dose with 6 mg/L addition of Fe(III) (applied as ferric chloride) at the rapid mix, post rapid mix, after the 1st flocculation cell, after the 2nd flocculation cell, and pre-settling. These results are presented in conjunction with the results from the previously presented data in which a 6 mg/L Fe(II) dose was applied to water containing 2 mg/L of ClO@ (22 mg/L of alum as coagulant). These data 2 suggest that settleable ferric hydroxide solids form more readily from pre-formed Fe(III) than those formed when Fe(II) is oxidized to Fe(III). Specifically, Fe(III) addition at the pre-settling location yielded settled water turbidities of 4 NTU. This is considerably less than the 7.4 NTU value observed when iron was added as Fe(II). These results suggest that the filterability of the iron may be linked not
only to the increased solids loading of the filter, but also to the chemical nature of the ferric hydroxide solids that are formed in the ClO@ reduction 2 process.
SUMMARY AND CONCLUSIONS
The results from a study on the impact of Fe(II) reduction of ClO@ on the performance of a pilot2 scale filtration process have been presented. Main conclusions resulting from the study are as follows: *
*
*
*
*
Filter headloss rate and effluent turbidity should not be adversely affected when Fe(II) is added for ClO@ 2 reduction if the ferrous compound is added prior to flocculation. Fe(III) solids that are formed during ClO@ 2 reduction process are an effective supplement to the coagulant that is being used during normal process operation. Fe(II) reduction of ClO@ yielded significant 2 negative impacts to the filtration process in terms of both run time and steady-state particle counts when the ferrous compound was applied postflocculation or post-settling. The suspended solids influent to the filter increased significantly regardless of whether the Fe(II) was applied at pre-settling or pre-filter locations. Bench-scale results indicate that Fe(III) solids formed by oxidizing Fe(II) appear to be significantly less settleable than iron solids formed from ferric chloride.
It should be noted that only one set of conditions was used throughout the course of this study. These included a source water that had a near neutral pH, low alkalinity and low TOC, and a pilot plant configuration that remained constant throughout the course of this study. Additional research should be conducted with a wider range of raw waters and
Ferrous reduction of chlorite
further pilot-scale testing should consider lower Fe(II) doses, the use of polymer filter aids, lower hydraulic loading rates, and different filter media. REFERENCES
APHA (1998) Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, American Water Works Association and Water Pollution Control Federation, Washington, DC. Dixon K. and Lee R. (1991) The effect of sulfur-based reducing Agents and GAC filtration on chlorine dioxide by-products. J. Am. Water Works Assoc. 83(5), 48–55. Ellenberger C. S., Hoehn R. C., Gallagher D. L., Knocke W. R., Via C. E., Wiseman E. V., Benninger R. W. and Rosenblatt A. (1998) Water quality impact of pure chlorine dioxide pretreatment at the Roanoke County (Virginia) water treatment plant. Proceedings from the 1998 AWWA Symposium, pp. 131–143. Federal Register (1998) National Primary Drinking Water Regulations: Disinfectants and Disinfection Byproducts; Final Rule. 40 CFR Parts 9, 141 and 142; Vol. 63 (No. 241), pp. 69,389-69,476. December 16. Finch G. R., Li H. and Belosevic M. (1999) Chlorine dioxide inactivation of Cryptosporidium Parvum. Proceedings of the Second European Symposium on Chlorine Dioxide, Paris, France. Gates D. A. (1998) The Chlorine Dioxide Handbook. American Water Works Association, Denver, CO. Gregory D., Carlson K. H., Gordon G. and Bubnis B. (1997) Removal of chlorite ion using sulfite ion. Proceedings of the 1997 AWWA WQTC, Denver, CO.
4473
Griese M. H., Hauser K., Berkemeier M. and Gordon G. (1991) Using reducing agents to eliminate chlorine dioxide and chlorite ion residuals in drinking water. J. Am. Water Works Assoc. 83(5), 56–61. Griese M. H., Kaczur J. J. and Gordon G. (1992) Combining methods for the reduction of oxychlorine residuals in drinking water. J. Am. Water Works Association. 84(11), 68–77. Hurst G. H. and Knocke W. R. (1997) Evaluating ferrous iron for chlorite removal. J. Am. Water Works Assoc. 89(8), 98–105. Iatrou A. and Knocke W. R. (1992) Removing chlorite by addition of ferrous iron. J. Am. Water Works Assoc. 84(11), 63–68. Rittman D. D. (1997) ‘‘Can You have Your Cake and Eat it Too’’ with Chlorine Dioxide? Water/Eng Manage. April, 30–35. Suzuki K. and Gordon G. (1978) Stochiometry and kinetics of the reaction between chlorine dioxide and sulfur(IV) in solution. Inorg. Chem. 17, 3315. US Environmental Protection Agency (1978) An Assessment of Ozone and Chlorine Dioxide Technologies for Treatment of Municipal Water Supplies. EPA-600/8-78-018. US Environmental Protection Agency (1999) Alternative Disinfectants and Oxidants Guidance Manual. EPA 815-R99-014. Werdehoff K. S. and Singer P. C. (1987) Chlorine dioxide effects on THMFP, TOXFP, and the formation of inorganic by-products. J. Am. Water Works Assoc. 79(9), 107–113. White G. C. (1999) The Handbook of Chlorination and Alternative Disinfectants, pp. 1153–1202. Wiley, New York.