• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Fixed-Bed Biological Treatment of Perchlorate-Contaminated DRINKING WATER

Posted on: Sunday, 18 September 2005, 03:01 CDT

A six-month pilot study was conducted to evaluate the use of fixed-bed (FXB) bioreactors to treat perchlorate-contaminated groundwater and to address some of the concerns associated with the full-scale implementation of such a process. The pilot data illustrated that perchlorate-reducing FXB bioreactors can be acclimated using organisms indigenous to the local aquifer; can achieve sustained perchlorate removal to below the analytical detection limit using reasonable contact times; can produce biologically stable effluent; do not foster the growth of pathogenic bacteria; and are robust with respect to system upsets. Based on this project, the California Department of Health Services conditionally approved FXB bioreactors for treating perchlorate- contaminated drinking water, making it possible to consider the full- scale design and implementation of FXB biological perchlorate treatment in California.

Perchlorate, a strong oxidant used in solid rocket fuel, munitions, fireworks, and airbag inflation systems, was relatively unknown to the general public before 1997. Since then, because of a 100-fold decrease in its limit of detection and its subsequent discovery in drinking water sources in more than 25 states (Brandehuber & Clark, 2004), perchlorate awareness among the general public has increased in the United States. Perchlorate is a concern because of its ability to disrupt thyroid hormone production when ingested, which may inhibit normal growth and development. Because associated toxicological and risk assessments are still being developed and evaluated, no maximum contaminant level has been set for perchlorate, although some states have established provisional perchlorate action levels as low as 1 g/L.

Both abiotic and biotic processes have been developed and evaluated for treating perchlorate-contaminated drinking water. Typical abiotic perchlorate treatment processes include ion exchange (Tripp et al, 2003; Gu et al, 2001), reverse osmosis/nanofiltration (Amy et al, 2003), electrodialysis reversal (Booth et al, 2000), and tailored granular activated carbon (GAC; Na et al, 2002). These processes separate perchlorate from the bulk solution by adsorption or diffusion-limited filtration. Typical biotic treatment processes include fixed bed (FXB) reactors (Min et al, 2004; Brown et al, 2002; Miller & Logan, 2000; Wallace et al, 1998), fluidized-bed reactors (Guarini, 2004), membrane biofilm reactors (Lehman et al, 2004; Nerenberg et al, 2002; Liu & Batista, 2000), and dynamic suspended growth reactors (Frankenberger et al, 2004). These systems reduce perchlorate to chloride and rely on the addition of an exogenous electron donor, which can be inorganic (autotrophic process) or organic (heterotrophic process).

Although several abiotic and biotic technologies can efficiently remove perchlorate from drinking water, only ion exchange is being applied for perchlorate removal in full-scale plants. In general, ion exchange is the least expensive option among the abiotic processes, although costs for tailored GAC treatment are not yet published. Cost estimates for biotic treatment processes indicate their potential to be relatively inexpensive options for treating perchlorate-contaminated drinking water. However, other concerns have limited their consideration for full-scale perchlorate treatment, including (1) the lack of a proven full-scale track record, (2) the need for process certification, (3) the potential for process instability, (4) the aversion to using exogenous microbial inocula, (5) the possible production of biologically unstable effluent, and (6) the possible fostering of pathogenic bacteria within a biological reactor.

OBJECTIVES

A sixth-month pilot study was performed at the Castaic Lake Water Agency (CLWA) in Santa Clarita, Calif., to evaluate the effectiveness of an FXB biological process for treating perchlorate- contaminated water from the Saugus Aquifer. The primary objective of the project was to address as many of the concerns as possible associated with the implementation of biotic treatment processes for treating perchlorate-contaminated drinking water. Specific objectives of the project were to

* verify that sufficient perchlorate-reducing biological activity could be developed in the pilot-scale reactors using microorganisms indigenous to the Saugus Aquifer;

* develop design criteria for empty-bed contact time (EBCT), backwashing procedures, electron donor (acetic acid) addition, and nutrient addition;

* evaluate the robustness of the FXB biological process with respect to system upsets, such as electron donor feed failure, process shut-downs, and variations in feed water quality;

* characterize the quality of bioreactor effluent; and

* pursue conditional acceptance of FXB biological technologies from the California Department of Health Services (CDHS) for the treatment of perchlorate-contaminated drinking water.

TABLE 1 Water quality in adjacent perchlorate-contaminated and non-perchlorate-contaminated wells

MATERIALS AND METHODS

Siting and feedwater. Perchlorate-contaminated wells were not used for testing because of the complications and costs of associated discharge requirements. Instead, a perchlorate-free operational well located near one of the perchlorate-contaminated wells was used as feed water for the system. Both wells draw water from the Saugus Aquifer and have comparable raw water quality, as illustrated in Table 1.

Perchlorate was spiked to achieve a target concentration of 50 g/ L for most of the study. The well water was transferred to four 6,500-gal (24,602-L) storage tanks located outside the chemical building at CLWA's Rio Vista Water Treatment Plant via a 6,000-gal (22,710-L) stainless-steel tanker truck, which was filled at the well site and transferred to the pilot site. The tanker truck was washed with hot water or a caustic solution and then rinsed before use to ensure a minimal effect on the raw water quality. The water was pumped from the four 6,500-gal (24,602-L) tanks through a heat exchanger1 (which ensured that any increase in water temperature during storage was eliminated) to a 250-gal (946-L) tank that fed the FXB system. Perchlorate stock solution was prepared as necessary by spiking reagent grade sodium perchlorate2 to 1 L of deionized, distilled water from the CLWA laboratory. A peristaltic pump3 dosed the perchlorate stock solution near the front end of the pilot feed line to allow for adequate mixing. Perchlorate was spiked to the influent at 50 g/L except during the perchlorate spiking tests, during which time perchlorate was spiked at between 5 and 300 g/L. Technical-grade acetic acid4 was used to prepare the stock electron- donor solution and was dosed to the pilot feed lines just before entry to the biological reactors using a peristaltic pump.3 American National Standards Institute/NSF 60-certified, food-grade phosphoric acid5 was also dosed to the pilot feed lines during a portion of the pilot study using a peristaltic pump.3 No other chemicals were added to the process system during the pilot study.

Pilot equipment and capabilities. The pilot skid comprised three granular media filters fed by three independent pumps. Filters were 4 in. (102 mm) in diameter and 14 ft (4.3 m) in height. The maximum flow per filter was 1.5 gpm (0.095 L/s), corresponding to a hydraulic loading rate of 17 gpm/sq ft (11.5 L/s/m^sup 2^). A schematic of the FXB pilot is shown in Figure 1. The FXB reactor was configured for down-flow operation. Initially, all three columns on the skid were used for pilot testing. However, after one month of piloting, two of the columns were shut down to streamline pilot operation.

FIGURE 1 Fixed-bed bioreactor schematic

Fffluent turbidity was monitored continuously by dedicated, inline turbidimeters. A pressure transducer mounted on the pilot skid monitored head loss development across the filter. The pilot skid was equipped with a data acquisition and control (DAC) system, which stored and managed data. For each column, the DAC continuously logged date and time, flow, head loss, and effluent turbidity. The skid included a computer monitor and keyboard interface to observe and control pilot operations.

The columns were filled to a depth of 7 ft (2.1 m) with virgin GAC.6 This allowed for a bed expansion of up to 50% during backwashing. The GAC, which was a bituminous-based carbon with an effective size of ~0.66 mm (0.24 in.), is commonly specified in the United States for granular media filtration and organic contaminant adsorption. The GAC was stored overnight in 5-gal (19-L) buckets to wet the surface and internal pores. The columns were filled with well water, and the slurried GAC was dosed slowly into the columns until the 7-ft (2.1-m) depth was reached. Once the columns were packed with GAC, they were backwashed with well water for a period of 24 h at a 45% bed expansion to remove carbon fines from the GAC bed.

Experimental design. Testing was divided into four phases.

Phase 1 (biological acclimation-six weeks). The purpose of this phase was to determine whether efficient perchlorate-reducing biological \activity could be developed in the GAC filters using microorganisms indigenous to the Saugus Aquifer. Once the filtration skid was mobilized, the columns began treating water from the Saugus Aquifer spiked with 50 g/L perchlorate. During the first two weeks of pilot-testing, acetic acid was fed to each column at a concentration twice that required to stoichiometrically reduce all influent dissolved oxygen (DO) and nitrate (O2 [implies] H2O, NO^sub 3^^sup -^ [implies] N^sub 2^). This ensured that the electron donor was not limiting. For calculation purposes, it was assumed that no electron donor was used for cell synthesis. An EBCT of 15 min was used during the first few weeks of the acclimation phase to allow for the rapid saturation of sorption sites on the surface of the GAC and to promote rapid growth of the microbial community in the FXB bioreactor.

Phase 2 (EBCT and acetic acid optimization-eight weeks). The purpose of this phase was to determine the minimum EBCT required to achieve perchlorate removal to below the 4-g/L method reporting limit (MRL). Acetic acid was initially added at 100% excess relative to the stoichiometric influent DO and nitrate demand, based on influent DO and nitrate concentrations of 7 mg/L and 15 mg/L, respectively (all nitrate values in this article are given as NO^sub 3^^sup -^). The EBCT and acetic acid dose were incrementally lowered and raised to locate threshold operating conditions.

Phase 3 (sustained removal-two weeks). The purpose of this phase was to demonstrate sustained perchlorate removal under optimal EBCT and acetic acid feed conditions. The optimal EBCT and acetic acid feed concentration determined during phase 2 testing were used throughout this phase.

Phase 4 (robustness characterization-nine weeks). The purpose of this phase was to determine how the FXB bioreactor responded to various process upsets. The optimal EBCT and acetic acid feed concentration determined during phase 2 testing were used throughout this phase. The following five process upsets were tested.

1) Backwashing. Two backwashing tests were conducted to evaluate perchlorate removal performance immediately after backwash episodes. Because backwashes remove perchlorate-reducing biomass from the reactor, perchlorate breakthrough could occur after backwashes until the biological community in the reactor has sufficient time to redevelop. Effluent samples were taken 0, 15, 30, 45, and 60 min after the FXB bioreactor was put into production following a backwash.

2) Perchlorate spiking. These tests were designed to evaluate perchlorate removal performance during transient perchlorate loading episodes. The feed perchlorate spike was increased from 50 g/L to 100 g/L for a 48-h period, and effluent perchlorate concentrations were monitored. The influent perchlorate concentration was then readjusted to 50 g/L, and the system was allowed to stabilize if necessary. A 48-h spiking test was also performed using 300 g/L perchlorate, followed by a 5-g/L perchlorate spiking test. The 5-g/ L perchlorate-spiking test was designed to demonstrate perchlorate removal performance at a low-end feed-perchlorate concentration, and initiate desorption of perchlorate from the carbon surface (if any) as the bed equilibrated with a lowered feed-perchlorate concentration. Along with the regular daily samples, effluent samples were taken at 0, 15, 30, 45, and 60 min after the filter was put into production following a change in feed-perchlorate concentration.

3) Nitrate spiking. This test was designed to evaluate perchlorate-removal performance during periods of transient nitrate- loading episodes. Nitrate was spiked to reach an influent concentration of 30 mg/L (the historic Saugus Aquifer maximum concentration) for an eight-day period. During the first 48 h of the test, the EBCT and acetic acid concentration were not adjusted to account for the increased nitrate in the feed. Subsequently, the EBCT and acetic acid dose were adjusted to account for the increased feed-nitrate concentration. Along with the regular daily samples, effluent samples were taken 0, 15, 30, 45, and 60 min after the filter was put into production following a change in feed-nitrate concentration.

4) Electron-donor feed failure simulation. This test was designed to evaluate perchlorate-removal performance during a period of simulated electron-donor feed system failure. The acetic acid feed was shut off for 24 h, and effluent perchlorate concentrations were monitored. The acetic acid feed was then restarted, and the time required to reestablish perchlorate removal to below the limit of detection (1 g/L) was evaluated. Effluent samples were taken 0, 15, 30, 45, and 60 min and 2, 3, 4, 5, and 24 h after the filter was put into production following the acetic acid feed system shut-off.

5) Total system shutdown. This test was designed to evaluate perchlorate-removal performance after the system was shut down for up to two weeks. The time required to reestablish efficient perchlorate removal to below detection in the FXB bioreactor was evaluated.

Backwashing. During Phase 1 testing (biological acclimation), various run times and backwashing procedures were tested to find an optimal protocol. Run times of at least 48 h were targeted to allow for high water production efficiency (i.e., ≥ 97%) without allowing overgrowth in the FXB, which would be difficult to clean during a backwash. Backwashes were initiated manually and were performed using effluent water stored in the pilot backwash tank (see Figure 1). Backwashes consisted of a 10-min air scour at ~0.1 to 0.2 SCFM (.0028 to .0057 m^sup 3^/min), followed by a 20-min backwash at 9 gpm/sq ft (6.1 L/s/m^sup 2^), which resulted in a 45% bed expansion.

Analytical methods. Water quality data were measured using bench- top equipment, in-line instruments, and laboratory analyses. For laboratory analyses, samples were collected in amber glass or polypropylene bottles and stored at 4 C until analyzed or shipped on ice.

US Environmental Protection Agency (USEPA) method 314.0 was used for perchlorate analyses at the CLWA laboratory (USEPA, 1999). The ion chromatograph7 included one analytical column8 and one guard column9 in series. The analytical method included a 65-mM potassium hydroxide eluent, a 1-mL/min flow rate, and a 1,000-L injection loop. The MRL for perchlorate was 4 g/L and the limit of detection was approximately 1 g/L.

Duplicate samples were prepared and analyzed weekly to indicate analytical precision. An onsite ion chromatograph7 was used to provide near real-time perchlorate-removal performance data and thus served as a guide for making timely pilot-plant operational adjustments. The system used one analytical column10 and one guard column11 in series.

A one-point calibration (20 g/L) was performed once per month. Effluent turbidity was measured inline using low range process turbidimeters.12 The turbidimeters were calibrated weekly using 0-, 20-, 100-, and 800-ntu primary calibration standards purchased from the manufacturer.

Feed water turbidity was measured daily using a benchtop field turbidimeter.13 Dissolved organic carbon (DOC) was measured according to method 415.1 (USEPA, 1999) using a total organic carbon (TOC) analyzer.14 Biodegradable organic carbon (BDOC) was measured using the method outlined by Servais et al (1987). Nitrate, nitrite, sulfate, and phosphate were analyzed using method 300.0A (USEPA, 1993). Hydrogen sulfide was measured using method 376.2 (USEPA, 1978). Onsite DO measurements were taken using a DO meter that was calibrated daily.15

A pH/conductivity meter with a built-in automatic temperature compensation device16 was used to measure pH and temperature in accordance with methods 4500-H+ B and 2550B, respectively. (Standard Methods, 1999). A two-point calibration of the pH meter was performed daily. Heterotrophic plate counts (HPCs) and total and fecal coliform counts were measured according to methods 9215B and 9221 D, respectively (Standard Methods, 1999).

RESULTS AND DISCUSSION

Biological acclimation. Figure 2 shows perchlorate removal performance during the acclimation, optimization, and sustained removal phases. Data graphs presented in this article show all perchlorate detections, even those below the 4 g/L MRL. Nondetections were given a value of zero. Perchlorate was initially removed by adsorption onto the virgin GAC and began to break through after 12 days as the carbon's perchlorate-adsorption capacity diminished. These data are consistent with other perchlorate/GAC breakthrough curves reported in the literature (Brown et al, 2002; Na et al, 2002).

During the subsequent month of biological acclimation, run times were less than 24 h because of rapid headloss development in the filter. The pilot feed lines experienced heavy biological growth, and the biofilm from the feed lines occasionally detached and clogged the bed. The acetic acid feed concentration was lowered during the acclimation phase to minimize growth. However, excessive biogrowth in the feed lines continued to generate rapid headloss build-up. Phosphoric acid was added at 0.2 mg/L PO^sub 4^-P to select away from filamentous organisms suspected of causing the overgrowth and clogging problems. Simultaneously, the electron donor feed was repositioned so that acetic acid was fed directly into the top of the column instead of upstream of the static mixer, thus minimizing biofilm growth in the feed lines. Consequently, biogrowth was isolated to the filter bed, headloss development rates decreased, and 48-h run times were achieved. Within 10 days of adding phosphoric acid and repositioning the acetic acid feed, steady perchlorate removal to below detection was achieved in the FXB bioreactor.

FIGURE 2 Fixed-bed bioreactor testing (phases 1-3)

The phosphoric acid feed was removed during the optimization tests without affecting perchlorate removal performance\, indicating that background phosphorus concentrations (~70 g/L) did not limit biological activity in the FXB. Therefore, repositioning the acetic acid feed was likely the key to effectively acclimating the FXB bioreactor to achieve biological perchlorate removal. If this electron donor-feed configuration had been used from the beginning, it is estimated that steady-state biological perchlorate removal to below detection would have been achieved within four weeks of start- up.

Optimization and sustained removal. EBCT and acetic acid dosing adjustments for the acclimation, optimization, and sustained- removal phases are plotted in Figure 3. During the two months following the acclimation period, testing focused on determining the minimum EBCT and acetic acid feed concentrations that allowed sustained perchlorate removal to below detection and ≥48-h run times. The lowest EBCT that allowed for consistent perchlorate removal to below detection was 15 min (0.30 gpm [0.019 L/s], 3.4 gpm/ sq ft [2.3 L/s/m^sup 2^]), though associated run times were 24 h or less. When the EBCT was increased to 20 min (0.23 gpm [0.015 L/s], 2.6 gpm/sq ft [1.8 L/s/m^sup 2^]) to lengthen run times, channeling was visually observed adjacent to the inner wall of the FXB column, allowing perchlorate to break through at 20 to 65%. When the EBCT was increased to 25 min (0.18 gpm [0.011 L/s], 2.1 gpm/sq ft [1.4 L/ s/m^sup 2^]), observed short-circuiting was minimal, perchlorate was consistently removed to below detection, and run times were maintained at 48 h. Thus, 25 min was established as the design EBCT.

FIGURE 3 EBCT and acetic acid adjustments

It is unknown why considerable short-circuiting was observed when the EBCT was 20 min but not observed at EBCTs of 15 or 25 min. On the other hand, it is possible that short-circuiting would not even be a significant issue with larger-scale FXB bioreactors, provided that the backwashing protocol was adequate. With 4-in. (0.102-m) diameter columns, the reactor-wall/carbon-interface area to carbon- bed volume ratio is relatively large, a condition favorable for short-circuiting along the inner wall of the reactor. As the cross- sectional area of the carbon bed increases, the reactor-wall/carbon- interface area to carbonbed volume ratio decreases, thereby creating conditions less favorable for short-circuiting along the inner wall of the reactor.

Though not tested during this project, it is also possible that specifying a larger GAC for the FXB bioreactor would decrease short- circuiting while maintaining efficient biological perchlorate removal. In general, the rate of head loss development in a granular- media filter/bioreactor will be slower as the effective size of the media increases. As head loss across the media bed decreases, there would be greater tendency for the water to flow uniformly across the bed instead of creating low-head flow channels. However, larger effective size also means decreased media-surface area for a given bed volume, which could effect the concentration of biomass in the bed and consequently decrease perchlorate removal efficiency. Further testing is required to confirm that a larger reactor and larger GAC would achieve better hydraulic performance without sacrificing perchlorate removal efficiency in a FXB bioreactor.

FIGURE 4 Perchlorate, DO, and nitrate trends

Acetic acid feed concentrations varied from 30 to 100% excess relative to stoichiometric DO and nitrate demand. During the 15-min EBCT experiments, 50% excess acetic acid (9.4 mg/L as carbon) was added, and perchlorate was removed to below detection. During the 25- min EBCT experiments, acetic acid was added at 7.8 mg/L as carbon, but the influent nitrate concentration decreased over this period, resulting in acetic acid excess between 35 and 83%. These conditions also resulted in consistent perchlorate removal to below detection. When excess acetic acid was decreased to 30% on days 101 and 102, perchlorate broke through at 32% and 14%, respectively. On day 103, acetic acid was increased to 50% excess, and perchlorate removal to below detection was again achieved.

After the optimization period, the FXB was run for two weeks using a 25-min EBCT and 7.8 mg/L acetic acid carbon (50-83% excess). Removal of perchlorate to below detection was observed throughout this period. Based on the optimization and sustained removal data, a design acetic acid dose was set at 50% excess relative to the stoichiometric demand of DO and nitrate in the feed water.

Dissolved oxygen and nitrate. All perchlorate-reducing bacteria described in the literature are facultative aerobes or microaerophiles (Coates et al, 1999; Wallace et al, 1998). Thus, DO can inhibit biological perchlorate reduction by acting as a competing electron acceptor. Nitrate is another highly oxidized compound that can compete with perchlorate for use as an electron acceptor. Research has confirmed the occurrence of this inhibition by demonstrating that DO and nitrate concentrations must be low to achieve perchlorate removal to below detection in biological reactors treating relatively low concentrations of perchlorate (Brown et al, 2002; Herman & Frankenberger, 1999). Therefore, influent and effluent DO and nitrate concentrations were monitored closely during pilot testing.

FIGURE 5 DOC trend

Figure 4 shows effluent DO, nitrate, and normalized perchlorate concentrations during the acclimation, optimization, and sustained removal testing phases. Even when the biological activity was not fully developed in the FXB, effluent DO concentrations were generally low during the acclimation phase. DO was likely reacting with the surface of the GAC initially and was being reduced biologically as bacteria colonized the bed. Influent DO concentrations ranged from 4 to 8 mg/L and were reduced to 0.2-0.3 mg/L throughout most of the pilot testing period. Influent nitrate concentrations were approximately 15 mg/L through day 88, steadily decreased to 6 mg/L by day 120, and were steady at 3-6 mg/L for the remainder of the testing period. The variation in influent nitrate concentration was likely caused by a change in the production rate for the feed water well, which began near day 88. Effluent nitrate concentrations were high initially, decreased as biological activity developed in the GAC bed during the acclimation phase, and remained at ≤0.3 mg/L throughout most of the pilot-testing period. Nitrate broke through on four occasions while EBCT and acetic acid addition were being optimized. Nitrite was not detected in the effluent of the FXB bioreactor during the acclimation, optimization, or sustained removal testing phases.

Organic carbon. Figure 5 shows DOC concentrations measured in the feed and effluent of the FXB bioreactor during the acclimation, optimization, and sustained removal pilot-testing phases. During the first month of testing, the acetic acid feed was positioned well upstream of the feed sample port, which caused heavy biogrowth in the feed lines and resulted in highly variable feed and effluent DOC concentrations. The acetic acid feed was repositioned so that acetic acid was dosed directly to the top of the column (downstream of the feed sample port), thus eliminating the ability to directly measure exogenous DOC in the FXB bioreactor influent. Therefore, diluted samples of the stock acetic acid solution had to be used to validate target acetic acid doses.

Several effluent DOC concentrations were measured during the acclimation phase of pilot testing; however, subsequent testing showed only a single effluent DOC measurement above the 0.7-mg/L MRL. More important, effluent BDOC was detected in only one sample during the optimization phase (0.04 mg/L) and was never detected during the sustained removal phase (Figure 6). The acetic acid dosed above the stoichiometric influent DO and nitrate demand was likely being used for cell synthesis and/or possibly for incomplete sulfate reduction, though no experiments were performed to confirm this. Regardless, these data are significant, as they reflect the ability of a FXB bioreactor to efficiently reduce perchlorate to below detection without producing biologically unstable product water.

Other water quality and operational parameters. The following additional water quality observations were made during FXB bioreactor piloting:

* Influent sulfate concentrations were approximately 140 to 250 mg/L, and measured sulfate reduction across the bed ranged from 0 to 3 mg/L (Figure 7). Effluent sulfide was not detected analytically (MRL = 0.1 mg/L), though it was occasionally detected by smell while taking an effluent sample.

* Influent pH values were between 6.7 and 8.1 and typically decreased by a few tenths of a pH unit across the bioreactor bed, presumably because of the formation of carbon dioxide during acetic acid oxidation.

FIGURE 6 BDOC trend

FIGURE 7 Sulfate trend

* As expected, HPCs increased across the bioreactor. Average feed and effluent HPCs were 60,000 mL^sup -1^ and 230,000 mL^sup -1^, respectively.

* With the exception of two data points, total coliforms were not detected in the feed or effluent of the FXB bioreactor. The two detections indicated that coliforms in the feed water were partially removed across the FXB bioreactor. No fecal coliforms were detected during pilot testing. These data suggest that pathogenic bacterial growth was not fostered within the microbial community of the FXB bioreactor.

* Average feed water turbidities were approximately 0.5 ntu. Average effluent turbidities were approximately 0.6 ntu.

* Head loss across the filter ranged from 2 to 30 ft (0.6 to 9.1 m). Under optimal operating EBCT and acetic acid conditions, headloss peaked at 25 ft (7.6 m) but was typically between 5-10 ft (1.5-3.0 m). For comparison, conventional gravity-driven and pressurized granular-media filters typically operate at ≤10 ft and ≤30 ft, respectively.

Robust\ness characterization. Once optimal EBCT and acetic acid feed concentrations were determined and sustained perchlorate removal to below detection had been demonstrated, a series of robustness characterization tests were performed to show how the FXB bioreactor responds to system upsets.

Backwashing. On days 62 and 116, samples were taken every 15 min for 1 h after the filter was put back into production following a backwash event. No perchlorate was detected in any sample from either of these tests.

Perchlorate spiking. Transient perchlorate-loading episodes did not affect perchlorate removal performance in the FXB bioreactor (Figure 8). The feed perchlorate concentration was varied in step changes from 50 g/L to 100 g/L to 50 g/L to 300 g/L to 5 g/L while the EBCT and feed acetic acid concentration were maintained at 25 min and 7.8 mg/L as carbon, respectively. No perchlorate was detected in the effluent, even for the high-resolution samples (i.e., samples taken at 0, 15, 30, 45, and 60 min) taken immediately after the influent perchlorate concentration was changed.

Nitrate spiking. The feed nitrate concentration was increased stepwise from 6 mg/L (background) to 29 mg/L. Even though EBCT and acetic acid feed concentration were not initially adjusted to account for the increase in feed nitrate concentration, effluent perchlorate was not detected in the high-resolution samples taken up to 1 h immediately following the nitrate spike (Figure 9), possibly indicating the occurrence of perchlorate adsorption to the GAC. However, samples taken ≥19 h after the step nitrate increase showed perchlorate, nitrate, and nitrite breakthrough. Increasing the acetic acid feed concentration to account for the 29-mg/L feed nitrate concentration (day two) did not immediately recover perchlorate removal performance to below detection. Eight days after the feed nitrate concentration was increased to 29 mg/L, complete removal of nitrate, nitrite, and perchlorate was again observed in the FXB.

FIGURE 8 Perchlorate spiking test

FIGURE 9 Nitrate spiking test

FXB bioreactor performance during the nitrate spiking test suggests that when large step increases in feed oxidant concentrations occur, the biological FXB requires a period of bioacclimation before perchlorate removal to below detection can be reestablished and maintained. The bacterial mass in the FXB, which was equilibrated to a feed nitrate concentration of 6 mg/L, was insufficient to handle a sudden increase in feed nitrate concentration of 23 mg/L. The bacterial community had to grow and equilibrate with a much higher nitrate concentration before it could generate the low-redox conditions necessary for complete perchlorate removal. Fortunately, large step changes in nitrate concentration like the one tested during this experiment are not typical for a groundwater system.

Natural, gradual nitrate fluctuations of up to a few mg/L over a 24-h period were observed throughout pilot testing and did not effect perchlorate removal performance in the FXB.

Electron donor feed failure simulation. To investigate the impact of an electron donor feed failure on perchlorate removal performance in the FXB, the acetic acid feed was shut off and effluent perchlorate concentrations were monitored. Samples taken 0, 15, 30, 45, and 60 min and 2, 3, 4, 5, and 24 h after the acetic acid pump was turned off showed no perchlorate breakthrough. The acetic acid pump was then restarted, and perchlorate continued to be removed to below detection. The following hypotheses may account for the exceptional perchlorate-removal performance in the FXB while no exogenous electron donor was being added to the system:

* The bacterial community in the FXB bioreactor was respiring endogenously. That is, bacteria were metabolizing their own cells for use as substrate. This often occurs when exogenous food sources are scarce.

* Perchlorate was being removed by adsorption while the acetic acid feed was off. The primary perchlorate removal mechanism in the FXB is biological reduction. However, if the microbial activity in the filter is insufficient to remove perchlorate, the GAC might adsorb perchlorate for a period of time, thus preventing breakthrough. If perchlorate diffusion through the biofilm is fast, adsorption could be occurring near the top of the GAC bed where the biological activity is highest. If perchlorate diffusion through the biofilm is slow, adsorption could be occurring in the lower portions of the GAC bed where there is less biological activity. Sorption sites may be biologically regenerated once the electron donor feed is restarted. This may be the case even for adsorption sites in the lower portions of the bed as GAC tends to migrate toward the top of the column during a backwash event, positioning it at depths of increased biological activity. This hypothesis could account for why there would be perchlorate-adsorption capacity available on the GAC after five months of pilot testing.

* Via adsorption and desorption, GAC was acting as an electron donor source while the acetic acid feed pump was turned off. A negative concentration gradient could have driven any previously adsorbed acetic acid to desorb once the acetic acid was no longer present in the bulk solution, thus temporarily providing the substrate necessary to achieve biological perchlorate reduction.

It is possible that any or all of the previously-described phenomena contributed to the 24-h removal of perchlorate to below detection in the FXB bioreactor while the acetic acid feed was shut off, though no experiments were performed to confirm this. These data suggest that in a full-scale application, there would be sufficient time to detect and remedy an electron donor feed failure before perchlorate breakthrough is observed.

System shutdown. The FXB bioreactor system was shut down for a two-week period in December 2003. It was backwashed and restarted at a 25-min EBCT and 7.8mg/L feed acetic acid carbon concentration. Samples were taken 0,15, 30, 45, and 60 min and 2, 24, and 48 h after the FXB bioreactor was put back into production. No effluent perchlorate was detected. This result is important, as it indicates the feasibility of maintaining the perchlorate removal capabilities of a redundant FXB bioreactor operating in stand-by mode during full- scale operation.

Technology acceptance. Based on the results of this pilot study, a comprehensive FXB biological perchlorate treatment engineering report was prepared and submitted to the CDHS technology acceptance application program. After a detailed review, CDHS granted conditional acceptance for FXB biological treatment of perchlorate- contaminated drinking water (CDHS, 2004), thereby making it possible for the first time to consider the full-scale design and implementation of FXB biological perchlorate treatment in California.

CONCLUSIONS

The treatment and disposal of perchlorate-laden residuals can be complicated and costly. Therefore, because of their ability to convert perchlorate to innocuous chloride and oxygen, biologically based processes may play an increasingly important role in the treatment of perchlorateladen drinking water. The pilot data from this project have shown that perchlorate-reducing FXB bioreactors can be acclimated using organisms indigenous to the Saugus Aquifer, can achieve sustained perchlorate removal to below the analytical detection limit using reasonable contact times, can produce biologically stable effluent, do not foster the growth of pathogenic bacteria, and are robust with respect to system upsets. As a result, FXB bioreactors are now conditionally approved by the CDHS for treating perchlorate-contaminated drinking water.

The only major concern related to the implementation of biological perchlorate treatment not addressed by this study relates to the lack of full-scale performance data. For now, the only way to alleviate this concern is to consider full-scale biological FXB systems that have been in operation in Europe for more than 20 years. Using acetic acid or ethanol as the electron donor, these systems effectively remove nitrate from drinking water under operating and water quality conditions comparable to those required for biological perchlorate removal (Brandebusch, 2004; Bonnelye, 1997). Together with this 20-year, full-scale biodenitrification track record, the results of this pilot study indicate that FXB biological treatment can be a feasible, effective, and sustainable process for removing perchlorate from drinking water.

ACKNOWLEDGMENT

The authors thank Steve McLean (director of operations at the time this work was completed), Paul Rowley, David Kimbrough, and other staff at the Castaic Lake Water Agency and Steve Cole and other staff from the Newhall County Water District for their expert assistance during this study. The authors also thank Laura Baumberger of Carollo Engineers for assistance in putting together the article.

If you have a comment about this article, please contact us at journal@awwa.org.

Groundwater contaminated with perchlorate is a health concern because of its ability to disrupt thyroid hormone production when ingested, which may inhibit normal growth and development. Although several abiotic and biotic technologies can efficiently remove perchlorate from drinking water, currently only ion exchange is being applied for perchlorate removal in full-scale plants, because it is the least expensive option among the abiotic processes. Cost estimates for biotic treatment processes indicate their potential to be relatively inexpensive options for treating perchlorate- contaminated drinking water. However, other concerns have limited their consideration for full-scale perchlorate treatment. In this article, the authors discuss the results of a six-month pilot study that was performed at the Castaic Lake Water Agency in Santa Clarita, Calif., to evaluate the effectiveness of fixed\-bed bioreactors to treat perchlorate-contaminated water. The pilot data showed that perchlorate-reducing fixed-bed bioreactors can be acclimated using organisms indigenous to the local aquifer; can achieve sustained perchlorate removal to below the analytical detection limit using reasonable contact times; can produce biologically stable effluent; do not foster the growth of pathogenic bacteria; and are robust in respect to system upsets.-KD

Because of a 100-fold decrease in its limit of detection and its subsequent discovery in drinking water sources in more than 25 states, perchlorate awareness among the general public has increased in the United States.

Cost estimates for biotic treatment processes indicate their potential to be relatively inexpensive options for treating perchlorate-contaminated drinking water.

Perchlorate-reducing fixed-bed bioreactors can . . . achieve sustained perchlorate removal to below the analytical detection limit using reasonable contact times.

FOOTNOTES

1 ITT Standard, Model SX200, Checktowaga, N.Y.

2 Fisher Scientific, Hanover Park, Ill.

3 Ismatec, Northbrook, Ill.

4 Eastman Chemical Company, Kingsport, Tenn.

5 Prayon, Inc., Augusta, Ga.

6 Calgon F-400, Pittsburgh, Pa.

7 Dionex DX-800 Process Analyzer, Sunnyvale, Calif.

8 Dionex 4 250 mm IonPacf AS16, Sunnyvale, Calif.

9 Dionex 4 50 mm IonPac AG16, Sunnyvale, Calif.

10 Dionex 2 250 mm IonPac AS16, Sunnyvale, Calif.

11 Dionex 2 50 mm IonPac AG16, Sunnyvale, Calif.

12 Hach 1720 D low-range process turbidimeters, Loveland, Colo.

13 Hach 2100 P field turbidimeter, Loveland, Colo.

14 Shimadzu TOC-5050 Total Organic Carbon Analyzer, Columbia, Md.

15 YSI 5500, Yellow Springs, Ohio.

16 Oakton 300 series, Vernon Hills, Ill.

REFERENCES

Amy, G. et al., 2003. Treatability of Perchlorate-containing Water by RO, NF, and UF Membranes. AWWA Res. Fdn. (AwwaRF), Denver.

Bonnelye, V. et al, 1997. Combined Advanced Oxidation of Pesticides and Biological Nitrate Removal in a 900-m3/hr Drinking Water Treatment Plant. Proc. 13th Ozone World Congress, Kyoto, Japan.

Booth, S. et al, 2000. Evaluating Electrodialysis Reversal (EDR) for Perchlorate Treatment. Proc. AWWA Ann. Conf., Denver.

Brandebusch, M. 2004. Personal communication.

Brandehuber, P. & Clark, S., 2004. Perchlorate Occurrence Mapping. Proc. AWWA WQTC, San Antonio.

Brown, J.C.; Snoeyink, V.L; & Kirisits, M.J., 2002. Abiotic and Biotic Perchlorate Removal in an Activated Carbon Filter. Jour. AWWA, 94:2:70.

California Department of Health Services, 2004. Conditional Acceptance of Fixedbed Biological Treatment for the Production of Drinking Water From Perchlorate-contaminated Water. Letter written to Carollo Engineers, November 15, 2004.

Coates, J.D. et al, 1999. Hydrocarbon Bioremediative Potential of (Per)chlorate-reducing Bacteria. Bioremediation Jour., 3:4:323.

Frankenberger, W.T. et al, 2004. Development of a Dynamic Suspended Bioreactor for Treatment of Perchlorate in Water. Proc. 2004 East Valley Water District/AwwaRF Water Quality Conf., Ontario, Calif.

Gu, B., et al, 2001. Regeneration of Perchlorate (CIO^sub 4^^sup - ^)-loaded Anion Exchange Resins by a Novel Tetrachloroferrate (FeCl^sub 4^^sup -^) Displacement Technique. Environ. Sci. & Technol., 35:3363.

Guarini, W.J., 2004. Ex Situ Biological Treatment of Perchlorate Using Biological Fluidized-bed Reactors: An update. Proc. East Valley Water District/AwwaRF Water Quality Conf., Ontario, Calif.

Herman, D.C. & Frankenberger Jr., W.T., 1999. Bacterial Reduction of Perchlorate and Nitrate in Water. Jour. Environ, Qual., 28:1018.

Lehman, G. et al, 2004. MBfR for Multicontaminant Groundwater Treatment and Remediation.Proc. East Valley Water District/AwwaRF Water Quality Conf., Ontario, Calif.

Liu, J. & Batista, J., 2000. A Hybrid (Membrane/Biological) System to Remove Perchlorate From Drinking Waters. Proc. Ann. Joint Services Pollution Prevention and Hazardous Waste Mgmt. Conf., San Antonio.

Miller, J.P. & Logan, B.E., 2000. Sustained Perchlorate Degradation in an Autotrophic, Gas Phase, Packed-bed Bioreactor. Environ. Sci. & Technol., 34:14:3018.

Min, B. et al, 2004. Perchlorate Removal in Sand and Plastic Media Bioreactors. Water Res., 38:47.

Na, C.; Cannon, F.S.; & Hagerup, B., 2002. Perchlorate Removal via Iron-Preloaded GAC and Borohydride Regeneration. Jour. AWWA, 94:11:90.

Nerenberg, R. et al, 2002. Perchlorate Reduction in a Hydrogen- Based Membrane-Biofilm Reactor. Jour. AWWA, 94:11:103.

Servais, P. et al, 1987. Determination of the Biodegradable Fraction of Dissolved Organic Matter in Waters. Water Res., 21:4:445.

Standard Methods for the Examination of Water and Wastewater, 1999 (20th ed.). APHA, AWWA, and WEF, Washington.

Tripp, A.R. et al, 2003. Treatment of Perchlorate in Groundwater by Ion Exchange Technology. AwwaRF, Denver.

USEPA (US Environmental Protection Agency), 1999. Method 314.0- Determination of Perchlorate in Drinking Water by Ion Chromatography. Office of Research and Development, Washington.

USEPA, 1999. Method 415.1-Determination of Total Organic Carbon in Water by Combustion or Oxidation. Office of Research and Development, Washington.

USEPA, 1993. Method 300.0A-Determination of Inorganic Substances in Environmental Samples by Ion Chromatography. Office of Research and Development, Washington.

USEPA, 1978. Method 376.2-Determination of Sulfide by Colorimetry (Methylene Blue). Office of Research and Development, Washington.

Wallace, W. et al, 1998. Perchlorate Reduction by a Mixed Culture in an Up-flow Anaerobic Fixed-bed Reactor. Jour. Ind. Microbiol. Biotechnol., 20:126.

ABOUT THE AUTHORS

Jess C. Brown (to whom correspondence should be addressed) is an engineer at Carollo Engineers, 401 N. Cattlemen Rd., Ste. 306, Sarasota, FL; e-mail jbrown@carollo.com. Brown has a BA environmental science and public policy from Harvard University in Cambridge, Mass. He also has a BS in civil engineering and MS and PhD degrees in environmental engineering from the University of Illinois at Urbana-Champaign. Brown manages Carollo Engineers' southeast advanced process engineering group and has seven years of experience in researching, testing, and designing biological perchlorate and nitrate treatment systems. Ryan D. Anderson, Joon H. Min, and Lina Boulos are engineers and David Prasifka and Graham J. G. Juby are engineers and parttiers at Carollo Engineers.

Copyright American Water Works Association Sep 2005

Source: American Water Works Association. Journal

More News in this Category