Electrochemical disinfection of surface water using mixed metal oxide anodes was evaluated in a pilot-scale demonstration. Disinfection rates, chlorine generation, energy demand, and generation of disinfection by-products were monitored over the 190-day study. Particular attention was given to the generation of trihalomethanes (THMs) and haloacetic acids (HAAs) during the electrochemical treatment cycle. In addition, the potential for generation of THMs and HAAs during post-treatment storage of the water was assessed. The electrochemical treatment system resulted in a 2- to 3-log removal of total heterotrophic bacteria, with values below detection (<1 CFU/mL) often observed. Disinfection occurred with only very low levels of observed chlorine generation (<0.1 mg/L), suggesting that alternate disinfection mechanisms likely played a significant role in the observed removal of bacteria. THM and HAA concentrations after treatment were consistently well below regulatory levels. Results also showed that electrochemical treatment significantly reduced the formation of bromoform when the water received subsequent exposure to hypochlorite. Removal of naturally occurring bromide in the water by the electrochemical system may have been the cause (in part) for this observed mitigation of bromoform formation. The formation of calcium scale on the cathode surface over time was the primary operational challenge.

INTRODUCTION

With growing demands on water resources, increased focus is on water reuse or utilization of surface waters as potable water sources. For these water sources, removal of chemical and biochemical oxygen demand (COD/BOD) and disinfection typically are the primary treatment objectives. Chlorine-based disinfection is by far the most common technology used for water treatment. However, the generation of haloacetic acids (HAAs) and trihalomethanes (THMs) resulting from chlorine-based disinfection has been well documented (Villanueva et al. 2003; Krasner et al. 2006). Implementation of the Stage 2 Disinfection Byproducts Rule (DBPR2) has applied increased regulatory pressure on water treatment operators in the United States to mitigate the generation of these disinfection by-products (DBPs). The DBPR2 sets a maximum contaminant level (MCL) for total THMs (TTHMs, defined as the sum of chloroform, bromodichloromethane, dibromochloromethane, and bromofrom) at 80 μg/L, and an MCL for the sum of five HAAs (HAA5, defined as the sum of monochloroacetic, dichloroacetic, trichloroacetic, bromoacetic, and dibromoacetic) at 60 μg/L. While strategies to mitigate DBPs through removal of organic carbon precursors via ion exchange (Tan et al. 2005) or sorption of DBPs onto activated carbon (Uyak et al. 2007) have been demonstrated, disinfection strategies that do not lead to DBP exceedances of the DBPR2 are preferred.

It is recognized that effective non-chlorine based chemical disinfection has been widely studied (e.g. ozone, peracetic acid). While TTHMs and HAA5 are typically not a concern for these disinfectants, generation of other classes of disinfection byproducts or increases in COD present challenges for these methods (e.g., Pisarenko et al. 2012; Cavallini et al. 2013). These methods also require the addition of chemicals. UV-based disinfection approaches also fail to provide any residual disinfection, which is required for many water treatment systems.

In efforts to provide disinfection without chemical addition of chlorine-based disinfectants, and to ultimately provide cost-effective disinfection without regulatory exceedances of THMs or HAAs, several laboratory studies have examined the use of electrochemical disinfection. These studies have shown that electrochemical treatment techniques using mixed metal oxide (MMO) anodes can be cost-effectively applied (Särkkä et al. 2008; Jeong et al. 2009), and without excessive generation of disinfection byproducts (Bagastyo et al. 2011; Ghebremichael et al. 2011). The mechanism for disinfection using MMO anodes (including Ti/IrO2) has been shown to be primarily due to the generation of oxidized chlorine species from chloride present in the water (Kraft 2008). Thus, increased chloride and conductivity levels in the water typically reduce the time and energy needed for disinfection. However, generation of other reactive oxidants such as ozone or hydroxyl radicals, generated at the anode surface (Comninellis 1994; Jeong et al. 2009), can contribute to disinfection via electrochemical treatment. Recent laboratory studies showed that several log removal of Escherichia coli using a Ti/IrO2 anode could be attained without exceedances of the DBPR2 (Schaefer et al. 2015). In addition, electrochemical treatment has been shown to be more effective than conventional chlorination approaches towards bacterial spores and more recalcitrant bacteria (Li et al. 2011; Mezule et al. 2014).

Formation of DBPs is often facilitated when chlorine-based chemicals are added to disinfected water that is being stored; this practice is used to prevent growth of microorganisms in tanks so that the water remains sufficiently disinfected. Chlorine-based oxidants can react with natural organic matter and naturally occurring bromide to form THMs and HAAs, as described by Sohn et al. (2004). Water that has been initially treated via an electrochemical system may be less prone to DBP formation upon subsequent exposure to chlorine-based disinfectants due to removal of halogens during the electrochemical process (Kimbrough & Suffet 2002) and/or removal of COD (Liu et al. 2012). However, this potential benefit has not been fully assessed at the pilot scale for an electrochemical treatment system.

Only limited information is available regarding pilot- or field-scale electrochemical treatment systems with respect to DBP formation. Katsoni et al. (2014a, 2014b) observed formation of THMs and other halogenated disinfection byproducts in wastewater and drinking water pilot studies employing the use of boron-doped diamond (BDD) anodes. THM formation was maintained below 80 μg/L in the electrochemical treatment of secondary effluents at a wastewater treatment plant and of the reverse osmosis concentrates only when low current densities were applied (Perez et al. 2010). Anglada et al. (2010), using a pilot scale system with BDD anodes to treat highly saline industrial wastewaters, also observed elevated THM generation. The formation of THM and haloacetonitriles, haloketones and 1,2-dichloroethane was observed in the electrochemical oxidation of landfill leachates (Anglada et al. 2011). Miao et al. (2015) examined the formation of chloroform and bromate in a pilot-scale electrochemical treatment system for river water using MMO anodes, but did not look at other THMs, HAAs, or perchlorate. Similar pilot scale studies using MMO anodes and analyzing for both THMs and HAA generation during disinfection have not been reported. Thus, the effectiveness of disinfection coupled with the generation of HAAs and THMs at the field scale has not been thoroughly demonstrated or assessed. The overall objective of this study was to assess the disinfection and disinfection by-product formation during a 6-month pilot-scale study using an electrochemical process. Overall system performance, including active chlorine generation and energy consumption, were also evaluated during this study.

METHODS

Demonstration location

The pilot demonstration was performed at Naval Air Station (NAS) Lemoore (Kings County/Fresno County, CA). The influent water source for the electrochemical treatment system was surface water from the California State Water Project aqueduct water, which was collected via a slip-stream from an existing water treatment system (water was diverted upstream of the existing water treatment system, thus the raw untreated surface water was used for this demonstration). Water quality parameters for the raw surface water are provided in Table 1. The skid-mounted treatment system was placed indoors within an existing treatment plant facility.

Table 1

Parameters for aqueduct water

Analyses Influent concentration Effluent concentration 
Nitrite (as N) (mg/L) <0.2 <0.2 
Chloride (mg/L) 110 ± 18 110 ± 17 
Sulfate (as S) (mg/L) 59 ± 14 57 ± 14 
Bromide (mg/L) 0.7 ± 0.3 0.4 ± 0.2 
Nitrate (as N) (mg/L) 0.4 ± 0.2 0.4 ± 0.2 
Phosphate (as P) (mg/L)a <0.2 <0.2 
Chlorate (mg/L) <0.2 <0.2 
Perchlorate (μg/L) 0.55 ± 0.15 0.58 ± 0.18 
TOC (mg/L) 5.9 ± 1.5 5.1 ± 0.6 
Total coliforms (CFU/mL) 5.2 NA 
Turbidity (NTU)b 0.43/0.44 <1.0 
Hardness (as Ca) (mg/L) 140/120 130 
Alkalinity (as Ca) (mg/L) 130/100 120 
Conductivity (μohms/cm) 685/551 681 
pH (standard units)c 8.3 ± 0.3 
Analyses Influent concentration Effluent concentration 
Nitrite (as N) (mg/L) <0.2 <0.2 
Chloride (mg/L) 110 ± 18 110 ± 17 
Sulfate (as S) (mg/L) 59 ± 14 57 ± 14 
Bromide (mg/L) 0.7 ± 0.3 0.4 ± 0.2 
Nitrate (as N) (mg/L) 0.4 ± 0.2 0.4 ± 0.2 
Phosphate (as P) (mg/L)a <0.2 <0.2 
Chlorate (mg/L) <0.2 <0.2 
Perchlorate (μg/L) 0.55 ± 0.15 0.58 ± 0.18 
TOC (mg/L) 5.9 ± 1.5 5.1 ± 0.6 
Total coliforms (CFU/mL) 5.2 NA 
Turbidity (NTU)b 0.43/0.44 <1.0 
Hardness (as Ca) (mg/L) 140/120 130 
Alkalinity (as Ca) (mg/L) 130/100 120 
Conductivity (μohms/cm) 685/551 681 
pH (standard units)c 8.3 ± 0.3 

Average values ± standard deviation are shown for the anions (n = 10), perchlorate (n = 4), pH (>4,000 on-line readings), and TOC (n = 3). Turbidity, hardness, alkalinity, and conductivity measured 51 days after start-up (influent and effluent), and 135 days after start-up (influent only). Total coliforms were measured by New Jersey Analytical Laboratories using SM9222B approximately 9 months prior to start-up. NA, not analyzed.

aOne detection of phosphate at 0.1 mg/L (influent and effluent) below the reporting limit of 0.2 mg/L was observed. The other 9 phosphate readings were non-detected (<0.2 mg/L).

bEstimated turbidity values in the influent below the reporting limit of 1.0 NTU.

cpH measured via on-line meter, and reflects pH from beginning to end of treatment cycle.

Pilot plant design and materials

The pilot demonstration was performed using a sequencing batch electrochemical treatment system. The demonstration period was approximately 190 days. System components are shown schematically in Figure 1. The primary system components include an influent flowmeter and 50 micron pre-filter, a treatment tank (100 L batch volume), a pump to recirculate water through the electrochemical cell, the electrochemical cell, DC power source (Keysight N5749A/903), recirculation and discharge flowmeters, and the collection tank to contain the treated water prior to drain disposal. The electrochemical cell used in this demonstration was the commercially available Multi-Purpose (MP) Flow Cell (ElectroCell North America, Inc.). A picture of the cell used in this pilot plant study is provided in the Supplemental Materials (Figure S1, available with the online version of this paper); additional information and schematics of this cell are provided at http://www.electrocell.com/electro-mp-cell.html. This undivided electrochemical cell consisted of six parallel plate electrodes (three anodes and three cathodes) within a polypropylene casing supported on a stainless steel frame. The electrodes were separated by a Viton gasket. The electrochemical cell consisted of 500 cm2 of Ti/IrO2 anodic surface area and 500 cm2 of stainless steel cathodic surface area. The distance between electrodes was 16 mm.
Figure 1

Schematic of the pilot-scale electrochemical treatment system.

Figure 1

Schematic of the pilot-scale electrochemical treatment system.

Operating conditions

Batch treatment cycles were 49 minutes, with an electrochemical cell contact time of 0.08 minutes (determined by dividing the electrochemical cell volume by the 100 L batch volume, and multiplying by the 49 minute batch time). During each batch cycle, water was recirculated through the electrochemical cell at a rate of 0.9 m3/h. The applied current to the electrochemical cell was 4 amps, which resulted in a current density of 8 mA/cm2. This current density was selected based on previous bench-scale electrochemical testing performed on the aqueduct water, where the 8 mA/cm2 current density generated the most residual chlorine among the current densities tested (Schaefer et al. 2015). The treatment tank was passively vented to the outside of the building as a precaution against any accumulation of gases (hydrogen, oxygen, or chlorine). The net treatment rate for the raw surface water, including the time needed to fill and discharge the treatment tank, was 0.11 m3/h. In addition to low and high level alarms for the treatment and collection tanks, the treatment tank was also equipped with a total free chlorine sensor and integrated temperature sensor (Hach 9184sc with sc200 controller), as well as a flow meter to measure the rate of recirculation flow.

As discussed in the results section, scaling at and near the cathode resulted in increased voltages and diminished system performance. To mitigate scaling, periodic system shutdowns were performed to rinse the electrochemical cell with approximately 0.15 L of a 10% (by volume) HCl solution. HCl was selected as it is commonly used for scale removal. The acid remained in the cell for approximately 15 to 30 minutes, after which time the cell was rinsed with water and the inlet and outlet tubing reconnected. During the last month of operation, a 31% (by volume) HCl solution was used to remove scale from the spacing screens used in the electrochemical cell (the cell was disassembled for this cleaning; this cleaning was performed weekly instead of the HCl acid rinse discussed above). It is noted that the cathodes themselves were rinsed with the 31% HCl acid solution during the last month of operation by submerging in the acid solution for 20 to 30 seconds. Following all acid rinsing procedures, all components contacted by the acid were thoroughly flushed with deionized water.

Monitoring and analytical methods

Temperature, pH, voltage, recirculation flow and pressure, and total free chlorine were monitored based on system automated readings. During the last several weeks of the demonstration, the N, N-diethyl-p-phenylene diamine (DPD) method was used off-line on grab samples to measure the sum of both free chlorine and other residual oxidants such as O3 or H2O2 (Schmalz et. al. 2009). Samples were analyzed via the DPD method approximately 24 hours after sample collection. Decay testing was performed, showing that storage of samples for 24 hours only resulted in approximately a 20% decrease in measured free chlorine levels as measured via the DPD method.

Total culturable heterotrophs were analyzed using a previously described plating technique (Schaefer et al. 2015). Microbial samples were preserved with sodium thiosulfate to quench any residual oxidant present in the electrochemically treated water. However, it is noted that previous testing showed that the storage of samples between sample collection and analysis (typically 24 hours) did not result in any additional disinfection (Schaefer et al. 2015). Total culturable heterotrophic plate counts (reported in units of CFU/mL) served as a surrogate measure for disinfection in this study. It is well recognized that heterotrophic plate counts can be used to assess disinfection for drinking water, and that heterotrophic plate counts serve as an indicator that is less susceptible to disinfection than total coliforms (LeChevallier et al. 1988; World Health Organization 2003; Allen et al. 2004). Thus, use of heterotrophic plate counts in this study serves as a conservative measure of the extent of disinfection relative to total coliform levels.

TTHMs and volatile organic compounds were analyzed using gas chromatography-mass spectrometry via EPA Method 8260B using a purge-and-trap. The detection limit for TTHMS and volatile organic compounds was 5 μg/L. The HAA5 were analyzed using a gas chromatograph with a flame ionization detector via EPA Method 552.3, which included adjustment of the pH to <0.5 and extraction using methyl tert-butyl ether. The analytical detection limit for HAA5 varied between 0.1 and 0.4 μg/L. Anions and perchlorate were analyzed via ion chromatography using EPA Methods 300.0 and 314.2, respectively. Anion detection limits were 200 μg/L; the perchlorate detection limit was 0.25 μg/L. In addition, influent and effluent samples were collected for analysis of turbidity (EPA Method SM2139B), alkalinity (EPA Method SM2320B), hardness (EPA Method SM2340B/C), total organic carbon (TOC) (EPA Method SM5310B,C,D), and conductivity (EPA Method 120.1; SW-846 9050A).

Following completion of the demonstration, the anode was removed and analyzed with a scanning electron microscope (SEM). Elemental analysis also was performed on the anode via energy dispersive X-ray spectroscopy (EDS). These analyses were performed using an FEI XL30 Environmental SEM with Bruker XFlash 4010 EDS Detector. The acceleration voltage used for SEM imaging was generally 30 kV. Comparisons were made using an unused anode.

Supplemental testing: disinfection by-product formation during post-treatment storage

Additional testing was performed on electrochemically treated water to assess the potential formation of TTHMs and HAA5 following disinfection within the treatment system. For these tests, system effluent samples were stored at room temperature in the dark in gas-tight 60 mL glass serum bottles with Teflon-lined butyl rubber stoppers and crimp seals. After 2 weeks of incubation, samples were analyzed for TTHMs and HAA5.

A separate set of experiments similar to that described in the previous paragraph was performed. The purpose of this second set of experiments was to determine if the electrochemically treated water mitigated the formation of DBPs upon subsequent exposure to hypochlorite while being stored. To assess the potential formation of TTHMs and HAA5 in electrochemically treated water that is amended with hypochlorite while being stored, the 2-week incubation experiments described in the previous paragraph were repeated, but were performed using both the influent (untreated) surface water as well as the electrochemically treated water. In addition, the water was amended with sodium hypochlorite solution (final concentration of 400 mg/L as hypochlorite) upon collection to determine if electrochemical treatment mitigated the formation of DBPs. This elevated hypochlorite level provided a means to attain a ‘worst case’ level of DBP generation that might occur over a longer storage period, and with multiple hypochlorite dosages.

RESULTS AND DISCUSSION

General operation

The system operated continuously for 190 days from start-up (November 19, 2014), with only brief (up to a few hours per event) shut-downs for the acid cleanings. The exceptions were a power outage that resulted in system shut-down between November 22, 2014 and November 24, 2014 (days 3 to 5), and a water level switch issue that resulted in ineffective operation of the system between March 16, 2015 and April 16, 2015 (days 120 to 148).

Influent and effluent for several water parameters are provided in Table 1. Comparison of influent and effluent parameters show no significant changes as a result of the electrochemical treatment process (i.e., <10% change). The exception is a 35% average decrease in bromide that was observed when comparing influent and effluent bromide concentrations (eight monitoring points during the demonstration period, which excludes two monitoring events when bromide levels were below the analytical detection limit). This decrease in bromide was likely due to transformation and volatilization of the bromide due to electrochemical treatment. pH changes during each treatment cycle were also negligible (generally <0.1 standard units for each 49 minute cycle). Results show that the temperature varied from approximately 19 to 30 °C (except for a 1 week period where the temperature decreased to 10 to 15 °C during days 49 to 56). Temperature fluctuations between 19 and 30 °C are not expected to substantially impact electrochemical disinfection (Jeong et al. 2006). These gradual fluctuations in temperature over the course of weeks likely reflect changes in the ambient temperature of the influent water. Temperature increases during the 49 minute cycle times were negligible (i.e., <2 °C).

Voltage levels measured during operation are shown in Figure S2 in the Supplemental Materials (available with the online version of this paper). They ranged between 10 and 15 V during the first 3 months of operation, and after a few initially high voltage readings, showed a slow increasing trend during this initial 3-month period. Due to the increased voltage, coupled with the diminished disinfection (discussed in the following section), a periodic acid rinsing was initiated on Day 85 to remove scale that was suspected to be the likely cause of the diminished performance.

Disinfection

Disinfection performance based on removal of total heterotrophic plate counts is summarized in Table 2. Effluent samples were collected within the last 5 minutes of the end of the selected batch cycle. Additional influent samples (data not shown) were collected prior to the treatment tank but after the 50 micron filter to confirm that no measureable removal of bacteria occurred due to filtration. Removal of total (culturable) heterotrophs from approximately 500 CFU/mL to <1 CFU/mL (∼3-log removal) was observed for all but one sampling event during the first 4 months of operation. The exception was the Day 69 sampling event, where only a 72% removal was observed. This diminished disinfection was likely due to the slow accumulation of scale; scale accumulation was not observed in the previous bench-scale studies (Schaefer et al. 2015). Upon initiation of periodic acid cleaning, effective disinfection was re-attained in the system.

Table 2

Disinfection results shown as a function of days since start-up

Days Influent (CFU/mL) Effluent (CFU/mL) 
0.2 534/512 <1/ < 1 
15 492/488 <1/ < 1 
21 477/480 <1/ < 1 
29 502/523 <1/ < 1 
56 526/531 <1/ < 1 
69 533/437 125/147 
85 510/527 <1/ < 1 
91 537/541 <1/ < 1 
98 512 <1 
106 530 <1 
154 531/519 <1/ < 1 
161 511 <1 
169 526 <1 
176 526 186 
182 544 147 
190 539 201 
Days Influent (CFU/mL) Effluent (CFU/mL) 
0.2 534/512 <1/ < 1 
15 492/488 <1/ < 1 
21 477/480 <1/ < 1 
29 502/523 <1/ < 1 
56 526/531 <1/ < 1 
69 533/437 125/147 
85 510/527 <1/ < 1 
91 537/541 <1/ < 1 
98 512 <1 
106 530 <1 
154 531/519 <1/ < 1 
161 511 <1 
169 526 <1 
176 526 186 
182 544 147 
190 539 201 

Samples were collected within 5 minutes of the end of the 49 minute cycle time. The system was started on November 19, 2014. Replicate values are shown, when collected.

While disinfection was generally effective during the first 4 months of the study, total heterotrophs were detected in the effluent for three sampling events during the last 3 weeks of the study (although there was no increasing trend in effluent total heterotrophs during the last 3 weeks, suggesting that there was no decreasing trend in disinfection rate during the last 3 weeks). It is suspected that the reduced acid contact time during the electrode cleaning that began on Day 154 may have resulted in insufficient scale removal, and ultimately diminished disinfection performance during the last few weeks of operation. The acid cleaning beginning on April 16 also was limited to the cathode only; it is unclear if not cleaning the anodes with acid had any adverse impacts, as the acid may have addressed any fouling/deactivation mechanisms not related to scale.

To further assess the rate of disinfection within the electrochemical system, samples were collected throughout the 49 minute treatment duration on four successive sampling dates: Day 161, Day 169, Day 176, and Day 182. Results of this sampling are shown in Figure S3 (Supplemental Materials, available with the online version of this paper). The rate of culturable heterotroph disinfection per specific charge (CFU/(A-min)), as determined by the slope of the linear regression in the figures, for the data collected on Days 161 and 169, are nearly identical, with an average value of 2.7 × 105 CFU/(A-min). This rate of removal is nearly identical (within 10%) to the rate of bacteria disinfection in the surface water (no added bacteria) observed in previous electrochemical bench-scale batch testing using the same surface water (Schaefer et al. 2015). However, the rates of disinfection on the Day 176 and Day 182 sampling events were approximately 30% less than in the previous 2 weeks. As discussed in the previous paragraph, this observed reduction in the disinfection rate may have been due to insufficient cleaning/contact time of the acid with the electrodes.

Free chlorine levels during the first 5 days of operation using the Hach 9184sc in-line meter showed maximum free chlorine in the range of 1 to 1.5 mg/L (measured at the end of the 49 minute treatment cycle). This value is consistent with the chlorine generation observed during the earlier bench-scale testing (using the same surface water and applied charge), which had a limited treatment duration (Schaefer et al. 2015). However, here, the maximum free chlorine levels decreased to <0.1 mg/L over time. To attain an improved measure of these low free chlorine levels, the DPD method was employed during the last several weeks of the study; samples were collected as a function of time during the treatment cycle. As shown in Figure 2, maximum free chlorine levels collected from Days 154 and 169 were in the range of 0.07 to 0.09 mg/L Cl, which are consistent with the on-line meter readings. These free chlorine levels were approximately 20 times less than the free chlorine generation (also determined using the DPD method) per specific charge (A-min/L) observed in the previously performed bench-scale studies using the same surface water (Schaefer et al. 2015).
Figure 2

Free chlorine levels as a function of the specific charge (Q) determined using the DPD method. Sampling was performed as a function of time during the 49 minute cycle for each of the days shown.

Figure 2

Free chlorine levels as a function of the specific charge (Q) determined using the DPD method. Sampling was performed as a function of time during the 49 minute cycle for each of the days shown.

Despite these low free chlorine levels, the measured rate of disinfection was similar in both the previous bench-scale (Schaefer et al. 2015) and the current pilot- scale studies, thereby suggesting that free chlorine generation alone was not responsible for the observed disinfection. Evidence for electrochemical disinfection (approximately a 1.5-log decrease in bacteria) was observed in our previous bench-scale experiments in the absence of reactive chlorine species; electrochemical disinfection results published by Mezule et al. (2014) also suggest that there are disinfection mechanisms not related to reactive chlorine species or other oxidants as measured via the DPD method that occur when using titanium-oxide based anodes. These alternate disinfection mechanisms for IrO2 anodes might include generation of low levels of ozone and/or higher oxide surface reactions due to hydroxyl radical interactions with the anode surface (Comninellis 1994; Jeong et al. 2009), and might explain why disinfection did not solely rely on active chlorine levels. No testing was performed to explicitly look for generation of ozone or hydroxyl radicals.

While the DPD and disinfection data collected between Days 154 and 169 (in comparison to the previously performed bench-scale studies) suggest that free chlorine levels were not the primary factor with respect to disinfection rate, the decreased levels of free chlorine (maximum of 0.01 to 0.05 mg/L Cl as shown in Figure 2) coupled with the decreased rates of disinfection (Table 2) during the last 3 weeks of operation (Days 176 and 190) suggest that free chlorine levels were important with respect to the observed rate of disinfection. Thus, both alternate disinfection mechanisms (described in the previous paragraph) as well as these relatively low free chlorine levels likely were contributing to the disinfection observed in the field demonstration. These data also suggest that, for the conditions of this pilot study, a free chlorine level greater than 0.05 mg/L is needed to achieve sufficient disinfection within the 49 minute treatment cycle.

While alternate disinfection mechanisms provide a plausible explanation for the disinfection observed with limited free chlorine generation in this pilot-scale demonstration, an explanation for the decrease in free chlorine generation during the 190-day pilot-scale demonstration is not readily available. Influent chloride levels remained relatively constant, so availability of chloride does not explain this observation. TOC concentrations entering the electrochemical reactor increased less than 1 mg/L during the last weeks of the study, thus it is unlikely that this small increase in TOC was responsible for the decrease in free chlorine, especially since the TOC levels did not decrease following electrochemical treatment. It is possible that ammonia or dissolved iron or manganese concentrations, or some other compound capable of consuming free chlorine that was not directly measured in this study, increased during the last few weeks of the study, and exerted an increased demand on the free chlorine. Another possibility is that the slow sorption of carbonate or bicarbonate species, or other compounds present in the water, may have hindered the formation of reactive chlorine species.

Disinfection by-products

Disinfection by-products results for TTHM and HAA5 are summarized in Table 3. Results indicated that there were no exceedances of the DBPR2 of 80 μg/L for TTHM and 60 μg/L for HAA5, with the exception of the 2-week incubated sample 3 weeks after start-up for TTHM (105 μg/L). While this exceedance appears to be anomalously elevated compared to all the other TTHM sample results, both the TTHM and HAA5 results show elevated levels during the first 3–4 weeks of operation. These elevated levels are likely due to the elevated free chlorine levels observed during the initial weeks of operation. It also is possible that elevated (up to 1,000 μg/L) levels of tetrahydrofuran (THF), 2-butanone, and acetone, which were observed in the influent and effluent during the first few weeks of sampling, contributed to the elevated TTHM and HAA5 levels. The presence of these organic compounds was most probably due to the presence of adhesives and sealants used for constructing the system and gluing piping. By the Day 29 sampling event, THF, 2-butanone, and acetone were below the analytical detection limit of 10 μg/L in the influent, and below the detection limit in the effluent by Day 56. However, the presence of these organic compounds in the first few weeks of operation may have contributed to the formation of the observed disinfection by-products. TTHM and HAA5 levels were generally below 10 μg/L after the first few weeks of the demonstration. Statistical analysis shows that these elevated disinfection by-product levels were highly correlated to both the free chlorine and the presence of the organic compounds, with linear correlation coefficients of 0.96 and 0.95, respectively. However, it is not possible to distinguish the impacts of free chlorine and the organic compounds (THF, 2-butanone, and acetone) on TTHM and HAA5 formation.

Table 3

Disinfection byproduct formation for effluent and 2-week incubated effluent

 TTHM (μg/L)
 
HAA5 (μg/L)
 
Days Effluent 2-week incubated effluent Effluent 2-week incubated effluent 
0.2 31/33 NS 36 NS 
15 9.3/9.8 NS 15 NS 
21 8.3/9.0 105 12 11 
29 6.2/6.3 14/15 2.0 2.0/2.0 
48 NS 15/7.2 NS <1.2/2.0 
56 5.1 NS 7.0 <1.2 
85 <5 NS <1.2 NS 
98 4.2 NS NS NS 
154 1.1 <5 NS <1.2 
169 1.8 1.0 1.6 2.6 
176 1.3 NS 7.2/6.9 NS 
182 2.9 NS 9.1/8.5 NS 
190 <5 0.9 NS <1.2 
 TTHM (μg/L)
 
HAA5 (μg/L)
 
Days Effluent 2-week incubated effluent Effluent 2-week incubated effluent 
0.2 31/33 NS 36 NS 
15 9.3/9.8 NS 15 NS 
21 8.3/9.0 105 12 11 
29 6.2/6.3 14/15 2.0 2.0/2.0 
48 NS 15/7.2 NS <1.2/2.0 
56 5.1 NS 7.0 <1.2 
85 <5 NS <1.2 NS 
98 4.2 NS NS NS 
154 1.1 <5 NS <1.2 
169 1.8 1.0 1.6 2.6 
176 1.3 NS 7.2/6.9 NS 
182 2.9 NS 9.1/8.5 NS 
190 <5 0.9 NS <1.2 

NS, not sampled.

The generation of TTHM in both the raw and electrochemically treated surface water during 2 weeks of incubation with 400 mg/L hypochlorite is shown in Figure 3. It is recognized that the extremely elevated levels of hypochlorite used for this portion of the testing likely resulted in drastically higher concentrations of DBPs than what would be expected using hypochlorite levels typically used for disinfection. However, this elevated hypochlorite dosage was intended to facilitate the assessment of TTHM and HAA5 formation following electrochemical chlorination, compared to chlorination on untreated water. There was no difference in HAA5 generation between the raw and electrochemically treated surface water. This suggests that (for the water characteristics and treatment applied in this study), electrochemical treatment provided no measurable mitigation of HAA formation for waters where hypochlorite might be used for residual disinfection downstream of the treatment system. However, the generation of the THMs was slightly but statistically less in the electrochemically treated water than in the raw surface water. The exception was bromoform, which was 45% less in the electrochemically treated water than in the raw water. Thus, electrochemical treatment was shown to provide mitigation of THM formation for waters exposed to hypochlorite for residual disinfection downstream of the treatment system. The difference in behavior between HAA5 and TTHM is not readily explained.
Figure 3

THM generation in both the raw (solid bars) and electrochemically treated (striped bars) surface water after 2 weeks of incubation with 400 mg/L hypochlorite. Average values of triplicate samples ±95% confidence intervals are shown. CF, chloroform; BDM, bromodichloromethane; DCM, dichlorobromomethane; BF, bromoform.

Figure 3

THM generation in both the raw (solid bars) and electrochemically treated (striped bars) surface water after 2 weeks of incubation with 400 mg/L hypochlorite. Average values of triplicate samples ±95% confidence intervals are shown. CF, chloroform; BDM, bromodichloromethane; DCM, dichlorobromomethane; BF, bromoform.

This electrochemical mitigation of THM formation, especially the bromoform, may be explained in part due to the partial removal of naturally occurring bromide during electrochemical treatment. On average, 35% of the naturally occurring bromide in the water was removed during treatment. Removal of bromide via transformation to Br2 and subsequent volatilization has been previously observed (Kimbrough & Suffet 2002). No measurable removal of chloride was observed during treatment. Removal of TOC during treatment could also contribute to mitigation of disinfection by-product removal. However, comparison of influent and effluent TOC concentrations showed that there was no measurable removal of TOC; removal of organics typically requires a greater specific charge than that required for disinfection. It is possible, however, that the TOC was transformed during electrochemical treatment via oxidative processes, altering its aromaticity and making it less susceptible to formation of THMs.

Monitoring for perchlorate was also performed during the first four sampling events. Results indicated that no measurable perchlorate generation (<1 μg/L) occurred. These results are consistent with our previous bench-scale study (Schaefer et al. 2015). Chlorate remained below the analytical detection limit (200 μg/L) in the effluent throughout the demonstration.

Anode surface

Post-demonstration analysis of the anode surface using SEM and EDS was performed; results were compared to SEM and EDS analyses performed on an unused anode. Visual inspection of the used anode showed approximately 12 white spots, with a diameter of approximately 1 mm. SEM analyses (Figure S4 in the Supplemental Materials, available with the online version of this paper) revealed more small mineral-like formations in the used anode than in the unused anode (though the field of view is larger since the magnification is lower), suggesting that electrochemical treatment may result in the deposition of minerals onto the surface of the anodes. EDS analysis (Table S1 in the Supplemental Materials, available with the online version of this paper) showed increases in sodium and calcium on the used anode (especially increased calcium at the location of the white spots), suggesting that the white spots are calcium deposits that developed during treatment. It is plausible that these calcium or other mineral deposits may have reduced the effectiveness of the anodes, and that improved acid washing (or periodic polarity reversal) may have removed these deposits.

Energy assessment

The energy demand per log-order of disinfection observed throughout most of the demonstration period was less than or equal to 0.21 W-h/(L order); during the last 3 weeks of the demonstration this value increased to approximately 0.30 W-h/(L order). It is noted that, based on our previous bench-scale study (Schaefer et al. 2015) using the same surface water, operating at a current density of 2 mA/cm2 (instead of the 8 mA/cm2 used in this demonstration to maximize chlorine residual) would have reduced the energy demand per log-order of disinfection by approximately a factor of 4 (to 0.05 W-h/(L order)). Raut et al. (2014) obtained values of 0.7–1.4 W-h/(L order) for E. coli spiked in synthetic urine using a BDD electrode setup. Disinfection energy using UV for treatment of E. coli has been observed at approximately 0.02 W-h/(L order) (Taghipour 2004; Madsen & Søgaard 2012), which is less than what we have observed in both the pilot- and bench-scale testing using the surface water. However, direct comparison of the energy demand for UV and electrochemical disinfection is not warranted. Differences in the water quality among studies can account for the observed energy differences. The presence of particulates passing filters as small as 10 to 20 microns has inhibited the effectiveness of UV with respect to disinfection (Taghipour 2004), thus UV disinfection in systems with suspended solids likely will require increased energy for disinfection. In addition, the energy required for electrochemical disinfection would substantially decrease if the water had a greater salinity. For example, Li et al. (2002) observed an energy demand of approximately 0.002 W-h/(L order) in a 0.8% salinity water. UV also provides no residual disinfectant to prevent microbial growth during storage; low levels of residual chlorine were generated during this electrochemical pilot study. Finally, UV disinfection may be far less complete than disinfection using electrochemical/chlorine-based disinfectants, as recent studies have shown that E. coli may be orders of magnitude more susceptible to regrowth when treated using UV than with chlorine-based disinfection (Zhang et al. 2015).

CONCLUSIONS

The results of this pilot-scale demonstration showed that electrochemical treatment can efficiently (with respect to energy consumption) provide disinfection without generation of disinfection by-products above regulatory levels. The results also suggest that electrochemical treatment reduced the potential and extent of bromoform formation (with only a trace reduction in the generation of other THMs) upon subsequent chlorination during storage of the treated water; the likely mechanism for this observation was removal of naturally occurring bromide from the surface water during the electrochemical treatment process.

Accumulation of calcium scale at and near the cathode surface was observed during the demonstration. While periodic (i.e., weekly) acid washing removed most of this scale, the acid washing technique employed during the last few weeks of the demonstration likely had insufficient contact time with the cathode surface for proper cleaning. In addition, mineral analysis of the anode surface showed that accumulation of calcium deposits occurred. Periodic (e.g., daily) polarity reversal could possibly provide means to reduce or prevent accumulation of scale within the cell, and on the electrode surfaces.

Operation of the electrochemical system at a lower current density is expected to substantially reduce the energy requirement for disinfection. However, operating at lower current densities likely will also reduce the already low levels of residual chlorine present, thus nearly eliminating any residual disinfection potential. Operating at a lower current density also would require a greater anodic surface area or greater treatment time in order to achieve the same treatment level (i.e., volume of water per unit time). Thus, the advantages and disadvantages of operating at a lower current density need to be considered in system design. Overall, these results suggest that electrochemical disinfection has potential to be a viable treatment option for drinking water, provided scaling or fouling of the electrodes can be prevented.

ACKNOWLEDGEMENTS

Support for this research was provided by the US Navy's Environmental Sustainability Development to Integration (NESDI) program, Project # 487, with technical input from Dr Nancy Ruiz. The authors also appreciate the assistance from Mr Chris Thompson of Greyter Water Systems, who provided technical design and fabrication support for the pumping and water storage components of the system. In addition, the authors recognize the significant contributions of Mr Matt Zwartjes, Mr Scott Gomarko, and Mr Bill Schwartz from Envirogen Technologies for their design support, as well as their operation and testing of the field unit at NAS Lemoore. Finally, the support from the staff at NAS Lemoore was greatly appreciated.

REFERENCES

REFERENCES
Allen
M. J.
Edberg
S. C.
Reasoner
D. J.
2004
Heterotrophic plate count bacteria – what is their significance in drinking water?
Int. J. Food Microbiol.
92
,
265
274
.
Anglada
A.
Ortiz
D.
Urtiaga
A. M.
Ortiz
I.
2010
Electrochemical oxidation of landfill leachates at pilot scale: evaluation of energy needs
.
Water Sci. Technol.
61
(9)
,
2211
2217
.
Bagastyo
A. Y.
Radjenovic
J.
Mu
Y.
Rozendal
R. A.
Batstone
D. J.
Rabaey
K.
2011
Electrochemical oxidation of reverse osmosis concentrate on mixed metal oxide (MMO) titanium coated electrodes
.
Water Res.
45
,
4951
4959
.
Cavallini
G. S.
de Campos
S. X.
de Souza
J. B.
Vidal
C. M. S.
2013
Evaluation of the physical-chemical characteristics of wastewater after disinfection with peracetic acid
.
Water Air Soil Pollut
.
224
,
1
11
.
Ghebremichael
K.
Muchelemba
E.
Petrusevski
B.
Amy
G.
2011
Electrochemically activated water as an alternative to chlorine for decentralized disinfection
.
J. Water Supply
60
,
210
218
.
Krasner
S. W.
Weinberg
H. S.
Richardson
S. D.
Pastor
S. J.
Chinn
R.
Sclimenti
M. J.
Onstad
G. D.
Thruston
A. D.
Jr
.
2006
Occurrence of a new generation of disinfection byproducts
.
Environ. Sci. Technol.
40
,
7175
7185
.
LeChevallier
M. W.
Cawthon
C. D.
Lee
R. G.
1988
Inactivation of biofilm bacteria
.
Appl. Environ. Microbiol.
54
,
2492
2499
.
Li
X. Y.
Ding
F.
Lo
P. S. Y.
Sin
S. H. P.
2002
Electrochemical disinfection of saline wastewater effluent
.
J. Environ. Eng.
128
,
697
704
.
Mezule
L.
Reimanis
M.
Krumplevska
V.
Ozolins
J.
Juhna
T.
2014
Comparing electrochemical disinfection with chlorination for inactivation of bacterial spores in drinking water
.
Water Sci. Technol.
14
(1)
,
158
164
.
Miao
J.
Chen
Y.
Yang
X. J.
Liu
Y.
Zhao
S.
Yang
Y.
Yang
C.
Wan
P.
2015
Laboratory and pilot plant studies on electrochemical pretreatment of drinking water
.
Desalin. Water Treat.
56
,
1
11
.
Perez
G.
Fernandez-Alba
A.
Urtiaga
A. M.
Ortiz
I.
2010
Electro-oxidation of reverse osmosis concentrates generated in tertiary water treatment
.
Water Res.
44
,
2763
2772
.
Raut
A. S.
Cunningham
G. B.
Parker
C. B.
Klem
E. J. D.
Stoner
B. R.
Deshusses
M. A.
Glass
J. T.
2014
Disinfection of E. coli contaminated urine using boron-doped diamond electrodes
.
J. Electrochem. Soc.
161
,
G81
G85
.
Särkkä
H.
Vepsäläinen
M.
Pulliainen
M.
Sillanpää
M.
2008
Electrochemical inactivation of paper mill bacteria with mixed metal oxide electrode
.
J. Hazard. Mater.
156
,
208
213
.
Uyak
V.
Yavuz
S.
Toroz
I.
Ozaydin
S.
Genceli
E. A.
2007
Disinfection by-products precursors removal by enhanced coagulation and PAC adsorption
.
Desalination
216
,
334
344
.
World Health Organization
2003
Heterotrophic Plate Counts and Drinking–water Saftey
.
IWA Publishing
,
London
,
UK
.

Supplementary data