JAWWA Full-Text

Lytle et al. | http://dx.doi.org/10.5942/jawwa.2015.107.0176Peer-Reviewed

E648

2015 © American Water Works AssociationJOURNAL AWWA DECEMBER 2015 | 107:12

Many regions in the United States have excessive levels of ammonia (NH4+) in their drinking water sources as a result of naturally occurring processes, agricultural and urban runoff, concentrated animal feeding operations, municipal wastewater treatment plants, and other sources. Ammonia is not regulated by the US Environmental Protection Agency (USEPA) as a con-taminant. Although ammonia does not pose a direct health con-cern at levels expected in drinking water, it may pose a concern when nitrification occurs in the drinking water distribution sys-tem. Nitrification, which is the conversion of the ammonia to nitrite and nitrate by bacteria, can lead to water quality issues such as accelerated corrosion, oxidant demand, taste and odor complaints, and elevated nitrite levels (Fleming et al. 2005, Bremer et al. 2001, Suffet et al. 1996, Wilczak et al. 1996, Ritt-man & Snoeyink 1984, Lee et al. 1980). USEPA’s regulatory limits for nitrite and nitrate (at the entry point to the distribution system) are 1.0 and 10 mg N/L, respectively.

Ammonia in water may also pose problems with water treat-ment effectiveness. For example, in source waters containing both ammonia and arsenic, the ammonia may negatively affect the removal of arsenic by creating a chlorine demand, therefore reducing the availability of chlorine needed to oxidize arsenic (Lytle et al. 2007). Lastly, water systems that have ammonia in their source water and need to maintain a free chlorine residual

will need to add chlorine to overcome the demand of ammonia, or face potential difficulties in meeting contact times necessary to achieve disinfection goals. The complete oxidation of source water ammonia before or as part of the water treatment process would eliminate potential negative impacts of nitrification on distribution system water quality.

BACKGROUNDStudy community. Palo, Iowa, is a small community of 1,026

people located 7 mi (11.3 km) west of Cedar Rapids. Before 2008, the community did not have centralized water treatment or a drinking water distribution system. Following extensive flooding of the region in 2008, plans were made to build the necessary infrastructure to supply the community with potable drinking water. The State of Iowa’s Department of Natural Resources (DNR) required that the treatment system be designed to address elevated levels of ammonia and iron in the source water. USEPA’s Office of Research and Development (ORD), supported by the Iowa DNR, and USEPA Region 7 conducted a pilot study to evaluate the ability of biological water treatment to meet drinking water goals for ammonia and iron removal (Lytle et al. 2013, 2012).

Elevated ammonia levels. Groundwater supplies in the Midwest are particularly affected by ammonia from natural sources,

Across the United States, high levels of ammonia in drinking water sources can be found. Although ammonia in water does not pose a direct health concern, ammonia nitrification can cause a number of issues and reduce the effectiveness of some treatment processes. An innovative biological ammonia-removal drinking water treatment process was developed and, after the success of a pilot study, a full-scale treatment system using the process was built in a small Iowa community. The treatment plant included a unique

aeration contactor design that is able to consistently reduce ammonia from 3.3 mg of nitrogen/L to nearly nondetectable after a biofilm acclimation period. Close system monitoring was performed to avoid excess nitrite release during acclimation, and phosphate was added to enhance biological activity on the basis of pilot study findings. The treatment system is robust, reliable, and relatively simple to operate. The operations and effectiveness of the treatment plant were documented in the study.

The Full-Scale Implementation of an Innovative Biological Ammonia Treatment Process

DARREN A. LYTLE,1 DAN WILLIAMS,1 CHRISTY MUHLEN,1 MAILY PHAM,1 KEITH KELTY,1 MATTHEW WILDMAN,2 GLENN LANG,3 MITCH WILCOX,4 AND MELISSA KOHNE5

1US Environmental Protection Agency, Office of Research and Development, National Risk Management Research Laboratory, Water Supply and Water Resources Division, Cincinnati, Ohio2HR Green Inc. Cedar Rapids, Iowa3City of Palo, Iowa4Pegasus Technical Services Inc., Cincinnati, Ohio5University of Cincinnati, Cincinnati, Ohio

Keywords: ammonia, biological treatment, drinking water, nitrification


E649


agricultural runoff, and other farming practices. For example, Iowa has a widespread distribution of ammonia in well waters across its communities (Figure 1). Initial water quality testing of the source groundwater in Palo (Table 1) showed that, on average, ammonia levels were 3.3 mg of nitrogen (N)/L.

Although the focus of this report is on ammonia contamination, it is relevant to note that initial site sampling indicated that the source water contained 0.82 mg/L of iron. USEPA recommended

a non-enforceable National Secondary Drinking Water Regulation Standard of 0.3 mg/L for iron, which is based on aesthetic and technical issues. Given the elevated ammonia levels in drinking water and the undesirable levels of iron, there was a clear need to establish an effective treatment approach to address both con-taminants. Furthermore, the Iowa DNR can require water systems to monitor nitrite and nitrate in their distribution systems should the DNR suspect that nitrification is occurring, and enforce the respective regulatory limits if they are exceeded.

Ammonia treatment options. The most commonly used water treatment options for addressing elevated ammonia in source waters are the formations of monochloramine and breakpoint chlorination. Breakpoint chlorination results in the removal of ammonia as nitrogen gas by a chemical reaction with chlorine, typically in the range of eight to 11 times the mg N/L of ammonia present (Blute et al. 2012). For Palo, this would be a chlorine dose of approximately 33 mg/L to achieve a free chlorine resid-ual of 1 mg/L. The formation of monochloramine involves the addition of chlorine to concentrations in which ammonia is not removed but rather bound to chlorine. Other removal ap -proaches, including ion exchange with zeolites, reverse osmosis, advanced oxidation, and air stripping, are capable of removing ammonia from water, but are relatively complex, expensive, or have limited applications.

Although often performed unintentionally, biological ammonia “removal” is another treatment approach to reduce source water ammonia (McGovern & Nagy 2010, Lytle et al. 2007). Techni-cally ammonia is not removed, but rather the process relies on bacteria to convert ammonia to nitrate. As a result, a more bio-logically stable water is produced, nitrification in the distribution system is not an issue, and free chlorine residual is easily achieved. Biological conversion of ammonia (NH3) to nitrate (NO3

–) involves a two-step sequence of reactions most commonly medi-ated by two genera of bacteria, Nitrosomonas and Nitrospira, although other genera can also mediate nitrification (e.g., Nitro-sococcus and Nitrobacter). These autotrophic bacteria derive energy for cellular functions from the oxidation of ammonia and nitrite, respectively. Nitrosomonas are responsible for the oxida-tion of ammonia, in the form of ammonium (NH4

+), to nitrite (NO2

–) according to the following reaction:

NH4+ + 1.5 O2 → NO2

– + H2O + 2H+ (1)

Nitrospira subsequently oxidizes nitrite to nitrate, as follows:

NO2– + 0.5 O2 → NO3

– (2)

By summing these equations, the overall nitrification reaction is obtained:

NH4+ + 2 O2 → NO3

– + 2 H+ + H2O (3)

These equations are net reactions involving a complex series of enzyme-catalyzed intermediate steps. Nitrification produces free protons, H+, which readily consume available bicarbonate ions (HCO3

–), thereby reducing the buffering capacity of the

FIGURE 1 Map of ammonia levels in Iowa based on groundwater

well analyses, 1998–2012

0–0.5 mg/L0.5–1.0 mg/L1.0–3.0 mg/L3.0–5.0 mg/L5.0–10.0 mg/L

Source: Provided by the State of Iowa

The star indicates the approximate location of the city of Palo.

Ammonia Levels

TABLE 1 Initial water quality analysis of drinking water source in Palo, Iowa

Parameter Level

Alkalinity 358 mg CaCO3/L

Fe 0.82 mg/L

Mn 0.01 mg/L

t-PO4 0.07 mg PO4/L

TOC 1.06 mg C/L

S 33 mg/L

Cl <5 mg/L

Mg 33 mg/L

t-NH3 3.3 mg N/L

NO2 0.04 mg N/L

NO3 0.02 mg N/L

pH 7.4

Temperature—°C 15.8

C—carbon, CaCO3—calcium carbonate, Cl—chloride, Fe—iron, Mg—magnesium, Mn—manganese, N—nitrogen, NO2—nitrite, NO3—nitrate, PO4—phosphate, S—sulfur, t-NH3—total ammonia, t-PO4—total phosphate, TOC—total organic carbon


E650


water. In addition, nitrifying bacteria consume CO2 to build new cells. The total consumption of alkalinity by nitrification is 7.1 mg as calcium carbonate (CaCO3) per mg NH4

+-N oxidized (USEPA 1975). The oxygen demand of nitrification is also sig-nificant. For complete nitrification, 4.6 mg O2 is required per mg NH4

+-N oxidized (USEPA 1993, 1975).Other factors that affect nitrification include phosphate con-

centration, pH, and water temperature. All organisms including nitrifying bacteria require phosphorus to build cell mass, with approximately 3% of dry weight consisting of phosphorus. Microorganisms use phosphate as the source of phosphorus for the synthesis of structural and physiological biomolecules such as deoxyribonucleic acid, phospholipids (membranes), teichoic acid (cell walls), and most important, as inorganic phosphorus in adenosine triphosphate (ATP) synthesis. Without ATP, the cellular metabolism (i.e., nitrification) cannot proceed, and the cells either become dormant or die. Some organisms are more sensitive to phosphate starvation than others, and in the case of nitrification, ammonia-oxidizing bacteria are less sensitive than nitrite-oxidizing bacteria (de Vet et al. 2012; Scherrenberg et al. 2012, 2011).

Numerous laboratory studies have cited the optimum pH for complete nitrification as between 7.4 and 8, although in practice, the bulk water pH may deviate from this value while nitrification remains high (Shammas 1986). Temperature can affect growth rate and metabolism by slowing or destroying necessary enzymes and proteins involved in physiological processes. Laboratory studies have demonstrated that the growth rate of nitrifying bacteria is negatively affected by temperatures below 10°C, although adjustments to the treatment process can be made to enhance nitrification in colder climates (Andersson et al. 2001).

Biological ammonia treatment limitations. The two-step micro-biological nitrification process requires oxygen (aerobic) to oxi-dize NH4

+ to NO2–, and then to NO3

–. The entire process requires approximately 4.6 mg of O2/mg of NH4

+-N in the source water. Elevated ammonia (3.3 mg N/L) and reduced iron (0.82 mg/L) (Table 1) levels combined with an already low oxygen content (3.6 mg O2/L) in Palo’s groundwater create an oxygen demand of more than 15 mg O2/L. Therefore, the introduction of oxygen (e.g., aeration) is a necessary feature of the biological ammonia treatment system. However, the traditional configuration of aeration followed by filtration (e.g., iron removal), including biologically active filtration, is not sufficient to address the oxy-gen demand to meet Palo’s treatment objectives.

The amount of oxygen that can be added to the water is con-trolled by the saturation limit of oxygen in water, which in most drinking waters—including the study community’s—is well below the total oxygen requirements of treatment. The authors, in their experience with microbiological systems that do not provide suf-ficient oxygen to a nitrifying system, have found that the result can be incomplete nitrification resulting in elevated nitrite levels in the finished water. Given that the drinking water standard for nitrite is only 1 mg N/L, concerns for potential exceedances exist where source water ammonia levels are greater than 1 mg N/L. Therefore, an innovative approach to introducing oxygen to the treatment system in Palo was necessary to meet the treatment objectives. Aerating with pure oxygen could provide super-saturated

oxygen conditions and sufficient oxygen; however, there are safety issues associated with onsite storage and use of pure oxy-gen, and potential filter-binding associated with gas bubbles being added at super saturation, although saturation thresholds change when the system is pressurized.

Biological ammonia treatment pilot study. From March 2011 to April 2012, a pilot-scale biological ammonia treatment plant was studied in Palo. Specifically, the pilot treatment system was based on a USEPA-patented process (Figure 2) to treat elevated levels of ammonia and iron in the source water (Patent No. U.S. 8,029,674). The process relies on bacteria to convert ammonia to nitrate. If the raw ammonia levels are lower than the nitrate drinking water limit of 10 mg N/L, the process can be effective and relatively simple.

The pilot system was designed and built by USEPA staff and installed in March 2011. The pilot system consisted of an aeration contactor followed by a granular (anthracite over sand) filter in series. The aeration contactor is the innovative process in the treatment train whereby air is bubbled through a biologically active gravel bed as source water passes through the column. As a result, saturated oxygen levels were maintained throughout the contactor despite being consumed in the nitrifying biofilm. This pilot system was operated on a continuous basis (24 h/day, 7 days/week), with the exception of a few instances where pumps were replaced or maintenance actions occurred (Lytle et al. 2012).

In April 2012, the pilot system was deemed a success and the decision to move toward full scale was made. The pilot system provided new and important findings that improved the drinking water field’s understanding of biological water treatment in gen-eral, and how to effectively operate such systems. Key findings included the need to add orthophosphate as a nutrient to aid the nitrification process, and the importance of maintaining saturated oxygen levels in the system (Lytle et al. 2013, 2012).

FULL-SCALE TREATMENT PLANT DESIGN AND OPERATIONStudy objective. The objective of this study was to document

the operation and treatment effectiveness of Palo’s innovative full-scale biological ammonia and iron-removal drinking water treatment plant. The treatment plant engineering design criteria and operating conditions are presented. The development of the project from pilot to full scale is discussed. Lastly, lessons learned from the project and future considerations when designing and operating biological treatment systems for ammonia removal are presented. The treatment plant represents one of the first full-scale biological drinking water treatment systems specifically designed to address elevated ammonia in the United States and the first to address more than 3 mg N/L of ammonia.

Palo project timeline. Before 2008, Palo had no community-wide public drinking water system, but instead was supplied drinking water through individual or neighborhood wells. In August 2010, an engineering evaluation of the water supply and treatment options recommended a new water treatment plant that could provide the city both potable water and sufficient fire protection.

As already noted, a year-long pilot study was conducted in Palo to assess the effectiveness of a biological treatment process to treat ammonia and iron in the source water between March 2011


E651


and April 2012. In December 2011, bids for the new well, water main installation, and water tower construction were awarded. The construction of the well, main, and tower were completed by July 2013. Before the completion of the treatment plant, a limited but growing number of connections in the city were supplied with groundwater treated with chlorine. Given the elevated ammonia in the source water, the system was likely a chloraminated system, although only chlorine was added. The bid for the water treat-ment plant construction went out in January 2013, and the Iowa DNR approved the pilot and plant design in April 2013. The water treatment plant was constructed between April 2013 and January 2014.

Water treatment system technology engineering and design specifications. The engineering design features and parameters of the full-scale plant were based on the pilot-study findings (Lytle et al. 2012). Table 2 summarizes the major engineering design and operating parameters of the pilot- and full-scale systems. The city receives water from two wells: the original well (“well 2”), which has a capacity of 320 gpm (1,211 L/min) and the new well devel-oped under the project (“well 3”), which has a capacity of 180 to 200 gpm (681 to 757 L/min). The wells are within approxi-mately 600 yd of each other and have similar water quality. Well 2 was used entirely during the pilot study.

The Palo water treatment plant design is based on a USEPA-patented process to treat elevated levels of ammonia as well as iron in the source water (Patent No. U.S. 8,029,674). The primary

processing units in the treatment train are the “aeration contac-tors,” blowers, and dual media filters. Also included in the treat-ment train are phosphate (blended phosphate was mistakenly purchased; straight orthophosphate is recommended) feed, chlorine (sodium hypochlorite) feed, and sodium hydroxide (caustic) feed. The new drinking water treatment system was designed to serve a population of 1,139 people with an average daily demand of 0.115 mgd, equivalent to 0.44 mil L/day, and a peak daily demand of 0.23 mgd (0.87 mil L/day). Using a system-design flow rate of 300 gpm (1,136 L/min), the treatment plant operates an average of approximately 6.4 h/day and as much as 12.8 h/day during peak demand. Following the two contactors are two finished water pumps with variable fre-quency drives to help control the level in the contactors. The two pumps feed into a common header that dosed the three dual-media pressure filters sized for 100 gpm each. The pressure filters are designed to remove any iron that is oxidized and passed through the contactors and provide additional biological nitrification capacity. The finished water then passes through a monitoring station and into the new 200,000-gal water tower before entering the new distribution system.

Aeration contactor design. The treatment plant uses two cylin-drical aluminum upflow aeration contactors (Figure 3, part A), designated as contactors 1 and 2, configured in parallel. Each contactor is 10.5 ft (3.2 m) in diameter and 10 ft (3.05 m) high, and filled with 45 in. (114 cm) of gravel with effective size of

Gravel-filled column= contactor(C1, C2, C3)

Each contactor contains a different size of media

Water from the source enters the

column from the top

Anthracite over sand= filter (F1, F2, F3)

Downflow

Water leaves pump at base of filler and goes to clear well

FIGURE 2 Schematic of the pilot biological ammonia removal treatment technology system

Pump Pump Pump Pump Pump PumpAir

bubbles

C1 F1 C2 F2 C3 F3

1 32


E652


¼ in. (0.65 cm). Air is delivered to the bottom of the contac-tor at a rate of 1.5 cfm/ft2 (0.40 m) through a network of horizontally slotted plastic distributers using a delta rotary positive-displacement blower unit (Figure 3, part B). Each contactor has a dedicated blower; however, both blowers can direct air flow into one contactor during the backwash pro-cess to increase air, scour the media, and dislodge any trapped sediment or loose biomass.

After phosphate addition, raw water is pumped into the bottom of each contactor at a rate as high as 2.2 gpm/ft2 (5.6 m/h), which results in an empty bed contact time (EBCT) of 12.8 min. However, under normal operation, water is pumped at a rate of 1.7 gpm/ft2 (4.3 m/h), a 16.5-min EBCT, so that water is upflow through the contactor and co-current to the air flow. This configuration differed from the pilot (counter-current operation) but is thought to reduce

any likelihood of flow restrictions or blockages through the contac-tor. Similar pilot studies performed elsewhere in Iowa at the time of design showed no difference in performance between counter- and co-current contactor operations (Lytle et al. 2014). Contactors can be backwashed with raw water, if necessary, at a rate of 15 gpm/ft2 (0.63 m/min) for 7 min. Backwashing will not expand the bed but remove loosely attached particles and biofilm from the gravel. On the basis of volatile gas measurements, specialized ven-tilation above the contactors was not required.

Blower design. Two rotary positive-displacement three-lobe blowers are used to deliver air to the contactors. The blower performance criteria include a maximum blower speed of 2,700 rpm, discharge pressure rise of 4.5 psig (31.03 kPa), and maximum motor of Hp 7.5 (5.6 kW). The blowers deliver 130 scfm at 4.5 psig (3.68 m3/min at 31.03 kPa). Air is pulled

TABLE 2 Engineering and operational design parameters of pilot and full-scale treatment systems

Contactor Design

Parameter Pilot Full Scale

Filter loading rate 2.2 gpm/ft2 (5.6 m/h) 2.2 gpm/ft2 (5.6 m/h)

Air flow rate 2.86 cfm/ft2 (0.87 m/min) 1.5 cfm/ft2 (0.46 m/min)

Backwash conditions:DurationBed expansionFlow rate

5 min0%51 gpm/ft2 (2.2 m/min)

7 min0%15 gpm/ft2 (0.63 m/min)

Contactor depth 30 in. (76 cm) 45 in. (114 cm)

Contactor effective size ¼ in. (0.65 cm) ¼ in. (0.65 cm)

Filter Design


Filter loading rate 2 gpm/ft2 (5 m/h) 2.2 gpm/ft2 (5.6 m/h)

Backwash conditions:DurationBed expansionFlow rate

15 min50%17 gpm/ft2 (0.72 m/min)

15 min50%17 gpm/ft2 (0.72 m/min)

Filter anthracite depth 20 in. (51 cm) 20 in. (51 cm)

Filter anthracite size 0.04 in. (0.1 cm) 0.04 in. (0.1 cm)

Filter sand depth 10 in. (25 cm) 10 in. (25 cm)

Filter sand size 0.018 in. (0.046 cm) 0.018 in. (0.046 cm)

Other Parameter Variables


Hours and days of operation 24 h/day, 7 days/week January to late April 2014: ~4 to 5 h/day, 5 days/week

Late April 2014 to March 2015: ~6 to 7 h/day, 7 days/week

Phosphate feed Orthophosphate: 0.3 mg PO4/L Blended phosphate (75% polyphosphate/25% orthophosphate): 0.3 mg PO4/L orthophosphate portion

Air/water flow configuration Counter-current Co-current

Filter backwash water Nonchlorinated Chlorinated

Filter type Gravity Pressure

PO4—phosphate


E653


from outside of the building and as a result is subject to sea-sonal temperature changes.

Filter design. The treatment plant uses three parallel 100 gpm (378 L/min) dual-media steel pressurized tanks (100 psig) (689.5 kPa) (designated as filters 1, 2, and 3) and operates at a loading rate of 2.2 gpm/ft2 (5.6 m/h; 8.5-min EBCT). This filter design differed from the pilot system in that the pilot filters were operated under gravity feed. Pressurized filters are commonly used all over to remove iron and, although not evaluated for ammonia removal, were not anticipated to be detrimental. The filters were designed to remove iron according to common practice. However, no chlorine was added ahead of the filters, allowing them to become biologically active to pro-vide additional ammonia and nitrite oxidation should the contactors not sufficiently address the ammonia load from the

raw water. Each filter is 6 ft, 6 in. (198 cm) in diameter and 5 ft (152 cm) high, and contains 20 in. (51 cm) of anthracite (0.04 in. [0.1 cm] effective size) over 10 in. (25 cm) of sand (0.018 in. [0.046 cm] effective size) (Figure 3, part D).

The filter backwash design rate is 17 gpm/ft2 (0.72 m/min) to achieve 50% filter bed expansion for 15 min. The frequency of filter backwash was established during full-scale operation to be once every three weeks with one filter backwashed per week. Filters are backwashed with finished chlorinated water (target concentration of 2 mg Cl2/L). The pilot-system filters were back-washed with nonchlorinated water to avoid destroying the nitrify-ing biofilm. The backwash procedure in the full-scale system is being examined to identify whether using chlorinated backwash water has a negative effect on biological activity within the filters. An Ohio research project documented that a full-scale biologically

FIGURE 3 Photographs of aeration contactors (A), aeration contactors and air blowers (B), aerated contactor surface (C), and pressure filters (D)

A B

C D


E654


active nitrifying water treatment plant using chlorinated finished water to backwash the filters showed no negative impact on the nitrifying ability of the filter (Lytle et al. 2007).

Chemical feed and other design features. The treatment plant has three chemical feed systems. Phosphate is fed using a positive displacement chemical metering pump1 ahead of the contactors to achieve a desired orthophosphate concentration of 0.3 mg PO4/L. Phosphate is used to provide the necessary nutrients to the nitrifying biomass as identified in the pilot study (Lytle et al. 2012). A blended phosphate (75%/25% poly/ortho phosphate) chemical is used in the full-scale plant, whereas the pilot system used a straight orthophosphate-feed chemical. The difference was not expected to pose an issue because the target phosphate dose is based on the orthophosphate fraction. A straight orthophos-phate chemical will be used in the future.

Sodium hypochlorite is fed after the filters to achieve a total chlorine residual of 2 mg Cl2/L using a positive displacement chemical metering pump.1 Lastly, sodium hydroxide can be fed ahead of the filters to compensate for potential pH drop in the contactor (as a result of nitrification) to avoid corrosion issues in premise plumbing. The desired pH of the finished water is 8.

Distribution system. Palo’s new distribution system consists of 8 mi of 6- and 8-in.-diameter C900 polyvinyl chloride (PVC) and ductile-iron valves, and has 385 connections. Treated water is fed into the distribution system and transported to a 200,000-gal elevated, spheroid storage tank via an 8-in. C900 PVC water main. The system is pressurized to 65 psi using the two finished water pumps in the water treatment plant.

System startup. Because the potential for elevated nitrite levels to be produced during contactor and filter acclimation periods exists, the system is incrementally started (contactor 2 and filters 1, 2, and 3 were started up initially). Treated water is monitored daily for ammonia, nitrite, and nitrate. When nitrite levels in the contactor effluent exceed 0.3 mg N/L, finished water is dis-charged to the sanitary sewer as a precaution, and chlorinated raw water from the wells is fed into the distribution system. The water treatment plant is drained to waste 4 to 5 h/day, five days a week, while an operator is available to monitor the tower. Once the nitrite levels in the treated water consistently drop below 0.1 mg/L, the treated water is returned to the distribution system and contactor 1 is placed in service.

Water treatment plant operation. The full-scale water treatment plant typically runs 8 to 12 h/day, 5 days/week. Raw water from Palo’s existing well and drinking water was not chlorinated or treated in any way before the installation of the water treatment system. The treated water was released into the city’s new water system and transported to a 200,000-gal (7,571-L) elevated, spheroid storage tank via a PVC water main 6 and 8 in. (15 and 20 cm) in diameter. Treated water and excess filter backwash water were routed to the on-site sanitary sewer.

MATERIALS AND METHODSOn-site analyses. Field operating and basic water quality

measurements were conducted by Palo’s water plant operators. On-site water quality measurements include temperature, dis-solved oxygen, nitrite, free and total chlorine, and pH. Dissolved

oxygen, pH, and temperature were measured using a portable meter2 with a dissolved oxygen probe3 and a pH probe.4 Head loss, flow rates, and other operating treatment system conditions were recorded.

Water quality analyses. Palo staff collected weekly water quality samples, made routine water quality measurements, and shipped them overnight on ice to the USEPA laboratory in Cincinnati, Ohio, for analysis. Water samples consisted of source water, raw water after phosphate addition, raw water after both contactor effluents, raw water after all three filter effluents, finished (treated) water (post-chlorine addition, entry point to the distri-bution system), and water from two distribution-system loca-tions. The following water sample volumes were collected on a weekly basis:

• 250 mL for inorganic analysis (e.g., alkalinity, ammonia, nitrite, nitrate, orthophosphate)

• 60 mL for metals analysis (e.g., iron, calcium, magnesium) • 250 mL for bacteriological analysis (heterotrophic plate counts [HPCs])

• 40 mL for total organic carbon analysisFree and total chlorine and pH of the samples were measured

with the understanding that they are not reliable, reportable numbers. Ammonia was also measured upon sample arrival using HACH DR 2700 spectrophotometer5 and appropriate test kit6 so that results could later be compared with laboratory results. Ammonia, nitrite, and nitrate analyses were typically performed in the laboratory on the same day the cooler arrived (within 24 h after field sampling) to ensure best results. Although not reported here, ammonia and nitrite were also measured in the field by the plant operator as part of the state’s requirements. Field results were generally in agreement with laboratory results when com-pared, indicating that water samples were stabile during ship-ment. All water analyses were performed according to USEPA or Standard Methods (1998).

RESULTSImportant dates. There were a number of operating changes and

other events that occurred over the course of the full-scale evalu-ation period that are worth noting. Palo’s full-scale water treat-ment plant started up on Jan. 15, 2014 (day 0), with the excep-tion of contactor 2, and treated water was pumped to the elevated storage tank and distribution system. Chlorine was added as sodium hypochlorite to achieve a goal of 2 mg Cl2/L of total chlorine residual. Contactor 2 startup was delayed to provide a safety backup in the event excess nitrite was generated while contactor 1 and the three filters transitioned through the bio-logical acclimation phase of operation. Piping to the elevated storage tank froze in late January, so the tank was taken out of service for nearly a month. During this time (day 31), filter 3 was taken off-line to delay nitrification progression in the filter rela-tive to the other two filters so it could be brought back on-line in the event nitrite levels increased in the finished water. The air blower was turned off to contactor 1 on day 35 for a day, in reaction to an increase in nitrite in the plant. The tower and filter 3 were put back in service on day 38. As a precaution, however, nonchlorinated finished water was wasted to the onsite sewer,


E655


and chlorinated raw water was sent directly to the distribution system as it had been before putting the plant into service.

After 98 days of operation, the plant flow rate was decreased from 150 gpm (568 L/min, 16-min EBCT) to 100 gpm (379 L/min, 24-min EBCT) to allow longer contact time through the system for nitrification to occur. At the same time, finished water was pumped into the distribution system. Lastly, contactor 2 was put into service after 139 days of operation, although initially at only 25% of the total flow (75% of the influent water going to contactor 1). The flow to contactor 2 was gradually increased with the eventual goal to split the plant flow evenly between the two contactors.

It is worth emphasizing that there was great awareness by the plant operator, Iowa DNR, and USEPA that nitrite levels could be elevated through the treatment plant during the biological acclimation phases. Nitrite, a regulated contaminant based on acute exposure, was not permitted to enter the distribution system at levels near the regulatory threshold of 1 mg N/L in any case. The plant operator measured nitrite in the finished water daily using a field colorimetric test kit7 that had an “over-concentration range” limit of 0.35 mg N/L. The operator responded to over-range readings by taking action to avoid sending “water of potential concern” to the distribution system. The study’s report-ing period ended after 406 days of operation (Feb. 25, 2015).

General water chemistry. The water quality analyses of the source water, contactor, and filter effluents; finished water; and distribution system water are summarized in Table 3. Palo can draw from two wells: the original (well 2) was solely used in the pilot study, and the new well (well 3), which is within several hundred yards of well 2, was used as the source at the start of the full-scale plant operation. However, shortly after system startup, the original well became the primary source of the city’s water supply as the full-scale plant went into operation. The source water was a relatively hard (326 mg CaCO3/L), high-alkalinity (355 mg CaCO3/L) groundwater with calcium and magnesium levels averaging 76 and 33 mg/L, respectively. The pH was 7.86, and iron levels averaged 0.12 mg/L, although the average was affected by a few random high concentrations measured early in the study as also reflected by the relatively large standard devia-tion. The pilot study and full-scale water quality measurements were nearly identical with the exception of iron. During the pilot study, the source water averaged 0.63 mg/L of iron (Lytle et al. 2012) (the discrepancy and possible implications will be discussed later), and the ammonia concentration averaged 3.3 mg N/L. Sulfate and chloride averaged 90 mg SO4/L (derived from sulfer analyses) and 5 mg/L, respectively. Orthophosphate was very low, averaging 0.026 mg PO4/L, and total phosphorus was at the detec-tion limit of 0.015 mg P/L. Manganese, nitrite, and nitrate were at or near the respective method detection limits, strontium aver-aged 1.0 mg/L, and total organic carbon averaged 1.1 mg C/L.

Removal of ammonia through aeration contactors. The first indica-tion of ammonia oxidation in contactor 1 was observed at 28 days of operation when a small amount of nitrite (0.1 mg N/L) and detectable nitrate were observed in the contactor effluent (Figure 4, part A). Over the next five weeks of operation, nitrite levels were no more than 0.2 mg N/L. By 35 days into operation,

nitrate in the contactor effluent became more evident (Figure 4, part A). Nitrate in the contactor effluent gradually increased between 35 days and 130 days to 1 mg N/L. During the same time, nitrite levels remained very low (typically less than 0.1 mg N/L). Between 139 days and 150 days of operation, ammonia increased rapidly back up to 2.5 mg N/L.

During the same period, contactor 2 was put into service (138 days into operation). Whether the events were linked is unknown, but the disruption in operation as a result of changing flow pat-terns could be important. Two significant changes occurred in contactor 1 when contactor 2 went on-line: (1) the flow rate dropped from 100 gpm (378 L/min) to 67 gpm (254 L/min), and (2) the air feed rate was reduced by 50% (changes will be detailed later). Contactor 1 recovered relatively quickly over the next 50 days with ammonia levels dropping below 1 mg N/L at the time of reporting. During this time, nitrite levels remained very low.

Although not in operation for the 138 days, the media in con-tactor 2 always remained submerged in raw water. Unexpectedly, ammonia levels leaving contactor 2 were very low from the startup date and ranged between 0.25 and 1 mg N/L (Figure 4, part B). Measurable levels of nitrite as high as 0.3 mg N/L were observed through the 60-day reporting period. Exactly why an acclimation period did not occur is unknown, but it is conceivable that nitrifying bacteria were able to colonize and acclimate during this period while contactor 2 sat idle. In addition, contactor 2 consistently performed better than contactor 1 from the startup date. The reason might be that contactor 2 had been operated at 1∕3 the flow rate (or 33 gpm [124.74 L/min]) of contactor 1 to ease it into operation. The flow rate of contactor 2 has gradually been increased over time and will eventually be equal to that of contactor 1, or 50 gpm (189 L/min).

Nitrogen balances comparing source water ammonia levels (primary nitrogen source representing total nitrogen in source water) and contactor effluent total nitrogen (sum of ammonia, nitrite, and nitrate) as a data-quality check are graphically com-pared in Figure 4, parts A and B. The closeness of the values (bar compared with raw) reflects the good sampling protocol as well as high quality and reliability of the laboratory data.

The expectation, based on the pilot study, has been that the contactors would eventually oxidize all of the ammonia in the source water to nitrate. After 231 days of operation, the contac-tors were consistently reducing ammonia levels to below 1 mg N/L to as low as 0.25 mg N/L. The acclimation period of the contac-tors has progressed slowly as expected on the basis of past work. Unfortunately, the pilot study conducted in Palo was not useful in projecting more precisely how long acclimation would take. The pilot system operated for more than 360 days and during that time, system acclimation was seriously delayed because of a lack of phosphate (identified as a necessary nutri-ent during the pilot) and oxygen delivery issues (Lytle et al. 2012). Once optimized, the system rapidly responded and performed as expected.

The USEPA pilot system was moved to a neighboring Iowa community with a source water nearly identical to that of Palo. The pilot system was operated under the optimal conditions identified during the Palo pilot study (Lytle et al. 2014). The new


E656


pilot system operated 24 h/day and 7 days/week. The new pilot contactors and filters acclimated (complete oxidation of ammonia to nitrate) within 50 days of operation. Other USEPA work has indicated that the acclimation time for a biologically active nitri-fying contactor or filter is linearly dependent on the number of hours of operation, provided its operation is at a similar loading

rate. For example, a filter that operates 12 h/day will take twice as long to acclimate as compared with one operating 24 h/day. The Palo full-scale system was operated only approximately 4 to 5 h/day during weekdays, which suggests that it will take more than 4.8 to 6 times (assuming no operation on weekends) longer to acclimate than the new pilot study suggested (240 to 300 days).

TABLE 3 Extensive water quality analyses of the source water, contactor, and filter effluents and finished waters over the Palo, Iowa, treatment plant study period

Analyte

Detection Limitmg/L Raw Raw PO4 Contactor 1 Contactor 2a Filter 1 Filter 2 Filter 3 Finished

Ca 0.01 79 ± 1.7(13)

79 ± 2(13)

79 ± 2(13)

— 78 ± 2(12)

78 ± 2(12)

78 ± 2(11)

78 ± 2(11)

Cl 5 5 ± 0(25)

5 ± 0(26)

5 ± 0(26)

5 ± 0(6)

11 ± 27(24)

10 ± 26(24)

9.2 ± 21(24)

5.4 ± 0.9(23)

Fe 0.001 0.15 ± 0.22(13)

0.06 ± 0.02(13)

0.03 ± 0.02(13)

— 0.008 ± 0.01(12)

0.006 ± 0.005(12)

0.005 ± 0.006(11)

0.005 ± 0.003(11)

K 0.3 5.0 ± 0.18(13)

5.0 ± 0.17(13)

5.1 ± 0.3(13)

— 5.3 ± 0.35(12)

5.2 ± 0.37(12)

5.3 ± 0.36(11)

5.1 ± 0.25(11)

Mg 0.005 34 ± 1.1(13)

34 ± 1.3(13)

34 ± 1.3(13)

— 34 ± 1.2(12)

34 ± 1.1(12)

34 ± 1.4(11)

34 ± 1.5(11)

Mn 0.001 0.002 ± 0.001(13)

0.002 ± 0.0007(13)

0.004 ± 0.001(13)

— 0.004 ± 0.003(12)

0.003 ± 0.002(12)

0.003 ± 0.002(11)

0.004 ± 0.003(11)

Na 0.03 32 ± 1.1(13)

32 ± 1.1(13)

33 ± 2(13)

— 50 ± 55(12)

49 ± 52(12)

49 ± 48(11)

34 ± 2.1(11)

t-NH3 0.03—mg N/L

3.2 ± 0.1(22)

3.2 ± 0.1(22)

2.6 ± 0.55(22)

0.57 ± 0.27(3)

1.6 ± 1.1(22)

1.5 ± 1.1(22)

1.5 ± 1(20)

1.6 ± 1.0(20)

NO2 0.01—mg N/L

0.01 ± 0.002(23)

0.01 ± 0.0039(23)

0.11 ± 0.08(23)

0.22 ± 0.04(3)

0.2 ± 0.2(22)

0.18 ± 0.15(22)

0.11 ± 0.09(21)

0.14 ± 0.15(21)

NO3 0.02—mg N/L

0.02 ± 0.001(23)

0.024 ± 0.018(23)

0.49 ± 0.41(22)

2.4 ± 0.3(3)

1.4 ± 1.2(21)

1.4 ± 1.2(21)

1.5 ± 1.1(20)

1.3 ± 1.1(20)

o-PO4 0.025—mg PO4/L

0.026 ± 0.004(15)

0.37 ± 0.43(15)

0.45 ± 0.36(15)

— 0.45 ± 0.45(14)

0.46 ± 0.49(14)

0.47 ± 0.46(13)

0.48 ± 0.49(13)

t-PO4a 0.015—mg

PO4/L0.015 ± 0.0007

(13)

0.51 ± 0.66(13)

0.55 ± 0.44(13)

— 0.56 ± 0.51(12)

0.57 ± 0.54(12)

0.6 ± 0.52(11)

0.59 ± 0.56(11)

Sr 0.001 1.1 ± 0.03(13)

1.1 ± 0.03(13)

1.1 ± 0.03(13)

— 1.1 ± 0.03(12)

1.1 ± 0.02(12)

1.1 ± 0.03(11)

1.1 ± 0.03(11)

S 0.003 30 ± 0.6(13)

30 ± 0.6(13)

31 ± 0.6(13)

— 31 ± 0.95(12)

31 ± 0.85(12)

31 ± 0.8(11)

30 ± 0.7(11)

SO4a 0.009—mg

SO4/L90 ± 1.8

(13)90 ± 1.8

(13)93 ± 1.8

(13)— 93 ± 2.85

(12)93 ± 2.55

(12)93 ± 2.4

(11)90 ± 2.1

(11)

Total alkalinity 1—mg CaCO3/L

355 ± 1.6(23)

352 ± 9.7(23)

348 ± 4.9(23)

332 ± 1.2(3)

344 ± 16(22)

344 ± 15(22)

341 ± 7.6(21)

350 ± 26(21)

TOC 0.1—mg C/L 1.1 ± 0.14(20)

1.0 ± 0.05(19)

1.0 ± 0.07(20)

1.1(1)

1.1 ± 0.12(17)

1.1 ± 0.18(18)

1.0 ± 0.07(17)

1.21 ± 0.69(17)

Heterotrophic plate count

1—cfu/mL 4,634 ± 7,466 (7)

1,571 ± 1,889 (7)

7,259 ± 3,788 (7)

10,450 ± 1,998(3)

10,923 ± 8,837

(7)

9,029 ± 5,107 (7)

13,217 ± 8,305(6)

5,910 ± 5,551 (6)

pH Units — 7.86 ± 0.1(2)

7.83 ± 0.2 (122)

7.87 ± 0.1 (57) 7.76 ± 0.2 (121)

7.75 ± 0.1 (121)

7.80 ± 01 (121)

7.78 ± 0.1 (121)

Temperature— °C

11.4 ± 0.1(2)

11.4 ± 0.2(5)

12.1 ± 0.85 (122)

13.8 ± 0.4 (57) 13.2 ± 1.1 (121)

13.4 ± 1.1 (121)

12.8 ± 1.0(121)

13.3 ± 0.8 (121)

Dissolved oxygen

mg O2/L — 9.10 ± 0.2(2)

8.96 ± 0.5 (122)

8.88 ± .2 (57) 7.53 ± 0.9 (121)

7.78 ± 0.6 (121)

7.42 ± 0.1 (121)

5.35 ± 1.8 (121)

C—carbon, Ca—calcium, CaCO3—calcium carbonate, Cl—chlorine, Fe—iron, K—potassium, Mg—magnesium, Mn—manganese, N—nitrogen, Na—sodium, NO2—nitrite, NO3—nitrate, O2—oxygen, o-PO4—orthophosphate, PO4—phosphate, S—sulfur, SO4—sulfate, Sr—strontium, t-NH3—total ammonia, TOC—total organic carbon, t-PO4—total phosphate

aDerived from P or S analyses

Average ± standard deviation; number of samples in parentheses; Dash = not analyzed


E657


After late April, the Palo system usage rate was increased to 6 to 7 h/day over the entire week, which should decrease the projec-tion time. On the basis of this discussion, the slow acclimation period was a function of operation time.

Some other considerations regarding the contactor engineering design and operation are worth noting. The contactors do not have individual water flow meters to monitor and regulate flow rate through them. Water flow through the contactors is con-trolled manually via gate valves that feed the contactors raw water following phosphate addition. Flow regulation could be improved by adding flow meters to each contactor to more accu-rately maintain equal flow through each. Second, although the air-flow rate to the contactors was based on the pilot study, the bubbles were much more finely and evenly dispersed through the contactor than in the pilot system. The air delivery dynamics could possibly affect biofilm attachment and biological growth mechanisms as well as beneficial nitrification. Unfortunately, the blower design does not incorporate controls to adjust air delivery rate. However, when contactor 2 was placed on-line, the decision was made to split air flow from one blower rather than add the second blower.

With regard to water quality, the iron levels in the source water were very low and much lower than during the pilot study despite using the same source water. This difference was thought to be related to the large difference in pumping rates between the pilot and full-scale systems. The pilot system was operated at a rate of 500 mL/min (as compared with 100 to 150 gpm [379 to 568 L/min] in the full-scale plant) and, at the time the pilot study was conducted, there was very little demand on the well since the distribution system was

not in place. Iron deposition onto the gravel media may provide some protection to the biofilm from shearing by the bubbles and improve biofilm attachment. White et al. (2012) examined biofilm on filter media from an iron and ammonia (<1.4 mg N/L) removal plant in Ohio. They observed that the biofilm was largely present under an inorganic (largely iron) layer that covered the anthracite and sand media. Lastly, phosphate was injected ahead of the contactors in the form of a blended phosphate (75:25 poly-/ortho-phosphate) to achieve a concentration goal of 0.3 mg PO4/L. This differed from the pilot study in which a straight orthophosphate was used. Only orthophosphate measurements were made at the plant and were used solely for process control as recommended. There is no evidence to suggest that the polymeric portion of chemical for-mulation (not identified in orthophosphate measurement) would affect nitrifying biofilm. In the future, the blended phosphate will be replaced with a straight orthophosphate.

Removal of ammonia through filters. The purpose of the dual-media filters that followed the contactors was to remove iron particles that developed in the contactors. The filters were also biologically active and provided additional ability to oxidize ammonia and nitrite that passed through the contactor. Ammo-nia, nitrite, and nitrate levels entering the filters were either equal to the level leaving contactor 1 or the combination of contactor 1 and 2 effluents (after 138 days).

In general, the patterns and progression of nitrification in the filters paralleled the contactors (Figure 5, parts A, B, and C). Addi-tional oxidation of the ammonia and nitrite was observed in the filters. Ammonia levels in all filters gradually decreased with time up to approximately 138 days. After 138 days, ammonia levels in

t-NH3NO3NO2Raw t-NH3

A4

3

2

1

0

Co

nce

ntr

atio

n—

mg

N/L

4

3

2

1

00 100 200 300 400

Elapsed Days0 100 200 300 400

Elapsed Days

Co

nce

ntr

atio

n—

mg

N/L

B

FIGURE 4 Nitrogen balance in contactor 1 effluent (A) and in contactor 2 effluent (B)

N—nitrogen, NO2—nitrite, NO3—nitrate, t-NH3—total ammonia


E658


all filters dropped to very low levels for the remainder of the data-collection period. On a few rare occasions, nitrite reached as high as 0.7 mg N/L in a single filter effluent but typically was very low. The filters improved overall water quality by removing any excess ammonia or nitrite from the contactor effluent. The finished-water ammonia, nitrite, and nitrate values paralleled the filters and reflected the combination of all three (Figure 5, part D).

As a data-quality check, nitrogen between source water ammonia levels (primary nitrogen source representing total nitrogen in source water) and filter-effluent total nitrogen (sum

of ammonia, nitrite, and nitrate) are graphically compared in Figure 5, parts A, B, and C. The closeness of the values reflects good sampling protocol as well as high quality and reliability of the laboratory data.

Some additional observations regarding filter operation are important to highlight. The filters are pressure filters, which are different from the gravity filters used in the pilot study. Pressure filters are obviously not detrimental to the biological nitrification process, determined by the effectiveness of the full-scale filters to oxidize ammonia and nitrite in this study. Palo’s treatment-plant

5

4

3

2

1

0

5

4

3

2

1

00 100 200 300 400 0 100 200 300 400

0 100 200 300 400

5

4

3

2

1

0

5

4

3

2

1

00 100 200 300 400

FIGURE 5 Nitrogen balance in filter 1 effluent (A), in filter 2 effluent (B), in filter 3 effluent (C), and in finished water (D)

Co

nce

ntr

atio

n—

mg

N/L

Co

nce

ntr

atio

n—

mg

N/L

Co

nce

ntr

atio

n—

mg

N/L

Co

nce

ntr

atio

n—

mg

N/L

A B

C D

Elapsed Days Elapsed Days

Elapsed Days Elapsed Days




E659


operator noted that very little head loss built up across the pres-sure filters. Since iron was very low, biomass was likely the major contributor to head loss, but apparently it was not significant enough to require frequent backwashing.

The filters were backwashed once every three weeks on a stag-gered schedule (i.e., one filter was backwashed per week) solely on the basis of time in service. A trend developed in which one filter a week had slightly higher nitrite levels than the others, but not high enough to raise concerns. Upon examination, it was discovered that the elevated nitrite filter effluent was always the last one backwashed. The observation suggests that chlorinated backwashing might have an impact on filter operation and bio-logical activity. As described in the contactor discussion in this article, iron deposition onto the filter media may provide some support to the nitrifying biofilm by improving biofilm attachment and protection against biofilm contact with chlorine during backwashing with chlorinated water.

Removal of iron in source water. As already noted, iron levels in the full-scale operation were much lower than during the pilot study (0.82 mg/L). Iron levels averaged 0.12 mg/L, although the average was affected by a few random elevated concentrations that were measured early in the study. The initial oxidation state of iron in the raw water was not determined, but it is reasonable to assume that iron was in the reduced Fe(II) form on the basis of water chemistry, dissolved oxygen, and local geology. The oxygen level in the contactors and the pH of the source water led to rapid oxidation of Fe(II) to Fe(III) (iron particles). Any iron particles should easily and reliably be removed by the pres-sure filters.

Iron levels in the filter effluents were well below the second-ary drinking water standard of 0.3 mg/L and often at the method detection limit of 0.001 mg/L. Average iron concentra-tions of the effluent from filters 1, 2, and 3 were 0.006, 0.005, and 0.004 mg/L, respectively. Finished-water iron levels aver-aged 0.004 mg/L.

Other water quality parameters. Other important water quality parameters were monitored during the study. Temperature is important from the perspective of microbial activity. Laboratory studies, for example, have demonstrated that the growth rate of nitrifying bacteria is negatively affected by temperatures below 10°C (Andersson et al. 2001). Although the air temperature in Palo can be very cold in the winter, the groundwater source dropped only to a low of 11°C (Figure 6, part A) during the winter. As the seasons changed from spring to summer, a gradual increase in temperature was observed to 14°C. The blowers used outdoor air to supply air to the contactors, and there was an initial question of whether cold outdoor air during winter months could decrease water temperature in the contactors and negatively affect biological activity. System performance indicated that air temperature changes did not affect the treatment system through the 2014 and 2015 winter seasons.

Total and free chlorine in the finished water were measured regularly on site by plant staff (Figure 6, part B). Total chlorine levels fluctuated around the 2-mg/L target and ranged between 1.3 to 2.3 mg/L over the first five weeks of operation. As expected, no free chlorine was present in the finished water because of the

high concentration of ammonia. Chlorine feed was discontinued at day 37 when the plant-finished water was sent to waste. Chlo-rine feed started up again on day 97 when the system-finished water was sent back to the distribution system. Between 97 and 150 days, total chlorine ranged between 0.9 and 1.7 mg Cl2/L, and free chlorine between 0.1 and 1 mg Cl2/L. A sharp drop of total chlorine occurred on day 150, and the operator reported difficulty keeping a residual over the next two weeks. The time frame corresponded to when ammonia was rapidly shifting to being completely oxidized through the plant. After 175 days, total chlorine was nearly identical to free chlorine. The absence of combined chlorine corresponded to the sharp disappearance of ammonia during the same time.

The pH was measured throughout the treatment plant over the course of the study. The biological nitrification process produces H+ (Eq 3) ions, which can decrease the pH of the water. Aeration can also increase or decrease the pH of water depending on the initial pH because of CO2 equilibria between the atmosphere and water. Very little pH variability occurred in the water through the plant or with time. Sodium hydroxide feed was on-line during the first several weeks of operation resulting in a finished water pH near 9. Because of feed-line blockage and clogging issues, likely associated with trapped air bubbles, the feed was taken off-line. The pH through the plant averaged 7.86 across all loca-tions sampled (Table 3). The absence of variability or influence of nitrification on pH through the plant was most likely because of the well-buffered, high-alkalinity source water.

Orthophosphate was identified in the pilot study as an essential nutrient necessary to the biological nitrification process in Palo’s water (Lytle et al. 2012). A target dose of 0.3 mg PO4/L was established on the basis of the pilot study. Field orthophosphate values varied over time and tended to be relatively high over the course of the study. Laboratory orthophosphate measurements were similar to field measurements, although laboratory data were still not completely reported at the time of document prep-aration. Average orthophosphate and total phosphate measure-ments of the water samples collected after the phosphate addition averaged 0.37 and 0.51 mg PO4/L (Table 3), respectively. Ortho-phosphate made up the majority of phosphate added although not in the ratio of the chemical formulation. Polyphosphate does degrade and will break down with time to orthophosphate, which may explain the apparent difference. Average ortho- and total phosphate concentrations were slightly greater in the contactor and filter effluents than in the raw water after phosphate addi-tion, which could be related to mixing issues. The ratio of total phosphate to orthophosphate, however, remained approximately the same (Table 3).

Oxygen is also a key parameter in the nitrification process, in which 4.6 mg O2/L is necessary to oxidize 1 mg N/L of ammonia to nitrate. Further, there is also a connection between oxygen levels and kinetic requirements associated with molecular diffu-sion. Providing adequate oxygen (to near-saturated oxygen levels) through the contactor, as well as phosphate, was enough for the contactor alone to achieve the desired ammonia reduction at a typical filter loading rate. Oxygen levels ranged between 8 and 8.5 mg O2/L over the first 120 days of operation (Figure 6, part C).


E660


Winterup to

3/20/2014

Springup to

6/21/2014

Springup to

9/23/2014

Autumnup to

12/21/2014

Winterup to

3/20/2015

16

14

12

10

8

6

4

2

00 100 200 300 400

3.0

2.5

2.0

1.5

1.0

0.5

0.0

300 350 400

0 50 100 150 200 250 300 350 400

420

400

380

360

340

320

0 50 100 150 200 250 300 350 400

10

8

6

4

2

0

0 50 100 150 200 250

Elapsed Days

Ch

lori

ne—

mg

Cl 2

/L

RawRaw PO4Contactor 1Contactor 2

Filter 1Filter 2Filter 3Finished

Tem

per

atu

re—

ºC

Tota

l Alk

alin

ity

Co

nce

ntr

atio

n—

mg

CaC

O3/

L

Dis

solv

ed O

xyg

en—

mg

O2 /L

Contactor 1Contactor 2Filter 1Finished

Elapsed Days

RawContactor 1Filter 1Finished

Elapsed DaysElapsed Days

CaCO3—calcium carbonate, Cl2—chlorine, O2—oxygen, PO4—phosphate

TotalFree

FIGURE 6 On-site temperature measurements throughout treatment plant (A), total and free chlorine in finished water (B), dissolved oxygen levels

throughout treatment (C), and total alkalinity throughout (D)

A B

C D


E661


After 120 days, oxygen levels in the contactors were less vari-able and decreased from 9.3 to 8.5 with time, and there were no differences between contactors. The decrease corresponded to the small increase in water temperature over the same period, which suggests that the drop in oxygen reflects the change in oxygen solubility in water. Oxygen levels in the effluents of contactors 1 and 2 averaged 9.0 and 8.9 mg O2/L, respectively (Table 3). Oxygen levels in the filter effluents were very similar and slightly lower than in contactor effluents (only shown for filter 1 in Figure 6, part C), which reflects active biological nitrification activity. Oxygen levels in the effluents of filters 1, 2, and 3 averaged 7.5, 7.8, and 7.4 mg O2/L, respectively, over the course of the study (Table 3). Finished-oxygen levels interest-ingly were very close to the filter 1 effluent values up to approx-imately 130 days. Between 130 days and 150 days, oxygen levels in the finished water location dropped to 2 mg O2/L. This was surprising because the three filter effluents that make up the finished water sample were at approximately 7 mg O2/L. This was also the period during which nitrification through the plant rapidly became established. The most reasonable explana-tion for the observation was that biofilm growth and nitrifica-tion occurred in long copper lines from the filters to the sample tap in the laboratory. After day 150, the operator left the tap open for extended periods before sampling.

Alkalinity in the source water averaged 355 mg CaCO3/L. Aver-age alkalinity in contactors 1 and 2 was 341 and 334 mg CaCO3/L,

respectively (Table 3). Alkalinity change is directly related to nitrification, and therefore closer examination of alkalinity trends are worthwhile (Figure 6, part D). At the beginning of the study, the alkalinity level in the raw water and each contactor was very similar. However, with time, as nitrification progressed through the plant, alkalinity levels decreased. Once acclimation through the plant was achieved, the alkalinity level in the finished water was approximately 24 mg CaCO3/L lower than in the raw water. This decrease is in proximity to the theoretically predicted drop of 7.1 mg CaCO3/L per 1 mg N/L of ammonia oxidized.

Assessment of bacterial population based on HPCs. Health effects are not directly associated with HPCs, but HPC analyses are used to measure the variety of bacteria that are common in water. In general, the lower the concentration of bacteria in drinking water, the better quality of the distribution water. USEPA has set a rec-ommended limit for HPCs of 500 cfu/mL for potable water.

Except for one measurement of 22,000 cfu/mL, the HPC levels of the source water samples were less than 6,000 cfu/mL, with several measurements less than 500 cfu/mL (Figure 7). HPC values varied from week to week with no apparent trend associated with time. Phosphate addition did not have an obvi-ous impact on the magnitude of HPC values. Contactor 1 effluent HPC values were much greater than for the raw water levels, ranging between 4,000 and 15,000 cfu/mL and averag-ing 7,300 cfu/mL (not shown). An obvious HPC correlation with time was not found that might correspond to contactor acclimation or development of biomass in the contactor. Contac-tor 2 HPCs were typically higher than in contactor 1 and aver-aged 10,500 cfu/mL, although the data were based on only three data-sampling events because of the delayed start (Figure 7). Filter effluent values were the highest and reached up to 25,000 cfu/mL. The HPC levels started low but increased quickly at approximately 50 days. Values tended to decrease with time from 10,000 to 15,000 cfu/mL (filter 2, shown in Figure 7). Finished- and distribution-system HPC levels were low (<500) provided total and free chlorine levels were significant. How-ever, when chlorine was low, HPC levels increased above the recommended limit and thus stressed the importance of meeting target chlorine goals.

Distribution system water quality. Water samples from two dif-ferent distribution system locations were collected regularly on a rotating basis and analyzed for HPC, ammonia, nitrite, and nitrate. HPC measurements at one of the locations have already been discussed. Ammonia levels at distribution site 1 reflected finished water ammonia levels and the progression of nitrification in the treatment plant (Figure 8, part A). Nitrite levels were very low and in most cases below the analytical detection limit of 0.01 mg N/L. No ammonia was present in the water after 150 days (all nitrogen was in the nitrate form).

A similar pattern was observed in water collected from distri-bution site 2 (Figure 8, part B). The only discrepancy was that complete nitrification occurred by 75 days, which was much quicker than at the plant and site 1. The observation suggests that nitrification could have occurred in the plumbing associated with the sampling location. A broader investigation is necessary to properly interpret the observation.

10,000

1,000

100

10

0

Het

ero

tro

ph

ic P

late

Co

un

t—cf

u/m

L

100 200 300 400

Raw waterContactor 2Filter 2Finished waterDistribution site

FIGURE 7 Heterotrophic plate count measurements through the

treatment plant and at one distribution system location

Elapsed Days


E662


The successful operation of the full-scale plant resulted in the decrease and eventual disappearance of ammonia in the distribution system. During plant startup and acclimation periods, nitrite was never a concern in the distribution system water samples.

Monitoring requirements. The Iowa DNR was actively involved in all phases of the Palo project. The DNR was updated through-out the pilot study and operation of the full-scale plant. The dis-cussions between USEPA and the DNR were helpful in developing

appropriate sampling requirements of the community once the full-scale system was brought on-line.

Given the nature of the biological treatment system in Palo, the Iowa DNR required additional routine water quality monitoring beyond federal drinking water regulatory requirements (Table 4). The requirements were deemed necessary because of the newness of the treatment process, a lack of full-scale long-term operating experience to draw from, and the possibility that elevated nitrite could be generated through the plant or in the distribution system

FIGURE 8 Nitrogen balance at distribution site 1 (A) and at distribution site 2 (B)

4

A B

3

2

1

0

Co

nce

ntr

atio

n—

mg

N/L

0 100 200 300 400Elapsed Days

4

3

2

1

0

Co

nce

ntr

atio

n—

mg

N/L

0 100 200 300 400Elapsed Days



TABLE 4 Water plant and distribution system monitoring requirements for Palo, Iowa, set by the Iowa DNR

Certified Laboratory Reporting

Location Parameter Frequency

Distribution systema Nitrite Monthly

Entry into distribution system (finished water) Nitrite Monthly

Raw water sample from each well Nitrite Monthly

Self-Monitoring

Location Parameter Frequency

Entry into distribution system (finished water) Ammonia Weekly

Entry into distribution system (finished water) Dissolved oxygen Daily

Entry into distribution system (finished water) Nitrite Daily

Distribution systema Nitrite Daily

Distribution systema Ammonia Weekly

DNR—Department of Natural Resources

aDistribution system sample rotated among several locations pre-identified by Iowa DNR field staff


E663


in the event of an unexpected failure. Specifically, the DNR requires the City of Palo to take a water sample for nitrite analy-sis from the raw sources (two wells), the entry point to the distri-bution system (finished water), and one location in the distribu-tion system on a monthly basis. Distribution sample locations are rotated among several sites pre-identified by Iowa DNR field staff. These samples must be analyzed by a certified laboratory and reported to the DNR.

The DNR also requires the city to self-monitor water quality regularly to ensure that the treatment system is operating properly (Table 4). Specifically, ammonia is monitored from the entry point to the distribution system (finished water) on a monthly basis. Dis-solved oxygen and nitrite are monitored at the same location on a daily basis. The DNR set a dissolved-oxygen target goal of 7 mg O2/L as an indicator of good treatment process control and a nitrite standard of 1 mg N/L in the finished water. Lastly, ammonia is monitored weekly and nitrite is monitored on a daily basis at various locations in the distribution system. As with reported monitoring requirements, distribution-sample locations are rotated among several sites pre-identified by Iowa DNR field staff.

The major concern of the project and sample-monitoring driver is nitrite generation. Figure 9 highlights nitrite levels at all sam-pling locations over the course of the study. With the exception of six nitrite measurements in the range of 0.5 to 0.7 mg N/L, all nitrite level measurements were less than 0.5 mg N/L.

Other. It is worth acknowledging the relative ease of operation and reliability of the water treatment plant. The city of Palo is a small water utility, and small drinking water treatment systems must be relatively simple, reliable, and effective. Operators of small municipal treatment systems are typically responsible for

many aspects of municipal operations. For example, Palo’s oper-ator was responsible for snow plowing, landscaping, and mowing of city grounds, and many other duties as assigned, while simul-taneously being responsible for operating the new water treat-ment plant. Aside from some common chemical feed startup problems, the plant ran largely without issue. Although busy with regular water sampling and monitoring throughout the plant, the operator did not report difficulties with running the plant.

CONCLUSIONSThe city of Palo developed and constructed its entire drinking

water treatment plant and distribution system over a period of approximately 3.5 years. The water treatment plant incorporated an innovative biological treatment system developed by USEPA to treat elevated ammonia (>3 mg N/L) and iron in the ground-water supplies. The biological treatment process was successfully demonstrated at the pilot-scale study during a year-long evalua-tion period, and the performance of the full-scale plant over the first 406 days of operation was evaluated. As a result of this effort, the following conclusions were drawn:

• The new and innovative biological drinking water treatment process to treat high ammonia levels was successfully scaled up and demonstrated at full scale.

• The biological ammonia treatment plant is robust, reliable, and relatively simple to operate.

• The biological ammonia treatment system successfully met the primary treatment objective of removing essentially all of the ammonia in the source water. Negligible concentra-tions of ammonia entered the distribution system once the filters were acclimated, thereby eliminating the possibility of nitrification in the distribution system.

• Iron in the source water was also successfully removed through the treatment plant, although the iron levels in the source water were much lower than in the pilot study, likely due to differences in well demand.

• By 231 days into the full-scale system study, the contac-tors were consistently reducing ammonia levels to between 0.25 and 1 mg N/L. The expectation is that eventually ammonia concentrations leaving the contactors will be negligible. Because the treatment plant operates only 8 to 12 h/day, 5 days/week, it is reasonable to assume that the contactors will continue to be in the biofilm acclimation process for many more weeks. The filters provide addi-tional ammonia oxidation and after 138 days, the filter effluent ammonia levels were negligible.

• Incorporating blended phosphate feed, co-current air/water flow in contactors, and pressure filters of the full-scale plant did not have a detrimental impact on plant operation. Filters were backwashed with chlorinated water, producing a rela-tively short-term, minor degradation effect on filter perfor-mance after being put back in service. By staggering filter backwash events and minimizing frequency, overall finished water quality was not notably affected.

• The success of developing a new biological treatment approach at a small water system was largely a result of a successful pilot study as well as regular and meaningful

RawContactor 1Filter 1Filter 2Filter 3

FIGURE 9 Nitrite concentration through treatment plant and

distribution system

Co

nce

ntr

atio

n—

mg

N/L

Elapsed Days

FinishedDistribution site 1Distribution site 2Contactor 2

1.0

0.8

0.6

0.4

0.2

0.00 50 100 150 200 250 300 350 400


E664


communication between USEPA, the State of Iowa DNR, the community, and engineering consultant during all phases of project development.

ACKNOWLEDGMENTThe authors thank Michelle Latham, Tom Sorg, and John

Olszewski of USEPA for technical and quality-assurance reviews. The authors also thank Andrew Roberts, Amber Seaman, Scott Lippett, Zach Radabaugh, and Colin White of the University of Cincinnati (Ohio) for water analyses and report preparation, and Roberta Campbell of USEPA for report review. The authors thank Jennifer Bunton of the Iowa DNR for initiating the project and providing valuable suggestions, and the city of Palo for support-ing the work. Lastly, the authors thank Brenda Groskinsky of USEPA Region 7 for project support and suggestions.

DISCLAIMERThe US Environmental Protection Agency (USEPA), through

its Office of Research and Development, funded and managed, or partially funded and collaborated in, the research described here. It has been subjected to USEPA’s administrative review and has been approved for external publication. The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of USEPA.

ENDNOTES1LMI (Milton Roy) Pulsafeeder Series2HQ40d, Hach Co., Loveland, Colo.3LD101, Hach Co.4PHC281, Hach Co.5DR 2700, Hach Co.6Method 8155, ammonia salicylate method, Hach Co.7DR 900, Hach Co.

ABOUT THE AUTHORSDarren A. Lytle (to whom correspondence may be addressed) is an environmental engineer for the Water Supply and Water Resources Division of the US Environmental Protection Agency’s (USEPA’s) National Risk Management Research Laboratory, 26 W. Martin Luther King Dr., Cincinnati, OH 45268 USA; [email protected]. He is also

acting branch chief for the Treatment Technology Evaluation Branch. Since beginning work at USEPA in 1991, Lytle’s primary goal has been to research the quality of drinking water. Over the years he has investigated and published works on drinking water systems, focusing on distribution-system corrosion control and water quality; filtration (emphasis on removal of microbial pathogens from water); USEPA Contaminant Candidate List contaminant removal studies; and iron and arsenic removal. Lytle holds a BS degree in civil engineering from the University of Akron (Ohio), an MS degree in environmental engineering from the University of Cincinnati (Ohio), and a PhD in environmental engineering from the University of Illinois. Dan Williams and Christy Muhlen are engineering technicians with USEPA. Maily Pham and Keith Kelty are chemists with USEPA. Matthew Wildman is an

engineer and project manager at HR Green Inc. in Cedar Rapids, Iowa. Glenn Lang is maintenance superintendent for the city of Palo, Iowa. Mitch Wilcox is a technical writer for Pegasus Technical Services Inc. in Cincinnati, Ohio. Melissa Kohne is an engineering student at the University of Cincinnati.

PEER REVIEWDate of submission: 05/18/2015Date of acceptance: 08/17/2015

REFERENCESAndersson, A.; Laurent, P.; Kihn, A.; Prévost, M.; & Servais, P., 2001. Impact of

Temperature on Nitrification in Biological Activated Carbon (BAC) Filters Used for Drinking Water Treatment. Water Research, 35:12:2923. http://dx.doi.org/10.1016/S0043-1354(00)00579-0.

Blute, N.; Ghosh, A.; & Lytle, D.A., 2012. Assessing Ammonia Treatment Options. Opflow, 38:5:14. http://dx.doi.org/10.5991/OPF.2012.38.0025.

Bremer, P.J.; Webster, B.J.; & Wells, D.B., 2001. Biocorrosion of Copper in Potable Water. Journal AWWA, 93:8:82.

de Vet, W.W.J.M; van Loosdrecht, M.C.M.; & Rietveld, L.C., 2012. Phosphorus Limitation in Nitrifying Groundwater Filters. Water Research, 46:4:1061. http://dx.doi.org/10.1016/j.watres.2011.11.075.

Fleming, K.K.; Harrington, G.W.; & Noguera, D.R., 2005. Nitrification Potential Curves: A New Strategy for Nitrification Prevention. Journal AWWA, 97:8:90.

Lee, S.H.; O’Connor, J.T.; & Banerji, S.K., 1980. Biologically Mediated Corrosion and Its Effects on Water Quality in Distribution Systems. Journal AWWA, 72:11:636.

Lytle, D.A.; Williams, D.; Muhlen, C.; & Bayner, M., 2014. Summary Report: Third Pilot Study of Biological Treatment Process for the Removal of Ammonia From a Small Drinking Water System: Application of Lessons Learned. US Environmental Protection Agency (USEPA), Office of Research and Development, Cincinnati [in review].

Lytle, D.A.; White, C.; Williams, D.; Koch, L.; & Nauman, E., 2013. Innovative Biological Water Treatment for the Removal of Elevated Ammonia. Journal AWWA, 105:9:E524. http://dx.doi.org/10.5942/jawwa.2013.105.0109.

Lytle, D.A.; White, C.; Williams, D.; Koch, L.; & Nauman, E., 2012. Summary Report: Results From the Pilot Study of an Innovative Biological Treatment Process for the Removal of Ammonia From a Small Drinking Water System in Iowa. EPA/600/R-12/655. USEPA, Cincinnati.

Lytle, D.A.; Sorg, T.J.; Wang, L.; Muhlen, C.; Rahrig, M.; & French, K., 2007. Biological Nitrification in Full-Scale and Pilot-Scale Iron Removal Drinking Water Treatment Plant Filters. Journal Water Supply: Research and Technology—AQUA, 56:2:125. http://dx.doi.org/10.2166/aqua.2007.092.

McGovern, B. & Nagy, R., 2010. Biological Removal of Iron, Ammonia, and Manganese Simultaneously Utilizing Pressurized Downflow Filtration. Proc. AWWA 2010 Water Quality Technology Conference, Savannah, Georgia.

Rittman, B.E. & Snoeyink, V.L., 1984. Achieving Biologically Stable Drinking Water. Journal AWWA, 76:10:106.

Scherrenberg, S.M.; Neef, R.; Menkveld, H.W.; van Nieuwenhuijzen, A.F.; & van der Graaf, J.H., 2012. Investigating Phosphorus Limitation in a Fixed Bed Filter With Phosphorus and Nitrogen Profile Measurements. Water Environment Research, 84:1:25. http://dx.doi.org/10.2175/106143011X13188633467076.

Scherrenberg, S.M.; Menkveld, H.W.; Bechger, M.; & van der Graaf, J.H., 2011. Minimising the Risk on Phosphorus Limitation; an Optimised Coagulant Dosage System. Water Science and Technology, 64:8:1708.

Shammas, N.K., 1986. Interactions of Temperature, pH, and Biomass on the Nitrification Process. Journal Water Pollution Control Federation, 58:1:52.

mailto:[email protected]

http://dx.doi.org/10.1016/S0043-1354(00)00579-0

http://dx.doi.org/10.5991/OPF.2012.38.0025

http://dx.doi.org/10.1016/j.watres.2011.11.075

http://dx.doi.org/10.2166/aqua.2007.092

http://dx.doi.org/10.2175/106143011X13188633467076

http://dx.doi.org/10.2175/106143011X13188633467076


E665


Standard Methods for the Examination of Water and Wastewater, 1998 (20th ed.). APHA, AWWA, & WEF, Washington.

Suffet, I.H.; Corado, A.; Chou, D.; McGuire, M.J.; & Butterworth, S., 1996. AWWA Taste and Odor Survey. Journal AWWA, 88:4:168.

USEPA (US Environmental Protection Agency), 1993. Manual: Nitrogen Control. EPA1625/R-93/010. USEPA, Washington.

USEPA, 1975. Process Design Manual for Nitrogen Control. USEPA, Washington.

White, C.P.; DeBry, R.W.; & Lytle, D.A., 2012. Microbial Ecology of a Full-Scale Biologically Active Filter for Drinking Water Treatment. Applied Environmental Microbiology, 78:17:6390. http://dx.doi.org/10.1128/AEM.00308-12.

Wilczak, A.; Jacangelo, J.G.; Marcinko, J.P.; Odell, L.H.; Kirmeyer, G.J.; & Wolfe, R.L., 1996. Occurrence of Nitrification in Chloraminated Distribution Systems. Journal AWWA, 88:7:74.

http://dx.doi.org/10.1128/AEM.00308-12

http://dx.doi.org/10.1128/AEM.00308-12

Documents

JAWWA Full-Text