oEPA
Untt«d States
Environmental Prntwiim
Agency
EPA/600/R-17/476 | December 2017 | www.epa.gov/research
Office
Water
Innovative Biological Treatment
Process for the Removal of
Ammonia, Arsenic, Iron and
Manganese from a Small Drinking
Water System in Gilbert, Iowa
of Research and Development
Systems Division
-------
EPA/600/R-17/476
Innovative Biological Treatment Process for the Removal of
Ammonia, Arsenic, Iron and Manganese from a Small Drinking Water
System in Gilbert, Iowa
Phase 1: Pilot Evaluation
Prepared by
Daniel J. Williams, Darren A. Lytle, Christy Muhlen, Maily Pham, Eugenia Riddick
U.S. Environmental Protection Agency
Office of Research and Development
National Risk Management Research Laboratory
Water Systems Division
Cincinnati, Ohio 45268, United States
and
Andrew Francis and Isaac Brewster
University of Cincinnati
Cincinnati, Ohio 45221, United States
and
Matt Staudinger
Oak Ridge Associated Universities
Oak Ridge, Tennessee 37831, United States
December 2017
-------
Notice
The U.S. Environmental Protection Agency, through its Office of Research and Development,
funded, managed, and collaborated in, the research described herein. It has been subjected to
the Agency's administrative review and has been approved for external publication. Any
opinions expressed in this paper are those of the author (s) and do not necessarily reflect the
views of the Agency, therefore, no official endorsement should be inferred. Any mention of trade
names or commercial products does not constitute endorsement or recommendation for use.
Acknowledgements
The EPA would like to acknowledge: Julie Sievers, Taroon Bidar and Tara Naber of the
Iowa Department of Natural Resources for providing valuable suggestions during the project
and comments on this document; Maria Lucente and John Arduini with the Ohio EPA for
reviewing this document; the City of Gilbert, Iowa, for supporting this project; Tad Stupp with
the City of Gilbert for operating and maintaining the pilot system, and providing valuable
project input; Steve Van Dyke with Fox Engineering for his valuable suggestions; Ronit Erlitzki,
Greg Gilles, Chris Clark, Victor Miller, and Rich Cavagnaro of AdEdge Water Technologies for
their important contributions to the project and support with the development of the study,
and partnership in moving toward developing a full-sale system; EPA Region 7 staff Brenda
Groskinsky (retired) and Amy Shields for supporting much of this research effort under EPA's
Regional Applied Research Effort (RARE) program as well as the EPA team responsible for
administering the RARE program; and Carolyn Carter of EPA's Oak Ridge Associated Universities
(ORAU) program.
i
-------
Table of Contents
1. Background 4
1.1 Ammonia in Drinking Water Sources 4
1.2 Community Water Source with Elevated Ammonia and Other Co-Occurring Contaminants
(Arsenic, Iron and Manganese) 4
1.3 Ammonia Treatment Options 5
2. Biological Water Treatment Technology Pilot Study 8
2.1 Collaboration 8
2.2 Treatment Approach 9
2.3 Pilot Technology Description 10
3. Operations, Materials, and Methods 14
3.1 Pilot System Operation 14
3.2 Water Quality Analysis 15
4. Results of the Pilot Study 17
4.1 Important Dates 17
4.2 General Water Chemistry 17
4.3 Removal of Ammonia in Source Water 17
4.4 Removal of Iron from Source Water 27
4.5 Removal of Manganese from Source Water 29
4.6 Removal of Arsenic from Source Water 31
4.7 Test Challenges: Redundancy Evaluation and Long-Term Shutdown 37
4.8 Other Water Quality Parameters 38
4.9 Assessment of Bacterial Population Based on HPCs 41
5. Discussion and Summary 42
5.1 Summary of Key Findings 43
6. References 45
2
-------
Figure 1. Map of ammonia levels in Iowa based on groundwater well analyses (1998-2012) provided by
the State of Iowa (star represent locations of current and past demonstration pilot studies) 7
Figure 2. Schematic of three pilot biological ammonia, iron, manganese and arsenic removal treatment
technology system ("NoMonia") 11
Figure 3. Pilot biological water treatment system for ammonia removal at the City of Gilbert, Iowa study
site 12
Figure 4. Column media for pilot biological water treatment system for ammonia removal at the Iowa
study site 12
Figure 5. Nitrogen content of treated water from contactor 1 20
Figure 6. Dissolved oxygen levels through pilot system 22
Figure 7. Contactor and filter flow and hydraulic loading rates 23
Figure 8: Nitrogen content of treated water from filter 1 25
Figure 9: Nitrogen content of treated water from contactor 2 as a function of depth into contactor 27
Figure 10. Iron in raw water and treated water through contactor 1, contactor 2 and filter 1 29
Figure 11. Manganese in raw water and treated water through contactor 1, contactor 2 and filter 1 31
Figure 12. Raw water arsenic speciation Error! Bookmark not defined.
Figure 13. Arsenic speciation through contactor 1 35
Figure 14. Arsenic speciation through filter 36
Figure 15. Total arsenic in raw, contactor 1, contactor 2 and filtered waters 37
Figure 16. Raw water pH, temperature and dissolved oxygen 39
Figure 17. Total alkalinity of raw, contactor 1, contactor 2 and filter 1 effluent 40
Figure 18. Heterotrophic plate counts (HPCs) in raw, contactor 1 effluent and filter 1 effluent 41
Table 1. Source water quality in Gilbert, Iowa 8
Table 2. Timeline of operational changes for contactor 1 and 2, and filter 1. 13
Table 3. Water quality analyses methods 16
Table 4: Water quality summary [average ± standard deviation (n)]. 19
Table 5. Final Design and Operating Parameters 42
3
-------
nss
1.1 Ammonia in Drinking Water Sources
Many regions in the United States have excessive levels of ammonia in their drinking
water sources (e.g., ground and surface waters) because 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 U.S. Environmental
Protection Agency (EPA) as a contaminant. Based on a 2003 World Health Organization (WHO)
assessment, ammonia levels in groundwater are typically below 0.2 milligrams per liter (mg/L),
and ammonia does not pose a direct health concern at levels expected in drinking water (WHO
2003); however, it may pose a concern when nitrification of significant levels of ammonia from
the source water occurs in the drinking water treatment plant and/or distribution system.
Nitrification, which is the conversion of the ammonia to nitrite and nitrate by bacteria, leads to
distribution system water quality issues, such as potential corrosion problems, oxidant demand,
taste and odor complaints, and elevated nitrite levels (Bremer et al.,2001; Fleming et al., 2005;
Lee et al., 1980; Odell etal., 1996; Rittman & Snoeyink, 1984; Suffet et al., 1996).
Ammonia in water may also pose problems with water treatment effectiveness. For
example, in source waters containing both ammonia and arsenic, ammonia may negatively
impact the removal of arsenic by creating a chlorine demand, therefore reducing the availability
of chlorine needed to oxidize the arsenic (Lytle et al., 2007). Lastly, water systems that have
ammonia in their source water and desire to maintain a free chlorine residual will need to add
additional chlorine to overcome the demand of ammonia. While chemical cost and added
operational complexity are an issue, excessive chlorine addition can pose disinfection by-
product issues as well. The complete oxidation of source water ammonia prior to or as part of
the water treatment process would eliminate the potential negative impacts on treatment
effectiveness and nitrification on distribution system water quality.
1.2 Community Water Source with Elevated Ammonia and Other Co-Occurring Contaminants
(Arsenic, Iron and Manganese)
Many regions in the Midwest are particularly impacted by ammonia in their source
waters from natural geology, agricultural runoff, and other farming practices. For example, the
State of Iowa has a widespread distribution of ammonia in well waters across its communities
(Figure 1). Water quality testing of the source groundwater in one small Iowa community,
Gilbert (population approximately 1082) (Figure 1) showed that, on average, ammonia levels
were 2.9 mg as nitrogen N)/L (Table 1). Although ammonia in water is not regulated, the State
of Iowa Department of Natural Resources (IDNR) can require water systems in the state to
4
-------
monitor nitrite and nitrate at their points of entry to the distribution system and in their
distribution systems should they suspect that nitrification of the source water ammonia is
occurring. Nitrite and nitrate have drinking water standards or maximum contaminant levels
(MCLs) of 1 (in Iowa 1.0) and 10 mg N/L, respectively, measured at the point of entry to the
distribution system. The Iowa IDNR extends the MCL sampling location definition to drinking
water in the distribution when nitrification is a concern. Particularly worrisome are waters with
ammonia levels greater than 1 mg N/L where incomplete nitrification can lead to exceedances
of the lower nitrite MCL.
Complicating matters, the groundwater source in Gilbert also contains elevated levels of
arsenic (0.023 mg/L), iron (2.9 mg/L) and manganese (0.08 mg/L), which are all at levels greater
than their respective primary or secondary regulatory MCLs. The impact of arsenic on human
health is well known. In 2001, the EPA reduced the MCL for arsenic from 0.05 mg/L to 0.010
mg/L (USEPA, 2001). The MCL reduction was prompted by new health effects research, which
concluded that extended human exposure to this element can cause severe health-related
illnesses, including various types of cancer, at much lower levels than previously believed
(Hopenhayn-Rich et al., 1996; 1998; Smith et al., 1998). As with ammonia, iron in drinking water
does not pose a direct health concern. However, there is an EPA recommended, non-
enforceable iron secondary MCL of 0.3 mg/L, based on aesthetic issues, rather than health-
based concerns. Iron in the water can cause a metallic taste, discoloration of the water, staining
of faucet and fixtures, and sediment build-up. Similarly, manganese levels at this site do not
pose health concerns but do present an aesthetic challenge associate with discolored water,
and staining of faucets and fixtures. As a result, a secondary MCL (SMCL) for manganese of 0.05
mg/L is in place. The SMCL is based on staining and taste considerations. It is not a federally
enforceable regulation, but is intended as a guideline for States.
Given the negative issues associated with high ammonia, iron and manganese
concentrations in drinking water, and with the health risks associated with arsenic and nitrite,
there was a clear need to identify an effective treatment approach to remove these
contaminants from Gilbert's drinking water while considering constraints on the small water
system. Treatment effectiveness as well as ease of operation, reduced operating costs, and
reliability are important design considerations that must be evaluated when recommending a
treatment approach.
1.3 Ammonia Treatment Options
The most commonly used water treatment options for addressing elevated ammonia in
source waters are the formation 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 8 to 11 times the mg N/L ammonia present. For a community
5
-------
with a water source such as the Gilbert, this would require a very high chlorine dose of
approximately 29 mg/L to breakpoint ammonia and achieve a 1 mg/L free chlorine residual. The
formation of monochloramine involves the addition of chlorine to concentrations where
ammonia is not removed but rather bound to chlorine. Other approaches including ion
exchange with zeolites, reverse osmosis (RO), advanced oxidation, and air stripping, can
remove ammonia from water, but are relatively complex, expensive, or have limited
applications, mainly when additional contaminants such as iron, manganese, and arsenic, are
present.
Although often performed unintentionally, biological ammonia "removal"1 is another
treatment approach to reduce source water ammonia. The process relies on bacteria to convert
ammonia to nitrate. As a result, a more biologically-stable water is produced, nitrification in the
distribution system is not an issue because ammonia has already been converted to nitrate, and
free chlorine residual is easily achieved. Biological conversion of ammonia (NH3) to nitrate (NO3"
) involves a two-step sequence of reactions mediated by two different genera of bacteria:
Nitrosomonas and Nitrospira. These autotrophic bacteria derive energy for cellular functions
from the oxidation of ammonia and nitrite, respectively. Nitrosomonas are responsible for the
oxidation of ammonia, in the form of ammonium (NH4+), to nitrite (NO2) according to the
reaction:
NH4+ + 1.5 02 -> N02" + H20 + 2H+ ^
Nitrospira subsequently oxidizes nitrite to nitrate, as follows:
N02" + 0.5 02 -> NO3- W
By summing these equations, the overall nitrification reaction is obtained:
IW + 2 02 -> NO3- + 2 H+ + H2O
It should be noted that 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 water. In
addition, nitrifying bacteria consume CO2 to build new cells. The total consumption of alkalinity
by nitrification is 7.1 mg as CaC03 per mg NhV- N oxidized (US EPA, 1975). The oxygen demand
of nitrification is also significant. For complete nitrification, 4.6 mg O2 is required per mg NhV-
N oxidized (US EPA, 1975; US EPA 1993).
1 The terms "removal" and "oxidation" are used interchangeably throughout this document. "Removal" is used to
represent the conversion of ammonia to nitrate and/or nitrite by biological oxidation even though treatment does not
physically remove ammonia-nitrogen but rather converts the form of nitrogen (i.e., total of ammonia, nitrite, and
nitrate).
6
-------
Other factors that affect nitrification include orthophosphate concentration, 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 (DNA); phospholipids (membranes), teichoic acid
(cell walls), and most importantly, 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 ah, 2011; Scherrenberg et al.,
2012).
Numerous laboratory studies have cited the optimum pH for complete nitrification is
between 7.4 and 8.0; although in practice, the bulk water pH may deviate from this value while
nitrification remains high (Shammas, 1986). Temperature can impact 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 impacted by temperatures below 10°C, although adjustments to the treatment
process can be made to enhance nitrification in colder climates (Andersson, et al., 2001).
CD
OO
Ammonia Levels jO q
O 0-.5 mg/L
O -5-1.0 mg/L
O 1.0-3.0 mg/L
O 3.0-5.0 mg/L
# 5.0-10.0 mg/L
Figure 1. Map of ammonia levels in Iowa based on groundwater well analyses (1998-2012)
provided by the State of Iowa (star represent locations of current and past EPA demonstration
pilot studies) (Red star is location of Gilbert, Iowa).
7
-------
Parameter
Raw
Arsenic
23 ng/L
Alkalinity
410 mg CaC03/L
Fe
2.94 mg/L
Mn
0.08 mg/L
P
0.32 mg/L
TOC
2.74 mg/L
S
0.12 mg/L
cr
7.4 mg/L
Mg
26.30 mg/L
NH4
2.91 mg-N/L
no2
0.01 mg-N/L
no3
0.02 mg-N/L
P04
0.43 mg PO4/L
pH
7.63
Temp °C
13.3
Table 1. Source water quality in Gilbert, Iowa
C,v;, v V;:C'S";
2.1 Collaboration
The City of Gilbert, a small community in Iowa, and their Engineering firm, Fox
Engineering, invited the EPA's Office of Research and Development (ORD) to conduct a pilot
demonstration of an innovative biological treatment approach to address elevated source
water ammonia concentrations and other co-contaminants. The City was interested in an EPA-
patented biological treatment approach (Figure 2) to address elevated levels of ammonia as
well as iron in the source water (Patent No. US 8, 029,674). This treatment approach has been
demonstrated at the pilot-scale at several locations in Iowa and elsewhere, and in the case of
Palo, Iowa, led to the construction of a full-scale implementation (US EPA, 2014).
EPA established a Cooperative Research and Development Agreement (CRADA, October
30, 2014) with AdEdge Water Technologies, LLC to develop a commercially available full-scale
biological drinking water treatment system marketed by AdEdge as "NoMonia" based on the
treatment approach and pilot study.
The treatment system relies on naturally occurring bacteria for the conversion of
ammonia to nitrate; provided the raw ammonia levels are lower than the nitrate MCL of 10 mg
8
-------
N/L, the approach can be effective and relatively simple. In addition, biological activity was
expected to play a necessary role in the oxidation and removal of arsenic, iron, and manganese.
An EPA patent-pending aeration pilot skid system (U.S. Application serial 14/459,277)
was designed and built by EPA staff, and installed in Gilbert on August 2016 (Figure 3) to
demonstrate the ammonia treatment approach. The City's plant operator, was trained on the
system operation and maintenance, as well as water sample collection and analysis.
Lastly, IDNR and EPA Region 7 were stakeholders of this pilot in Gilbert. IDNR was
provided periodic project updates, and commented on data and the draft project report.
2.2 Treatment Approach
The introduction of oxygen through an aeration treatment step is critical to the
successful microbiological conversion of ammonia to nitrate. Nitrification is a two-step,
microbiological process that requires oxygen (aerobic) to oxidize NH4 to NO2, and then to NO3.
The entire process requires approximately 4.5 mg of 02/mg of NH4-N in the source water.
Because the groundwater in the study community has low oxygen (1.3 mg O2/L) and elevated
ammonia of 2.9 mg N/L as well as reduced forms of iron (Fe2+), manganese (Mn2+) and arsenic
(As3+) (Table 1) that also exerts an oxygen demand, more than 13.5 mg O2/L (without
considering oxygen gradients in a fixed bed reactor or kinetic constraints) would be necessary
to address the demand. Aeration consisting of a continuous supply of adequate concentrations
of dissolved oxygen 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 oxygen demand to meet
the treatment objectives of the community's water system.
The amount of oxygen that can be added to the water is controlled 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 EPA's experience with microbiological
systems that do not provide sufficient oxygen to a nitrifying system has shown that the result is
incomplete nitrification or the production of elevated nitrite levels in the finished water. Given
the drinking water standard for nitrite is only 1 mg N/L (1.0 in Iowa), 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 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 flammable
gases and filter binding associated with gas bubbles can also be an issue.
9
-------
2.3 Pilot Technology Description
The ammonia biological removal treatment pilot system is based on the EPA patented
design (US 8,029,674 B2 awarded on 10/2/2011) seen in Figure 2. The Gilbert pilot study varied
slightly using one pair of 3-inch (7.62 cm) diameter columns in series built from clear PVC and
other common plumbing materials (Figure 3). Each pair consisted of one column or "Aerated
Contactor" filled with 55 inches (139.7 cm) of medium gravel having a nominal 1/2" diameter
(Figure 4) in series with a second column or "filter" filled with anthracite (10 inches [25.4 cm]
deep) over ADGS+silica sand-based media with a manganese dioxide coating (30 inches [76.2
cm] deep). The contactor was aerated from the bottom, such that air bubbles flow upward co-
current to the water flow (up-flow) using a diffuser (U.S. Patent Application Serial 14/459,277)
connected to a gas pump at a rate of 2.5 L/min (0.66 gpm).
In this configuration, water in the aerated contactor was always saturated with respect
to dissolved oxygen throughout the gravel media bed despite the demand from the nitrification
process, and iron, manganese and arsenic oxidation. The gravel in the contactor served to
support growth of nitrifying (and other important oxidizing) bacteria where nitrification and
other biological oxidation processes occur. Gravel allowed bacteria attachment and growth yet
eliminated the potential for "clogging" of the media, reduced backwashing frequency, and
allowed air bubbles to move through the contactor. Oxidation of ferrous iron in the source
water also occurs in the contactor, but minimal iron removal is expected to occur. Contactor
loading rates were adjusted during the pilot study (Table 2). The filter was intended to remove
arsenic-containing iron particles, manganese and bacteria, and can also provide biological
oxidation of excess ammonia and/or nitrite that exit the contactor because of incomplete
nitrification. The filter serves as a polishing step and back-up against disruption in the operation
of the contactor. Non-chlorinated effluent water from the filter is routed to a clear well, that
when full, can be used to backwash the contactor and filter, or overflow to the sanitary sewer.
10
-------
Water from the source
enters the column
from the top
Downflow
Gravel-filled column
= Contactor
(Labeled CI,
C2, & C3)
Each contactor
contains a different
size of media
PUMP J
PUMP
PUMP
PUMP
Anthracite over sand
= Filter
^Labeled F1,F2, & F3)
a
A|r (-1- -k r
bubbles I PUMP ) [pump
Water leaves pump at base of
filter and goes to clear well
Figure 2. Schematic of a "down-flow" (or counter-current to air) three pilot biological ammonia,
iron, manganese and arsenic removal treatment technology system ("NoMonia"). In this study
source water was pumped into the bottom of contactor "up-flow" (or co-current to air).
11
-------
Figure 3. Pilot biological water treatment system used to evaluate ammonia, arsenic, iron and
manganese removal in Gilbert, Iowa.
Small gravel
C2
0:
[£
Medium gravel
CI
50 mm
Figure 4, Column granular media for pilot biological water treatment system used to evaluate
ammonia, arsenic, iron and manganese removal in Gilbert, Iowa.
12
-------
Table 2. Timeline of operational changes for contactor 1 and 2, and filter 1.
Date
Elapsed Days
Description
8/17/2016
0
Pilot Start up
8/17/2016
0
Contactor 1 loading rate average 2.7 gpm/ft2
8/17/2016
0
Filter 1 loading rate 2.15 gpm/ ft2
8/19/2016
2
Needle valve removed from raw feed
8/31/2016
14
Filter 1 ran dry
9/2/2016
16
Possible mud ball formation top of filter
9/9/2016
23
Air flow increased in contactor (low DO)
9/13/2016
22
Start acclimation
10/13/16
54
Changed loading rate of filter to 1.8 gpm/ ft2, contactor to 2.3gpm/
ft2
11/28/16
100
Backwash of contactor
12/12/2016
114
Air flow increased in contactor (low DO)
1/10/2016
143
Filter 1 ran dry
01/20/17
153
Backwash of contactor
2/14/2017
178
Contactor flow rate was 370 ml/min (2.0 gpm/ ft2)
2/21/2017
185
Contactor was at 430 ml/min (2.3 gpm/ ft2)
2/28/2017
192
Filter 1 ran dry
3/1/2017
193
Filter 1 ran dry
4/10/2017
230
Changed contactor flow rate to 375 ml/min (~2.0 gpm/ ft2)
4/17/2017
243
Contactor 2 startup 450 ml/min (~2.4 gpm/ft2)
5/06/2017
262
Challenge test #1: Contactor flow doubled 820 ml/min (4.4 gpm/
ft2) ortho-P04 feed adjusted accordingly. Filter 1 shut off for
weekend
5/9/2017
265
Filter 1 back online loading rate of 1.8 gpm/ft2
5/11/2017
267
Filter 1 ran dry
5/19/2017
275
Filter 1 ran dry
5/30/2017
286
Contactor flow back to 400 ml/min (~2.15 gpm/ ft2)
6/01/2017
288
Backwash Contactor 1
6/14/2017
301
Challenge test #2: Contactor 1 and Filter 1 shutdown (air was shut
off)
6/23/2017
310
Contactor 1 and Filter 1 back on-line with air
7/12/2017
329
Backwash Contactor 1
7/26/2017
343
Interstage pH adjustment started
8/8/2017
356
Backwash Contactor 1 weekly samples collected 60 minutes after
backwash (Cl/Fl)
13
-------
3.1 Pilot System Operation
The pilot system (Figure 3) contactor (contactor 1, CI) was operated approximately 7
hours a day, 7 days per week for nearly a year beginning on August 17, 2016. Raw water from
the Gilbert's existing well was not chlorinated or treated in any way prior to supplying the pilot
system. Treated water and excess filter backwash water was routed to the on-site sanitary
sewer.
Field operating and water quality measurements were collected by the Gilbert's water
plant operator and included flowrates, temperature, dissolved oxygen, and pH. Dissolved
oxygen, pH, and temperature were measured using an HQ40d meter with an LD101 dissolved
oxygen probe and PHC281 pH probe (Hach Company, Loveland, CO). Gilbert's water plant
operator also conducted field tests to determine the concentrations of iron, manganese,
arsenic, ammonia, nitrate, and nitrite in addition to water samples that were collected and sent
to the EPA on a weekly basis. The filter was backwashed using filter effluent water
approximately every 24 hours of operation although longer frequency was evaluated
successfully (up to 110 hours). Backwashing was achieved by expanding the bed by 50% for 15
minutes. The contactor was first backwashed at 100 days, then again at 153 days using raw
water. In following months of the pilot study, the contactor was placed on a monthly backwash
cycle. Contactor gravel did not expand during backwashing. A total volume of 12.5 gallons (47.3
L) was used to backwash the contactor for approximately 5 minutes at rate of 2.5 gallon/min
(gpm) (9.45 L/min).
Many parameters were varied to optimize nitrification; these included changes to
increase dissolved oxygen levels and reduce loading rate. A second contactor (contactor 2, C2)
was brought online April 17, 2017 to evaluate the impact of smaller gravel or increased
contactor surface area on ammonia levels. Changes to pilot system operation, water quality,
and other notable conditions are summarized in Table 2. Filter loading rate changes were made
by adjusting the flowrate through the pilot columns by valve adjustment. For example,
contactors began the study with a loading rate of 2.4 gpm/ft2 (5.87 m/hr) and ended the study
at 2.2 gpm/ft2 (5.38 m/hr). Filters averaged 1.8 gpm/ ft2 (3.67 m/hr) over the duration of the
study.
Since nitrifying bacteria require phosphorus to build cell mass, a phosphate chemical
feed with a target dose of 0.3 mg orthophosphate PO4/L based on previous pilot studies was
installed in-line from initial startup of study. Microorganisms use phosphate as the source of
phosphorus for the synthesis of structural and physiological biomolecules such as
14
-------
deoxyribonucleic acid (DNA), phospholipids (membranes), teichoic acid (cell walls), and most
importantly, 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 at., 2011; Scherrenberg et at., 2012). Orthophosphate was
provided by the EPA in the form of technical grade Na3P04-12H20 (Fisher Scientific) dissolved in
deionized water. This solution was added to 20 L of deionized water in a carboy and injected
into contactor 1 (and later contactor 2) at 2 mL/min via a peristaltic pump.
3.2 Water Quality Analysis
Gilbert's water plant operator collected weekly water quality samples, while making routine
measurements and shipped them on ice overnight to the EPA Office of Research and
Development (ORD) in Cincinnati for analysis. Water samples were collected from the raw
water and effluent of contactor and filter. The following water samples were collected on a
weekly basis:
250 mL for inorganic analysis
60 mL for metals analysis
40 mL for organic carbon analysis
250 mL for bacteria analysis (heterotrophic plate counts [HPC's])
60 mL for arsenic speciation (i.e., As3+and As5+) w/EDTA
Upon arriving at EPA, the samples along with the chain of custody, were removed from the
cooler, preserved accordingly, and submitted for analysis. Ammonia, nitrite, and nitrate analysis
were typically performed on the same day the cooler arrived (approximately 24 hours after
sampling). All water analyses were performed according to EPA or Standard Methods (Table 3).
15
-------
Table 3. Water quality analyses methods
Analysis
Method
Method #
Reference
Total Alkalinity
Potentiometric Titration
2320 B.4.6
Std. Methods1
Ammonia (as N)
Automated Colorimetric
350.1
EPA Methods2
Chloride
Potentiometric Titration
4500-CI D
Std. Methods1
Nitrate & Nitrite
(as N)
Automated
Colorimetric
353.2
EPA Methods2
Orthophosphate
Automated Colorimetric
365.1
EPA Methods2
As, Pb, U, Se, Bi
ICP-MS
200.8
EPA Methods2
Al, As, Ba, Be, Bi, Ca, Cd, Cr, Cu,
Fe, K, Mg, Mn, Na, Ni, P, Pb, S, Sb,
Sulfate, Si, Silica, Sn, Zn
ICP-AES
200.7
EPA Methods2
TOC
Combustion
5310 C
Std. Methods1
Temperature
Thermocouple
17.1
EPA Methods2
HPC
Culture
9215 C
Std. Methods1
1 Standard Methods for the Examination of Water and Wastewater," 21st Edition (2005).
2 USEPA, "Methods for the Determination of Metals in Environmental Samples," EPA-600/14-91-010 (1994).
16
-------
4.1 Important Dates
There are many operating changes and other events that occurred over the course of
the pilot study that are worth noting because they had a direct impact on the results and
proceeding discussions. Events including changes in contactor dissolved oxygen, flowrates
(loading rates), backwash events, have been documented (listed in Table 2) and will be referred
to when appropriate.
4.2 General Water Chemistry
Extensive water quality analysis of the site's source water, as well as the pilot contactor
and filter effluent over the entire pilot study, is summarized in Table 4. The source water was a
very hard, high alkalinity groundwater with calcium and magnesium levels averaging 69 and 26
mg/L, respectively, total hardness of 280 mg CaCOs/L, and a total alkalinity of 410 mg CaCOs/L.
The pH averaged 7.68, and sulfate, chloride, and silica averaged 94 mg SO4/L, 5 mg/L and 7.1
mg Si02 /L, respectively. Iron and manganese levels averaged 2.9 mg/L and 0.079 mg/L,
respectively, and ammonia averaged 2.9 mg N/L. Orthophosphate was on average 0.395 mg
PO4/L, and total phosphorus was 0.316 mg P/L. Nitrite (average 0.009 mg N/L) and nitrate
(0.021 mg N/L) were at or near the respective method detection limits and total organic carbon
(TOC) averaged 2.74 mg C/L.
4.3 Removal of Ammonia in Source Water
Contactor 1. Ammonia levels in contactor 1 decreased over the first 20 days of
operation from nearly 3 mg N/L to approximately 2.2 mg N/L where levels remained for the
following 50 days (Figure 5). During this period, nitrite peaked early to 0.4 mg N/L on day 20
then dropped back to near non-detectable levels as the contactor acclimated with nitrite
oxidizing bacteria. Nitrite peaked for a short period of time due to the lag in acclimation of
nitrite oxidizing bacteria (i.e., no nitrite was available prior to encourage activity). Nitrate levels
steadily increased during this same time eventually to a concentration that nearly equaled the
amount of oxidized ammonia. Between 65 and 70 days, ammonia levels unexpectedly increased
back to 2.7 mg N/L while nitrate levels decreased by a similar amount. Based on past work,
biological ammonia oxidizing contactors operated under similar conditions and water
chemistries totally acclimated (achieved complete oxidation of ammonia) within 30 days when
operated 24 hours per day (Lytle. et al., 2007). When operated for a fraction of a day, the
acclimation time can be approximated by multiplying 30 days by the reciprocal of the fraction
of operation. In the Gilbert pilot, the pilot operated 8 hours (1/3) of 24-hour day so the
17
-------
contactor was anticipated to be totally acclimated by 90 days. Given the observed slow rate of
acclimation and reversal in progress, other parameters necessary for nitrification were closely
examined. Oxygen is a critically important parameter identified in past work so dissolved
oxygen (DO) levels during the first 70 days of operations were closely examined.
18
-------
Analyte
Detection limit (mg/L)
Raw
Contactor 1
Contactor 2
Filter 1
As
0.4 ng/L
22.8 ±2.0 (50)
14.0 ±3.0 (60)
15.0 ±3.0 (48)
8.0 ±3.0 (67)
Ca
0.01
69.1 ± 1.5(50)
68.6 ±1.6 (60)
68.7 ±1.5 (48)
68.2 ±1.6 (67)
CI
5
7.4 ±6.1 (41)
7.8 ±1.9(46)
na
8.6 ±4.7 (46)
Fe
0.001
2.9 ±0.3 (50)
1.2 ±0.6 (60)
1.1 ±0.3 (48)
0.02 ±0.1 (67)
K
0.3
4.4 ±0.1 (50)
4.4 ±0.1 (51)
4.4 ±0.2 (48)
4.4 ±0.2 (67)
Mg
0.005
26.3 ± 0.6 (50)
26.3 ± 0.7 (60)
26.2 ±0.5 (48)
26.2 ± 0.7 (67)
Mn
0.001
0.08 ± 0.003 (50)
0.05 ± 0.02 (60)
0.06 ± 0.02 (48)
0.01 ± 0.01 (67)
Na
0.03
39.4 ± 1.1 (50)
39.3 ±1.1 (60)
39.3 ±0.7 (48)
39.3 ±1.2 (67)
nh3
0.03 (mg-N/L)
2.9 ±0.1 (50)
1.6 ±0.7 (47)
1.6 ±1.0 (48)
0.4 ±0.7 (47)
no2
0.01 (mg-N/L)
0.01 ± 0.0 (46)
0.2 ±0.1 (47)
0.2 ±0.1 (48)
0.23 ± 0.4 (47)
no3
0.02 (mg-N/L)
0.04 ± 0.03 (43)
1.3 ±0.7 (46)
1.1 ±0.9 (48)
2.3 ± 0.9 (46)
0-P04
0.025 (mg PO4/L)
0.4 ±0.2 (46)
0.4 ±0.7 (46)
0.4 ±0.08 (42)
0.2 ±0.1 (46)
p
0.005 (mg P/L)
0.3 ±0.03 (50)
0.2 ± 0.08 (60)
0.2 ±0.05 (48)
0.1 ±0.02 (67)
s
0.003
0.12 ±0.1 (50)
0.09 ±0.01 (60)
0.1 ±0.01 (48)
0.09 ±0.02 (67)
Sr
0.001
0.9 ±0.03 (50)
0.9 ± 0.02 (60)
0.9 ±0.01 (48)
0.9 ±0.03 (67)
Total Alkalinity
1 (mg-CaC03/L)
410.7 ± 2.3 (49)
398.7 ±7.3 (50)
397.5 ± 7.4 (48)
389.8 ±8.7 (50)
TOC
0.1 (mg-C/L)
2.74 ±0.2 (42)
2.80 ±0.12 (41)
2.84 ±0.14 (48)
2.78 ±0.11 (39)
PH
0.1
7.68 ±0.17 (50)
8.07 ± 0.29 (55)
8.16 ±0.13 (14)
8.03 ± 0.28 (55)
DO
0.01 (mg-02/L)
1.1 ±0.4 (50)
8.94 ±2.21 (55)
9.38 ±0.4 (14)
8.83 ± 1.29 (55)
Temperature
o.rc
14.2 ±2.3 (50)
15.9 ±2.2 (55)
16.6 ± 1.7 (14)
16.3 ±2.5 (55)
Table 4: Water quality summary [average ± standard deviation (n)].
-------
Increase Loading Rate to Contactor 1
Raw Water NH3
0 100 200 300
Elapsed Time (Days)
Figure 5. Nitrogen content of treated water from contactor 1.
Oxygen is a critical parameter in the nitrification process, where 4.6 mg O2/L is
necessary to microbiologically oxidize 1 mg N/L ammonia to nitrate. Further, there is also a
connection between oxygen levels and kinetic limitations associated with molecular diffusion.
Oxygen levels in the raw water were generally less than 2 mg/L over the course of the entire
study (Figure 6). The contactor oxygen level was increased to 8.4 mg/L at the contactor start-
up, after which it steadily dropped over the initial 20 days of operation that corresponded to
the initiation of nitrification. Oxygen levels remained relatively steady between 20 and 70 days
at approximately 7.2 mg/L (Figure 6). During this time ammonia levels leaving the contactor
also remained steady. Problems with the air-feed system between 70 and 80 days resulted in a
large drop in oxygen to 3.6 mg/L leaving the contactor. The drop directly corresponded to the
-------
sudden observed increase in ammonia. Adjustments to the oxygen feed rate were made at 114
days resulting in a dissolved oxygen increase to 9.6 mg/L where the level roughly stayed for the
remainder of the study. The increase in dissolved oxygen resulted in an immediate decrease in
ammonia levels (and corresponding nitrate increase) dropping to nearly 1.2 mg N/L within 14
days after the oxygen adjustment. Nitrate produced in the contactor before DO increase was
an average of 0.249 mg N/L and was increased to an average of 1.32 mg N/L after the DO
increase. Although significant and rapid improvement was observed (i.e., more ammonia was
oxidized), bacterial acclimation progress was still not totally complete. With constant and
elevated dissolved oxygen levels, acclimation continued and by 220 days, ammonia levels were
below 1 mg N/L and reached an eventual low of 0.5 to 0.6 mg N/L. Nitrite levels remained
consistently below 0.4 mg N/L and nitrate made up the concentration difference between the
raw water ammonia and contactor effluent ammonia and nitrite levels. The contactor took
approximately 105 days to fully stabilize in regard to reaching maximum ammonia removal
after dissolved oxygen levels were controlled and optimally maintained. Under these conditions
contactor 1 reached steady levels of about 60% reduction in ammonia concentration. It is worth
noting that Nitrite spiking above the 1.0 mg/L MCL was not observed at any point of time after
acclimation.
Contactor loading rate is also a very important parameter with respect to contactor and
filter performance. To somewhat complicate the interpretation, loading rate (flowrate through
contactor) was adjusted particularly over the first 80 days of operation from as high as 3
gpm/ft2 to 1.5 gpm/ft2 (Figure 7) in effort to improve the performance of contactor 1 prior to
becoming aware of the reduction in DO. Since oxygen levels were being adjusted to accelerate
acclimation at the same time, it was difficult to clearly quantify the relative impact of loading
rate on performance, although clearly, increasing DO had the most dramatic impact on
ammonia removal. After 80 days, the loading rate settled in at approximately 2.2 gpm/ft2 until
approximately 220 days to 250 days where it decreased to nearly 2.0 gpm/ft2 (4.8 m/hr) (Figure
7). The loading rate decrease resulted increase ammonia oxidation on average of 0.5 mg N/L
coming out of contactor between 230 days and end of pilot. During this time, ammonia levels
in the contactor effluent did appear correspondingly decrease as bacteria were given more time
in the contactor to accomplish ammonia oxidation.
21
-------
-1iii|iiii|iiii|iin
Raw
Contactor
Filter m T
~, I
1 ti
W:
\i
i
I I I I I I I I I I I I I I I I I LI
0 100 200 300
Elapsed Days
Figure 6. Dissolved oxygen levels through pilot system.
-------
£
©
1000
Contactor Flow Rate
800 -
O Contactor 1 Loading Rate
- Filter 1 Flow Rate
-O Filter 1 Loading Rate
600 -
4001
200
100
200
300
Elapsed Time (days)
Figure 7. Contactor and filter flow and hydraulic loading rates.
Previous work (Lytle et al., 2007) indicated that the complete oxidation of ammonia to
nitrate, or complete acclimation of bacteria after start-up of a new biologically active nitrifying
filter could take as little as 30 days for a system operated 24 hours per day, seven days a week
for a system with similar ammonia levels as Gilbert. It was also reported in subsequent work
that the acclimation time (time required to reach optimized ammonia oxidation) was
proportional to the daily hours of operation (i.e., a system operated 12 hours per day would
take twice as long to fully acclimate or 60 days). Gilbert's pilot operated 7 hours per day
suggesting a period of more than 90-days to reach steady state. The Gilbert pilot was in
operation for approximately 105 days between the time when oxygen levels were corrected
and ammonia levels approached a stable low value which is in agreement with past
observations.
23
-------
Filter. The primary intent of the filter that followed the contactors was to remove iron
particles that contained arsenic and manganese that formed in the contactor. The filter was
also biologically-active, and served as a secondary back-up barrier by oxidizing any excess
ammonia and nitrite that may have passed through the contactor. Ammonia, nitrite, and nitrate
levels entering Filter 1 were those exiting Contactor 1 (Figure 5). Ammonia oxidation to nitrite
began shortly after the pilot was initiated and rapidly increased to a peak of 2.3 mg N/L by 20
days and dropped off by 40 days (Figure 8). Such a spike is typical as there is a lag in the growth
of nitrite oxidizing bacteria until significant nitrite levels are present to trigger their activity. The
peak must be watched closely as it can briefly increase above the nitrite MCL of 1 mg N/L.
Fortunately the peak is short-lived and nitrite can be oxidized with chlorine if needed.
Considering that DO concentrations were not optimized at the beginning of the pilot (only ~ 7.5
mg/L), this peak would be expected to be even shorter under adequate DO concentrations.
Between 40 and 80 days (the time when oxygen levels were relatively low), nitrite varied but
never exceeded 0.6 mg N/L, illustrating that oxygen levels leaving the contactor impact the
filter as well. After 114 days, nitrite levels were very low and never were greater than 0.3 mg
N/L (Figure 8). Nitrate corresponded to changes in ammonia and nitrite to complete mass
balance. After 114 days, nitrate accounted for 96% of the total nitrogen leaving the filters. The
filter loading rate at the beginning of the study up to 42 days was 2.1 gpm/ft2(5.0 m/hr), and
1.8 gpm/ft2 (4.1 m/hr) for the remainder of the study (Figure 7). It is worth noting that the filter
functioned successfully as a polishing stage by completing the removal of ammonia that was
not removed in the contactor.
24
-------
3.5
3.0
2.0
^ 2.5
DC
B
a
"-C
-
=
4>
a 1.5
©
U
-
4>
1.0
0.5
Increase Loading Rate to Contactor 1
100 200 300
Elapsed Time (Days)
Figure 8: Nitrogen content of filter 1
nh3
nq2
no3
Raw Water NH3
Contactor 2. A second contactor installed on 4/17/2017 was constructed to the same design
(diameter bed depth, etc.,) and received the same raw water as Contactor 1. However, it had
two sample ports located within the gravel media bed, and was loaded with smaller nominal %
inch diameter gravel with a 55" gravel bed depth (including support layers consisting of 4" of
large-sized gravel and 4" of medium-sized gravel). The water sample taps were positioned on
the side of contactor 2 protruding 1" into the media bed to facilitate a true media bed sample
and to provide diagnostic performance are various depths if desired. The lowest contactor tap
(25") was located at an elevation equivalent to the depth where the surface area of the
gravel was equal to the surface area of 55" of Vi" gravel (designated tap CI) in the Contactor 1.
The second tap was located (37.5") at half the depth between CI and the media surface. A
contactor effluent sample (C3) was also collected.
25
-------
Ammonia levels decreased relatively constantly through contactor 2 (C3 location) from 3 mg
N/L to non-detectable levels by 110 days (Figure 9). During this time, nitrite levels remained low
and never exceeded 0.45 mg N/L. Nitrification at location C3 reflected biological activity
through the entire contactor. The time necessary for complete acclimation was on target to the
estimated 90 days based on the hours of daily operation.
As time went on, the contactor became fully acclimated with bacteria as reflected by the
progression of nitrification through Contactor 2. Nitrification progression through contactor
locations 2 and 1 lagged shortly behind contactor effluent (C3). Interestingly, more than 90% of
the ammonia was oxidized at location CI (first 25 inches of gravel) by 110 days. The results
clearly illustrate the benefits of added surface area for smaller gravel versus medium gravel.
Although acclimation rate was not impacted, treated ammonia levels were improved.
Contactor 2 performance rivaled Fe, Mn removal with no backwashing required as of 110 days.
26
-------
/ \ Effluent £
3.0 -
25.0"
40 60
Elapsed Days
| nh3
i no2
I NO,
Figure 9: Nitrogen content of treated water from contactor 2 as a function of depth into
contactor.
4.4 Removal of Iron from Source Water
The contactor was designed to be a main point where nitrification occurred, and iron,
arsenic and manganese could be oxidized. The contactor was not intended to remove particles,
such as iron particles, from the source water. The oxidation state of iron in the source water
27
-------
was not determined, but it is reasonable to assume that the reduced Fe (II) form was prevalent
based on water chemistry, low dissolved oxygen, and local geology (Figure 10). The elevated
oxygen concentration and pH in the contactor likely resulted in rapid oxidation of Fe(ll) to Fe(lll)
particles before the water entered the contactor gravel. Although Fe(ll) oxidation kinetics are
rapid under the pilot conditions, it is possible that some biological iron oxidation took place in
the contactor. Iron particles that exit the contactor should be readily removed by the polishing
filter which are commonly designed for such purposes.
Iron in the source water averaged 2.91 mg/L (±0.20 standard deviation) (Table 4) and
was relatively consistent across the entire evaluation (Figure 10). Interestingly, the contactor
removed considerable levels of iron (approximately 59%) with the effluent iron averaging 1.2
mg/L (±0.6 standard deviation) (Table 4). The contactor effluent iron levels were variable but
stayed within a wide range of approximately 0.5 mg/L to 2 mg/L (Figure 10). Although iron was
trapped in the gravel and likely became incorporated into the biofilm structure, no degradation
in contactor performance, flow restriction, or any obvious negative impact was observed.
Nonetheless, the contactor was backwashed routinely more frequently the past pilots to
removed accumulated iron. Specifically, the contactor was backwashed monthly at a rate of 2.5
gpm for 5 minutes.
The filter iron effluent averaged 0.02 mg/L (±0.1 standard deviation) (Table 4).
Regardless of the iron content in the contactor effluent, iron levels in filter effluent waters were
at or below the detection limit (Figure 10). Outstanding and consistent removal of iron was
observed through the system from the very start-up.
Iron removal through the filters was not impacted by filter loading rates (Figures 7 and
10). Filters were operated between 1.6 gpm/ft2 (3.8 m/hr) and 2.1 gpm/ft2 (5.1 m/hr). Filter
flowrates had to be lower than contactor flowrate only due to limitations in pilot design and
this observation will be taken into consideration when the design of the full-scale system is
finalized. At the completion of the study, Filter 1 was operated at a loading rate of
approximately 1.8 gpm/ft2 (4.2m/hr).
28
-------
Increase Loading Rate to Contactor 1
t 1 1 1 1 1 1 1 1 1 1 r
Raw
A Contactor 1
~ Contactor 2
~ Filter 1
1 1 1 r
« IMMMIMIIWUMI
Elapsed Time (Days)
Figure 10. Iron in raw water and treated water through contactor 1, contactor 2 and filter 1
4.5 Removal of Manganese from Source Water
The oxidation state of manganese in the source water was not determined, but it is
reasonable to assume that the reduced Mn(ll) form was prevalent based on water chemistry,
low dissolved oxygen, and local geology (Figure 11). Manganese oxidation to Mn(IV) and solid
Mn02 is not feasible without the addition of permanganate, chlorine or other strong oxidation
or through biological oxidation processes. Unlike iron, elevated oxygen concentration alone will
not oxidize soluble Mn(ll).
Manganese in the source water averaged 0.080 mg/L (±0.003 standard deviation) (Table
4) and was relatively consistent across the entire study period (Figure 11). Interestingly, the
contactor removed considerable levels of manganese (approximately 36% on average) with the
effluent manganese averaging 0.05 mg/L (±0.01 standard deviation) (Table 4). The contactor
effluent manganese levels were very variable, however, and corresponded closely with
29
-------
dissolved oxygen concentration (Figure 11). Manganese levels dropped steadily to 0.06 mg/L
over the first 80 days while a relatively stable dissolved oxygen level was maintained (Figure 6).
Although manganese is assumed to be trapped in the gravel and likely became incorporated
into the biofilm structure, no degradation in contactor performance, flow restriction, or any
obvious negative impact was observed. Nonetheless, the contactor was backwashed routinely
more frequently than past pilots to removed accumulated manganese (the contactor was
backwashed monthly at a rate of 2.5 gpm for 5 minutes). The sudden drop in dissolved oxygen
experienced at 80 days resulted in an immediate increase in manganese (Figure 11).
The filter manganese effluent concentration averaged 0.01 mg/L (±0.01 standard
deviation) (Table 4). The filter reduced manganese levels beyond the contactor throughout the
study except for the time when oxygen control was lost (day 80). Reestablishment of oxygen
levels resulted in an improvement of manganese levels. After oxygen levels were increased (115
days), manganese levels decreased to near the detection limit for the remainder of the
evaluation. Outstanding and consistent removal of manganese was observed but oxygen
control is critical.
30
-------
0.10
O/j
0.08
s
"-C
-
=
w
c
©
U
4>
VI
4>
=
«
O/j
=
«
0.06
0.04
0.02
Raw
Contactor 1
Contactor 2
Filter 1
Increase Loading Rate to Contactor 1
Elapsed Time (Days)
Figure 11. Manganese in raw water and treated water through contactor 1, contactor 2 and
filter 1.
4.6 Removal of Arsenic from Source Water
Studies have demonstrated the effectiveness of removing arsenic from aqueous systems
with natural iron. However, most of those studies required a strong oxidant such as chlorine,
potassium permanganate or iron-based, chemical coagulation treatment (adsorptive media) to
remove arsenic. In addition, those studies have shown that the sorption of arsenic is affected
by many factors such as pH, water quality, amount and form of iron. In this pilot study, air
pumped into the contactor supporting bacterial growth was the source of arsenic oxidation,
although some bacterial oxidation cannot be ruled out.
31
-------
Typically, the oxidation state of inorganic arsenic in groundwater is in the form of
arsenite, As(lll), and arsenate, As(V), in surface water. Speciation of As(lll) and As(V) was
performed by varying a Reversed Phase-High Performance Liquid Chromatography-lnductively
Coupled-Mass Spectrometry (RP-HPLC-ICP-MS) method published by Almassalkhi, (2009).
Separation of As(lll) and As(V) was achieved using an Agilent 1260 Infinity Series (HPLC)
outfitted with an Agilent Zorbax Eclipse XDB C-18 analytical column. The HPLC was coupled to
an Agilent 7700x Inductively Coupled Plasma Mass Spectrometer (ICP-MS) for arsenic75 mass
detection and quantification using EPA Method 200.8. All arsenic speciation solutions and
samples were prepared daily in HPLC ammonium phosphate ((NH4)H2P04), tetrabutyl-
ammonium hydroxide (TBAH) mobile phase (pH 6.0). Calibration standards were prepared at
concentrations of 1, 10, 25, 50, 75, 100 and 150 |ag-L 1 and As(lll/V) samples were diluted to ~
75 |ag-L 1 in (NH4)H2P04, TBAH (pH 6.0) mobile. The Detection Limit (MDL) calculated for the
HPLC-ICP-MS As (lll/V) speciation method was 0.148 and 0.155 |ag-L1, respectively, for As(lll)
and As(V). Samples where preserved with EDTA (Ethylenediaminetetraacetic acid) at time of
collection. Samples were also filtered using a nylon 0.2 |am nylon syringe filter prior to loading
into spectrometer.
Source water arsenic levels were dominantly in the As(lll) (Figure 12) oxidation state. As
oxygen was introduced into the contactor, arsenic oxidizing bacteria acclimated the gravel and
began to convert As (III) to the pentavalent form As(V) and more easily removed oxidation state
(Figure 13). Oxidation was evident shortly after start-up, suggesting arsenic oxidizing bacteria
were rapid growers. As(V) accounted for as much as 65% of the arsenic that passed the
contactor during the first 50 days of operation. Just after 50 days, a rapid shift in arsenic
speciation was noted that resulted in as much as 93% of the arsenic in the oxidized As(V) form.
The shift occurred at the same time as the hydraulic loading rate was lowered. All the arsenic
leaving the filter was in As(V) (Figure 14), indicating effective biological arsenic oxidation.
Total arsenic (soluble and particulate As[lll] and As[V]) as determined by ICP-AES are
shown in Figure 15. Total arsenic in the source water, averaged 23 |ag/L (±2.4 |ag standard
deviation) (Table 4) (Figure 15). The contactor removed considerable levels of arsenic
(approximately 60% on average) with the effluent arsenic averaging 14.0 ug/L (±3.0 ug standard
deviation) (Table 4). At 100 days, the contactor oxygen concentration was increased resulting
in a higher pH (8.7); thus, a slight increase in arsenic levels (Figure 15) was observed.
The filter arsenic effluent concentration averaged 8.0 ug/L (±3.0 ug/L standard
deviation) (Table 4). The filter arsenic levels were at or below the arsenic MCL of 10 ug/L for
most of the study with the exception of a few sample collections. It is possible that some of the
higher arsenic valves can be attributed to operational issues. At 200 days, the contactor air
concentration was increased which released large amounts of floe particulates onto the filter.
32
-------
On day 268, the filter ran dry just prior to sample collection. Extending the filter run time may
have resulted in higher arsenic values (day 285).
It is important to note the difference between arsenic speciation (Figures 12,13,14) and
total arsenic (Figure 15) levels. Slight variances when comparing these values are attributed to
the differences in each analytical method. The arsenic speciation method only detects soluble
arsenic because the method requires sample filtration before loading onto the mass
spectrophotometer. However, total arsenic by ICP-AES detects both soluble and particulate
arsenic because the sample is not filtered.
To reduce filter effluent arsenic concentrations, filter influent water (contactor 1
effluent) was pH adjusted (day 343). pH adjustment consisted of installing a chemical injection
and inline mixer prior to entering the top of filter. Muriatic acid (31.5%, Sun Belt Chemical,
Palm Coast, FL.) was used to adjust pH from 8.21 to approximately 7.5. Results indicated that
adjusting pH after the contactor did not decrease filter arsenic concentrations (Figure 15). The
more beneficial location to adjust pH to benefit arsenic removal is before the contactor, and
iron and arsenic are oxidized with the understanding that aeration will counter to some degree
pH reduction. Optimal pH adjustment to 7.0 to 7.2 before the contactor would potentially
increase arsenic attachment to iron particles thus increasing removal efficiency through the
contactor and onto the polishing filter. However, the pilot study ended before the acid feed
location (pH adjustment) before contactor could be evaluated.
33
-------
As (soluble HI}
¦¦ As (soluble V)~
- As(in + v) .
0 100 200 300
Elapsed Time (Days)
Figure 12. Raw water arsenic speciation.
34
-------
i 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 r
As (soluble HI)
¦ As (soluble V)
As(m + v)
100 200
Elapsed Time (Days)
Figure 13. Arsenic speciation through contactor 1.
35
-------
~
As (soluble HI)
As (soluble V)
As (HI + V)
> l»iinliinrtiiniiiii
100 200 300
Elapsed Time (Days)
Figure 14. Arsenic speciation through filter.
-------
Interstage pH adjustment to Filter
35
Increase Loading Rate to Contactor 1
~iiii|iiii|iir
- Raw
-¦ Contactor 1
Contactor 2
Filter
i 25
Elapsed Time (Days)
Figure 15. Total arsenic in raw, contactor 1, contactor 2 and filtered waters.
4.7 Test Challenges: Redundancy Evaluation and Long-Term Shutdown
Challenge test #1 - failure of one contactor: The loading rate of Contactor 1 was doubled
to over 4.4 gpm/ft2 after 260 days (Figure 7) for 25 days to simulate the scenario in which one
of two operating contactors fails (i.e., treatment redundancy). During this time, ammonia
immediately increased by approximately 1.5 mg N/L to nearly 2 mg N/L and nitrate decreased
by an equivalent amount (Figure 5). Nitrite did not change. Although the filter's loading rate did
not change, ammonia and nitrite levels combined increased by a total of nearly 1 mg N/L while
nitrate decreased by an equivalent amount.
37
-------
Iron levels through the contactor and filter were not impacted by the loading rate
increase (Figure 10) with the exception of an iron spike on the first day of the change.
Manganese levels increased by approximately 0.02 mg/L out of the contactor during the change
in loading rate (Figure 11). Manganese removal through the filter were also not negatively
impacted by the loading rate change except for a spike with iron on the first day of the loading
rate increase. Total arsenic through the contactor did not noticeably change (Figure 15). Two of
the three arsenic levels through the filter during this time, however, were above the MCL. The
results reflect the reduced contact time in the contactor. Upon returning to the original loading
rate, all water quality parameters rapidly returned to previous levels.
Challenge Test #2 - intermittent operation: Contactor 1 and Filter 1 were shut down for
9 days (day 301) to simulate a scenario in which both contactor and filter were out of service
for an extended amount of time. During this time, the air pump supplying oxygen to contactor 1
was also turned off. The results indicated no negative impact on contactor ammonia oxidation
performance, ammonia (0.565 mg/L), nitrite (0.172 mg/L), and nitrate (2.261 mg/L) were
observed. Filter 1 also showed very little impact from the shutdown. Oxidation levels
observed were ammonia (0.035 mg/L), nitrite (0.004 mg/L), and nitrate (2.967 mg/L).
4.8 Other Water Quality Parameters
Source water dissolved oxygen levels averaged 1.1 ± 0.4 mg/L over the course of the
study (Figure 16 and Table 4). The source water temperature averaged 14.2 ± 2.3° C and did
experience some seasonal variability ranging between 11°C to 21°C over the course of the pilot.
Although the expectation would be that the biological system would perform better in the
warmer months of the year, it was not evident that temperature during the pilot influenced
performance. The pilot study demonstrated that biological treatment will work in colder
regions, provided groundwater is the source of drinking water and the facility is adequately
heated. TOC as was not removed throughout the pilot study. TOC in the source water
averaged 2.7± 0.2 mg C/L and 2.8 mg C/L in the contactor and filter. .
38
-------
Increase Loading Rate to Contactor 1
o
Cn|
O
Cxi
O
1- o
©
7.8
1 7.6
s
X
a
7.4
7.2
o
O-Q00
30
25
20
fS
o
Otj
e,
=
4>
otj
X
o
¦a
>
15 -S
10
Elapsed Time (Days)
Figure 16. Raw water pH, temperature and dissolved oxygen.
¦a
=
-------
ammonia oxidized) for Gilbert's source water after complete oxidation of ammonia (2.9 mg
N/L) is achieved.
Increase Loading Rate to Contactor 1
-| 1 1 1 1 1 1 1 1 1 1 1-i 1 1 1 1 r
410
o
u
«
u
Otj
£
o
H
400
390 -
i i
\
380 -
Contactor 1
¦ Contactor 2
Filter 1
I I I I I I I I L
_L
_l I ; I l_
_L
_l I L
0
300
100 200
Elapsed Time (Days)
Figure 17. Total alkalinity of raw, contactor 1, contactor 2 and filter 1 effluent.
40
-------
Increase Loading Rate to Contactor 1
1e+6 rr
T
T
Raw
Contactor 1 (medium gravel)
Contactor 2 (small gravel)
Filter 1
1e+5 -
P
^ 1e+4
U
a.
B
1e+3 -
1e+2
\
100
200
300
Elapsed Time (Days)
Figure 18. Heterotrophic plate counts (HPCs) in raw, contactor 1 effluent and filter 1 effluent.
4.9 Assessment of Bacterial Population Based on HPCs
Heterotrophic plate count (HPC) measurements in the raw source water, contactor, and
filter effluent waters were performed on a routine basis as an indicator of microbial activity
although they do not directly reflect nitrifying bacteria. Raw water HPCs generally fell between
500 and 9500 CFU/mL (Figure 18). During the same time, HPCs in both the contactor and filter
were approximately an order of magnitude greater in concentrations indicating biological
activity (although not necessarily associated with nitrifying bacteria) within both systems. HPC
levels leaving the contactor and filter were very similar for the first 250 days of operation.
Beyond 250 days, filtered HPC levels were lower than contactor effluent levels. There did not
appear to be any important trends from the HPC data particularly as it relates to operational
41
-------
considerations. The random variability of HPC measurements tended to decrease with time
(most apparent after 250 days of operation) and might suggest a stabilization of the system.
There also appeared to be a trend with temperature in which greater HPC levels were observed
when the water was warmer.
The release of bacteria from the system will occur with any biological treatment
approach. Appropriate and effective disinfection must be in place to adequately inactivate the
microbiological community shed from the system.
The biological treatment pilot study demonstrated the ability of biological treatment to
effectively reduce ammonia, iron, manganese and arsenic from the source water to
concentrations below their primary and secondary MCLs. The development of biological
activity, and subsequent complete oxidation of ammonia to nitrate in the system, was
established in the expected time based on past work once the oxygen and loading rate
parameters were optimized for a system only operating 8 hours a day. Although the site's
water quality was challenging because it included high ammonia, iron, manganese and arsenic
levels, the pilot study proved to be valuable in identifying engineering and design criteria in
support of future full-scale implementation. For example, dissolved oxygen throughout the
contactor, loading rate targets, monthly backwash of contactor, and phosphate feed were all
identified as important factors affecting performance.
Table 5. Final Design and Operating Parameters
Parameter Contactor Filter
Filter loading rate
m/hr 5.4(1.2-10.5) 4.9(1.22-5.4)
gpm/ft2 2.2(0.5-4.3) 1.8(0.5-2.2)
Air flowrate
L/min 2.5
cfm/ft2 2.86
Backwash conditions
duration, min 5 15
bed expansion, % 0 50
m/hr 124 41.5
42
-------
gpm/ft2 51 17
Contactor
depth, cm 139.7
depth, inches 55
effective size, mm 12.7 (6.35 - 31.8)
effective size, inches 0.5 (0.25 -1.25)
Filter
anthracite depth, cm 25.4
anthracite depth, inches 10
Anthracite, mm - 0.97
anthracite, inches - 0.04
ADGS+silica sand depth,
cm 76.2
ADGS+silica sand depth,
inches 30
ADGS+silica sand, mm - (0.30-.35)
ADGS+silica sand, inches - (0.012-0.014)
By the end of the pilot study, complete oxidation of the source water ammonia (2.9 mg N/L) to
nitrate was achieved in Filter 1 and removal of arsenic (22.8 mg As/L), iron (2.9 mg Fe/L) and
manganese (0.08 mg Mn/L) through the anthracite/ ADGS+silica sand filter followed. Other
operating and maintenance parameters are summarized in Table 5.
5.1 Summary of Key Findings
The biological treatment pilot study produced several very important findings that will aid in
the design and installation of a full-scale water treatment plant. The following findings are
highlighted:
The innovative biological treatment system effectively reduced the levels of ammonia, iron,
manganese and arsenic to below the desired level of primary and secondary MCLs.
Although arsenic was consistently removed below the MCL of 10 |ag/L, further optimization
could be explored such as pH adjustment (lowering of pH) before the contactor to enhance
arsenic adsorption to iron oxy-hydroxides.
Biological acclimation of contactor and filter can vary depending on pilot run time, DO, and
other key parameters and is defined as the time once nitrogen species equilibrium is
reached.
43
-------
Once optimized, contactor 1 achieved approximately 83% ammonia reduction (to levels as
low as 0.5 mg N/L) using medium (1/2-inch diameter) gravel. Contactor 2 using small (1/4-
inch diameter) gravel achieved nearly 100% ammonia reduction because of added surface
area for biological attachment and growth. Despite relatively high iron and manganese
levels in the source water, and the unexpected reduction of iron and manganese in the
contactors, no clogging, flow restriction or short circuiting were observed in the contactors.
Nonetheless, a monthly routine contactor backwash regime was followed.
A dual media (10 inches [25.4 cm] anthracite/30 inches [76.2 cm] ADGS+silica sand) filter
after contactor 1 provided additional ammonia/nitrite oxidation, and achieved excellent
and consistent iron, arsenic and manganese removal once the system was fully acclimated
and optimized.
Orthophosphate is an important biological nutrient and was necessary to for microbial
acclimation, particularly with regards to nitrite oxidizing bacteria. A dose of 0.3 mg PO4/L
onto the contactor was used in the pilot.
Maintaining saturated dissolved oxygen levels in the contactor was critical to the pilot's
operation and effectiveness at achieving desired ammonia oxidation and iron removal. A
drop in dissolved oxygen levels resulted in delayed oxidation of ammonia in the contactor
and release of nitrite. Dissolved oxygen monitoring was a good process measurement tool
and must be incorporated into full-scale operation. Diffuser design will also be very
important engineering aspect of the full-scale system.
Contactor and filter loading rates were important operating variables, although the pilot
system was more sensitive to oxygen concentration. The pilot demonstrated that a
contactor and filter operated in series at loading rates of 2.2 gpm/ft2 (5.03 m/hr) and 1.8
gpm/ft2 (4.14 m/hr), respectively, met desired finished water quality objectives. The
contactor performance was affected with respect to ammonia reduction when the loading
rate was doubled to evaluate redundancy considerations, yet, ammonia increase was not
significant, and more even more important, there was no spiking on N02 above the MCL
which implies that in case of failure or maintenance, the system can still generate safe
drinking water.
The system was robust in that it recovered rapidly after long-term (1 week) and short-term
(18 hours per day) shutdown periods, and changes in loading rates and minor seasonal
water changes.
Alkalinity decrease following nitrification in the systems was predicted by theoretical
considerations and could be used as an additional process monitoring tool.
44
-------
The Filter was backwashed on average of 24 hours of run time by achieving 50% bed
expansion.
Almassalkhi, B. A. (2009). Arsenic Speciation, Detection, and Quantification in Drinking Water
using High Performance Liquid Chromatography and Inductively Coupled Plasma Mass
Spectrometry. University of Cincinnati.
Andersson, A.; Laurent, P.; Kihn, A.; Prevost, M.; Servais, P. (2001). Impact of Temperature on
Nitrification in Biological Activated Carbon (BAC) Filters used for Drinking Water Treatment.
Water Research. 35:2923-2934.
Bremer, P.J.; Webster, B.J.; Wells, D.B. (2001). Biocorrosion of Copper in Potable Water. Journal
AWWA. 93, 82-91.
de Vet, W.W.J. M; van Loosdrecht, M.C.M.; Rietveld, L.C. (2012). Phosphorus Limitation in
Nitrifying Groundwater Filters. Water Research. 46\ 1061-1069.
Fleming, K.K.; Harrington, G.W.; Noguera, D.R. (2005). Nitrification Potential Curves: A New
Strategy for Nitrification Prevention. Journal A WWA. 97, 90-99.
Hopenhayn-Rich, C., M.L. Biggs, A. Fuchs, R. Bergoglio, E.E. Tello, H. Nicolli, and A.H. Smith.
1996. Bladder cancer mortality associated with arsenic in drinking water in Argentina.
Epidemiology 7:117-124.
Hopenhayn-Rich C., M.L. Biggs, and A.H. Smith. 1998. Lung and kidney cancer mortality
associated with arsenic in drinking water in Cordoba, Argentina. Int. J. Epidemiol. 27:561-569.
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, 636-645.
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. Jour. Water
Supply: Res. and Tech.: AQUA. 56, 125-136.
Odell, L.H.; Kammeyer, G.J.; Wilczak, A.; Jacangelo, J.G.; Marcinko, J.P.; Wolfe, R.L. (1996).
Occurrence of Nitrification in Chloraminated Distribution Systems. Journal AWWA. 88, 74.
45
-------
Rittman, B.E.; Snoeyink, V.L. (1984). Achieving Biologically Stable Drinking Water. Journal
AWWA. 76, 106.
Scherrenberg, S. M.; Menkveld, H.W.; Bechger, M.; van der Graaf, J.H. (2011). Minimizing the
Risk on Phosphorus Limitation; An Optimized Coagulant Dosage System. Water Sci Technol.
64:1708-1715.
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 Environ Res. 84:25-33.
Shammas, N.K. (1986). Interactions of Temperature, pH, and Biomass on the Nitrification
Process. J. Water Pollut. Control Fed. 58, 52-58.
Smith, A.H., M. Goycolea, R. Haque, and M.L. Biggs. 1998. Marked increase in bladder and lung
cancer mortality in a region of northern Chile due to arsenic in drinking water. Am. J. Epidemiol.
147:660-669.
Suffet, I.H.; Corado, A.; Chou, D.; McGuire, M.J.; Butterworth, S. (1996). AWWA Taste and Odor
Survey. Journal AWWA. 88, 168-180.
U.S. Environmental Protection Agency (US EPA). (2014). Engineering Design and Operation
Report: Biological Treatment Process for the Removal of Ammonia from a Small Drinking Water
System Iowa: Pilot to Full-Scale. Office of Research and Development, Washington, DC
EPA/600/R-14/336.
U.S. Environmental Protection Agency (US EPA). (1975). Process Design Manual for Nitrogen
Control, Technology Transfer, US EPA, Washington D.C.
U.S. Environmental Protection Agency (US EPA). (1993). Manual: Nitrogen Control. Office of
Research and Development. Office of Water, US EPA, Washington D.C., EPA-1625/R-93/010.
U.S. Environmental Protection Agency (US EPA). (1983). Methods for Chemical Analysis of
Water and Wastes. US EPA, Washington D.C., EPA-600/14-79-020.
U.S. Environmental Protection Agency (US EPA). (1994). Methods for the Determination of
Metals in Environmental Samples. US EPA, Washington D.C., EPA-600/14-91-010.
U.S. Environmental Protection Agency (US EPA). (2001). Final Arsenic Rule. Fed. Reg., Jan. 22.
WHO. (2003). Ammonia in Drinking-Water. Background Document for Preparation of WHO
Guidelines for Drinking-Water Quality. Geneva, World Health Organization
(WHO/SDE/WSH/03.04/1).
Standard Methods for the Examination of Water and Wastewater. (2005). (21st ed.). APHA,
AWWA, and WEF, Washington.
46
-------
47
------- |