Arsenic Removal from
Drinking Water by Iron Removal
USEPA Demonstration Project at Climax, MN
Project Summary
Wendy E. Condit, Abraham S.C. Chen
A project to demonstrate Kinetico's Macrolite® pressure
filtration process' ability to remove arsenic from drinking
water was conducted at a water system in Climax, MN. The
project objectives were to evaluate: (1) the effectiveness of
Kinetico's Macrolite® pressure filtration process in removing
arsenic to meet the new arsenic maximum contaminant level
(MCL) of 10 micrograms per liter (|ag/L), (2) the reliability
of the treatment system, (3) the required system operation
and maintenance (O&M) and operator's skills, and 4) the
capital and O&M costs of the technology. The project also
characterized water in the distribution system and process
residuals produced by the treatment system.
Introduction
Amended in 1996, the Safe Drinking Water Act (SDWA)
required that the United States Environmental Protection
Agency (EPA) develop an arsenic research strategy and
publish a proposal to revise the arsenic MCL. On March 25,
2003, EPA revised the rule text to express the MCL as 0.010
mg/L (10 |Jg/L) and to require all community and non-
transient, noncommunity water systems to comply with the
new standard by January 23, 2006 (EPA, 2003).
In October 2001, EPA announced an initiative for additional
research and development of cost-effective technologies
to help small community water systems (those with less
than 10,000 customers) meet the new arsenic MCL, and
to provide technical assistance to small system operators
to reduce compliance costs. As part of this Arsenic Rule
Implementation Research Program, EPA's Office of Research
and Development proposed a project to conduct a series
of full-scale, onsite demonstrations of arsenic removal
technologies, process modifications, and engineering
approaches applicable to small systems.
Site Information
The water system in Climax, MN supplies drinking water to
264 community members. Two wells in a Quaternary Buried
Artesian aquifer provide the groundwater. Each well is 141
feet deep with 15 feet of slotted screen. Well 1 has a capacity
of 140 gallons per minute (gpm), and Well 2 has a capacity
of 160 gpm. The wells are alternated monthly to meet the
peak daily demand of 105,000 gallons per day (gpd). Both
pumps are used during fire emergencies with a full capacity
of 300 gpm. The treatment system originally consisted of
a gas chlorine feed to reach a target chlorine residual level
of 0.6 mg/L. The water is fluoridated to a target level of 1.0
mg/L.
-------
Source water samples were collected on July 30, 2003
from the West Well. The results of the source water
analyses are presented in Table 1. Based on the July 30,
2003 sampling results, as much as 90 percent of the total
arsenic existed as arsenic (III) and 10 percent existed as
particulate arsenic. Almost all of the iron concentration in
the source water existed as soluble iron. A rule of thumb
is that the soluble iron concentration should be at least
20 times the soluble arsenic concentration for effective
removal of arsenic onto iron solids. The results from the
July 30, 2003 sampling event indicated that the soluble
iron level was approximately 16 times the soluble arsenic
level. Because the natural iron content in the source
water was close to the target ratio, the initial plan was to
operate the system without supplemental iron addition.
Arsenic Treatment System
The treatment train for the Climax system includes
oxidation, co-precipitation/adsorption, and Macrolite®*
pressure filtration (see Figure 1 for the process flowchart
and sampling locations and Figure 2 for a photograph
of the Macrolite® FM-236-AS Arsenic Removal System).
Macrolite® is a low-density, spherical, and chemically
inert ceramic media that is designed for a high-rate
filtration up to 10 gpm per square foot. The media,
manufactured by Kinetico, is approved for use in
drinking water applications under NSF Standard 61.
Table 2 summarizes the design features of the Macrolite®
pressure filtration system. The major process steps and
system components include:
• Oxidation—The liquid sodium hypochlorite feed
system consisted of a day tank and a chemical feed
pump with a maximum capacity of 6 gallons per
hour. The operator tracked the operation of the
Table 1. Climax, MN Source Water Quality
Parameter
Value
Total Alkalinity (as CaCO3;
Hardness (asCaCO3)
Chloride
Fluoride
Sulfate
Silica (as SiO2
Orthophosphate (as PO4
As (Total)
As (Soluble)
As(lll)
As(V)
Total Fe
TotalAI
Total Mn
TotalV
Total Na
mg/L
mg/L
mg/L
mg/L
mg/L
304.0
227.6
190.0
,7
120.0
LEGEND
Unit Process/
System Component
Sampling Location
Process Flow
Backwash Flow
CHLORINE OXIDATION
CONTACT
TANK
CONTACT
TANK
BACKWASH
At Backwash
TO SANITARY
SEWER
After Contact Tanks
MACROLITE I MACROLITE
PRESSURE 1 PRESSURE
FILTRATION j FILTRATION
TANK A 1 TANKB
After Tank A • I After Tank B
After Tanks A and B Combined
DISTRIBUTION
SYSTEM
Figure 1. Process Flow (140 gpm) Diagram and
Sampling Locations
chemical feed system by measuring free and total
chlorine across the treatment train.
• Co-Precipitation/Adsorption with Supplemental
Iron Addition—Beginning on January 3, 2005, an
iron addition system was used to inject a target dose
of 0.5 mg/L of iron after the prechlorination tap
* The mention of trade names does not constitute endorsement.
Figure 2. Macrolite® Pressure Filtration System
(Control Panel [1], Macrolite® Filters [2,3],and Contact Tanks [4,5])
-------
Table 2. Design Specifications for the Macrolite® FM-236-AS Pressure Filtration System
Parameter Value
Prechlorination Dosage (mg/L [as CI2]) 1.2
Supplemental Iron Dosage (mg/L [as Fe]) 0.5
Remarks
The chlorine dosage was adjusted to provide a chlorine residual in
the distribution water of 0.6 mg/L.
Implemented on January 3,2005
No. Vessels
Vessel Size (inch)
Contact Time (minutes/vessel)
Arranged in parallel
42Dx72H 345 gallons each tank
5 —
No. Vessels
Vessel Size (inch)
Media Quantity (ftVvessel)
Filtration Rate (gpm/ft2
Pressure Drop (psi)
Backwash Initiating Pressure (psi)
Throughput before Backwash (gallons)
Backwash Hydraulic Loading (gpm/ft2)
Backwash Duration (minutes)
Wastewater Production (gallons)
System Design Flowrate (gpm)
Maximum Daily Production (gpd)
Hydraulic Utilization (%)
Arranged in parallel
264 gallons each tank
24-inch bed depth of 40/60 mesh Macrolite® media in each vessel
Across a clean bed
Across bed at end of filter run
Variable
8 to 10
Variable
Variable
140
201,600
52
Based on PLC settings for pressure, run time, or standby set points.
Based on PLC settings for minimum and maximum backwash times
(e.g., 7 to 15 minutes from factory set points).
See above
A flow-limiting device regulated flow through each vessel to less
than 70 gpm to prevent filter overrun and system damage.
Based on peak flow, 24 hours per day
Estimated based on a historic peak daily demand of 105,000 gpd
using a ferric chloride solution. The iron addition
system included one 55-gallon polyethylene tank
with containment, an overhead mixer, a 2.5-gallons
per hour chemical metering pump, and a 600-
pound capacity drum scale. The operator used daily
readings of the weight of the day tank to measure the
consumption of ferric chloride solution.
Contact—The two 345-gallon contact tanks were
constructed of fiberglass-reinforced plastic and had
6-inch top and bottom flanges. The water passed
through the contact tanks in an upflow configuration.
Pressure Filtration—Pressure filtration involved
downflow filtration through two vessels arranged in
parallel. The vessels, equipped with 6-inch top and
bottom flanges, were mounted on a polyurethane-
coated steel frame. The Macrolite® media in each
vessel was underlain by a fine garnet fill layered
1 inch above the 0.006-inch slotted stainless steel
wedge-wire underdrain.
Backwash Operations—The filter was automatically
backwashed in an upflow configuration when
the pressure drop across the bed had reached 20
pounds per square inch (psi). The backwash was
also triggered by the length of time the unit had
been in service and/or in stand-by mode. During
backwash, the water in one of the filtration vessels
was first drained from the vessel and the filter was
then sparged with air for 2 minutes at a pressure of
100 pounds per square inch gauge. After a 5-minute
settling period, the filtration vessel was backwashed
with treated water at a flowrate of approximately 55
gpm (8 gpm per square foot) until the turbidity of
the backwash water had reached a target threshold
level of 6 nephelometric turbidity units (NTU) based
on the factory setting. The backwash was conducted
one vessel at a time and the resulting wastewater was
sent to a sump and then to the sanitary sewer. After
backwash, the filtration vessel underwent a filter-
to-waste cycle for 5 minutes before returning to the
service mode.
The Macrolite® treatment system is fully automated with
an operator interface, programmable logic controller
(PLC), and modem housed in a central control panel.
The control panel is connected to various instruments
used to track system performance, including inlet
and outlet pressure after each filter, system flowrate,
backwash flowrate, and backwash turbidity with a
Hach™ high range turbidimeter. All major functions of
the treatment system are automated and would require
only minimal operator oversight and intervention if
all functions are operating as intended. Under normal
operating conditions, the skill set required to operate
the system was limited to observation of the process
equipment integrity and operating parameters such as
pressure, flow, and system alarms. The daily demand on
the operator was about 30 minutes to visually inspect
the system and record the operating parameters on the
-------
log sheets. The operator also performed O&M activities
such as cleaning the turbidimeter photo cell, monitoring
backwash operational issues, and working with the
vendor to troubleshoot and perform minor on-site
repairs. All plumbing for the system is Schedule 80 PVC
and the skidded unit is pre-plumbed with the necessary
isolation valves, check valves, sampling ports, and other
features. A 5-hp, 60- gallon vertical air compressor is
included in the system.
System Performance
The performance of the Macrolite® FM-236-AS Arsenic
Removal System was evaluated based on analyses
of water samples collected from the treatment plant,
distribution system, and backwash lines. The treatment
plant water was sampled on 53 occasions (including
four duplicate sampling events) during the one-year
demonstration period.
Arsenic Removal. The total arsenic levels across
the treatment train over the duration of the one-year
period are illustrated in Figure 3. Total arsenic levels in
raw water ranged from 31.2 to 51.4 |ag/L. From August
11, 2004 to January 2, 2005, total arsenic levels in the
treated water ranged from 9.7 to 19.0 |ag/L, averaging
14.7 |Jg/L. Insufficient natural iron was present in the
raw water to achieve effective arsenic removal to below
the 10 |-ig/L MCL. After supplemental iron addition was
implemented, total arsenic levels in the treated water
were reduced to 6.0 to 9.3 |ag/L, averaging 7.4 |ag/L, with
no exceedances of arsenic above the 10 |-ig/L level for the
remainder of the study period.
Figure 4 shows the arsenic speciation results. The total
arsenic concentration in the raw water averaged 36.5
|jg/L that consisted predominately of soluble arsenic. Of
the soluble fraction, 35.8 |jg/L was arsenic (III) and 2.1
|jg/L arsenic (V). These results compared well with those
of the July 30, 2003 source water sampling.
After prechlorination and the contact tanks, the soluble
fraction of the arsenic decreased to an average 14.7 |jg/L
and the particulate fraction to 24.1 |ag/L. Of the soluble
arsenic, 12.2 |jg/L (83 %) was arsenic (V) indicating
effective oxidation of arsenic (III) to arsenic (V) with
chlorine. Most of the particulate arsenic was removed
At Well head
After Contact
After Tank A
After Tank B
After Tanks A and B Combined
MCL
Date
Figure 3. Total Arsenic Concentrations Across
Treatment Train
by the filters. Because of insufficient natural iron in the
raw water, however, the arsenic concentration in the
combined filter effluent ranged from 9.7 |jg/l to 19.0 |-ig/L
and averaged 14.1 |jg/L (primarily soluble arsenic) that
was above the 10 |-ig/L arsenic MCL.
After the start of supplement iron addition, the
particulate arsenic of the water from the contact tanks
averaged 23.4 |jg/L and the soluble arsenic averaged 11.7
|jg/L with the soluble arsenic being 83% arsenic (V). The
Arsenic Species at the Wellhead
60
50
40
30
20
10
CIAs (particulate)
• As(V)
• As (III)
MCL
Date
Arsenic Species after Contact Tanks
As Concentration (|ig/L)
->• w co .&. en c
D O O O O C
0,
r~
J
— 1
•
i
• As (particulate)
Iron addition began BAs(V)
on 01/03/05 • As (III)
— |
-
^j
~
J
— 1
•
|
y
MCL
Date
Arsenic Species afterTanks A and B Combined
50'
40.
30'
20'
10-
fl.
CIAs (particulate)
Iron addition began '' '''
on01/03/05 BAsflll)
B D 0 D _ y S
MCL
Date
Figure 4. Concentrations of Arsenic Species
-------
total arsenic in the combined filter effluent averaged 7.4
|jg/L with 6.5 |ag/L (88%) of the effluent arsenic being
soluble arsenic. Particulate arsenic levels in the treated
water were low, ranging from less than 0.1 to 2.4 |-ig/L
and averaging 0.9 |ag/L. The reduction of some of the
soluble arsenic (V) through the filters suggests that the
iron particles accumulated within the filters had some
additional adsorptive capacity for arsenic (V) removal.
Iron Removal. Figure 5 shows the total iron levels
across the treatment train over the duration of the study
period. Total iron levels in the raw water ranged from
361 to 1,209 |-ig/L and averaged 540 |-ig/L. Iron in the raw
water existed primarily in the soluble form and averaged
485 |ag/L. The average soluble iron and soluble arsenic
levels in the source water corresponded to an iron:arsenic
ratio of 13:1, which was below the target ratio of 20:1 for
effective arsenic removal. As expected, iron existed solely
in the particulate form after prechlorination and the
contact tanks.
Manganese Removal. Total manganese levels in
raw water ranged from 112 to 218 |-ig/L with an outlier
at 505 |-ig/L. Manganese in raw water existed primarily
in the soluble form at levels ranging from 112 to 145
|jg/L. After prechlorination and the contact tanks, soluble
manganese concentrations decreased to 59.0 to 89.1 |ag/L.
An average of 42 percent of the soluble manganese was
precipitated to particulate manganese. Unlike MnOx-
coated media, Macrolite® does not promote Mn(II)
removal via adsorption with the presence of chlorine.
Only particulate manganese was filtered out by the
Macrolite® filters, leaving soluble manganese in the
treated water at levels ranging from 55.5 to 91.5 |ag/L.
Other Water Quality Parameters. Dissolved
oxygen levels remained low across the treatment train,
with average values ranging from 1.3 to 1.7 mg/L,
but oxidation-reduction potential values significantly
increased after chlorine addition ranging from -63 to
-128 mV before chlorination to 121 to 382 mV after
chlorination. The pH values of the raw water and treated
water had average values of 7.5 and 7.4, respectively.
Average alkalinity results ranged from 313 to 326
2500 T
At Wellhead
After Contact
After Tank A
After Tank B
After Tanks A and B Combined
^^^^^^^^
rC^ ^ i^ xN1 K"V ""V 0) fy Vs
-------
Table 3. Summary of Backwash Parameters
Table 4. Backwash Water Sampling Results
Backwash Parameters
Mini- Me- Maxi-
Initial Field Settings (08/11/04-01/14/05)(a)
Backwash Duration (min) 18 18 53(c)
Water Quantity Generated (gal) 800 900 2,650(c)
Modified Field Setting (01/14/05-08/12/05)*'
Backwash Duration (min) 5 10 306(c)
Water Quantity Generated (gal) 250 500 15,300(c)
(a) Backwash events: 70 for Vessel A and 71 for Vessel B.
(b) Backwash events: 119forVessel A and llSforVessel B, not
including multiple successive events caused by backwash
malfunctions.
Backwash control malfunctions caused repeat backwash
cycles to occur on the same day.
(c)
startup, and one sample exceeded the 1,300 |jg/L action
level. The treatment system did not appear to impact
the factors that can increase the solubility of copper in
drinking water in contact with plumbing fixtures (e.g.,
low pH, high temperature, and soft water with fewer
dissolved minerals).
Backwashing, Table 3 summarizes data related to
the backwash duration and backwash water quantity
produced under the initial and modified field settings
from August 11, 2004 through January 14, 2005 and from
January 14, 2005 through August 12, 2005, respectively.
The backwash flowrate for both time periods was
approximately 50 gpm or 7 gpm/ft2, which is lower than
the 8 to 10 gpm/ft2 design value. The backwash flowrate
was lowered in the field at startup to avoid media loss
that was observed when a higher flowrate was used such
as the factory set point of 75 gpm.
Table 4 summarizes the analytical results from the
twelve backwash water sampling events. Prior to iron
addition, soluble arsenic and iron concentrations in
the backwash water averaged 16.0 |jg/L and 21.0 |jg/
L, respectively. After iron addition, soluble arsenic
concentrations decreased and averaged 8.4 |ag/L, while
soluble iron concentrations increased and averaged
75.4 |jg/L (excluding the July 27, 2005 data that had
uncharacteristically high soluble arsenic, iron, and
manganese). After iron addition, the soluble iron levels
in the backwash water increased due to equilibrium with
the higher total iron levels (e.g., iron particulates) in the
backwash water. However, the soluble arsenic levels
decreased, due to increased adsorption onto the iron
particulates. For the last sampling event on November 15,
2005, total suspended solids (TSS) and total arsenic, iron,
and manganese also were analyzed for the composite
sample collected. The results showed total iron levels in
the backwash water at 74.2 to 97.6 mg/L and total arsenic
levels at 1.42 to 1.85 mg/L. TSS levels in the backwash
water ranged from 188 to 278 mg/L.
The Toxicity Characteristic leaching Procedure (TCLP)
results of the backwash solids showed no detectable
arsenic concentrations in the leachate. Concentrations
of cadmium, chromium, lead, mercury, selenium, and
silicon were also below the detection limit. Only barium
6 • • • •
arameier Unit
PH
S.U.
Turbidity NTU
Total
Dissolved
Solids
mg/L
Soluble As ug/L
Soluble Fe ug/L
Soluble Mn ug/L
Value
Minimum
Maximum
Average
Minimum
Maximum
Average
Minimum
Maximum
Average
Minimum
Maximum
Average
Minimum
Maximum
Average
Minimum
Maximum
Average
Pre Iron Post Iron
ddition Addition
7.1
7.9
7.6
7.6
60
41.6
758
990
840
12.3
21.6
16.0
<25
39.9
21.0
24.9
413
119.65
7.3
7.6
7.5
14
140
93.4
646
940
786
6.4
25.6
10.5
25.7
771
164.3
65.6
118
79.6
One-half of the detection limit was used for non-detect samples
for calculations.
showed detectable concentrations ranging from 0.189 to
0.231 mg/L. The TCLP regulatory limit set by EPA is 5
mg/L for arsenic and 100 mg/L for barium.
System Costs
The cost of the system was evaluated based on the
capital cost per gpm (or gpd) of design capacity and the
O&M cost per 1,000 gallons of water treated. The costs
associated with the building, sanitary sewer connections,
and other discharge-related infrastructure were not
included in the treatment system costs. These costs were
funded separately by the demonstration site.
Capital Costs. Table 5 summarizes the capital
investment for the Climax system. The equipment costs
include the costs for the Macrolite® media, contact tanks,
filtration skid, instrumentation and controls, labor
(including activities for the system shakedown), system
warranty, freight, and sales tax. The system warranty
included repair and/or replacement of any equipment
or installation workmanship for twelve months after
system startup. The engineering costs include the costs
for preparing a process design report and the required
engineering plans. The installation costs include the
costs for labor and materials for system unloading and
anchoring, plumbing, and mechanical and electrical
connections.
The total capital cost of $270,530 was normalized to the
system's rated capacity of 140 gpm (201,600 gpd), which
resulted in $1,932 per gpm ($1.34 per gpd). The total
capital cost of $270,530 was converted to a unit cost of
-------
Tables. Summary of Capital Investment
Description Cost
Equipment Costs (59%) $159,419
Media, Filter Skid,andTanks
Air Compressor
Control Panel
Labor
Warranty
Additional Flowmeter/Totalizers
Iron Addition Equipment
Freight and Sales Tax
$2,346
$11,837
$43,005
$11,950
$2,622
$5,259
$16,190
Labor
Subcontractor
osts (26%)
Labor
Travel
Subcontractor
Total Capital Investment (100%)
$38,094
$1,250
^^
$12,914
$6,163
$52,690
$270,530
$0.35/1,000 gallons, using a capital recovery factor of
0.9439 based on a 7 percent interest rate and a 20-year
return period (Chen et al., 2004). These calculations
assumed that the system operated 24 hours a day, 7 days
a week, at the system design flowrate of 140 gpm. The
system operated only 5.6 hours a day and produced
13,829,000 gallons of water during the study period. At
this reduced usage rate, the total unit cost was increased
to $1.85/1,000 gallons.
Operation and Maintenance Costs. Table 6
summarizes O&M costs, which include costs associated
with chemical supply, electricity, and labor. Because the
system was under warranty during the one-year study
period, no costs were incurred for repairs to the system.
Chlorination was performed prior to the demonstration
study so the incremental cost for the sodium hypochlorite
solution was assumed to be negligible. The usage rate
for the ferric chloride stock solution was approximately
80 gallons or 900 pounds per year. The incremental
power costs were estimated based on the change in
electric utility bills for a one-year timeframe before and
after the treatment plant installation and do not include
propane costs to heat the building. Under normal
operating conditions, the skill requirements to operate
the system were minimal, with a typical daily demand
on the operator of 30 minutes. The operator performed
activities such as cleaning the turbidimeter photocell,
monitoring backwash operational issues, and working
with the vendor to troubleshoot and perform on-site
repairs. Remote monitoring of the treatment system by
the vendor was effective in troubleshooting problems.
Based on this time commitment and a labor rate of
$21/hour, the labor cost was $0.22/1,000 gallons of
water treated. The total O&M cost was approximately
$0.29/1,000 gallons.
Table 6. O&M Costs
Cost Category
Volume processed
(1,000 gallons)
Ferric Chloride Unit Price
($/pound)
Ferric Chloride
Consumption Rate
(pounds/1,000 gallons)
Chemical cost ($71,000
gallons)
Value Assump
From 08/16/04
through 08/12/05
$0.40
35% ferric chloride
in a 600-lbdrum.
0.065 80 gallons or 900
pounds annually
Power use ($/1,000
gallons)
$0.03
$0.04
Labor (76%)
Based on
additional costs
after treatment
plant startup.
Average weekly labor
(hours)
Labor cost ($71,000 gallons)
Total O&M Cost
($1,000 gallons)
2.5
30 minutes/day;
5 days/week
Labor rate = $21/
hour
$0.29 —
$0.22
Conclusions
The Climax, MN demonstration project confirmed
that iron removal is an effective way to remove arsenic
from water. Additionally, when natural iron levels are
insufficient for desired arsenic removal, ferric iron can be
added to the water after the oxidant feed (for arsenic [III]
oxidation). Removing iron from source water improves
water quality in the distribution system and reduces
flushing frequency.
Battelle submitted the full report in fulfillment of EPA
Contract 68-C-00-185, Task Order 0019.
References
Chen, A., L. Wang, J. Oxenham, and W. Condit. 2004.
Capital Costs of Arsenic Removal Technologies: U.S. EPA
Arsenic Removal Technology Demonstration Program Round
1. EPA/600/R-04/201. U.S. EPANRMRL, Cincinnati, OH.
Lytle, D. 2005. Coagulation/Filtration: Iron Removal Processes
Full-Scale Experience. EPA Workshop on Arsenic Removal
from Drinking Water. Cincinnati, Ohio. August 16-18.
U.S. Environmental Protection Agency. 2001. National
Primary Drinking Water Regulations: Arsenic and
Clarifications to Compliance and New Source
Contaminants Monitoring. Federal Register, 66:14:6975.
January 22.
U.S. Environmental Protection Agency. 2003. Minor
Clarification of the National Primary Drinking Water
Regulation for Arsenic. Federal Register, 40 CFR Part 141.
March 25.
-------
f
I
V \t
&.
\
LU
C9
Please make my changes on the below label,detach or copy,
and return to the address in the upper left-hand corner.
If you do not wish to receive these reports CHECK HERED;
detach, or copy this cover, and return to the address in the
upper left-hand corner.
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
National Risk Management
Research Laboratory
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
EPA/600/S-06/152
February 2007
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free.
------- |