EPA/600/R-08/051
                                                                April 2008
    Assessing Arsenic Removal by Metal (Hydr)Oxide
Adsorptive Media Using Rapid Small Scale Column Tests


                                by

                  Paul K. Westerhoff and Troy M. Benn

                       Arizona State University
                      Tempe, Arizona 85287-5306

                               and

                         Abraham S.C. Chen
                             Lili Wang
                         Lydia J. Gumming

                              Battelle
                      Columbus, OH 43201-2693
                       Contract No. 68-C-00-185
                         Task Order No. 0025
                                for

                           Thomas J. Sorg
                        Task Order Manager

               Water Supply and Water Resources Division
             National Risk Management Research Laboratory
                        Cincinnati, OH 45268
             National Risk Management Research Laboratory
                   Office of Research and Development
                  U.S. Environmental Protection Agency
                        Cincinnati, OH 45268

-------
                                       DISCLAIMER
The work reported in this document is funded by the United States Environmental Protection Agency
(EPA) under Task Order (TO) No. 0025 of Contract No. 68-C-00-185 to Battelle. It has been subjected to
the Agency's peer and administrative reviews and has been approved for publication as an EPA document.
Any opinions expressed in this paper are those of the author(s) and do not, necessarily, reflect the office
positions and policies of the EPA.  Any mention of products or trade names does not constitute
recommendation for use by the EPA.

-------
                                         FOREWORD
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the Nation's
land, air, and water resources. Under a mandate of national environmental laws, the Agency strives to
formulate and implement actions leading to a compatible balance between human activities and the ability
of natural systems to support and nurture life. To meet this mandate, EPA's research program is
providing data and technical support for solving environmental problems today and building a science
knowledge base necessary to manage our ecological resources wisely, understand how pollutants affect
our health, and prevent or reduce environmental risks in the future.

The National Risk Management Research Laboratory (NRMRL) is the Agency's center for investigation
of technological and management approaches for preventing and reducing risks from pollution that
threaten human health and the environment.  The focus of the Laboratory's research program is on
methods and their cost-effectiveness for prevention and control of pollution to air, land, water, and
subsurface resources; protection of water quality in public water systems; remediation of contaminated
sites, sediments and ground water; prevention and control of indoor air pollution; and restoration of
ecosystems.  NRMRL collaborates with both public and private sector partners to foster technologies that
reduce the cost of compliance and to anticipate emerging problems. NRMRL's research provides
solutions to environmental problems by: developing and promoting technologies that protect and improve
the environment; advancing scientific and engineering information to support regulatory and policy
decisions; and providing the technical support and information transfer to ensure implementation of
environmental regulations and strategies at the national, state, and community levels.

This publication has been produced as part of the Laboratory's strategic long-term research plan. It is
published and made available by EPA's Office of Research and Development to assist the user
community and to link researchers with their clients.

                                                    Sally Gutierrez, Director
                                                   National Risk Management Research Laboratory
                                               in

-------
                                         ABSTRACT
The lowering of the maximum contaminant level (MCL) for arsenic in drinking water from 50 to 10 (ig/L
has posed significant technical and financial challenges to water treatment facilities throughout the nation.
To assist small water systems (< 10,000 customers) in meeting the new standard, U.S. Environmental
Protection Agency (EPA) announced in October 2001 an initiative, i.e., the Arsenic Rule Implementation
Research Program, to conduct, among others, a series of full-scale, on-site demonstrations of arsenic
removal technologies, process modifications, and engineering approaches applicable to small systems. Of
the 40 project sites under the Round 1 and Round 2 demonstration program, 23 selected adsorptive media
technology because of its ease of operation.

The conventional way of selecting adsorptive media has been based on the results of long-term pilot-plant
studies. To reduce time required and save cost, it was desirable to develop new or utilize existing rapid,
small-scale methods to evaluate media performance.  Preliminary studies have been recently conducted
using a rapid small-scale column test (RSSCT) method that was originally developed for evaluating  the
performance of granular activated carbon. The results of these studies have shown that the RSSCT
method, which usually requires only three to four weeks of testing, has the potential to predict  the
performance of full-scale systems. If proven to be  true, this method would provide the water industry
with a lower cost alternative to develop performance data necessary for full-scale system design.

Battelle was contracted by EPA to evaluate the usefulness of this short-term predictive method. Side-by-
side tests were conducted using RSSCTs and pilot/full-scale systems either in the field or in the
laboratory.  The test locations included six EPA arsenic removal technology demonstration sites and one
EPA pilot-scale test site. For each location, RSSCTs were conducted using at least three parallel test
columns packed with different adsorptive media to compare arsenic breakthrough of the small-scale
columns to the pilot/full-scale systems.

A total of eight commercially available, NSF International (NSF)-certified adsorptive media were tested,
including three iron oxide-based media (i.e., E33, ARM 200, and Kemlron), one iron hydroxide-based
media (i.e., granular ferric hydroxide or GFH), one iron modified activated alumina (i.e., AAFS50),  two
titania-based media (i.e., MetsorbG and AdSorbsia granular titania oxide [GTO]), and one hybrid ion
exchange resin (HlX)-based media (i.e., ArsenXnp). Virgin media were crushed and sieved to obtain the
100 x 140 mesh fraction, which was packed in 1.1  cm x 30.5 cm glass columns. The amount of media
packed into each column was determined through the use of proportional diffusivity scaling equations
(see Equations 1.1 and 1.2 in Table 1-1).  The columns were thoroughly backwashed to remove media
fines and entrained air before use.

Of the media tested, full-scale performance data were available for direct comparison for AAFS50, E33,
GFH, and ArsenXnp. RSSCTs proved to be a reasonably reliable approach for comparing media run
lengths and adsorptive capacities for arsenate (As[V]), but over-predicted the  capacities to remove
arsenite (As[III]).  Key operational parameters, including reduced Reynolds-Schmidt product (ReSc)
values and empty bed contact time (EBCT), were evaluated to minimize the run time and volume of water
required to conduct RSSCTs. Under the conditions tested for most media, an ReSc value of 2,000
appeared to be appropriate for RSSCT column design. A reduced ReSc value of 1,000 was needed for
titania-based media to cope with operational difficulties related to excessive pressure buildup,  bed
compaction (up to 50%), and media disintegration. RSSCT columns scaled to a reduced full-scale EBCT
of 2.5 to 3.0 min could produce similar results as those scaled to the whole-length full-scale EBCTs.

Water quality at each test site varied and, consequently, the ability to remove arsenic by a given media
also varied. Arsenic occurred as As(V) at concentrations of 21.5 to 61 |o,g/L in five of the seven source
                                               IV

-------
waters tested and pH values were between 7.2 and 7.7. The two source waters containing As(III) had pH
values of 8.1 and 8.5 and As(III) concentrations of 64 and 22.5 |o,g/L. (One site utilized a solid-phase
oxidizing media to convert the As(III) into As(V) and RSSCTs were conducted to study both As(V) and
As(III) removal.)  All the waters contained arsenic plus at least one other important element of interest,
such as uranium, antimony, vanadium, silica, and/or phosphate. Arsenic adsorption capacities across the
different source waters and different media tested, based on reaching an RSSCT effluent concentration of
10 |o,g/L, ranged from 0.05 to 2.0 mg As/g of dry media. The influent pH value as well as the presence of
silica, vanadium, phosphate, and/or calcium all appeared to impact the relative performance of media on
different source waters. By conducting side-by-side RSSCTs, significant differences were observed
among the media to  remove arsenic and these co-occurring elements.

-------
                                      CONTENTS

DISCLAIMER	ii
FOREWORD	iii
ABSTRACT	iv
FIGURES	vii
TABLES	viii
ACRONYMS AND ABBREVIATIONS	ix
ACKNOWLEDGEMENTS	xii

1.0  INTRODUCTION	1
    1.1  Background	1
        1.1.1   Arsenic Interaction with Metal (Hydr)Oxide Surfaces	1
        1.1.2   Mass Transport of Ions in Porous Adsorptive Media	3
        1.1.3   Rapid Small Scale Column Tests for Arsenic Removal by Porous Adsorbents	3
    1.2  Objectives	5
    1.3  Report Organization	5

2.0  SUMMARY AND CONCLUSIONS	7

3.0  MATERIALS AND METHODS	9
    3.1  RSSCT Apparatus	9
    3.2  Preparation of RSSCT Adsorptive Media	9
    3.3  Preparation of RSSCT Columns	9
    3.4  RSSCT Field Setup	14
    3.5  RSSCT Laboratory Setup	14
    3.6  Sampling and Analyses	16

4.0  RESULTS AND DISCUSSION	19
    4.1  Source Water Quality	19
    4.2  Valley Vista, AZ (Site 1)	19
        4.2.1   Field RSSCTs (Site IF)	19
        4.2.2   Laboratory RSSCTs (Site 1L)	21
        4.2.3   Comparison of Field/Laboratory RSSCT and Full-Scale System Results	24
    4.3  Rimrock, AZ (Site 2)	24
        4.3.1   Arsenic Removal	24
        4.3.2   Removal of Other Elements	26
    4.4  Licking Valley High School in Newark, OH (Site 3)	26
        4.4.1   Comparison of As(III) Removal by RSSCT and Pilot-Scale Columns	26
        4.4.2   Removal of Other Elements	28
    4.5  Village of Lyman in NE (Site 4)	28
        4.5.1   Arsenic Removal	30
        4.5.2   Uranium Removal	30
        4.5.3   Removal of Other Elements	30
    4.6  Upper Bodfish in Lake Isabella, CA (Site 5)	30
        4.6.1   Arsenic Removal	33
        4.6.2   Uranium Removal	33
    4.7  South Truckee Meadows General Improvement District (STMGID) in Washoe County,
        NV(Site6)	36
        4.7.1   Arsenic Removal	36
        4.7.2   Antimony Removal	38
                                           VI

-------
         4.7.3  Removal of Other Elements	39
    4.8   Spring Brook Mobile Home Park in Wales, ME (Site?)	41
         4.8.1  Arsenic Removal	41
         4.8.2  Removal of Other Elements	43
    4.9   Media Adsorptive Capacity for Arsenic	46
    4.10  Comparison of Arsenic Removal for Different Source Waters	46

5.0  REFERENCES	49
                                        FIGURES

Figure 1-1.   Arsenate (As[V]) and Arsenite (As[III]) Speciation	2
Figure 3-1.   Illustration of Stand-Alone RSSCT Apparatus Dimensions (upper) and Layout with
            Column and Pump (lower)	10
Figure 3 -2.   Schematic (upper) and Photograph (lower) of Typical Packed Columns	11
Figure 3-3.   Photographs of Water Delivery System (Including Inlet Piping, Fiber Glass Filter
            Cartridge [top] and Five-Port Manifold [bottom]) for Licking Valley High School
            RSSCT Columns	15
Figure 3-4.   Photograph of RSSCT System and Sampling Points at Spring Brook Mobile Home
            Park in Wales, ME	16
Figure 3-5.   Schematic Arrangement for RSSCT System and Sampling Points	17
Figure 4-1.   As Breakthrough Curves from Field RSSCT Columns at Valley Vista, AZ	21
Figure 4-2.   Breakthrough of V, P, Si, and Al from Field RS SCT Columns at Valley Vista, AZ	22
Figure 4-3.   As Breakthrough Curves from Valley Vista, AZ Laboratory RSSCT Columns	23
Figure 4-4.   Comparison of Field and Laboratory RS SCT Results for Valley Vista, AZ	23
Figure 4-5.   Comparison of Field/Laboratory RSSCTs with Full-Scale AAFS50 System at
            Valley Vista, AZ	24
Figure 4-6.   As Breakthrough Curves from Laboratory RSSCT Columns and Full-Scale E33
            System at Rimrock, AZ	25
Figure 4-7.   Breakthrough of V, P, Si, and Al from Rimrock, AZ Laboratory RSSCT Columns	27
Figure 4-8.   Comparison of RSSCT and Pilot Results at Licking Valley High School in
            Newark, OH	28
Figure 4-9.   Breakthrough of P, Si, Fe, and Al from Field RSSCT Columns at Licking Valley
            High School in Newark, OH	29
Figure 4-10.  Arsenic Breakthrough Curves from Lyman, NE Laboratory RSSCT Columns	31
Figure 4-11.  Comparison of Two ReSc Values on Lyman, NE Titania-Based Media RSSCTs	31
Figure 4-12.  Breakthrough of U, V, Fe, and Mn from Lyman, NE Laboratory RSSCT Columns	32
Figure 4-13.  Arsenic Breakthrough Curves from Upper Bodfish,  CA Laboratory RSSCT
            Columns	34
Figure 4-14.  Comparison of Two ReSc Values on Upper Bodfish, CA Titania-Based Media
            RSSCTs	34
Figure 4-15.  Uranium Breakthrough from Upper Bodfish, CA Laboratory RSSCT Columns	35
Figure 4-16.  Comparison of Two ReSc Values on Upper Bodfish, CA Titania-Based Media
            RSSCTs	35
Figure 4-17.  Arsenic Breakthrough Curves from STMGID, NV Laboratory RSSCT Columns	37
Figure 4-18.  Comparison of GFH RSSCT and Full-Scale System Results for STMGID in
            Washoe County inNV	37
Figure 4-19.  Antimony Breakthrough Curves from STMGID, NV Laboratory RSSCT Columns	39
Figure 4-20.  Breakthrough of V, P, Si, and Fe from STMGID, NV Laboratory RSSCT Columns
            Scaled to a Full-Scale EBCT of 3 min	40
                                           vn

-------
Figure 4-21.  As(V) Breakthrough Curves from Field RSSCT Columns at SBMHP in Wales, ME
            (Parti)	42
Figure 4-22.  As(V) Breakthrough Curves from Field RSSCT Columns at SBMHP in Wales, ME
            (Part 2)	42
Figure 4-23.  As(III) Breakthrough Curves from Field RSSCT Columns at SBMHP in
            Wales, ME	43
Figure 4-24.  Breakthrough of Si and P from Field As(V) RSSCT Columns at SBMHP in
            Wales, ME	44
Figure 4-25.  Breakthrough of Si and P from Field As(III) RSSCT Columns at SBMHP in
            Wales, ME	45
Figure 4-26.  Total and Soluble Ti Breakthrough Curves from As(III) RSSCT Columns at
            SPMHP in Wales, ME	46
                                         TABLES

Table 1-1.  RSSCT Scaling Equations	4
Table 1-2.  RSSCT Site-Specific Study Objectives	6
Table 3 -1.  Physical Properties of Eight Commercially-Available Adsorptive Media Evaluated	12
Table 3-2.  Design Conditions of Pilot- and Full-Scale Systems and Corresponding RSSCT
           Columns	13
Table 3-3.  Summary of Sampling Frequency and Analysis	17
Table 3-4.  Analytical Methods,  Sample Volumes,  Containers,  Preservations, and  Holding
           Times	18
Table 4-1.  Average Influent Water Quality Data	20
Table 4-2.  Throughput Before Reaching Arsenic and Uranium MCLs for Upper Bodfish
           Ground-water in Lake Isabella, California	36
Table 4-3.  Throughput Before Reaching Arsenic and Antimony MCLs for STMGID, NV
           Groundwater	39
Table 4-4.  Comparison of Throughput for As(III) and As(V) Removal from SBMHP
           Groundwater in Wales, ME	43
Table 4-5.  Summary of Arsenic Adsorption  Capacities	47
Table 4-6.  Summary of Media Life at Six Arsenic Demonstration Sites	48
                                            Vlll

-------
                            ACRONYMS AND ABBREVIATIONS
Pb             apparent bulk density (g/L)
Al             aluminum
As             arsenic
As(III)         arsenite
As(V)         Arsenate
As0            initial arsenate concentration (|Jg/L)
As             final arsenate concentration (ng/L)
Asb            arsenate concentration in the bulk solution (ng/L)
Ass            arsenate concentration at solid/liquid interface
ASU          Arizona State University (Tempe, AZ)
ATS           Aquatic Treatment Systems
Bi

BV
Biot number =-

bed volume
CAs            equilibrium concentration between solid and liquid phase (|o,g/mg)
Ca            calcium
Cl"            chloride

dp             media particle diameter (cm)
Dp            pore diffusion coefficient (cm2/sec)
Ds            surface diffusion coefficient (cm2/sec)
DL            free liquid diffusivity (cm2/sec)

E33           granular ferric (hydr)oxide (product of Severn Trent Services)
EBCT         empty bed contact time
EPA           United States Environmental Protection Agency

Fe            iron

GAC          granular activated carbon
GFH          granular ferric hydroxide (product of Siemens, formally US Filter)
gpm           gallons per minute
GTO          granular titania oxide (product of DOW Chemical)

HC1           hydrochloric acid
HOPE         high-density polyethylene
HIX           hybrid ion exchange
FiNO3         nitric acid

ICP-MS        inductively coupled plasma-mass spectrometry

ID            identification

K             Freundlich coefficient
Kf            external mass transfer coefficient (cm/sec)
                                              IX

-------
LVHS
Licking Valley High School
m
MCL
MDL
Mn

n
NO3
NOM
NSF
mass of the media
maximum contaminant level
method detection limit
manganese

Freundlich coefficient
pitrate
natural organic matter
NSF International
P
PO43
PVC

tfavg
QA/QC
QAPP
RP
Re

ReSc
RPD
RSSCT
phosphorous
phosphate
polyvinyl chloride

average adsorptive capacity (|o,g/mg)
adsorptive capacity (|o,g/mg)
maximum solid phase concentration (|o,g/mg)
quality assurance/quality control
quality assurance project plan

radial coordinate from the center of the particle
radius of particle (cm)
Reynolds Number =
                  2 * Rp *p* v
Reynolds-Schmidt Number
relative percent difference
rapid small scale column test
Sh

Sc

St

Sb
SBMHP
Si
SiO2
SO42
STMGID
                   2 * R  * K
Sherwood Number = - - - —
                     "mol

Schmidt Number = — — —
                 P*Dmol
                K * (1— e ) * T
Stanton Number = — -
antinomy
Spring Brook Mobile Home Park
silicon
silica
sulfate
South Truckee Meadows General Improvement District
t
TO
TOC

U
uss
time
task order
total organic carbon

uranium
US Sieve Size

-------
e              void fraction
V              vanadium
H              viscosity
Vr             volume of the reactor (L)
v              interstitial velocity (cm/sec)
                                                 XI

-------
                                 ACKNOWLEDGEMENTS
We wish to thank the water operation staff of several water systems for their support in obtaining water
samples for the study. The water systems involved included Valley Vista and Rimrock with Arizona
Water Company in Arizona, Licking Valley High School (LVHS) in Ohio; Village of Lyman in
Nebraska; Upper Bodfish (Lake Isabella) with California Water Service Company (Cal Water) in
California; South Truckee Meadows General Improvement District (STMGID) in Nevada, and Spring
Brook Mobile Home Park (SBMHP) in Wales, Maine. Mohammad Badruzzaman worked as a graduate
student at the Arizona State University (ASU) at three of the sites. Inductively Coupled Plasma-Mass
Spectrometry (ICP-MS) analytical assistance for samples collected by ASU was provided by Panjai
Prapaipong in Professor Shock's Laboratory at ASU.
                                            xn

-------
                                     1.0  INTRODUCTION
Mini-columns were scaled to mimic the performance of continuous-flow, full- and pilot-scale arsenic
treatment systems using a rapid small scale column test (RSSCT) approach originally developed for
removal of organic compounds by granular activated carbon (GAC) and recently verified for arsenic
removal by porous metal (hydr)oxides (Crittenden et al., 1991; Badruzzaman et al. 2004; Westerhoff et al.,
2005). Several commercially available adsorptive media including iron-, alumina-, titania-, and hybrid
ion exchange resin-based products were evaluated in parallel RSSCT columns using a common influent
water source. Water for the testing was obtained from sites where the U.S. Environmental Protection
Agency (EPA) had installed or planned to install adsorptive media arsenic removal systems as part of the
Arsenic Treatment Technology Demonstration Program (Wang et al., 2004) under EPA's Arsenic Rule
Implementation Research Program.  This report describes the fabrication and testing of the RSSCTs and
results from their use for removing arsenic and other inorganic constituents/co-contaminants from
groundwater at seven different sites.

1.1        Background

Battelle is currently contracted by EPA to conduct 50 full-scale, long-term, on-site demonstrations of
arsenic removal technologies applicable to small systems.  This project is part of an EPA initiative to
assist small community water systems (< 10,000 customers) in complying with the new arsenic standard.
Twenty-three of the 40 demonstration sites use adsorptive media arsenic removal technology, which was
selected because of its ease of use for treating water in small systems.  The conventional means for
selecting adsorptive media has been based upon the results of long-term pilot-plant studies. Because the
time required for evaluating the capacities of new adsorptive media usually is lengthy (e.g., nine to 12
months), the cost for the pilot plant studies can be substantial.  To reduce the  cost and time to evaluate
media performance, several preliminary research studies have been conducted using the RSSCT method
for arsenic removal (Thomson and Anderson, 2004; Westerhoff et al., 2005).  The results of these
preliminary studies have  shown that the RSSCT method, which usually requires only three to four weeks
of testing, has the potential to predict the performance of full-scale systems. If this proves  true, the
method would provide the water industry with a lower cost method to develop performance data
necessary  for full-scale design of arsenic adsorptive media systems.  To evaluate the usefulness of this
short-term predictive method, side-by-side tests  using RSSCTs and pilot/full-scale systems are required.

1.1.1       Arsenic Interaction with Metal (Hydr)Oxide Surfaces.  Arsenate (H3AsO4,  H2AsO4,
HAsO42", or AsO43") is the dominant species in oxygenated water, present in anionic forms  over the pH
range of 2 to  14 (Figure 1-1).  Over the pH range of most groundwater, arsenate is present as H2AsO4" and
HAsO42".  Arsenite occurs under more reducing conditions and is present in a non-ionic form (H3AsO3)
below pH around  9.2 (McNeill and Edwards, 1997a; McNeill and Edwards, 1997b). Chlorine,
permanganate, or  ozone readily oxidizes arsenite to arsenate, while oxidation by monochloramine,
chlorine dioxide, or oxygen is less effective (Ghurye and Clifford, 2001).

Arsenic (i.e.,  arsenate and arsenite) can associate with iron surfaces either by  forming inner-sphere or
outer-sphere complexes (Goldberg and Johnston, 2001; Raven et al., 1998; Wilkie and Hering, 1996).
Surface chemistry is important in arsenic removal by metal oxides.  The surfaces of metal oxides are
collections of unfilled metal-oxygen bonds that hydrate in water. Electrostatic attraction of anionic
species is favored onto positively charged surface sites. At the pH of zero point of charge (pHZPC), an
equal number of positively and negatively charged surface sites exist, and  proportionally more positively
charged surface sites at pH levels below pHZPC.  Therefore, pHZPC is one indicator for the potential of
removing anionic  arsenic species. Iron (hydr)oxides have pKa] and pKa2 values of ~7.3 and 8.9,
respectively, resulting in a pHZPC on the order of 8.0.  For aluminum and titanium oxides, the pHZPC

-------
                128
                                             PH
                 120
                  Figure 1-1.  Arsenate (As[V]) and Arsenite (As[III]) Speciation
values are around 7. Anionic arsenic species are generally removed more effectively than non-ionic
arsenic species by most metal-oxide adsorbents.

Columbic forces favor association of anionic arsenate with positively charged sites on metal oxides (e.g.,
MeOH2+). In addition to electrostatic bonding, arsenic also forms with some surfaces covalent bonds
including monomolecular monodentates and bidentates.  Whereas electrostatic bonds form rapidly (in

-------
seconds) and depend largely on the charge difference between arsenate and the media surface, covalent
bonds form less rapidly and depend on their respective molecular structure.  Covalent bonds are
irreversible and stronger than electrostatic attractions.  As covalent bonds form, surface sites can become
available for electrostatic bonding again. The kinetics of bond formation may affect the optimal contact
time required for a specific media in a column operation.

Silica is a major anion that exerts significant impact on arsenate removal by porous metal-oxide
adsorptive media (Smith and Edwards, 2005). A few batch and  column studies document that silica
reduces arsenic adsorptive capacity of ferric oxides/hydroxides and activated alumina (Meng et al., 2002;
Meng et al., 2000).  Several mechanisms have been proposed to describe the role of silica in iron-silica
and iron-arsenic-silica systems, including that (1) adsorption of silica may change the surface properties
of adsorbents by lowering the iso-electric point or pH^; (2) silica may compete for arsenic adsorption
sites; (3) polymerization of silica may accelerate silica sorption and lower the available surface sites for
arsenic adsorption; and (4) reactions of silica with divalent cations such as calcium, magnesium and
barium may form precipitates.

1.1.2      Mass Transport of Ions in Porous Adsorptive Media. Iron, aluminum, titanium,
zirconium, and other metal oxide-based adsorptive media have been manufactured to remove arsenic from
water supplies over the past decade. Most of these media have surface areas over 100 m2/g and have a
continuum of micro- and macro-pores (Badruzzaman et al., 2004).  Mass transport of arsenic from
solution onto these porous adsorptive media has been described  by film diffusion and intraparticle surface
and/or pore diffusion (Badruzzaman et al., 2004; Lin and Wu, 2001).  Despite the formation of strong
bonds between arsenic and metal oxides, surface diffusion is still possible (Axe and Trivedi, 2002).
However, intraparticle transport is probably a combination of surface and pore diffusion.

Intraparticle mass transport is believed to be the rate limiting step in adsorption removal of arsenic. As a
result, adsorbed arsenic concentrations (i.e., mg As/g of adsorbent) are the highest on the external surface
of an adsorbent particle and decline towards the center of the particle. The concentration gradient causes
arsenic to migrate from the external surface into the porous particle. Over time, adsorbed arsenic
concentrations in the particle increase, thus decreasing the concentration gradient and, therefore,
decreasing the arsenic removal from solution. In a packed-bed adsorptive media system, a mass transfer
zone, where active adsorption occurs, is established initially at the inlet of the bed and gradually migrates
deeper into the bed. Arsenic eventually breaks through the bed with increasing concentrations in the
system effluent.

1.1.3      Rapid Small Scale Column Tests for Arsenic Removal by Porous Adsorbents.
Procedures have been developed and applied over the  last two decades for RSSCTs that simulate pilot-
and full-scale performance of GACs for organic micropollutants and natural organic matter (NOM)
removal. RSSCT bench-scale testing is a method where dimensionless mathematical parameters are used
to scale down a pilot- or a full-scale adsorber based on applicable adsorbate  transport mechanisms.  The
advantage of RSSCTs is that breakthrough curves can be obtained in a fraction of time with a fraction of
water that would be required for pilot-scale testing. Theoretically, RSSCT and pilot/full-scale adsorbers
would produce identical breakthrough curves, but in reality, they differ based on water quality, biological
processes, and/or RSSCT scaling assumptions.

In the development of scaling equations, three conditions are required in order to maintain similarity
between large- and small-scale systems (Crittenden et al., 1991; Crittenden et al.,  1987; Crittenden et al.,
1986). First, boundary conditions for the large- and small-scale systems must occur at the  same
dimensionless coordinate values in dimensionless differential equations. Second, dimensionless
parameters  in the differential equations must be equal  for the large- and small-scale systems. Finally, no

-------
                              Table 1-1. RSSCT Scaling Equations
                Scaling Assumption
                                  Relationships
Equation
  No.
Proportional Diffusivity
(PD:x= 1)
                                      DLC
                        EBCTS>
                        EBCT,
                                                 d
                                                   p,LC
                                                                             1.1
                                                                             1.2
Constant Diffusivity
(CD:x = 0)
                                      EBCTS(
                                      EBCT,,
                                      V
                                                 •* p,LC
                                                                             1.3
                                                                             1.4
              General Relationships
                        Re =


                        Sc = -

                                              p,LC
                                      RescxSc
                                      ReLCxSc
                                              JU
                                                                             1.5
                                                                             1.6
                                                                             1.7
               Note: Subscripts "SC" and "LC" indicating small column (i.e., RSSCT column)
                 and large column (i.e., pilot/full-scale column), respectively.
               EBCT = empty bed contact time    pL = liquid density
               dp = media diameter              |a = viscosity
               t = time                         DL = liquid diffusivity of arsenic
               V = loading rate                  D = effective surface diffusivity
               Re = Reynolds number            ReSc = Reynolds-Schmidt product
               Sc = Schmidt number
change in mechanism may occur while reducing the size of the system. RSSCTs may be designed using
equations 1.3 and 1.4 in Table 1-1 if the effective surface diffusivity is independent of particle size and,
therefore, identical between the large-scale and RSSCT columns. If the effective surface diffusivities are
not identical between the large-scale and RSSCT columns, perfect similarity cannot be guaranteed.
However, if it is assumed that the effective surface diffusivity is linearly proportional to the particle
radius and that the intraparticle diffusion is the controlling mechanism, an RSSCT may be designed using
equations 1.1 and 1.2 and perform similarly to the large-scale column.  The Reynolds number, Re, is a
dimensionless ratio of the inertial forces over the viscous forces in a fluid and the Schmidt number,  Sc, is
a dimensionless ratio of the diffusion of momentum over the diffusion of mass.  The product of the
Reynolds number and the Schmidt number (ReSc) may be used to determine the minimum Reynolds
number for the RSSCT such that the effects of dispersion are not important (dispersion is not important if

-------
ReSc is in the mechanical dispersion region from 200,000 to 200). Equations 1.5 through 1.8 are used to
reduce loading rates in the small column and minimize potential for bed compaction, while maintaining a
minimum Reynolds-Schmidt product.

Both proportional and constant diffusivity scaling relationships have recently been applied for arsenic
removal by porous metal (hydr)oxide adsorptive media (Badruzzaman, 2005; Badruzzaman and
Westerhoff, 2005; Sperlich et al, 2005; Westerhoff et al., 2005; Badruzzaman et al., 2004). Each
research group appears to make different assumptions and scale to different types of larger scale loading
rates and other design parameters. Based upon ASU's work with multiple comparisons between pilot-
scale and RSSCT arsenic breakthrough curves, proportional diffusivity scaling equations appear to
accurately mimic performance of larger-scale columns without leading to excessive pressure development
within the RSSCT columns, as is the case with constant diffusivity-based RSSCTs. As such, proportional
diffusivity scaling relationships were used in this study.

The size of an RSSCT column is based on the  scaling approach (constant or proportional diffusivity) and
the ratio of the adsorptive media size used in the large-scale column to that in the RSSCT column.
Selecting smaller media sizes or a lower ReSc value for the RSSCT columns would decrease the volume
of water and duration of the test. However, smaller media sizes can significantly increase pressure drop
across the RSSCT columns. Excessive pressure  accumulation can compress "softer" media and lead to
failure of the RSSCT testing. In previous work,  100  x 140 and 140 x  170 mesh  sizes performed well for
most media (Westerhoff et al., 2005).  Another assumption for the use of RSSCT is that the media is
homogeneous. Therefore, crushing the media  would not alter the mechanisms governing the adsorption
on the uncrushed media.  This assumption appears to be valid for most of the commercially available
metal (hydr)oxide adsorptive media.

1.2        Objectives

The overall objective of this project is to evaluate the usefulness of RSSCT by performing side-by-side
comparison between the results of RSSCT and pilot- or full-scale adsorptive media systems either in the
field or in the laboratory.  The test locations included six EPA arsenic removal technology demonstration
sites (i.e., Valley Vista and Rimrock in Arizona; Village of Lyman in Nebraska; Upper Bodfish in Lake
Isabella, California; South Truckee Meadows General Improvement District [STMGID] in Washoe
County, Nevada; and Spring Brook Mobile Home Park [SBMHP] in Wales, Maine) and one pilot-scale
test site at Licking Valley High School (LVHS) in Newark, Ohio. The LVHS test site was selected
because extensive pilot-scale tests had been performed for various commercially available adsorptive
media and because the groundwater at LVHS contained high levels of arsenite (i.e., 60 to 80  (ig/L),
whereas the other technology demonstration sites had mostly arsenate. For each location, RSSCT tests
were conducted using three to four parallel columns packed with different adsorptive media.  Specific
objectives were developed for each site as shown in Table 1-2.

1.3        Report Organization

This report is organized into four chapters.  Chapter 1 is the introduction. Chapter 2 summarizes the
conclusions from the study. Chapter 3 describes the methodologies for sizing, preparing, and operating
RSSCTs, and for sampling and analyses. Chapter 4 presents the key results from the study.

-------
Table 1-2. RSSCT Site-Specific Study Objectives
Site No.
Sitel
Site 2
SiteS
Site 4
SiteS
Site 6
Site?
Location
Valley Vista, Arizona
Rimrock, Arizona
LVHS in Newark,
Ohio
Village of Lyman,
Nebraska
Upper Bodfish in
Lake Isabella,
California
STMGIDinWashoe
County, Nevada
SBMHP in Wales,
ME
Objectives
1) To predict the performance of the full-scale AAFS50 system for arsenic
removal, 2) to compare AAFS50 arsenic removal capacity to that of two iron-
based media, i.e., E33 and GFH, 3) to evaluate the validity of conducting
RSSCTs in the laboratory - a controlled environment - by collecting and
transporting water from the field, and 4) to examine the use of a lower ReSc
value (i.e., 500 versus 2,000) for RSSCT column design
1) To compare the performance of laboratory RSSCT columns against that of
the full-scale E33 system, 2) to compare E33, AAFS50, GFH, and ArsenXnp
for arsenic removal, and 3) to examine the use of a lower ReSc value (i.e.,
1,000 versus 2,000) for RSSCT column design
To compare results of RSSCT with those of several previously performed
pilot-scale tests on AAFS50, GFH, E33 and ArsenXnp for arsenite removal
1) To evaluate the removal of co-occurring uranium and arsenic by titania-
based and other media, and 2) to evaluate an approach (i.e., by reducing ReSc
values) that could help reduce operational problems associated with titania-
based media
To evaluate the removal of co-occurring uranium and arsenic by hybrid ion
exchange resin (HlX)-based media, ArsenXnp, and other media
1) To predict the full-scale GFH system performance for arsenic and
antimony removal, 2) to evaluate multiple adsorptive media for arsenic and
antimony removal, and 3) to examine the use of a shorter-than-full-scale
EBCT for RSSCT scaling (a technique that can significantly shorten the
required RSSCT run time)
1) To simulate the full-scale system, 2) to evaluate alternative media for full-
scale implementation, and 3) to evaluate the removal capacity of four
adsorptive media to remove arsenite

-------
                            2.0 SUMMARY AND CONCLUSIONS
The following summarizes the key findings from this study related to the use of RSSCTs to
predict/evaluate the ability of metal hydr(oxide) media to remove arsenic and other co-contaminants.

Key operational parameters, including reduced ReSc values and EBCT, were evaluated to minimize the
run time and volume of water required to conduct RSSCTs.  The ReSc parameter essentially fixes mass
transfer within the packed bed. All previous work had been conducted with an ReSc value of 2,000.
Decreasing the ReSc value would decrease the volume of water and run time required for the RSSCTs
and shorten the length of packed columns, hence leading to less pressure drop across the packed beds.
For iron-based media, decreasing the ReSc value from 2,000 to 1,000 or 500 resulted in pre-matured
arsenic breakthrough; therefore, a ReSc value of <2,000 would not be recommended. For titania-based
media, an ReSc value of 2,000 or 1,000 gave comparable results, thus a reduced value could be used to
cope with operational problems, such as excessive pressure buildup and bed compaction, due to inherent
media properties.

In addition to ReSc values, the selection of the full-scale EBCT to scale the RSSCT column would affect
the RSSCT run time and volume of water required.  In parallel tests, EBCTs of 6.3 and 3 min with GFH
resulted in nearly identical arsenic breakthrough curves when plotted again throughput.  In separate
parallel tests with titania-based media (MetsorbG), EBCTs of 5.3 and 2.5 min resulted in nearly identical
arsenic breakthrough curves. Along with related work on other projects, it appears that selection of a
shorter full-scale EBCT can be used for scaling RSSCT columns.

RSSCTs conducted in the  field at the well site corresponded reasonably well with those conducted in the
laboratory. The laboratory setting offered a more controlled environment.  Because water was generally
collected just one time for laboratory RSSCTs, these RSSCT tests had less varying, but less representative,
influent water quality than those conducted in field where influent water was refilled every one to three
days.

Water quality at each site tested varied and consequently the ability to remove arsenic by a given media
also varied. The removal of arsenic from water by adsorptive media was affected by the pH and presence
of other metals and competitive anions, such as silica and phosphorous. Arsenic occurred as As(V) at
concentrations of 21.5 to 61.0 |o,g/L and pH values between 7.2 and 7.7 in five of the seven source waters
tested. Two source waters containing As(III) had arsenic concentrations of 64 and 22.5 |o,g/L and pH
values of 8.1 and 8.5.

For As(V), the RSSCT method proved to be a reliable approach for comparing the  ability of different
adsorptive media for its removal.  RSSCT arsenic breakthrough curves corresponded reasonably well with
those  observed in full-scale systems installed as part of the EPA arsenic removal demonstrations in Valley
Vista and Rimrock in Arizona, the Village of Lyman in Nebraska, Upper Bodfish in Lake Isabella, CA,
and STMGID in Washoe County, Nevada.

For As(III), the number of bed volumes (BV) treated to reach 10 |o,g/L in the RSSCT effluent was notably
higher than that for As(V) and that of the pilot-scale system performance at LVHS in Newark, OH. The
results, however, did predict the relative performance of the media to one another based on the results of
the pilot tests conducted at LVHS. Additional work is needed to determine if the scaling equations that
are proven adequate for As(V) need to be modified for A(III) RSSCTs.

RSSCT may be used as  a standard testing protocol to assess media performance for different source
waters, compared to the inherent variability experienced  by pilot tests. For the six  demonstration

-------
locations, five source waters had pH values ranging from 7.2 to 7.7, while the other one had a much
higher pH value which averaged 8.6. All waters had varying concentrations of other constituents, which
could pose detrimental effects on arsenic removal. For example, the source water at STMGID was
considered the most difficult water to treat because of the presence of the highest phosphate and silica
concentrations in that water. The source water at Lyman, NE also was difficult to treat due to the
presence of the highest amount of vanadium, sulfate, and TOC and the second highest phosphorous
concentration.  The SBMHP site had a high pH (i.e., 8.7), which significantly reduced the arsenic removal
capacity.

The source water at STMGID contained arsenic and antimony, both of which were above their respective
MCLs of 10 and 6 |o,g/L.  Co-removal of antimony essentially did not occur, with antimony breaking
through before arsenic. Although operational problems were encountered, it appeared that the only media
that exhibited the potential for antimony removal was titania-based media. It was not clear if antimony
removal was inhibited by competing ions present or if the media tested, in fact, were not capable of
significant antimony removal.

The source water at Lyman and Upper Bodfish had levels of arsenic and uranium above the respective
MCLs.  The only media that exhibited removal capacity for uranium was ArsenXnp, which removed
uranium most likely via the anion exchange sites on its resin matrix.  Uranium began to break through
from the Lyman column after 20,000 BV and never broke through from the Upper Bodfish column as the
study ended at about 50,000 BV. The difference probably was due to the presence of competing anions  in
the Lyman water.

All media tested removed some vanadium.  GFH media removed more vanadium than any other media.
Phosphate was present at concentrations ranging from 15 to 162 |og/L.  All media removed some
phosphate. GFH media removed more phosphate than any other media. None of the media removed any
significant amount of the nitrate, chloride, fluoride or silica present in the waters under the conditions
tested.

-------
                              3.0  MATERIALS AND METHODS
This section discusses the materials and methods used for the design and operation of RSSCTs in the field
and laboratory.

3.1        RSSCT Apparatus

A series of stand-alone RSSCT apparatuses, each consisting of a glass column packed with adsorptive
media and a piston pump (Figure 3-1), were fabricated for the laboratory and field testing. The 1.1-cm
(inner diameter) x 30.5 cm (length) glass column (Ace Glass, Vineland, NJ) was used to house a media
bed packed between the top and bottom layers of glass wool and 5 mm-diameter borosilicate glass beads
and sealed by a Teflon® end cap on each end. The piston pump (Model QG150 or QG50, FMI Inc.,
Syosset, NY)  equipped with a stainless steel or ceramic pump head (Model Q2CSC, FMI Inc., Syosset,
NY) was used to deliver water via 3.2 mm Teflon® tubing.  The glass beads in the column helped disperse
the incoming flow from the top of the column and provide a base layer of support for the media bed.
Figure 3-2 shows a schematic of a packed RSSCT column and a photograph of four typical packed
columns.

3.2        Preparation of RSSCT Adsorptive Media

Table 3-1 provides the physical properties of eight adsorptive media evaluated in this study. The types of
media tested include iron oxide-based media (i.e., E33, ARM 200, and Kemlron), iron hydroxide-based
media (GFH), iron modified activated alumina (AAFS50), titania-based media (MetsorbG and GTO), and
hybrid ion exchange resin (HlX)-based media (ArsenXnp). Arithmetic means were used to calculate
respective media diameters unless noted. Virgin media (as received) were crushed using a mortar and a
pestle.  Crushed media were sorted using a series of stainless steel U.S. standard mesh sieves to obtain the
100 x 140 mesh fraction. The media in the 140 mesh sieve were rinsed with distilled water until fines no
longer filtered through the sieve.  The washed media in the 140 mesh sieve were transferred to Nalgene
bottles containing  distilled water and stored until use.

An assumption for the use of RSSCT is that the media is homogeneous. Crushing the media, therefore,
would not alter the adsorption mechanisms that govern the adsorption on the uncrushed media.  While it
is valid for most of the commercially-available metal (hydr)oxide media, this assumption may not be
valid for ArsenXnp, which is an iron-impregnated anion exchange resin. Recent work at ASU suggests
that iron concentrations  might be significantly higher near the exterior surface of resin beads than at the
center of the beads. As  such, crushing the ArsenXnp media for RSSCT would redistribute the  iron
concentrations, thus effectively "normalizing" the iron contents throughout the crushed fractions. It is
possible that higher iron contents near the exterior surface of the uncrushed beads could create a more
favorable adsorbed arsenic concentration gradient, thus causing better arsenic removal by a pilot- or full-
scale packed bed than by a RSSCT column.  It is also possible that some of the impregnated iron particles
were sloughed off from  the resin beads during media crushing and washed away along with media fines
during media washing, thereby causing less arsenic removal by an RSSCT column.

3.3        Preparation of RSSCT Columns

Each RSSCT column was prepared by placing an appropriate amount of a washed and sieved 100 x 140
mesh adsorptive media into a glass column containing layers of glass beads and glass wool in the bottom
of the column. The amount of media packed into the column and the corresponding bed depth was
determined through the use of proportional diffusivity scaling equations (i.e., Equations 1.1 and 1.2 in
Table 1-1).

-------
            MOUNTING HOLES SHALL BE
            LARGE ENOUGH TO ACCEPT
            Jt* HARDWARE (TYP. OF 6)
                                                              3/4"

                 -LAMINATED PLYWOOD TO HE
                  SECURLY ATTACHED BY ft"
                  HARDWARE
                                                                         FRAME SHALL BE
                                                                         CONSTRUCTED OF
                                                                         1" STEEL TUBING -
                                                                          1-2"
                                              SCALE;  Jt" - 1" (ON 8 J4" X 11" PAPER)
                          BATTELLE MEMORIAL INSTITUTE
                           VERIFICATION  OF
                      MINI-COLUMN ARSENIC TEST
 RSSCT APPARATUS
                                                       SCALE: AS NOTED QN
                                                                                 MSfZDNA S1ATE U
03/2004
1 OF 1
                                                                          _^?t**u
                                             OF RSSCT
                                          NOT TO SCALE

                         SAT1TLLE yEMORIAL INSHTUTE
                          VERIFICATION OF
                     MINI-COLUMN  ARSENIC TEST
RSSCT APPARATUS

SCALE AS NGTS3 DM DKAVilNS
OJ/Z004
1 OF 1
Figure 3-1. Illustration of Stand-Alone RSSCT Apparatus Dimensions (upper)
                     and Layout with Column and Pump (lower)
                                              10

-------
                               Direction of
                               flow
Teflon End Cap
        0-Ring


    Glass Beads
     Glass Wool

         Sieved
      Adsorbent
          Media

     Glass Wool
    Glass Beads
        0-Ring
Teflon End Cap
                                        1.1  x  30.5  cm
                                        Glass column
                                        «••—u^^^
   Figure 3-2. Schematic (upper) and Photograph (lower) of
                Typical Packed Columns
                          11

-------
   Table 3-1. Physical Properties of Eight Commercially-Available Adsorptive Media Evaluated
Media Type
Iron oxide-based
Iron hydroxide-based
Iron-modified,
alumina-based
Titania-based
Hybrid ion exchange
resin-based
Media
E33
ARM 200
Kemlron
CFH-12
GFH
AAFS50
MetsorbG
Adsorbsia
GTO
ArsenXnp
Supplier/
Manufacturer
Severn Trent
Services
Engelhard
Kemlron
Siemens
(formally USFilter)
Alcan
Graver/Hydroglobe
DOW Chemical
Purolite/Solmetex
Media Diameter
Product Literature
0.5-2.0 mm (10x35 USS)
0.42-1.68 mm (12x40 USS)
0.30-1.68 mm (12x50 USS)
1.0-2.0mm (10x18 USS)
0.32-2.0 mm (-10x50 USS)
0.3-0.6 mm (28x48 Tyler)
0.3-1. 19 mm (16x50 USS)
0.25-2.0 mm (10x60 USS)
0.3-1.2 mm (16x50 USS)
RSSCT
Design(a)
(mm)
1.16*
1.06
1.0
1.5
1.16
0.85*
0.67*
0.67*
0.75,
0.57(c)
BET
Surface
Area(b)
(m2/g)
133
(142)
NA
NA
259
(127)
267
(220)
125
70
(200-300)
82
USS = US Sieve Size, NA = Not Available
(a) Values calculated by taking arithmetic mean, except for those denoted by *
(b) Data in parentheses reported in product literature
(c) 0.57 mm used in laboratory Reno test and 0.75 mm used in others. ASU measured particle diameter to be 0.53
   to 0.61 mm with an arithmetic mean of 0.57 mm.
To prevent entrainment of air during column preparation, the media were transferred to the RSSCT
column filled with water. The media in the column were thoroughly backwashed until the effluent was
visibly clear of fines.  To compress the media, the exterior of the glass column was gently tapped while
the media particles settled. Additional media were added and backwashed until the desired bed depth was
achieved.  The mass of the media added to the column was estimated from the bulk density and the
volume of the packed bed. Layers of glass wool and glass beads were then placed on top of the column
before the top Teflon® end-cap was carefully attached.

The flowrate of each piston pump was calibrated prior to connecting it to a column. Distilled water was
pumped through the column for approximately 15 min to confirm the flowrate and to allow additional
compaction of the media bed. After the valves on the top and bottom end-caps had been tightly closed,
the preparation of the RSSCT column was complete and the column was ready for use (the valves on the
top and bottom end-caps had to be tightly closed to  prevent drainage during storage or transport of
RSSCT columns). If air bubbles were observed in a packed media bed, the packing procedure had to be
repeated to remove the entrapped air.

Table 3-2 summarizes the key design and operational parameters for the pilot- and full-scale  systems and
the corresponding RSSCTs for all media. In most cases, an ReSc value of 2,000 was used for column
design. In selected tests with E33, this parameter was adjusted to a value of 500 and 1,000 for
comparison. For MetsorbG and Adsorbsia GTO, it  was necessary to use an ReSc value of 1,000 to
minimize bed compaction. In some cases, reduced EBCTs of a full-scale system had to be used in the
scaling equations to overcome operational difficulties, such as a 30 to 50% decrease in bed depth
                                              12

-------
Table 3-2. Design Conditions of Pilot- and Full-Scale Systems and Corresponding RSSCT Columns
Site ID
Site 1 F


Site 1 L


Site 2


SiteS



Site 4





SiteS





Site6




Site 7

Testl
Test 2
Tests



Site Location
Test Duration
Valley Vista, AZ
(field tests)
09/25/04-10/22/04
Valley Vista, AZ
(lab tests)
10/20/04-11/24/04
Rimrock, AZ
(lab tests)
01/27/05-03/03/05
Licking Valley
School District, OH
(field tests)
05/25/05-06/05/05
Lyman, NE
(lab tests)
05/31/05-06/27/05



Lake Isabella, CA
(lab tests)
09/23/05-11/09/05



STMGID, NV
(lab tests)
01/24/06-02/10/06


Wales, ME
(field tests)
01/24/06-01/31/06
03/14/06-03/23/06
05/10/06-05/23/06



Adsorptive Media
Tested
AAFS50
GFH
E33
AAFS50
GFH
E33
AAFS50
GFH
E33
AAFS50
GFH
E33
ArsenXnp
ArsenXnp
E33
Adsorbsia GTO
MetsorbG
Adsorbsia GTO
MetsorbG
E33
GFH
ArsenXnp
Adsorbsia GTO
Adsorbsia GTO
MetsorbG
ARM 200
Adsorbsia GTO
ArsenXnp
GFH
GFH
AAFS50
Adsorbsia GTO
ARM 200
ARM 200
AsXnp
GFH
Kemlron
E33
Pilot- and Full-scale Design Conditions'3'
Effective
Media Vessel
Diameter Loading EBCT Diameter Bed Depth
(mm) Rate (m/h) (min) ReSc (cm) (cm)
0.85 12.8 4.5 12085 91 85
1.16 12.8 4.5 16500 91 85
1.16 12.8 4.5 16500 91 85
0.85 12.8 4.5 12085 91 85
1.16 12.8 4.5 16500 91 85
1.16 12.8 4.5 16500 91 85
0.85 15.6 4.5 14700 91 85
1.16 15.6 4.5 20059 91 85
1.16 15.6 4.5 20059 91 85
0.85 3.7 5.0 3489 5 31
1.16 3.7 5.0 4761 5 31
1.16 3.7 5.0 4761 5 31
0.75 3.7 5.0 3078 5 31
0.75 17.0 3.0 14175 244 85
1.16 17.0 3.0 21924 244 85
0.67 17.0 3.0 12663 244 85
0.67 17.0 3.0 12663 244 85
0.67 17.0 3.0 12663 244 85
0.67 17.0 3.0 12663 244 85
1.16 9.7 5.3 12435 107 85
1.16 9.7 5.3 12435 107 85
0.75 9.7 5.3 8040 107 85
0.67 9.7 2.5 7182 107 85
0.67 9.7 2.5 7182 107 85
0.67 9.7 2.5 7182 107 85
1.06 10.0 3.0 11745 168 103
0.67 10.0 3.0 7424 168 103
0.57 10.0 3.0 6316 168 103
1.16 10.0 3.0 12853 168 103
1.16 10.0 6.2 12853 168 103
0.85 24.2 2.2 23262 25 85
0.67 24.2 2.2 18336 25 85
1.06 24.2 2.2 29009 25 85
1.00 24.2 2.2 27367 25 85
0.75 24.2 2.2 20525 25 85
1.16 24.2 2.2 31746 25 85
1.5 24.2 2.2 41050 25 85
1.16 24.2 2.2 31746 25 85
RSSCT Design Conditions
Effective
Media Vessel BV
Diameter Loading EBCT Diameter Bed Depth Mass of Processed
(mm) Rate (m/h) (min) ReSc (cm) (cm) Media (g) (#)
0.128 14.1 0.68 2000 1.1 15.9 23.1 64,462
0.128 14.1 0.50 2000 1.1 11.6 16.9 75,579
0.128 14.1 0.50 2000 1.1 11.6 16.1 93,905
0.128 14.1 0.68 2000 1.1 15.9 23.1 51,191
0.128 14.1 0.50 2000 1.1 11.6 16.9 80,102
0.128 14.1 0.50 2000 1.1 11.6 16.1 88,627
0.128 14.1 0.68 2000 1.1 15.9 23.1 64,552
0.128 14.1 0.50 2000 1.1 11.6 16.9 92,293
0.128 14.1 0.50 2000 1.1 11.6 16.1 92,293
0.128 14.1 0.75 2000 1.1 17.6 25.6 17,238
0.128 14.1 0.55 2000 1.1 12.9 18.8 23,269
0.128 14.1 0.55 2000 1.1 12.9 17.9 23,269
0.128 14.1 0.85 2000 1.1 20.0 25.5 25,242
0.128 14.1 0.51 2000 1.1 11.9 14.7 54,147
0.128 14.1 0.33 2000 1.1 7.7 8.0 47,694
0.128 7.0 0.57 1000 1.1 6.6 9.4 48,251
0.128 7.0 0.57 1000 1.1 6.6 9.4 48,251
0.128 14.1 0.28 2000 1.1 6.6 18.8 65,979
0.128 14.1 0.28 2000 1.1 6.6 18.8 55,265
0.128 14.1 0.58 2000 1.1 13.7 19.4 74,123
0.128 14.1 0.58 2000 1.1 13.7 19.4 74,123
0.128 14.1 0.90 2000 1.1 21.2 30.0 50,927
0.128 11.7 0.48 1000 1.1 5.6 7.9 62,990
0.128 14.1 0.48 2000 1.1 11.2 15.8 45,759
0.128 14.1 0.48 1000 1.1 5.6 7.9 23,649
0.128 14.1 0.36 2000 1.1 8.5 8.9 39,695
0.128 7.0 0.57 1000 1.1 6.7 7.0 20,397
0.128 14.1 0.67 2000 1.1 15.8 19.5 31,803
0.128 14.1 0.33 2000 1.1 7.8 9.6 43,440
0.128 14.1 0.68 2000 1.1 16 19.8 31,317
0.128 14.1 0.3 2000 1.1 7.3 8.7 66,177
0.128 7.0 0.38 1000 1.1 5.1 4.5 45,730
0.128 14.1 0.24 2000 1.1 5.9 5.9 80,643
0.128 14.1 0.26 2000 1.1 6.1 6.3 68,937
0.128 14.1 0.34 2000 1.1 8.3 9.9 58,132
0.128 14.1 0.22 2000 1.1 5.2 6.4 86,516
0.128 14.1 0.17 2000 1.1 4.1 4.9 102,697
0.128 14.1 0.22 2000 1.1 5.3 5.7 80,284
(a) Actual field conditions may vary from design conditions for loading rate, EBCT, and bed depth

-------
observed over the duration of RSSCT tests. The cause of the bed compaction and/or loss of media from
the column might be attributed to excessive pressure development within the columns.  These operational
problems were experienced only with titania-based media, which have significantly lower structural
integrity as compared to iron-based media.  The scaling adjustments of titania-based media columns will
be justified by the results of RSSCT tests in Sections 4.5 to 4.8.  The results of several unsuccessfully
operated titania-based columns are not included in this report.

3.4        RSSCT Field Setup

RSSCT tests were conducted in the field at Valley Vista, LVHS, and SBMHP.  The Valley Vista site had
a full-scale arsenic treatment system consisting of two 36 in x 72 in columns in series, each packed with
16.7 ft3 of AAFS50 media (Valigore et al., 2006).  The average flowrate to the system was 36 gal/min
(gpm) and the average EBCT per column was 3.5 min (not 4.5 min as designed).  Three sets of RSSCT
apparatus were set up in the immediate vicinity to the full-scale system inlet piping and media vessles
under a sun shed. Water to the RSSCT columns was pumped continuously from a 50-gal Nalgene
container wrapped with atarp to minimize temperature variations and algae growth.  The 50-gal container
was refilled every three days with water after pre-chlorination and filtered with a  lO-jom glass fiber
cartridge  filter. Effluent from the RSSCT columns was discharged to an on-site holding tank used to
reclaim backwash wastewater to the head of the treatment train.

At LVHS, four sets of RSSCT apparatus were placed in the school's utility room  by a polyurethane-
coated unistrut frame on which a pilot system consisting of six 2 in  x 4 ft glass columns (in parallel) was
mounted.  The average flowrate to the pilot system was 0.033 gpm (which yielded a hydraulic loading of
1.5 gpm/ft2) and the average EBCT was 5 min. Water from the pilot system inlet piping was directed to a
10-|o,m glass fiber cartridge filter and a five-port polyvinyl  chloride (PVC)  manifold. Water was then
pumped continuously from four of the five ports on the manifold to four RSSCT columns. The fifth port
on the manifold was used for pressure release and influent water sampling. The RSSCT effluent was
discharged directly to the sewer.  Figure 3-3 presents photographs of the RSSCT water delivery system.

At SBMHP, four sets of RSSCT apparatus were installed adjacent to the full-scale adsorptive media
system, which consisted of two parallel treatment trains, each consisting of one 25-(im cartridge filter,
one 10 in x 54 in oxidation column, and three 10 in x 54 in adsorption columns (Lipps et. al., 2006).  The
average flowrate through each train was 5.2 gpm and the average EBCT per column was 2.2 min. A 55-
gal polypropylene tank was used to store water withdrawn every one to two days  from either the raw
water tap (that contained mostly As[III]), or one of the oxidation columns (that contained mostly As[V]).
A 10-|om glass fiber cartridge filter was used to filter raw water from the well during the test on As (III)
removal.  Water was pumped from the storage tank to the four RSSCT columns and then discharged to
the sewer. A photograph of the RSSCT setup is presented  in Figure 3-4.

3.5        RSSCT Laboratory Setup

Additional RSSCT tests were conducted in the ASU laboratories at a constant ambient temperature of
about 18°C. Water samples were collected and transported to ASU from Valley Vista, Rimrock, the
Village of Lyman, Upper Bodfish, and STMGID.  The Rimrock system consisted of two 36 in x 72 in
columns in series, each packed with 22 ft3 of E33 media (Wang et al., 2005).  The average flowrate to the
system was 31.5 gpm and the average EBCT per column was 5.2 min (not 4.5 min as designed).
Although never installed due to lack of adsorptive  capacity for uranium, the Metsorb system proposed for
the Lyman site had two 96 in x 96 in vessel in series, each loaded with  140 ft3 of Metsorb media. The
average system flowrate was 350 gpm and the EBCT was 3 min per vessel. The system installed at Upper
Bodfish had one 42 in x 60 in work column and one 42 in  x 60 in standby column, each loaded with 27
ft3 of ArsenXnp. The design flowrate to the system was 38  gpm and the EBCT was 5.3 min. The
                                              14

-------
•£•£11,
 Figure 3-3. Photographs of Water Delivery System (Including Inlet Piping
   Fiber Glass Filter Cartridge [top] and Five-Port Manifold [bottom])
        for Licking Valley High School RSSCT Columns
                      15

-------
                                                          Interm^liate
                                                          Water |jbrage
                Figure 3-4. Photograph of RSSCT System and Sampling Points at
                        Spring Brook Mobile Home Park in Wales, ME
STMGID system had three parallel 66 in x 72 in columns, each bedded with 80 ft3 of GFH for arsenic and
antimony removal.  The average system flowrate was 275 gpm and the average EBCT was 6.5 min.

For each laboratory RSSCT test, approximately 500 to 1,000 gal of groundwater was collected from a
field location. The groundwater was pumped from a well site, filtered using an in-line, lO-jom glass fiber
cartridge filter, stored in 5 5-gal drums double-lined with high-density polyethylene (HDPE) containment
bags, and shipped with drum liners sealed and drums securely closed via trucks to ASU laboratories. The
drums were stored at room temperature in ASU laboratories and sealed until use.  In the laboratory, water
was supplied to RSSCT columns either directly from the containment bags inside the 55-gal drums or
from a 330-gal Nalgene tank with water transferred from the 55-gal drums via a peristaltic pump. RSSCT
effluent was discharged to the sewer.
3.6
Sampling and Analyses
Water samples were collected from the inlet to the RSSCT columns and effluent from each RSSCT
column over the duration of each test. Figure 3-5 illustrates the schematic arrangment of a typical RSSCT
system and sampling points. Unless mentioned otherwise, influent samples were taken directly from the
respective influent water containers and effluent samples taken from the exit tubing of each column.
Sampling location, frequency, and analytes are presented in Table 3-3.  In most cases, samples were
collected three times per week during the test. Co-contaminants, such as antimony, uranium, and
vanadium, were analyzed if present in inlet water.

Table 3-4 summarizes the analytical methods, sample volumes, container types, preservations required,
and holding times for all analytes.  Samples for pH and temperature were taken in clean wide-mouth
plastic containers and measured immediately using an YSI 60 (by ASU) or WTW Multi 340i (by Battelle)
                                             16

-------
                                    Influent Sample Point
                                                                      Effluent #4
                                                                       Sampling
                                                                        Point
           Figure 3-5. Schematic Arrangement for RSSCT System and Sampling Points
                    Table 3-3.  Summary of Sampling Frequency and Analysis
Sample
Location
Inlet to
RSSCT
Columns
Effluent
from Each
RSSCT
Column
No. of
Samples
3
1
1
o
5
i
i
Sampling
Frequency
3 times/week
Weekly
Once during first
and last week of test
3 times/week
Weekly
Weekly
Analytes
On-site: pH, temperature
Off-site : As, Fe, Mn, Al, Si, P, Ca, Sb(a), U(a), Va)
Off-site: Alkalinity, Cl, F, sulfate, TOC
On-site/Off-site: As speciation
On-site: pH, temperature
Off-site : As, Fe, Mn, Al, Si, P, Ca, Sb(a), U(a), Va)
Off-site: Alkalinity, Cl, F, sulfate, TOC
On-site/Off-site: As speciation(b)
      TOC = total organic carbon
      (a) Analyzed only if contaminant present in groundwater
      (b) Performed weekly only if arsenite was the predominating arsenic species in influent
handheld meter.  Samples for alkalinity, chloride, fluoride, sulfate, and silica were collected in a single 1-
L Nalgene bottle without preservative.  Samples for total organic carbon (TOC) were collected in 40-mL
glass vials containing hydrochloric acid (HC1) for preservation. Samples for metals including As, Fe, Mn,
Al, P, Ca, Sb, U, and/or V were collected in 60-mL Nalgene (by ASU) or 250-mL HOPE bottles (by
Battelle) containing ultra pure nitric acid (HNO3) for preservation. Arsenic speciation was performed
using arsenic speciation kits prepared at Battelle or ASU laboratories according to the procedures detailed
in Appendix A of the EPA-endorsed Quality Assurance Project Plan (QAPP) for this project (Battelle,
2004).
                                               17

-------
  Table 3-4. Analytical Methods, Sample Volumes, Containers, Preservations, and Holding Times
Analyte
As, Fe, Mn,
Al, P, Ca, Sb,
u,v
As
Speciation(a)
pH
Temperature
Alkalinity
Cl
F
SO4
Si
TOC
Analytical
Method
EPA 200.8 (Battelle)
EPA 200.9 and
SM3111(Caonly)
(ASU)
EPA 200.8 (Battelle)
EPA 200.9 (ASU)
YSI 60 handheld meter
or equivalent^
YSI 60 handheld meter
or equivalent*'
EPA310.1(ASU/AAL)
EPA 300.0 (ASU/AAL)
EPA 300.0 (ASU/AAL)
EPA 300.0 (ASU/AAL)
EPA 200.7 (AAL)
SM3111(ASU)
SM5310B (ASU/AAL)
Sample
Volume
250 mL
10 mL
125 mL
125 mL
50 mL
50 mL
200 mL
50 mL
50 mL
50 mL
200 mL
200 mL
40 mL
Sample
Container
HOPE bottles
HOPE bottles
Certified clean
HOPE bottles
Certified clean
HOPE bottles
Plastic
Plastic
Plastic
Plastic
Plastic
Plastic
Plastic
Plastic
Glass
Preservation
Cool, 4°C
HNO3forpH<2
Cool, 4°C
HNO3forpH<2
Cool, 4°C
HNO3forpH<2
Cool, 4°C
HNO3forpH<2
None
None
Cool, 4°C
Cool, 4°C
Cool, 4°C
Cool, 4°C
Cool, 4°C
HNO3forpH<2
Cool, 4°C
Cool, 4°C
HClforpH<2
Holding
Time
6 months
6 months
6 months
6 months
Immediately
immediately
14 days
28 days
28 days
28 days
6 months
28 days
14 days
      HOPE = high density polyethylene; SM = Standard Methods (APHA and WEF, 1998);
      AAL = American Analytical Laboratory (Columbus, OH); TOC = total organic carbon
      (a) After on-site speciation using a field speciation sampling kit
      (b) Analysis performed on site using an handheld meter
All the samples collected from the laboratory RSSCTs were taken to ASU's laboratories for analyses
immediately after sample collection.  For the samples collected in the field for off-site analyses, they were
packed carefully in sample coolers with wet ice. Samples taken at Valley Vista and LVHS were shipped
to ASU for analysis. Samples collected at SBMHP were shipped overnight via FedEx to Battelle.  Upon
receipt, the respective sample custodians verified that all samples indicated on the chain-of-custody forms
were included and intact and the samples were logged into the laboratory sample receipt logs.

The analytical procedures described in Section 4.0 of the EPA-endorsed QAPP were followed by ASU
and Battelle ICP-MS laboratories and relevant wet chemistry laboratories.  Laboratory quality
assurance/quality control (QA/QC) of all methods followed the prescribed guidelines.  Data quality in terms
of precision, accuracy, method detection limits (MDL), and completeness met the criteria established in the
QAPP (i.e., relative percent difference [RPD] of 25%, percent recovery of 75 to 125%, and completeness of
80%).  Field measurements of pH and temperature were conducted by each sampler using an YSI 60 or
WTW Multi 340i handheld meter, which was calibrated following the procedures provided in the user's
manual.
                                              18

-------
                               4.0 RESULTS AND DISCUSSION
This section presents the results of RSSCTs conducted with multiple commercially available adsorptive
media. The focus is on arsenic removal, although removal of other elements also is discussed to aid in the
understanding of arsenic removal.

4.1        Source Water Quality

Table 4-1 presents the inlet water quality data for the seven test locations where pilot/full-scale system
results were compared with RSSCT results.  The data presented in the table represent the average of
RSSCT influent results sampled over the course of each test.

4.2        Valley Vista, AZ (Site 1)

The objectives of the Valley Vista, AZ RSSCT tests were (1) to predict the performance of the full-scale
AAFS50 system to remove arsenic, (2) to compare AAFS50 arsenic removal capacity to that of two iron-
based media, i.e., E33 and GFH, (3) to evaluate the validity of conducting RSSCTs in the laboratory - a
controlled environment - by collecting and transporting water from the field, and (4) to examine the
possibility of using a lower ReSc value for the design of RSSCT columns. The comparison  of field and
laboratory RSSCT results was accomplished by conducting equivalent sets of RSSCT experiments both in
the field and in the laboratory.

4.2.1       Field RSSCTs (Site IF). Figure 4-1 presents arsenic breakthrough curves from the RSSCT
columns packed with AAFS50, E33, and GFH adsorptive media. Although scattered rather  extensively,
the AAFS50 breakthrough data showed a trend of reaching complete arsenic breakthrough after
approximately 20,000 BV of throughput. In contrast, effluent arsenic concentrations were <1 |o,g/L for
both E33 and GFH at 20,000 BV and reached 10 ng/L after approximately 44,000 and 48,000 BV,
respectively. Two data points on the E33 arsenic breakthrough curve had abnormally high arsenic
concentrations; both of these samples also contained unusually high iron concentrations. Iron particles
containing arsenic might have been sampled following the RSSCT column due to different sampling
practices or migration of iron particles out of the  packed bed. During the field RSSCT test, the 50-gal
Nalgene feed tank was refilled every three days.  Natural variations in water quality over the course of the
test might have resulted in somewhat different water quality in the feed tank, as evidenced by different pH
values, ranging from 7.45 to 8.02, measured in the feed tank. These variations in influent water quality
might explain some of the data scattering observed.

While E33 and GFH had very similar arsenic breakthrough curves, these two media exhibited differences
in their abilities to remove vanadium (Figure 4-2). While vanadium concentrations following the E33
column reached the influent level of 15  (ig/L at approximately 56,000 BV, the concentrations following
the GFH column remained as low as 1 (ig/L after 76,000 BV. Differences among the media were less
apparent for phosphorous and silica. GFH removed slightly more phosphorous than E33 and AAFS50.
All three media had little ability to remove silica, with effluent silica levels reaching the influent level of
18.5 mg/L within the first few thousand BV. It is interesting to note that slightly lower calcium and
alkalinity levels and pH values (-0.3 pH units) were observed across the E33 and GFH columns  during
the first few thousand BV of RSSCT operations.  This observation was consistent with prior work
performed at ASU when calcium silicates formed on the surface of iron (hydr)oxide media during the first
part of column operations.  AAFS50 had somewhat higher aluminum concentrations in its effluent (6 to
14 |og/L) compared to the influent concentration of 6 |o,g/L (Figure 4-2). None of the media  removed
chloride, fluoride, sulfate, or TOC.
                                              19

-------
                                                    Table 4-1. Average Influent Water Quality Data
Analyte
Unit
Sampling Date
As(III)
As(V)
As(total)
Fe
Mn
Al
P
Ca
Sb
U
V
PH
Temperature
Alkalinity
Cl
F
SO4
Si (as SiO2)
TOC
^g/L
HB/L
W?/L
^g/L
HB/L
mg/L
HB/L
HB/L
W?/L
S.U.
°C
mg/L(b)
mg/L
mg/L
mg/L
mg/L
mg/L
Site IF
Site 1L
Valley Vista,
AZ
(Well 2)
09/25/04-
10/22/04
0.5
39.5
40.0
4
0.2
6
19
40
NA
NA
15
7.8
20
160
12
0.2
11
18
1.9
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
8.0
15
NA
NA
NA
NA
NA
NA
Site 2
Rimrock,
AZ
(Well #2)
01/27/05-
03/03/05
1.0
61.0
62.0
o
3
0.4
2
23
79
NA
NA
9.0
7.5
19
370
32
0.1
11
27
1.6
Site3
LVHS,
OH
05/25/05-
06/05/05
64.0
1.5
65.5
1,420
183
5
51
57
NA
NA
<1.0
8.1
18
505
9
0.1
8
6
NA
Site 4
Lyman,
NE
(Well #3)
05/11/05
<1.0
21.5
21.5
38
70
8
54
77
NA
40
37
7.7
17
342
36
0.1
476
NA
3.8
SiteS
Upper
Bodfish,
CA
(Well
CH2-A)
09/13/05
<1.0
43.0
43.0
5
1.2
4
15
30
2.1
56
0.1
7.2
18
87
12
1.3
39
8
0.4
Site 6
STMGID,
NV
(Well #9)
01/03/06
<1.0
51.0
51.0
9
<1.0
4
162
5.1
13
<1
4.0
7.4
18
83
10
<0.1
12
23(o)
0.9
Site 7
SBMHP,
Wales, ME
After
Oxidation
(Parti)
01/24/06-
01/31/06
0.8(a)
40.2(a)
39.2(a)
<25
0.1
37
33
18
NA
NA
NA
8.6
10
69
8
0.4
17
11
NA
After
Oxidation
(Part 2)
05/10/06-
05/23/06
0.1(a)
40.8(a)
38.6(a)
<25
O.I
28
30
14
NA
NA
NA
8.7
10
68
11
0.5
19
10
NA
Raw Water
03/14/06-
03/23/06
22.5(a)
13.0(a)
34.3(a)
<25
7.1
<10
24
16
NA
NA
NA
8.5
7
64
8
0.5
19
11
NA
to
o
             NA = Not analyzed; LVHS = Licking Valley High School; SMHP = Spring Brook Mobile Home Park
             (a) As(III) and As(V) averaged based on two sets of speciation data; As(total) averaged based on entire dataset
             (b) asCaCO3
             (c) Data lower than those measured at influent to full-scale system by Battelle, i.e., 73.6 mg/L (as SiO2) on average

-------
                 LU
 50

 45

 40

 35

 30

 25

 20

 15


S°
35

  0
                          Influent As Concentration = 40 ug/L
AA     A
      A
                                               IAA
                                                      n
      <>n
   on
AAAFS-50
DGFH
OE33
                               20,000    40,000    60,000    80,000   100,000
                                       Bed Volumes Treated
                   Figure 4-1.  As Breakthrough Curves from Field RSSCT
                                Columns at Valley Vista, AZ
4.2.2       Laboratory RSSCTs (Site 1L). Arsenic breakthrough curves from four RSSCT columns
(including one each for AAFS50, E33, and GFH, and one additional for E33 with a lower ReSc value)
using Valley Vista, AZ source water were obtained in the laboratory and are presented in Figure 4-3.
Comparison of these to the field-produced breakthrough curves (Figure 4-1) showed the same pattern of
arsenic breakthrough. (Throughput for both E33 and GFH media at 10 (ig/L was 40,000 BV, somewhat
shorter than the 44,000 and 48,000 BV observed in the field). However, the curves produced in the
laboratory were much smoother than those from the field.  The smoother curves obtained might be caused
by the use of only two separate batches of water in the laboratory, compared to the many batches of
influent water refilled every three days in the field. Thus, laboratory-operated RSSCT columns would
have more consistent pH, temperature, arsenic concentration, and other constituents in the feed water
compared to the field-operated RSSCT columns.

Figure 4-4 overlays the field and laboratory breakthrough curves for both E33 and GFH media. At a 95%
confidence level and excluding the two E33 outliers in the field, there was no statistical difference
between the two sets of curves. Therefore, transporting groundwater to a centralized laboratory for
RSSCT tests would be as valid as conducting the tests on site, provided that representative samples may
be collected and transported to the laboratory.

To conduct RSSCTs in the laboratory, it would be beneficial if the volume of water to be collected and
transported could be significantly reduced. Two RSSCT columns packed with E33, therefore, were
conducted with a ReSc value of either 2,000 or 500.  A value of 500 reduces the volume of water needed
for treating the same number of bed volumes by a factor of four. As shown in Figure 4-3, arsenic
breakthrough at  10 (ig/L occurred at approximately 26,000 BV with a ReSc value of 500, compared to
approximately 40,000 BV with a ReSc value of 2,000. Thus, reducing the ReSc value from 2,000
(equivalent to a Reynolds number of 2.2) to 500 would not be recommended for E33.  Similar work was
previously performed for GFH and led to a similar conclusion (Badruzzaman, 2005).
                                             21

-------
^u -
=1.
•1 15
(D

§
c
o -in -
O IU
E
^
ro
ro 5
>
0)
LU n v
Influent Cone = 15 |jg/L
A
AAA

A ,


o
o o o
.
?s ^
. v \?





A




o
o








^JJM~I /~cy\gs n , n







n m
A AAFS50
DGFH


0 E33 (ReSc=2000)

n D D





OU "
5 25 ;
c
0
'ro 20 :
'E
sA —
A ^
. AO AAAFS50
A^ n
n n OGFH
0 E33 (ReSc=2000)

iarsn4_i^c^v^\ n , m , , . . . .
                             20,000    40,000    60,000    80,000    100,000
                                      Bed Volumes Treated
                                   20,000     40,000    60,000    80,000    100,000
                                            Bed Volumes Treated
to
to

~ 181
— 1
I5 161
	 ~
O \"t ~.
1 12;

(11
0 10 :
o :
0 8 :
1 6^
W :
c 4i
0) '
t 2 E
LJJ
n
Influent Cone = 18.5 mg/L A
0 A 0 OK> 040 .
AA1
-n n

-A





">
i

3


o^ A ^ct> n

AK>DO D OOO

























A AAFS50
OGFH










0 E33 (ReSc=2000)









^^
— i
D)

C
0
15
'E
0)
o
c
o
O
E
c
'E
^
<
'E
0)

^
LLJ
                             20,000    40,000    60,000
                                      Bed Volumes Treated
80,000    100,000
                                                                                    10 -
Influent Cone = 6 ug/L
0 A A
A A A
>
A A
R> 0
] ° °0<> OD c
on DO
AAAFS50
DGFH
OE33
(ReSc=2000)

20,000    40,000    60,000    80,000    100,000
         Bed Volumes Treated
                            Figure 4-2. Breakthrough of V, P, Si, and Al from Field RSSCT Columns at Valley Vista, AZ

-------



-2~
•fc)
^
o
5
O
<
§*~"

£
LU



^\j
45 :
:
40 :
:
35 :
:
30 :
25 :
20 :
:
15 :

10 :
;
5 :
n C
Influent Cone = 39.5 ug/L



A A
A A
^T^ n

o o °o °
A <0 0
O

o
^l AAAFS50
O E33 (ReSc=2000)
™ DGFH
O g «• o E33 (ReSc=500)
jfr T rUffriTnTNn 	 . 	 . 	 . 	 1 	 . 	 . 	 . 	 . 	 1 	 . 	 . 	 . 	 . 	 1 	 . 	 . 	 . 	 . 	
           20,000    40,000    60,000
                    Bed volumes treated
80,000    100,000
Figure 4-3. As Breakthrough Curves from Valley Vista, AZ
              Laboratory RSSCT Columns

^
"ra
5
0
o
O
0
0)
<
0)
j£
Hi



*j\j -
45 ^
I
40 :
:
35 :
30 \
25 :
20 :
:
15 :
10 ^
5 :
n 4
Influent Cone = 39.5 ug/L
^
*E 33 (Field)
OE33(Lab) *
AGFH(Field)
AGFH(Lab)
AA
A *<>*
tA^ A^
A * 0A°
O A

^*A
*A>A
L^A4,/Mt^4Wm
            20,000     40,000     60,000
                    Bed Volumes Treated
80,000    100,000
 Figure 4-4. Comparison of Field and Laboratory RSSCT
              Results for Valley Vista, AZ
                          23

-------
                     Influent Cone = 39.5 ug/L

                                       D
4.2.3       Comparison of Field/Laboratory RSSCT and Full-Scale System Results. Figure 4-5
overlays and compares the arsenic breakthrough curves for AAFS50 from the field and laboratory RSSCT
columns and the full-scale system. Although the lab and field RSSCT data were scattered more than
those of the full-scale system, both of them predicted rapid arsenic breakthrough across the lead tank of
the full-scale system at 10 (ig/L between 5,000 and 10,000 BV. Note that for the breakthrough curve
produced by the full-scale treatment system,  acid addition was applied after approximately 30,000 BV of
system operations, thereby reducing the pH of the influent and improving arsenic removal (data not
plotted on the figure). pH adjustment was not applied to the RSSCT columns, thus no enhanced arsenic
removal was observed.
      50

      45

      40

      35

      30

      25

      20

      15

      10

       5

       0
             •5)
o
O
o
'c
             LU
                                       A
                                                                           D
                               D
                                                              *Tank 1
                                                              D RSSCT (Field)
                                                              A RSSCT (Lab)
                            10       20       30       40       50
                                    Bed Volumes Treated (in thousands)
                                                            60
                                                                   70
                      Figure 4-5.  Comparison of Field/Laboratory RSSCTs with
                            Full-Scale AAFS50 System at Valley Vista, AZ
4.3
Rimrock, AZ (Site 2)
The purpose of the Rimrock, AZ tests was to compare arsenic breakthrough from laboratory RSSCT
columns against the full-scale E33 system. To compliment the results of the Valley Vista tests, RSSCT
columns using AAFS50, GFH, and ArsenXnp also were performed for this site. Similar to the Valley
Vista tests, the effect of a lower ReSc value on arsenic breakthrough also was investigated.

4.3.1       Arsenic Removal.  Figure 4-6 presents arsenic breakthrough curves from E33, GFH, and
AAFS50 RSSCT columns and from the lead vessel of the full-scale E33 system.  The E33 and GFH
RSSCT columns were operated for nearly 100,000 BV before shutdown. Arsenic broke through at 10
(ig/L from the E33 RSSCT column after approximately 40,000 BV, which was nearly identical to that
through the full-scale system. GFH performed better than E33, removing arsenic to the  10 (ig/L level for
about 52,000 BV.  AAFS50 showed comparatively little arsenic removal capacity, reaching 46 (ig/L (the
first data point) at approximately 8,000 BV.
                                             24

-------
In addition to the two iron-based and one iron-modified alumina-based media, an RSSCT column packed
with ArsenXnp containing 25% iron was tested. The arsenic breakthrough curve for ArsenXnp is presented
in Figure 4-6 and compared with those of other adsorptive media.  ArsenXnp exhibited a sharp arsenic
breakthrough at 10 (ig/L after approximately 30,000 BV, slightly less than the 40,000 BV for E33.

An additional E33 RSSCT column also was conducted to evaluate an ReSc  value of 1,000 (data in Figure
4-6 used a ReSc value of 2,000 for E33 and other media).  Similar to the laboratory results using Valley
Vista source water, the lower ReSc  value resulted in earlier arsenic breakthrough and was, therefore, not
valid (especially given the excellent comparison between the RSSCT and the full-scale system results).
Interestingly, a ReSc value of 500 for E33 using Valley Vista source water had approximately the same
net effect as a value of 1,000 using Rimrock source water; both shortened the number of BV at 10 (ig/L
by approximately 45 to 50%.  This suggests that the length of the packed bed using a lower ReSc value is
not long enough to capture the mass transfer zone.

To summarize, the full-scale and RSSCT arsenic breakthrough curves for E33 were nearly identical, thus
supporting the use of the RSSCT approach to predict the full-scale performance of iron-based adsorptive
media. The iron-based adsorptive media treated a significantly larger number of BV than alumina-based
media. Furthermore, GFH outperformed E33 for arsenic removal by about  12,000 BV.  ArsenXnp media
was found to have a treatment capacity much higher than that of AAFS50, but less than GFH and E33.
Noteworthy of the ArsenXnp media  is the steepness of its arsenic breakthrough curve compared to that of
an iron-based media. This suggests relatively fast movement of the mass transfer zone through the
packed resin-based media bed.

Jl^^
"5)
^
c
o
"OB

§
c
o
O
o
'c
0
e
-i— •
< D
O NP"l
,-, )gj(
y$$®® AAAFS50
•vyo E GFH
°XV<^ OE33
s^^g Dn X E33 Full Scale
>^^*id«^!^E D OHIX
                               20,000    40,000    60,000    80,000    100,000
                                        Bed Volumes Treated
               Figure 4-6. As Breakthrough Curves from Laboratory RSSCT Columns
                             and Full-Scale E33 System at Rimrock, AZ
                                             25

-------
4.3.2       Removal of Other Elements. Breakthrough curves for vanadium, phosphorous, silica and
aluminum are presented in Figure 4-7. Similar to the Valley Vista RSSCT results, GFH removed nearly
all vanadium throughout the entire RSSCT test. However, complete vanadium breakthrough (i.e., full
exhaustion of vanadium adsorption capacity) was observed at approximately 64,000 BV for E33 and less
than 20,000 BV for AAFS50. All media removed some phosphorous, with iron-based media having
much more capacities than alumina-based media.  Some silica was removed immediately after the column
runs but it reached nearly complete breakthrough soon after. Aluminum was released from the AAFS50
media throughout the entire RSSCT run, producing 9 to 20 (ig/L of aluminum in the column effluent.
The iron-based media did not release aluminum. There was no removal or release of fluoride, chloride, or
sulfate by any of the media.

4.4        Licking Valley High School in Newark, OH (Site 3)

A unique characteristic of the LVHS groundwater was that arsenic was present almost entirely in the
reduced form (Table 4-1). The objective of the RSSCT tests was to compare RSSCT results with those of
the previous  pilot-scale tests on several adsorptive media for As(III) removal. Therefore, RSSCTs were
conducted on-site to ensure that water containing As(III) was fed to the columns for testing.

4.4.1       Comparison of As(III) Removal by RSSCT and Pilot-Scale Columns. RSSCTs were
conducted with AAFS50, GFH, E33 and ArsenXnp. As shown in Figure 4-8, arsenic broke through at 10
jig/L after approximately 2,500 BV for AAFS50, 11,000 BV for E33, and > 23,000 BV for GFH.
Although the RSSCTs predicted the same order of performance as the pilot tests, they appeared to have
significantly over-predicted the run length for all media. For example, the corresponding BV obtained
from the pilot-scale tests was 650 BV for AAFS50, 6,500 BV for ArsenXnp, 4,700 BV for E33, and
21,000 BV for GFH.  The discrepancies observed could be due to a number of factors, including varying
influent water quality, different loading rates, and improper use of scaling relationship for As(III) removal.
The average  influent As(III) and As(V) concentrations through the duration of the tests from May 25 to
June 5, 2005 were 64.0 and 1.5  (ig/L, respectively, which were comparable to the historical data since
2000. The average influent iron concentration of 1,420 (ig/L was within the range of  1,077 to 1,725 (ig/L
reported historically since 2000. Since the samples were not filtered on site, it was suspected that the
soluble iron (not measured) might be lower due to aeration and precipitation of some soluble iron in the
leaky water supply/delivery system, as evidenced by the iron precipitates found in the plastic and Teflon®
tubing, in-line glass fiber cartridge filter, and the five-port manifold as shown in Figure 3-3. Nonetheless,
partially aeration of the influent water did not appear to have impacted the amount of As(III) in that water.

The loading rate for the pilot tests was 1.5 gpm/ft2 (or 3.7 m/hr), which was significantly lower than that
commonly used for full-scale systems (i.e., 4 to 8 gpm/ft2 or 9.8 to 19.6 m/hr) for which the RSSCTs were
initially validated against. The lower loading rate was caused by the pilot column design, which used a 2
in diameter by 12 in long media bed to achieve an intended EBCT of 5 min.  To maintain this EBCT but
increase the loading rate (for example, three times) would require the bed depth to increase three times.
As a result, the throughput to reach the 10 (ig/L breakthrough level should remain the same even with a
three-time higher loading rate.

It is possible that the logarithms developed for scaling down a pilot- or a full-scale arsenic removal
system may be applied only to arsenate, which is a charged species as discussed in Section 1.1.1.
Arsenite, at an average pH value of 8.1 at LVHS, is not charged and, thus, may involve a different set of
transport mechanisms for its adsorption on porous metal (hydr)oxides.  Future studies are warranted to
determine how different arsenic species may affect the validity of the scaling equations discussed in
Section 1.1.3 and if any modifications would be required to better predict pilot- and full-scale
performance involving As(III).
                                              26

-------
G~
^)
3.
c 8-
"ro
£ R
8 6

o
O
1 4-
ro
c
ro
> 2
'E
0)
LU n -
Influent Cone = 9 |jg/L ^
AAA A AA ^ °
AAA> AA A00*
A* 0


0
AAAFS50
DGFH
OE33
O
r\ , v&\. m R t& n n, nmn n ,n, . n rn .D.
3~
5 25 :
c
0
'ro 20 :
c
(D

§ 15:
o
0.
ra ID-
'S
1-
'E c
0) O
3
n ;
n -
Influent Cone = 23 |jg/L ^> O O
OO
o o D
A AAA
A A 0A D °
A AAA n

AA ODD

AAA
A A D D AAAFS50 D GFH
d<> D
«] C^ pn O cP O E33
                20,000     40,000    60,000
                         Bed Volumes Treated
                             80,000   100,000
                                                                           20,000    40,000     60,000    80,000    100,000
                                                                                    Bed Volumes Treated
D)
_§
C
O
c
(U
o
c
o
o
w
'
(U
30

25

20

15

10

 5

 0
         ; Influent Cone = 27 mg/L
         A  At
         n
                    n
                    u
20,000
                          40,000    60,000
                         Bed Volumes Treated
                                  8  n
 AAAFS50
 DGFH
 OE33
 i  I  i =
80,000    100,000
                                                 c
                                                 0)
                                                 o
                                                 c
                                                 o
                                                 O
20
18

14
12
10
 8
 6
 4
 2
 0
                             Influent Cone = 2 ug/L

                                           A
                                                     A    A
                                                         A
                                                                                n
                                                                               n
                                                                                                      n
                                                                                                   n  o
                                                                           20,000     40,000     60,000    80,000    100,000
                                                                                    Bed Volumes Treated
            Figure 4-7. Breakthrough of V, P, Si, and Al from Rimrock, AZ Laboratory RSSCT Columns

-------



^^
!s>
5
o
o
0
o
'c
CD
-i— »
c
0>

LU




-
-
60
-
50 :
-

40 :_
-
-
30 :
-

-
20 :
-

10 :_

o -
A
A A
/\ A ^ A
A
A
L^
A O A A
A
A AA A

0
o/ ^
A A

A *

A A 4*
A A
/
J /S. |— | ^ [— | r-|
	 KaiJ ^ ^g ^—LJ-! ^ p—^™^ ! ^^^^^^^———^
• GFH (RSSCT)
• E33(RSSCT)
AAAFS-SO(RSSCT)
D GFH (Pilot)
0 E33(Pilot)
AAAFS-SO(Pilot)

^



^




D D
D


                        5,000         10,000        15,000
                                     Bed Volumes Treated
                                                       20,000
25,000
                    Figure 4-8.  Comparison of RSSCT and Pilot Results at
                          Licking Valley High School in Newark, OH
4.4.2       Removal of Other Elements.  Figure 4-9 presents the breakthrough curves for phosphorous,
silica, iron, and aluminum.  Vanadium concentrations at this site were <1 ug/L and, therefore, not plotted.
Phosphorous was removed by all media from 51 to <4 ug/L. The average influent silica concentration
was 6 mg/L, while the effluent silica concentrations varied between 1 and 8 mg/L and did not exhibit a
clear breakthrough pattern.  The average influent iron concentration was 1,420 (ig/L.  As shown
previously by the corresponding pilot-scale tests, iron, existing almost entirely in the soluble form, was
removed initially by all media and began to break through after certain run length.  Iron concentrations
remained extremely low for E33 and GFH even after 20,000 BV. It is not clear how the iron  removed
would affect arsenic removal over time.
4.5
Village of Lyman in NE (Site 4)
Groundwater at the Lyman, NE arsenic removal demonstration site contained naturally occurring arsenic
and uranium, both above their respective MCLs of 10 and 30 u.g/L. The water also had high
concentrations of alkalinity (342 mg/L [as CaCO3]) and sulfate (476 mg/L).  The vendor selected for the
demonstration study proposed to install an adsorption system using MetsorbG, a titania-based media
manufactured by Graver/Hydroglobe, to remove both arsenic and uranium. Before installing the system,
a decision was made to use RSSCTs to test the efficiency of MetsorbG and three other adsorptive media
(i.e., E33, ArsenXnp, and Adsorbsia GTO) to remove uranium in concert with arsenic.  Therefore, the
RSSCTs conducted on Lyman, NE water had the primary goal of evaluating the co-removal of arsenic
and uranium. A secondary goal was to evaluate an approach (i.e., by reducing ReSc values) that could
help reduce operational problems associated with the titania-based media tested; the operational problems
were encountered when the RSSCT columns experienced increased pressure drop and lost > 40% of the
beds resulting from bed compaction and/or loss of media into the RSSCT effluent.
                                              28

-------
                                    on
                                          o
                                          o

                                    DO+ °
                                          o
                                          CM
    X
                                     Q*
                                  X
                                     on
                                   X
                                          o
                                          o
                                          o

                                          in
                                          CM
                                        8  •»

                                        °-   S
                                        o  —
                                        o  P
                                        o  >
                                        O  T3

                                        ^~  m
                                          o
                                          o
oooooooo
1^    CD    in    ^    CO    CM    i-
  (l/Brl) uo!}Bj}U90uoo wnujiuniv luanyg
o
o
1— 1


0 D
o
in
w
^~ CO •)/
*^? rf\ ^ O]
<£ t_J LU I
•^ n o x
•* C
in
" CM



Q
o
- O
o
CM
- O
o
3 °-
1 in
^ r ^
=3, vx P §
o x « ^
CM * X .
V 1
- i
0 ?
1 ^
' 0
o
8
in

3OOOOOOOO1
3OOOOOOO












Tl
is Treatei
1
"o
^>
T3
s






                                                                                                          ffi
                                                                                                          o
                                                                                   uoj|
                                                                                                             o
                                                                                                             o

                                                                                                             o
 6/j

3


"«
                                                                                                             S/j
                                                                                                             s

                                                                                                             -2
                                                                                                             u
                                                                                                             s
                                                                                                             S
                                                                                                             s
                                                                                                          H

                                                                                                          U
                                                                                                          t/3
                                                                                                          tf3
   •4DOX
                                 D
                n
          X
II

o
c
o
O

tr
0
                                                                  o  n
                                                              0
                                                              v
                                                                 D
                                                                 D   0

                                                                 0   D
                                                                          n   si  o
                                                                              o
                                                                              X
                                                                                     a
                                                                                     X
                                                                                                   o
                                                                                                   o
                                                                                                 -- O
                                                                                                 o  -="

                                                                                                 o'  "o
                                                                                                 •^   
-------
4.5.1      Arsenic Removal. Four adsorptive media were evaluated, including two titania-based media,
i.e., MetsorbG and Adsorbsia GTO, and E33 and ArsenXnp.  E33 exhibited the best arsenic removal
(approximately 25,000 BV at 10 ng/L), followed closely by Adsorbsia GTO (approximately 22,000 BV)
and more distantly by ArsenXnp and MetsorbG (approximately 16,000 BV) (Figure 4-10). The Adsorbsia
GTO and MetsorbG media were run at ReSc values of 2000 (Figure 4-10) and 1000 (comparison in
Figure 4-11).  Unlike the iron-based media studied for Valley Vista and Rimrock, reducing the ReSc
value from 2,000 to 1,000 for titania-based media had minimal effect on the resulting arsenic
breakthrough curves, suggesting that the length of the mass transfer zone probably is shorter with
Adsorbsia GTO and MetsorbG than with E33 and GFH.

4.5.2       Uranium Removal.  Of particular interest in the Lyman water was the co-occurrence of
arsenic and uranium above their MCLs. Figure 4-12 presents uranium breakthrough curves along with
those for vanadium, iron, and manganese.  Contrary to the claims made  by the vendor, MetsorbG did not
show much adsorptive capacity (i.e., < 500 BV to reach 30 (ig/L) for uranium. The other titania-based
media, Adsorbia GTO, and the only iron-based media tested, E33, also did not show any adsorptive
capacity for uranium.  ArsenXnp, however, exhibited a significant capacity for uranium, treating
approximately 20,000 BV before reaching the 30-(ig/L MCL. A chromatographic-like peaking of
uranium in the column effluent occurred between 20,000 to 50,000 BV, where the uranium concentrations
in the RSSCT effluent were greater than those in the influent. This suggests that the uranium previously
removed was displaced into the column effluent by other anionic species with higher selectivity or higher
concentrations, such as sulfate (476 mg/L). It is hypothesized that uranium was removed mainly by the
anionic resin via an ion exchange process and that competing anions, such as sulfate, might have
displaced the uranium on the resin.

As noted above, the groundwater at this site contained a relatively high level of alkalinity (i.e., 342 mg/L
[as CaCO3]). Uranium might have reacted with carbonate to form uranium carbonate complexes (Fuller
et al., 2002; Gu et al., 2004), thus limiting its adsorbability on titanium oxide surface.

4.5.3       Removal of Other Elements. As also shown in Figure 4-12, the two titania-based media
removed significantly more vanadium than E33 or HIX did.  Vanadium began to break through from E33
and ArsenXnp at approximately 17,000 BV, but did not breakthrough from MetsorbG and Adsorbsia GTO
until approximately 44,000 BV. No discernable trends in iron removal were observed.  Manganese
exhibited unique trends as its removal began to increase later during the tests (Figure 4-12). During tests
at the other sites, low manganese concentrations precluded any discernable trends.  At this site, GTO and
MetsorbG removed a significant portion of the manganese, while E33 and HIX did not.

4.6         Upper Bodfish in Lake Isabella, CA (Site 5)

Similar to Lyman, NE, the groundwater from Upper Bodfish, CA contained arsenic and uranium at 43
and 56 (ig/L, respectively, which were above the respective MCLs. Therefore, the main objective of the
RSSCT tests was to evaluate media capacities (including ArsenXnp, MetsorbG, and Adsorbsia GTO) to
remove both arsenic and uranium. The data produced were used to assess the applicability of installing a
full-scale ArsenXnp arsenic removal system, as proposed by the selected vendor for this arsenic removal
demonstration site.  The average alkalinity value of the groundwater was 87 mg/L (as CaCO3 at pH = 7.2)
and the average  sulfate concentration  was  39 mg/L, which were considerably lower than the 342 mg/L (as
CaCO3 at pH = 7.7) and 476 mg/L present in the Lyman groundwater. Another objective of these
RSSCTs was to  evaluate the capacity  of two iron-based media, i.e., E33 and GFH, to remove arsenic and
uranium. Due to pressure buildup and media compaction with the titania-based media, only data from the
RSSCT columns scaled to a 2.5 min full-scale EBCT (rather than the 5.3 min EBCT) were included in
Table 3-2 and discussed in Sections 4.61 and 4.62.
                                              30

-------
O)
o
-

CD
O

O
O
g
'c
CD
CD
LU
ou
25 :
20 :

15 :

10

5 :


Influent Cone = 22 ug/L
X*
nxMX)lt3**4> n
x o
>v A A A
r^^ ^s A ^^
X A A
D
y*w
X ^ AE33
sin p DHIX


D






<^A OGTO
>tx> X MetsorbG
                10,000   20,000   30,000   40,000   50,000  60,000

                          Bed Volumes Treated

      Figure 4-10. Arsenic Breakthrough Curves from Lyman, NE
                   Laboratory RSSCT Columns
o
"co
i_
"c
CD
0
O
O
.0
CD
W) o
<^
c 0
CD
LJJ

"co
E
n

,
I
-
-
0.8
.
0.6


0.4
-
0.2

_ _ 1 .
^1 ^ft ]£j^M A ^^
r~r^ A
nil nfcA
DD A
n AA
n AA
1 1 •

D H
^
n J AGTO(ReSc=1000)
A • MetsorbG (ReSc=1000)
^ AGTO(ReSc=2000)
.-tf A D MetsorbG (ReSc=2000)
n JJiaiB^ 	 I .!....!...!... I
               10,000  20,000 30,000 40,000  50,000  60,000 70,000


                           Bed Volumes Treated

      Figure 4-11. Comparison of Two ReSc Values on Lyman, NE
                   Titania-Based Media RSSCTs
                               31

-------
OJ
to
C7U ~
^ on
o> BU :
~Z 70 ;
.0 ;
1 60 :
'E :
ri)
S 50 :
d
o
" 40-
E :
•1 30 :
(D
5 20 ^
'E
S 10^
S ou
0 i
Influent Cone = 40 |jg/L
r-i
D n

n
n
n n

&\ t$tx"! >fx*x" xx* $x ** *
^ A xK x* xK xK "x /N /^ A A ^ ^ A
n
AE33
n MIX
D OGTO
w—i — I I — I I — I ,| — I i — i xj/ IWl^tQnrhrn
3 I. I Q fj Q LJ — t — t — \ — t — t — t — t — ^ — t — t — t — ** ivieisoiuo














tj ;
^ 4n
D) ^U
S 35 ;
O ;
| so ;
(D
£ 25 ;
o
1 2"
| 15i
I 10;
^
^ 5 ;
LU r\ s
Influent Cone =











AA
m 0 Baj
10,000 20,000 30,000 40,000 50,000 60,000
Bed Volumes Treated
151
	 |
D)
^-
o -in -
•~ i *-'
ro
"c
8 ]
c
O

i s-
— X
(D
E '
LU
L
0-

Influent Cone = 38 ug/L A ^33
O MIX
OGTO
X MetsorbG
-/N »j*
X
D X
> on
x ^
rx n

,-, X X l-l XU- XIX
| 	 | W 1 — 1 /N J~^ >X

ft V X'^xu'N' O
'--' l
SJm ^%HeU3
10,000
38 |jg/L

J

A
*
n

n
A

n
A
n
!r~>«R~~>R /N WC"
20,000 30,000


. A A A

n n
n n





V|X
X
jix wN(- XV


























40,000 50,000 60,000
Bed Volumes Treated
















80 -I
1 70 i
c :
•B 60 :
ro
S 50 :
o
c :
(S 40 :
(D
(/)
(D on
^ OU
ro
D)
c 9D -
ro ^u _

1 ^
s oi



Influent Cone - 70 |jg/L
A A n* * * . A





i





]
fe
10,000 20,000 30,000 40,000 50,000 60,000

n


L
L




^
A>
X
M!»!!|{>KO^

10,000
^ —
D\-\
LJ
n
J

n



X)^ XjX
X /Ix X X^
^/ ^x


20,000 30,000

D MIX
OGTO
X MetsorbG

LQ n
n
A n





*****x
































40,000 50,000 60,000
                                   Bed Volumes Treated
                                                                                             Bed Volumes Treated
                         Figure 4-12. Breakthrough of U, V, Fe, and Mn from Lyman, NE Laboratory RSSCT Columns

-------
4.6.1      Arsenic Removal. Figure 4-13 presents the arsenic breakthrough curves from ArsenXnp, E33,
GFH, MetsorbG, and Adsorbsia GTO.  All RSSCT columns were scaled to a 5.3 min full-scale EBCT
except for the two titania-based media, which were scaled to 2.5 min.  As expected, the two iron-based
media, i.e., GFH and E33, exhibited the best arsenic removal, with the run length extending to
approximately 50,000 BV by GFH and 44,000 BV by E33 at 10 (ig/L. ArsenXnp reached the 10 ng/L
level at approximately 28,000 BV. Similar to what was observed at Rimrock and LVHS (Figures 4-6 and
4-8), ArsenXnp exhibited a sharp breakthrough curve.  MetsorbG and Adsorbsia GTO had the lowest
adsorptive capacities, reaching 10 |o,g/L at approximately 21,000 and 16,000 BV, respectively.  Once
again, different ReSc values for the titania-based media did not significantly influence arsenic
breakthrough; arsenic broke through from both Adsorbsia GTO RSSCT columns at about 15,000 to
17,000 BV (Figure 4-14). Lowering the ReSc value from 2000 to 1000 alleviated the excess pressure
problem in the Adsorbsia GTO RSSCT column, while producing equivalent arsenic breakthrough curves.

4.6.2       Uranium Removal.  Figure 4-15 presents the uranium breakthrough curves from ArsenXnp,
E33, GFH, and MetsorbG.  As shown in the figure, ArsenXnp continued to remove uranium to less than 1
(ig/L as sampling was discontinued at about 50,000 BV when arsenic, which was monitored daily, had
completely broken through  from the column. It  is well known that anionic exchange resin (the resin
matrix used for producing ArsenXnp) has very large exchange capacity for uranium, especially when the
concentrations of competing anions, such as sulfate and bicarbonate/carbonate, are low in the influent
water. As noted above, the  Upper Bodfish groundwater had only 39 mg/L of sulfate and 87 mg/L (as
CaCO3) of alkalinity, compared to the 476 and 342 mg/L, respectively, found in the Lyman groundwater.
This probably explains why the chromatographic-like uranium peaking observed previously for the
Lyman groundwater did not occur for the Upper Bodfish groundwater.

Effluent uranium concentrations from GFH and  E33 exceeded the 30 |o,g/L MCL at 25,000 and 12,000
BV, respectively. Sampling of effluent uranium concentrations from the MetsorbG RSSCT column
discontinued after approximately 24,000 BV due to excessive pressure accumulation and media
compaction in the column, even though the column was designed using a reduced ReSc value of 1,000.
At 24,000 BV, the effluent uranium concentration was 24 |og/L, indicating higher adsorptive capacity than
that of GFH and E33. Additional RSSCTs conducted with two separate ReSc values (i.e., 1,000 and
2,000) for Adsorbsia GTO indicated that uranium breakthrough at 30 |o,g/L would not occur until 26,000
(with ReSc = 2,000) to 40,000 BV (with ReSc = 1,000) of operation (Figure 4-16). Therefore, while
scaling Adsorbsia GTO RSSCT columns to lower ReSc values could produce matching arsenic
breakthrough curves, this approach might not be valid for predicting uranium adsorption.

Unlike the observations made for the Lyman groundwater, the RSSCT tests on the  Upper Bodfish
groundwater demonstrated that several media were capable of removing arsenic and uranium
simultaneously.  Whereas ArsenXnp was the only media capable of remove uranium from the Lyman
groundwater, the other iron- and titania-based media exhibited significant removal capacities for uranium
from the Upper Bodfish groundwater. ArsenXnp removed uranium better than the other adsorptive media,
although GFH or E33 removed arsenic better. The ArsenXnp RSSCT predicted rather accurately the
performance of the full-scale system, which reached the arsenic MCL  at approximately 33,000  BV and
never had any detectable uranium breakthrough  (i.e., < 0.1 (ig/L). Table 4-2 summarizes the number of
BV for each media before exceeding the respective MCLs for uranium and arsenic. E33 and GFH
removed comparable amounts of arsenic, but GFH removed more uranium than E33.

As uranium is removed from the source water, it is concentrated onto the adsorptive media and may
ultimately affect the transportation and disposal  of the spent media.  Accumulation of over 0.05% (by
weight) of uranium on a media, known as "Unimportant Limit", may require a licensed broker to
transport the spent media to a  special disposal facility, such as an U.S. Ecology burial site in Grandview,
                                             33

-------
g
0
o
o
O
.o
0
0
LU
JU
~
-
40
AE33
DHIX
OGFH
X MetsorbG (ReSc=1000**)
0 GTO (ReSc= 1000**)
n

n .
n _ ^_ ^
- Influent Cone = 41 ug/L n o u ~u £
30
;
"
20
"
_
10
-
n ?
n
u o
O
n"


0 *
o * ^
_ X^ A O
\ iHafSvfcn rj*n /^i C\ O

A
A 0

A A
O
O
0


                   20,000       40,000       60,000
                         Bed Volumes Treated

      Figure 4-13.  Arsenic Breakthrough Curves from Upper
             Bodfish, CA Laboratory RSSCT Columns
80,000
ou
^ 40-
c
g
E
"£ ^n
!,*
_ 0 * 0 GTO (ReSc= 1000)
® ^) X • GTO (ReSc=2000)
E * Xx X MetsorbG (ReSc= 1000)
LJJ A. . »- ^Ix
              10,000  20,000  30,000 40,000  50,000  60,000 70,000


                         Bed Volumes Treated

      Figure 4-14.  Comparison of Two ReSc Values on Upper
            Bodfish, CA Titania-Based Media RSSCTs
                              34

-------
	 1
•&
3.
g
"ro
-i-»
c
CD
O
O
O
E
CD
I)
"c
CD
1
111






G*

^
O
"ro
"E
CD
0
o
O
c
i»_
.3
CD
^
c
CD
UJ
ou
70 :
60 :
-
50 :

_
40 :
30 -_
-
20 :
10 :

0 t
Influent Cone = 56 ug/L
A A
A A A A A 6 ^^
o

O °

A 0 0
o
X> X A E33
^( D MIX
x OGFH
X MetsorbG (ReSc=1 000**)
? ra^n nn rnn nnnrn n m
2T£L< I 	 I LJ Lil \U LJ I 	 I !l 	 I LJ Lll I 	 I * I " i iii














20,000 40,000 60,000 80,000
Bed Volumes Treated
Figure 4-15. Uranium Breakthrough from Upper Bodfish, CA
Laboratory RSSCT Columns
80
—j /-i
/( 1 -
/ \J
60

50


40


30
on
£-\J


Influent Uranium: 56 ug/L

"GTO media started to exit RSSCT at 42,000 BV




^
.• A *A
• • A
- A
• .
A A A
A AGTO(ReSc=1000)
10 4- • A \ A • GTO (ReSc=2000)
j A AMetsorbG(ReSc=1000)
n •,AA^mi , , , i , , , , i , , 	 '

















              10,000 20,000  30,000  40,000  50,000  60,000  70,000

                         Bed Volumes Treated

Figure 4-16. Comparison of Two ReSc Values on Upper Bodfish, CA
                 Titania-Based Media RSSCTs
                             35

-------
            Table 4-2. Throughput Before Reaching Arsenic and Uranium MCLs for
                    Upper Bodfish Groundwater in Lake Isabella, California
Media
ArsenXnp(HIX)
E33
GFH
MetsorbG
Adsorbsia GTO
Throughput (BV)
Arsenic
28,000
44,000
50,000
21,000
16,000
Uranium
> 50,000
12,000
25,000
> 24,000(a)
26,000
               (a)   Column failed at approximately 24,000 B V due to pressure buildup and
                    bed compaction
Idaho.  Over 50,000 BV of uranium removal would result in approximately 2 mg U/g of ArsenXnp (or
0.2%) accumulated on the media, exceeding the 0.05% "Unimportant Limit".  Alternatively, the spent
media may be regenerated in situ, which would produce a uranium-laden brine that also would require
special disposal at a licensed burial site.

4.7        South Truckee Meadows General Improvement District (STMGID)
           in Washoe County, NV (Site 6)

The well (No. 9) that supplies water to the GFH adsorption system at the STMGID facility, an EPA
arsenic removal demonstration site in Washoe County, Nevada, requires treatment to remove both arsenic
and antimony.  The MCL for antimony is 6 |og/L, and the average influent antimony concentration in the
Well No. 9 groundwater is 13 |og/L. Because the antimony concentrations are relatively low and because
the facility has some blending capacity with arsenic and antimony-free groundwater, the GFH adsorption
system is used primarily to remove arsenic, which fluctuates in concentration between 40 and 70 |o,g/L.
The objectives of the RSSCT tests were to predict the performance of the full-scale GFH system for both
arsenic and antimony  removal and to compare the adsorptive capacities and run lengths for four
adsorptive media, i.e., GFH, ArsenXnp, ARM 200, and Adsorbsia GTO.  The comparison of media
capacities  and run lengths was prompted by the unexpected early arsenic breakthrough from the full-scale
GFH system, i.e., around 8,000 BV at 10 (ig/L. Another objective of the tests was to demonstrate that
arsenic breakthrough curves for GFH would not be affected by scaling the RSSCT columns to a shorter
full-scale EBCT, a technique that can significantly shorten the required RSSCT run time.

4.7.1       Arsenic Removal. Figure 4-17 presents the arsenic breakthrough curves from four RSSCT
columns scaled to a short full-scale EBCT of 3 min. As shown by the breakthrough curves, all four media
had a rather short run  length before reaching 10 (ig/L. The longest run length  was 11,000 BV, which was
achieved by GFH, the media that is being evaluated at the STMGID, NV site. Although somewhat longer,
the RSSCT accurately predicted the unexpectedly short run length (i.e., 8,000  BV) experienced by the
full-scale system.  The run lengths achieved by the other three media were progressively shorter,
decreasing from approximately 9,000 BV for ArsenXnp, to 8,000 BV for ARM 200, and to 4,000 BV for
Adsorbsia GTO.  Similar to the problems encountered elsewhere, the Adsorbsia GTO RSSCT column
experienced pressure buildup and bed compaction. Therefore, the column run had to be ceased at about
21,000 BV.

To validate the short full-scale EBCT used, one additional RSSCT was  conducted with GFH using  a full-
scale EBCT of 6.2 min, which was close to the average EBCT of 6.5 min experienced at the site. In
doing so, the loading rates to both  columns were equal and only the media bed depth was changed from
7.7 cm (for 3 min EBCT) to 16 cm (for 6.2 min EBCT). Figure 4-18 demonstrates that, on a bed volume
                                             36

-------
CD
-b

0
o

o
O
o
'c
0
0
LU
uu
50 :
-
40 :
-
~
30 :
-
20 :
-
10 :

n i
Influent Cone = 51 ug/L A A A
A A D D
A D D D
/N

n x
X
A 0
n ^
0
o # AHIX
O * DARM200

X









OGTO(ReSc=1000)
u ~ XGFH
KAST* A

                 10,000    20,000    30,000    40,000    50,000

                         Bed Volumes Treated
  Note: Scaled to a full-scale EBCT of 3 min
    Figure 4-17. Arsenic Breakthrough Curves from STMGID, NV

                  Laboratory RSSCT Columns

Q.
C
O
*^_«
CD
-b
c
0
o
o
o
o
'c
0
-!-•
0
fc
LU
uu
50 :_
-

40 ^
-
"
30
-
;

20 :
-
10 :

A

n4j A
D
A
AD
D

X Aj

rj
<§£ A
s^1 AGFH (Simulated EBCT = 3 min)
^AU DGFH (Simulated EBCT =6.2 min)
*L ^J^tfN1 X Full Scale (Vessel A)
/-> SH^iiM'.rn 	
                 10,000     20,000    30,000    40,000


                          Bed Volumes Treated



       Figure 4-18. Comparison of GFH RSSCT and Full-Scale

        System Results for STMGID in Washoe County in NV
50,000
                              37

-------
treated basis, there was no difference between the two arsenic breakthrough curves; similar results also
were observed for other oxy-anions (data not shown).  RSSCTs conducted with shorter EBCTs would
require less water to be collected and transported and shorter operational run time. For example, the
RSSCT that simulated a 3-min full-scale EBCT had a RSSCT EBCTSC of 20 sec and required 10 days of
operation to complete the experiment, compared to a RSSCT EBCTSC of 40 sec and 17 days of operation
for the RSSCT scaled to the 6.2 min full-scale EBCT. Therefore, as long as it is within the applicable
range, a shorter full-sea le EBCT may be utilized to scale the RSSCT experiments.

Also plotted in Figure 4-18 is the arsenic breakthrough curve from the full-scale system, which reached
the 10-(ig/L level at about 8,000 BV. The slight difference between the laboratory RSSCT and full-scale
results most likely were caused by the varying influent water quality. The water sample collected during
the laboratory RSSCT testing contained 51 (ig/L of arsenic (existing entirely as As[V]), 0.16 mg/L of
phosphorous, and 23 mg/L of silica (as SiO2) and had apH value of 7.4 (the silica analysis was suspected
to be inaccurate due to instrumentation problems).  The influent water fed to the full-scale system from
system startup through just before reaching 10 (ig/L arsenic breakthrough during September 27, 2005
through March 28, 2006 contained 35.0 to 80.1  (ig/L (averaged 62.9 (ig/L) of arsenic, 0.25 to 0.46 mg/L
(averaged 0.35 mg/L) of phosphorous, and 51.5 to 95.1 mg/L (averaged 72.5) of silica; the pH values
during this period ranged from 6.5 to 7.9 and averaged 7.2.  The lower arsenic and phosphorous levels in
the RSSCT influent very likely have resulted in the later arsenic breakthrough from the RSSCT column.

The RSSCT and full-scale system results clearly indicated that the Well No. 9 water at STMGID was
difficult to treat.  Careful review of the influent water quality data suggested that the only constituents that
could have exerted  such a strong impact on arsenic adsorption were silica and phosphorous.  As noted
above, silica and phosphorous concentrations in raw water averaged 72.5 and 0.35 mg/L, respectively.
Silica at high concentrations, as reviewed in Section 1.1.1, can significantly impact arsenate adsorption by
porous metal-oxide media (Smith and Edwards  2005; Meng et al., 2000; Meng et al., 2002).  The
mechanisms involved may include changes of media surface properties, such as lowering the iso-electric
point or pH^ due to silica adsorption; competitive adsorption of silica on available adsorption sites; loss
of available adsorption sites caused by polymerization of silica; and reactions of silica with divalent
cations such as calcium, magnesium and barium to form precipitates. Phosphorous if charged also can
compete with arsenic for available adsorption sites.  The full-scale system data indicated that phosphorous
was completely removed by GFH during the first 4,500 BV and partially removed when arsenic
breakthrough had reached the 10 (ig/L.  The removal of phosphorous is further discussed in Section 4.7.3.

An additional RSSCT was later conducted for MetsorbG. The 100 x 140 mesh RSSCT media was
prepared by sieving, without crushing, a 60 x 150 mesh material supplied by the vendor (note that a
16x50 mesh material was used for full-scale applications). The RSSCT column was scaled to a full-
scale EBCT of 3  min and operated with a ReSc value of 1,000.  There was no bed compaction or other
operation problems over the 8-day RSSCT run. Arsenic did breakthrough relatively early, i.e., < 1 (ig/L
at 44 BV, 4 ng/L at 1,050 BV, 29 ng/L at 2,500 BV, 37 jig/L at 8,200 BV, and 43 ng/L at 11,000 BV.
Thus, MetsorbG treated even fewer BV than the four media already tested.

4.7.2       Antimony Removal. Figure 4-19 presents the antimony breakthrough curves from the four
RSSCT columns discussed above. In addition to the 3 min EBCT, GFH also was scaled to a full-scale
EBCT of 6.2 min. As shown by the breakthrough curves, GFH, ARM 200, and ArsenXnp exhibited little
or no adsorptive capacity for antimony. GFH reached the 6 (ig/L MCL within only 2,000 BV, similar to
that (i.e., 3,000 BV) observed at the full-scale demonstration site. Adsorbsia GTO was the only media
that appeared to have significant adsorptive capacity for antimony, removing it to the  6 (ig/L level at
approximately 20,000 BV. However, because of the operational difficulties experienced by this media,
the observed capacity might not be representative of the true capacity that the media would possess.
Table 4-3 summarizes the number of BV treated until breakthrough for arsenic and antimony.
                                              38

-------
4.7.3       Removal of Other Elements. Figure 4-20 presents the breakthrough curves for vanadium,
phosphorous, silica, and iron.  The water contained low levels of vanadium (i.e., 4 ug/L) and all media
except Adsorbsia GTO removed vanadium almost completely during the entire RSSCT study period.  The
water contained moderately high levels of phosphate (0.16 mg/L). Partial to complete phosphorous
removal was achieved by all media during the first 10,000 to 20,000 BV, consistent with what was
observed for the full-scale system. The shape of the breakthrough curves was similar to that for the
arsenic breakthrough curve for each media, although phosphorous broke through at least 5,000 to 10,000
BV earlier than arsenic. This suggests that phosphorous may compete with arsenic for adsorption sites on
the media. Silica levels were moderately high (23 mg/L [as SiO2]  - note that average influent silica
concentration to the full-scale system was 73.6 mg/L) and all the media removed some silica during the
first few thousand BV of operation (Figure 4-21).  Silica removal by the media will change the surface
chemistry properties (i.e., lower the zeta potential on the media) and in general will detrimentally impact
arsenic removal. Iron concentrations in the RSSCT effluent were comparable among the different media.
             O)
             o
            O
             o

            '
             o>
            I
1 \J
14 ;
12 '-
10 '-
8 :
-
6 ;
4 '-
_
Influent Cone = 1 3 ug/L
^^l^^^^C^S"1 ^
L-A/ rr^-
r~i A ^\

0
0

0 A AHIX
v ° D ARM20
0 A . OGTO(F
2 :- v VV XGFH
] O OGFH(6

n x





0
teSc=1000)

3 min)
                              10,000     20,000     30,000
                                        Bed Volumes Treated
40,000
50,000
               Note:  Scaled to a Full-Scale EBCT of 3 min unless mentioned otherwise

  Figure 4-19. Antimony Breakthrough Curves from STMGID, NV Laboratory RSSCT Columns
            Table 4-3. Throughput Before Reaching Arsenic and Antimony MCLs for
                                  STMGID, NV Groundwater
Media
GFH
ArsenXnp
ARM 200
Adsorbsia GTO
MetsorbG
Throughput (BV)
Arsenic
11,000
9,000
8,000
4,000
< 2,000
Antimony
< 2,000
Low
Low
20,000(a)
NA
               (a)  Result unreliable due to operational difficulties with RSSCT column
                                              39

-------
o
o
o



X


D

__
ijzj

xn*

Q
x


O
o

II
0
W
o /S
o !£,
C\l
xl ££
X < O O
•4 n o x




^x

=d °r o
1 * °
S -^ o
II <^<
o ^ri 1 1
c O
8 o ^ x
Jo g
"c ^<
	 B
o
in


o
o
o
o'
•*



o
o
o
o'
CO


o
o
- O
o'
C\l

o
o
- O
o'

I











-o
ro
0
i_
l —

|

O
-D
0
m






ooooooooooo
OCOCD-*CNOCOCD-*CN
o
o
1— 1

^^
o
o
o
II
o
o £
CM ~~~
..SOX
X ce i- LJ-
-[— ^- f n f n
•* D O X



X


n x

n x

D i
	
^
X
n x
o n ^
„ •*
— ' TN O
2 n ^
» X D ^
O 1— 1 .-^
c ^?
o ^
° n o
1 x Do
I o -«
o"
- in



- O
o
- O
o'
^


- O
o
- O
o'
CO

. o
o
- O
o'
C\l

o
o
- O
o


ri
T o in o '













-o
ro
0

'
1
2
-o
$
CD







                                                                                                                   O
                                                                                                                  U

                                                                                                                  H
                                                                                                                  U
                                                                                                                  Vl
                                                                                                                  VI
                                                                                                                   o
                                                                                                                   o
                                                                                                                  J2

                                                                                                                  J

                                                                                                                  ^  C
                                                                                                                  z  'a
                                                                                     uoj|
                                                                                                                   s  .a
                                                                                                                   o  "S
o
o

II
o
W
0
 ^ 2 O X
 x ce i- LL-
 x < o o
 •^ n o x
                                        *
                               0
                                0
                                            o
                                            o
                                            o
                                            o'
                                            in
                                            0
                                            o
                                    8  I
                                    §  |
                                    CO  H
                                            o  —
                                            o  >

                                            °  I
                                            o
                                            o
                                            o

















_l
^
CO
CM
II

0
c
o
O
'c
0
^
!^
c


3
->
o
o
o


X


1— 1
XJ
DX


•4 x

^~l
vy
>
X
ox

o
o
o

M
o
w
o£
^ O T

X < OO





n
^•b
o

0
*»<
0
0

^ SS f>
I—I
in

o
o
o
o'


o
o
o
o
CO

o
o
o
o'
CM


o
Q
o
o'




	 ^i teat ! \/ r
in o in o in o '
CN C\l T- T-











T3
0
ro
0

H
v>
3
^
T3
0
CD












                                                                                                            R  R
                                                                                                           .-  o
K/1  W
•**  u
!_ C^3
 O


 S/j

                                                                                                                   S/j
   (l/Brl) uo!}Bj}U90uoo LunjpeueA luanyg
                                                          40

-------
4.8        Spring Brook Mobile Home Park in Wales, ME (Site 7)

Three rounds of field RSSCTs were conducted at SBMHP in Wales, ME, where groundwater had a high
average pH value of 8.7.  One objective of the RSSCTs was to simulate the full-scale system, which uses
a solid-phase oxidizing media (A/P Complex 2002) to convert As(III) to As(V), followed by arsenic
adsorption via A/I Complex 2000 media. The A/I Complex 2000 media consists of an activated alumina
substrate onto which a ferric oxide coating is chemically "grafted." Because of the proprietary nature of
the full-scale media, AAFS50, an iron modified activated alumina media, was used in place of the A/I
Complex 2000 media. Another objective of the RSSCTs was to evaluate alternative media for full-scale
implementation. Arsenic removal capacities were evaluated for four iron-based media (GFH, E33, ARM
200, and Kemlron), one HlX-based media (ArsenXnp), and one titania-based media (Adsorbsia GTO).
The ARM 200 was tested a second time because the vendor had reduced the media particle size and bulk
density and it was desired to use the new media for the RSSCT. Additionally, RSSCTs were used to
evaluate the removal capacity of four adsorptive media (i.e., GFH, ARM 200, and Kemlron, and
Adsorbsia GTO) for arsenite removal.  This was accomplished by conducting one  round of RSSCTs with
filtered well water prior to the treatment with the solid-phase oxidizing media.

4.8.1       Arsenic Removal. Two rounds of RSSCTs were performed on As(V) removal. Figure 4-21
shows the breakthrough curves for AAFS50, ARM 200 (12 x  40 mesh; 46 to 50 lb/ft3), ArsenXnp, and
Adsorbsia GTO (Part 1) and Figure 4-22 shows the breakthrough curves for E33, GFH, ARM 200 (12 x
50 mesh; 35 to 40 lb/ft3),  and Kemlron (Part 2). For the most part, the iron-based adsorptive media
achieved the highest number of BV until breakthrough at 10 (ig/L (i.e., 25,000 BV with Kemlron, 23,000
BV with GFH, and 20,000 BV with E33).  The number of BV for ARM 200 was significantly lower,
ranging from 17,500, for the old media, to 13,000 BV, for the new media.  ArsenXnp achieved 18,000 BV,
which was better than the 6,700 BV achieved by the iron modified activated alumina. Both ArsenXnp and
AAFS50 exhibited steep arsenic breakthrough curves compared to that of the iron-based media. This
suggests relatively short mass transfer zone through the ArsenXnp and AAFS50 media beds.  It should be
noted that there were operational difficulties with the Adsorbsia GTO media, resulting in the media
breaking down and particles exiting the RSSCT column. By the end of the test, the volume of the media
present in the column was reduced by 40%.  Based on discussion with the media vendor, the process of
grinding the media appeared to affect the binding agent. The full-scale system using the Aquatic
Treatment Systems (ATS) A/I Complex 2000 media produced a similar breakthrough curve to that of
AAFS50, which was selected to represent the media. The  full scale performance curve for A/I Complex
2000 media is included in Figure 4-21 for comparison. The AAFS50 RSSCT appears to be able to predict
the full-scale performance of the A/I Complex 2000 iron-modified activated alumina media, which
reached 10 (ig/L after approximately 7,000 BV.

The RSSCT and full-scale test results indicate that the number of BV achieved by all media was
significantly less than that observed previously at other demonstration sites.  The low removal capacity is
attributed to the high pH value of the SBMHP groundwater of 8.7, which is close to the respective pH^
of the metal (hydr)oxides tested.

Figure 4-23 presents the breakthrough curves from the RSSCT performed using water containing arsenite.
Table 4-4 compares the bed volumes treated until breakthrough at 10 (ig/L for As (V) and  As (III).  As
shown in the table, for GFH, ARM 200, and Kemlron, the number of BV until the 10 |o,g/L breakthrough
was greater for As(III) than As(V). Although site-specific full-scale performance results are not available
for comparison, it is highly unusual for metal (hydr)oxide adsorptive media to show higher adsorptive
capacities for As(III) than for As(V). Similar to the results obtained at LVHS (Site 3), it is believed that
the RSSCT method overestimated the media's adsorptive capacities for As (III)  removal.  It appears that
the scaling equations used to design the columns need to be modified to account for the different sorption
properties/mechanisms between As(III) and As(V). Additional study is required to resolve this issue.
                                              41

-------
Effluent Arsenic Concentration (|j,g/L)
->• iv) co -P>. en c
3 O O O O O C
Average Influent Cone = 39.2 |j,g/L co^parison'^ue^to^e8^ aS
proprietary nature of the full-
A scale media.
A x x * A
AAAAA* *
AAAAA x X
A ^
A \^
A
A * xX
A~ * x
ICw^tk*^ X^
*
XARM200
xHIX
AAAFS-50
A A/I Complex 2000 (Full-Scale)

--
            10,000  20,000  30,000  40,000  50,000  60,000 70,000 80,000
                            Bed Volumes Treated

          Figure 4-21. As(V) Breakthrough Curves from Field RSSCT
                 Columns at SBMHP in Wales, ME (Part 1)
  60
240

§
§30
o

§20
fc
LU
       Average Influent Cone = 38.6 u,g/L
            D
                -D-
                                                      n
n
                                    n
n
                 D
         A
          X
                 •  A
                 X
                                       X
                         X
   o m—Pkx
     0    10,000  20,000 30,000  40,000 50,000  60,000 70,000  80,000
                          Bed Volumes Treated
          Figure 4-22. As(V) Breakthrough Curves from Field RSSCT
                 Columns at SBMHP in Wales, ME (Part 2)
                                    42

-------
            60

          Ij"
          550
          c
          o
          240
          -i—'
          c
          O>
          o
          §30
          o

          120
           O>

          I
10
     Average Influent Cone = 34.3 |j,g/L
                      »•      .   •

               0     10,000  20,000  30,000 40,000  50,000  60,000  70,000  80,000
                                     Bed Volumes Treated
             Figure 4-23.  As(lll) Breakthrough Curves from Field RSSCT Columns
                                   at SBMHP in Wales, ME
               Table 4-4. Comparison of Throughput for As(III) and As(V) Removal
                            from SBMHP Groundwater in Wales, ME
Media
Kemiron
GFH
ARM 200
Adsorbsia GTO(C)
Throughput (BV)
A(V) Removal
25,000
23,000
17,500(a)/13,000(b)
12,500
As(III) Removal
24,600
36,000
26,000(a)
10,000
               (a)  Old media with 12 x 40 mesh size and 46 to 50 lb/ft3 bulk density
               (b)  New media with 12 * 50 mesh size and 35 to 40 Ib/ft3bulk density
               (c)  Although present, data not reliable because of operational difficulties
                   encountered
4.8.2       Removal of Other Elements. Breakthrough curves for phosphorous and silica for the three
rounds of RSSCT tests are presented in Figures 4-24 (for As[V]) and 4-25 (for As[III]). As shown by the
figures, all media removed some phosphorous. Silica was partially removed during the first few thousand
BV.  There was no removal or release of fluoride, chloride, or sulfate by any of the media. In light of the
significant bed loss, the influent and effluent of the Adsorbsia GTO column also were monitored for total
and soluble titanium.  Filtered samples showed a slight increase in dissolved titanium (average 2 ug/L),
but the majority of the titanium measured in the effluent was participate (Figure 4-26), which could be
visibly observed in the effluent of the RSSCT column. The data confirmed the field observation that the
media in the RSSCT column disintegrated during the course of RSSCT test.  As noted above, grinding of
the media might have affected the integrity of the media, causing the crushed factions to fall apart.
                                              43

-------
5~ 19
D) 12 "
c -in
o lu
I
"c 8
8 °
g -
O R
8
W 4
.,_, 4
c
0)
3
qz 9
LU
n -

D«n£ttSnPA* * D A
/ A
^

*GTO
XARM200
DHIX
Average Influent Cone = 11.0 mg/L AAAFS-50

                                                                     14
          10,000  20,000  30,000  40,000  50,000  60,000 70,000 80,000
                         Bed Volumes Treated
                                                                     12 --
                                                                   o
                                                                   I
                                                                   I
                                                                   o
                                                                   O
                                                                   I
                                                                   w
                                                                      8 --
                                                                      6 --
                                                                      2 --
                                                                             Average Influent Cone = 10.0 mg/L
                                                                             10,000 20,000 30,000 40,000 50,000  60,000  70,000  80,000
                                                                                           Bed Volumes Treated
  45
_40 --
535
o 30
'•a
0
o
o 20
O
i 15
5E
L1J
       Average Influent Cone = 33.2 Lig/L
                      X
                                    -X-
                                 D
                            n
   5 --
   0 A<3C4»M1X—D
          10,000 20,000 30,000 40,000 50,000 60,000 70,000  80,000
                         Bed Volumes Treated

^J
*O)OC
£ on
"(0
-toe
0
O
O
Q- HE
^ lo
0
e 10
UJ

n i
Average Influent Cone = 30.0 Lig/L
O
D ..
°*7
n *
n
*A.
X
n -^ A x
n

K ["BtV ,[~1^A , Vf
X

*E33
nARM 200
AGFH
X Kemlron

                                                                            10,000 20,000  30,000 40,000  50,000 60,000 70,000 80,000
                                                                                          Bed Volumes Treated
             Figure 4-24. Breakthrough of Si and P from Field As(V) RSSCT Columns at SBMHP in Wales, ME
                                               (Part 1 on Left; Part 2 on Right)

-------
  10
D)
E
^—f

o
"ro
g  8 +
o
o
O  64-
03
O
LJJ
   2  :
                        X
                             "~X"
            Average Influent Cone = 1 1 .2 mg/L
                                                 • ARM200
                                                 AGFH
                                                 X Kemlron
                                                 ^GTO
     0    10,000 20,000  30,000  40,000 50,000  60,000  70,000 80,000
                          Bed Volumes Treated
  25

 o
  20
 515
  10
 c
 0
e 5
LU
       Average Influent Cone = 24 ng/L
                                                   AGFH
                                                   X Kemlron
                                                   »GTO
                                          X
                                                       X
                   , *

                   A
X
     0    10,000 20,000  30,000  40,000 50,000  60,000  70,000 80,000
                          Bed Volumes Treated
    Figure 4-25. Breakthrough of Si and P from Field As(III) RSSCT Columns
                         at SBMHP in Wales, ME
                                  45

-------
              800

              700
             rsoo
            CD
            c<
            o
            o
              100

                0
                           Ti
                        AGIO, Total Ti
                      Average Influent Cone = 1 .41 ^g/L
                           10,000      20,000      30,000
                                      Bed Volumes Treated
                                                   40,000
50,000
4.9
           Figure 4-26. Total and Soluble Ti Breakthrough Curves from As(III) RSSCT
                               Columns at SPMHP in Wales, ME
Media Adsorptive Capacity for Arsenic
To further analyze the data, the adsorptive capacity of each media (u.g As/mg of dry media) was
calculated using the arsenic breakthrough curve. Because breakthrough curves finished at varying
degrees of complete arsenic breakthrough (i.e., different As/As0 values), the adsorptive capacity
associated with an effluent arsenic concentration of 10 u.g/L was selected for comparison among media.
The area above the arsenic breakthrough curve but below the influent level was integrated and divided by
the mass of media in a RSSCT column. The mass of dry media in the RSSCT column was based upon
drying the media for 24 hr at 105°C or estimated based on packed bed geometry and dry bulk density of
each media (reported in Table 3-2). Loss of media during the tests was not accounted for. The mass of
media in a pilot- or full-scale system was estimated in a similar way.

Table 4-5 summarizes the number of BV treated and arsenic adsorptive capacities for the media tested at
all sites. Adsorptive capacities ranged from 0.05 to 2.0 mg As/g of dry media.  For the iron media, based
upon independent measurements of iron content in the media, E33  and GFH contained  177 mg Fe/g of
dry media and 228 mg Fe/g of dry media, respectively.  For the sites where E33 and GFH were tested
side-by-side, both media exhibited similar arsenic adsorptive capacities on the order of 3 to 10 mg As/g of
Fe (dry weight).
4.10
Comparison of Arsenic Removal for Different Source Waters
One benefit of conducting RSSCT is its ability to assess media performance for different source waters
using a standard testing protocol, compared to the inherent variability experienced in pilot tests. Based on
the throughput data in Table 4-5 for water containing As(V), a rank order may be established
                                              46

-------
                    Table 4-5. Summary of Arsenic Adsorption Capacities




Site
Valley Vista, AZ



Rimrock, AZ




Licking Valley High
School, Newark, OH






Village of Lyman, NE



Upper Bodfish, Lake
Isabella, CA




South Truckee Meadows
General Improvement
District, Washoe County,
NV

Spring Brook Mobile
Home Park, Wales, ME














Adsorptive
Media
AAFS50
AAFS50
E33
GFH
AAFS50
E33
E33
GFH
ArsenXnp
AAFS50
AAFS50
E33
E33
GFH
GFH
ArsenXnp
ArsenXnp
E33
ArsenXnp
MetsorbG
Adsorbsia GTO
E33
GFH
ArsenXnp
ArsenXnp
MetsorbG
Adsorbsia GTO
GFH
GFH
ARM 200
ArsenXnp
Adsorbsia GTO
A/I Complex 2000
AAFS50(b)
ESS*-1
GFH00
GFH(C)
ARM200(b)(d)
ARM200(b)(e)
ARM200(c)(d)
Kemlron(b)
Kemlron(c)
ArsenXnp(b)
Adsorbsia GTO®
Adsorbsia GTO(c)


RSSCT vs.
Full-/Pilot-Scale
System
RSSCT
Full-Scale
RSSCT
RSSCT
RSSCT
RSSCT
Full-Scale
RSSCT
RSSCT
RSSCT
Pilot-Scale
RSSCT
Pilot-Scale
RSSCT
Pilot-Scale
RSSCT
Pilot-Scale
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
Full-Scale
RSSCT
RSSCT
RSSCT
Full-Scale
RSSCT
RSSCT
RSSCT
Full-Scale
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
RSSCT
Throughput
to Reach 10 jig/L
Effluent
Concentration
(BV)
5,000-10,000
7,000
44,000
48,000
a few thousand
40,000
40,000
52,000
30,000
2,500
650
11,000
4,700
> 23,000
21,000
7,000
6,500
25,000
16,000
16,000
22,000
44,000
50,000
28,000
33,000
21,000
16,000
11,000
8,000
8,000
9,000
4,000
7,000
6,700
20,000
23,000
36,000
17,500
13,000
26,000
25,000
24,600
18,000
12,500
10,000
Adsorptive
Capacity at 10
|ig/L Effluent
Concentration
(mg As/g)
0.3
0.12(a)
1.1
1.2
0.13
1.8
1 .2(a)
2.0
1.3
0.05

1

> 1

0.3

0.4
0.2
0.2
0.3
1.4
1.4
0.9

0.6
0.5
0.4

0.3
0.3
0.2
0.28
0.19
0.62
0.61

0.57
0.41

0.65

0.51
0.40

(a)  Full-scale estimates based upon approximate masses of media in the vessel and a density of
    2.5 g/cm3
(b)  Influent containing primarily As(V)
(c)  Influent containing primarily As(III)
(d)  Old media with 12 x 40 mesh size and 46 to 50 Ib/ft3bulk density
(e)  New media with 12 x 50 mesh size and 35 to 40 Ib/ft3bulk density
                                              47

-------
among the source waters at the six EPA demonstration sites from the easiest to treat (most BV treated) to
the most difficult to treat (fewest BV treated) (Table 4-6).

Five of the six source waters had pH values ranging from 7.2 to 7.7, while the other one had a much
higher pH value which averaged 8.6. All waters had varying concentrations of other constituents (see
Table 4-1). For example, the most difficult water to treat at STMGID had the highest phosphate and
silica concentrations, which are known to have significantly decreased arsenic adsorption capacities for
all media tested.  The water also had the lowest calcium concentration; calcium can increase media
surface charge and possibly improve arsenic removal (Smith and Edwards, 2005). The second most
difficult to treat group included water at Lyman, NE and SBMHP in Wales, ME.  The Lyman, NE water
had the highest vanadium, sulfate, and TOC concentrations and the second highest phosphorous
concentration. Vanadium and phosphorous directly compete with arsenic for available adsorption sites.
Sulfate at high concentrations also may be detrimental to media adsorption. The challenge to adsorptive
media technology at the SBMHP site was the high pH value of the source water, which significantly
reduces the amount of charged sites available for arsenic adsorption.  Therefore, the three most difficult
water to treat appeared to have more challenging water chemistry for adsorptive media.

Two RSCCTs were performed on water containing arsenic primarily in the form of As(III).  In both cases,
the number of BV to reach 10 |o,g/L in the RSSCT effluent was notably higher than that achieved by the
respective pilot-scale system.  Although full-scale performance results were not available for comparison,
it is highly unusual for adsorptive media to show a higher adsorptive capacity for As (III). As a result of
these two tests, it is believed that the RSSCT method over estimated the media life for pilot/full-scale
application for As (III) treatment with the adsorptive technology.
              Table 4-6. Summary of Media Life at Six Arsenic Demonstration Sites
Ranking
for Ease
to Treat
1
2
3
Demonstration
Site
Rimrock, AZ
Valley Vista, AZ
Upper Bodfish in Lake Isabella, CA
Spring Brook Mobile Home Park in Wales,
ME
Village of Lyman, NE
South Truckee Meadows General
Improvement District, Washoe County, NV
Influent As (V)
Concentration
(Hg/L)
61.0
39.5
43.0
40.8(a)
21.5
51.0
Throughput to
lO^g/L
(BV)
A few thousand
to 52,000
7,000 to
48,000
16,000 to
50,000
6,700 to
25,000
16,000 to
25,000
4,000 to
11,000
        (a) As(III) in raw water converted to As(V) using a solid-phase oxidizing media prior to removal
           via adsorptive media
                                               48

-------
                                     5.0 REFERENCES
APHA, A. and WEF.  1998.  "Standard Methods forthe Examination of Water And Wastewater (19th
       Edition)." American Public Health Association, Washington, B.C.

Axe, L. and Trivedi, P. 2002. "Intraparticle Surface Diffusion of Metal Contaminants and Their
       Attenuation in Microporous Amorphous Al, Fe, and Mn Oxides."  Journal of Colloid and
       Interface Science, 247(2): 259-265.
Badruzzaman, M. 2005.  "Mass Transport Scaling and the Role of Silica on Arsenic Adsorption onto
       Porous Iron Oxide (Hydroxide)," Ph.D. Dissertation, Arizona State University, Tempe, AZ.
Badruzzaman, M. and Westerhoff, P.  2005. "The Application of Rapid Small-Scale Column Tests in
       Iron-Based Packed Bed Arsenic Treatment  Systems." Advances in Arsenic Research, 268-283.
Badruzzaman, M., Westerhoff, P., and Knappe, D.R.U.  2004.  "Intraparticle Diffusion and Adsorption of
       Arsenate onto Granular Ferric Hydroxide (GFH)."  Water Research, 38(18): 4002-4012.

Battelle.  2004.  "Quality Assurance Project Plan for Verification of Mini-Column Arsenic Test (QA ID
       No. 355-Q-2-1)." Prepared under Contract  No.  68-C-00-185, Task Order No. 0025, for U.S.
       Environmental Protection Agency, National Risk Management Research Laboratory, Cincinnati,
       OH.

Battelle.  2005.  "Quality Assurance Project Plan for Verification of Mini-Column Arsenic Test (QA ID
       No. 355-Q-2-1) - Addendum No.l." Prepared under Contract No. 68-C-00-185, Task Order No.
       0025, for U.S. Environmental Protection Agency, National Risk Management Research
       Laboratory, Cincinnati, OH.
Crittenden, J. C., Berrigan, J. K., and Hand, D. W.  1986. "Design of Rapid Small-Scale Adsorption Tests
       For a Constant Diffusivity." Journal Water  Pollution Control Federation, 58(4), 312-319.
Crittenden, J. C., Berrigan, J. K., Hand, D. W., and  Lykins, B.  1987. "Design of Rapid Fixed-Bed
       Adsorption Tests for Nonconstant Diffusivities." Journal of Environmental Engine ering-ASCE,
       113(2), 243-259.

Crittenden, J.C., Reddy, P.S., Arora, H., Trynoski, J., Hand, D.W., Perram, D.L., and Summers, R.S.
       1991. "Predicting GAC Performance with Rapid Small-Scale Column Tests." Journal American
       Water Works Association, 83(1): 77-87.
Fuller, C.C., Bargar, J.R., Davis,  J.A., and Piana, M.J. 2002.  "Mechanisms of Uranium Interactions with
       Hydroxyapatite: Implications for Groundwater Remediation." Environmental Science &
       Technology, 36(2): 158-165.
Ghurye, G. and Clifford, D.  2001.  "Laboratory Study on the Oxidation of Arsenic III to Arsenic V."
       EPA/600/R-01/021, U.S. Environmental Protection Agency, National Risk Management Research
       Laboratory, Cincinnati, OH.
Goldberg, S., and Johnston, C. T. 2001. "Mechanisms of Arsenic Adsorption on Amorphous Oxides
       Evaluated Using Macroscopic Measurements, Vibrational Spectroscopy, and Surface
       Complexation Modeling." Journal of Colloid and Interface Science, 234(1), 204-216.
Gu, B.H., Ku, Y.K., and Jardine, P.M. 2004. "Sorption and Binary Exchange of Nitrate, Sulfate, and
       Uranium on an Anion-Exchange Resin." Environmental Science & Technology, 38(11): 3184-
       3188.

Lin, T.F.  and Wu, J.K. 2001. "Adsorption of Arsenite and Arsenate Within Activated Alumina Grains:
       Equilibrium and Kinetics."  Water Research, 35(8): 2049-2057.

Lipps, J.P., Chen, A.S.C., and Wang, L.  2006.  "Arsenic Removal from Drinking Water by Adsorptive
       Media.  U.S. EPA Demonstration Project at Spring Brook Mobile  Home Park in Wales, ME.  Six-
                                              49

-------
       Month Evaluation Report." EPA/600/R-06/090. U.S. Environmental Protection Agency,
       National Risk Management Research Laboratory, Cincinnati, OH.
McNeill, L.S. and Edwards, M.  1997a. "Arsenic Removal during Precipitative Softening." Journal of
       Environmental Engine ering-ASCE, 123(5): 453-460.
McNeill, L.S. and Edwards, M.  1997b. "Predicting As Removal during Metal Hydroxide Precipitation."
       Journal American Water Works Association, 89(1): 75-86.

Meng, X.G., Bang, S., and Korfiatis, G.P.  2000. "Effects of Silicate, Sulfate, and Carbonate on Arsenic
       Removal by Ferric Chloride." Water Research, 34(4): 1255-1261.
Meng, X.G., Korfiatis, G.P., Bang, S.B., and Bang, K.W. 2002. "Combined Effects of Anions on
       Arsenic Removal by Iron Hydroxides." Toxicology Letters, 133(1): 103-111.

Raven, K. P., Jain, A., and Loeppert, R. H. 1998. "Arsenite and Arsenate Adsorption on Ferrihydrite:
       Kinetics, Equilibrium, and Adsorption Envelopes." Environmental Science & Technology, 32(3),
       344-349.

Smith, S.D. and Edwards, M.  2005. "The Influence of Silica and Calcium on Arsenate Sorption to Oxide
       Surfaces." Journal of Water Supply: Research and Technology - AQUA,54(4): 201-211.
Sperlich, A., Werner, A., Genz, A., Amy, G., Worch, E., and Jekel, M.  2005. "Breakthrough Behavior of
       Granular Ferric Hydroxide (GFH) Fixed-Bed Adsorption Filters: Modeling and Experimental
       Approaches."  Water Research, 39(6):  1190-1198.
Thomson, B. and Anderson, J. 2004.  "Evaluation  of Two Arsenic Adsorption Media Using a Rapid
       Small Scale Column Testing (RSSCT) Procedure." Final report to Kemiron Corporation.
Valigore, J.M., Wang, L., and Chen, A.S.C. 2006.  "Arsenic Removal from Drinking Water by
       Adsorptive Media.  U.S. EPA Demonstration Project at Valley Vista, AZ.  Six-Month Evaluation
       Report." EPA/600/R-06/083.  U.S. Environmental Protection Agency, National Risk
       Management Research Laboratory, Cincinnati, OH.
Wang, L., Condit, W.E., and Chen, A.S.C. 2004. "Technology Selection and System Design. U.S. EPA
       Arsenic Removal Technology Demonstration Program, Round 1." EPA 600-R-05-001.  U.S.
       Environmental Protection Agency, National Risk Management Research Laboratory, Cincinnati,
       OH.
Wang, L., Valigore, J.M., and Chen, A.S.C. 2005.  "Arsenic Removal from Drinking Water by
       Adsorptive Media.  U.S. EPA Demonstration Project at Rimrock, AZ. Six-Month Evaluation
       Report." EPA/600/R-05/159.  U.S. Environmental Protection Agency, National Risk
       Management Research Laboratory, Cincinnati, OH.

Westerhoff, P., Highfield, D., Badruzzaman, M., and Yoon, Y. 2005.  "Rapid Small-Scale Column Tests
       for Arsenate Removal in Iron Oxide Packed Bed Columns." Journal of Environmental
       Engineering-ASCE, 131(2): 262-271.

Wilkie, J. A., and Hering, J. G. 1996. "Adsorption  of arsenic onto hydrous ferric oxide: Effects of
       Adsorbate/Adsorbent Ratios and Co-occurring Solutes." Colloids and Surfaces a-
       Physicochemical and Engineering Aspects, 107, 97-110.
                                             50

-------