|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
DIRECT MEASUREMENT OF WATER AND SOLUTE MASS FLUXES USING A
PASSIVE SURFACE WATER FLUX METER
JULIE C. PADOWSKI
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
Julie C. Padowski
This thesis is dedicated to my siblings, Jeannie and Nick, who have also chosen the path
of higher learning. Good luck.
I would especially like to thank my major advisor, Dr. James W. Jawitz, for all the
time, effort and patience he invested in me and my work, and my committee members Dr.
Kirk Hatfield and Dr. Michael Annable for their advice and support.
I sincerely appreciate the expertise and humor of Dr. Jaeyhun Cho and his many
hours of laboratory assistance, as well as the guidance I received from Dr. Mark Newman
and Dr. Huaguo Wang. I would also like to thank the employees of the Coastal
Engineering Laboratory for access to their equipment.
I am especially grateful for the help and support of my fellow graduate students, D.
Perkins, J. Bhadha, W. Ha, E. Atkinson. In particular, I would like to thank A. Olsen for
the invaluable scholarly wisdom he imparted on me during all the hours he hovered
annoyingly behind my desk and M.K. Hamilton for all her wonderful emotional support.
Special thanks go to all the friends I have met here, especially S. Anderson, S. Curtis, and
S. Muller and to my parents and siblings for keeping me positive and focused with their
support and encouragement.
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iv
LIST OF TABLES ....................................................... ............ .............. .. vii
L IST O F FIG U R E S .............. ............................ ............. ........... ... ........ viii
A B ST R A C T .......... ..... ...................................................................................... x
1 IN TR O D U C TIO N ........................ .... ........................ ........ ..... ................
1.1 Background ............................... ................................ ...1
1.1.1 Total Maximum Daily Loads (TMDLs) ..........................................3
1.1.2 Current Methods for Determining Contaminant Loads ................4
1.2 Stu dy R ation ale .......... .......................................................... ....... ...... ... 6
1.3 O bjectiv es ....................................................... 9
2 THEORETICAL BACKGROUND AND DEVICE DESIGN............................10
2 .1 In tro d u ctio n ........................ .. ............... .. ........................................... 10
2.1.1 Flow Field Determination......... .............. ..................... 12
2.1.2 Flux Determination Under Steady State Conditions....................14
2.1.3 Flux Determination Under Transient Conditions ........................17
2.2 Development of a Passive Surface Water Flux Meter ..............................18
2.2.1 PSFM Body Development..........................................................18
2.2.2 PSFM Cartridge Development............................. ..............20
2.3 PSFM H hydraulic A analysis ........................................ ...... ............... 21
2.3.1 Laboratory Experim ents...................................... ............... 21
2.3.2 Laboratory R results ...................................................................... 22
2 .3.3 F lum e E xperim ents ............................................. .....................24
2.3.4 Flum e R esults....................................... .......... .. .. ...... .. 25
3 WATER FLUX EXPERIMENTS ............................................................. 28
3.1 Introduction ................................................ ........................ ....... 28
3.2 Porous M edia Selection ........................................ ......... ............... 28
3.3 V isual T racers .................. .......................... .................... .. 29
3.3.1 T racer C riterion .......................................... ............ ............ 30
3.3.2 M materials and M ethods ................................................ ............... 31
3.3.3 R results and Conclusions ...................................... ............... 32
3 .4 O rg an ic T racers........... ............................ ...... ................ .. .... .... .. ..3 5
3 .4 .1 T racer C riterion .......... .................................. ........ ............ 36
3.4.2 M materials and M ethods.......................................... ............... 36
3.4.3 R results and Conclusions ...................................... ............... 39
4 SOLUTE FLUX EXPERIMENTS ........................................ ...................... 47
4.1 Introduction ................................................................ .... ......47
4.2 M materials and M ethods.......................................... ........................... 48
4.3 R results and C conclusions ........................................ ......................... 50
4.3.1 Colum n Experim ents ............................. ................................. 51
4.3.2 Steady-State Flume Experiments ..............................................52
4.3.3 Transient Flume Experiments............................ ................. 57
5 CONCLUSIONS AND IMPLICATIONS................................................60
L IST O F R E F E R E N C E S ......... ...... ........... ................. ...............................................63
B IO G R A PH IC A L SK E TCH ...................................................................... ..................67
LIST OF TABLES
3-1 Ion exchange resins tested for PSFM use..................................... ............... 29
3-2 Dyes tested for PSFM use ..................................................................... 31
3-3 Tracer criteria met by dye/resin combinations.......................... ..............33
3-4 Water flux measurements using a cylindrical PSFM and a dye tracer.....................35
3-5 Parameters used for linear segments of IPA and ethanol elution functions............43
4-1 Concentrations measured from Lewatit flux meters .............................................53
4-2 Concentrations measured from Lewatit/activated charcoal flux meters ..................55
4-3 Differences between true and estimated solute fluxes (JH and J) ...........................56
4-4 Comparison of actual (JADV) vs. transient (Jr) solute fluxes and associated
p aram eters. ........................................................ ................. 59
LIST OF FIGURES
2-1 Schem atics of PSFM s of different shapes........................ .................................. 10
2-2 Estimated head-velocity relationships for two different PSFM devices..................14
2-3 Linearized segments fit to a general non-linear elution curve ..............................15
2-4 Four PSFM devices used to calculate water and solute fluxes .............................19
2-5 Diagram of cartridge calibration apparatus............... ...............................................22
2-6 Calibrations for three different types of porous media ................. ............. .....23
2-7 Comparison of "true" water velocities using a resin/tracer PSFM ..........................25
2-8 Comparison of estimated vs. true stream velocities based on differences in head ..26
3-1 Comparison of specific discharge using two different methods ...........................34
3-2 Velocity profiles for the range of velocities tested in the flume ............................37
3-3 Cartridge packed with Lewatit S6328A anion exchange resin and activated
carbon loaded with a resident tracer .......................... ........... ............. .................. 39
3-4 Breakthrough curves for a resin/tracer elution study ............................................40
3-5 Comparison of true vs. pressure-based estimates of velocities using a cylindrical
PSFM w ith resin/tracer cartridges .............................................. ..........................41
3-6 Breakthrough curves for a activated carbon/tracer elution study.............................42
3-7 True vs. pressure-based estimate of water velocity .......................................44
3-8 True vs. tracer-based estimate of water velocity................................................45
4-1 Adsorption Isotherms for a) Lewatit S6328A and b) Dowex Marathon MSA.......50
4-2 Desorption data for a) Lewatit S6328A and b) Dowex Marathon MSA resin.........51
4-3 Mass sorbed with distance along a PSFM cartridge ............ ............... 52
4-4 Comparison of estimated vs. true solute fluxes using a resin/tracer combination...54
4-5 Comparison of estimated vs. true solute fluxes using a resin/activated carbon
com bination ............................................................... ... .... ........ 57
4-6 Diagram of procedures for transient solute flux experiments.............................58
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
DIRECT MEASUREMENT OF WATER AND SOLUTE MASS FLUXES USING A
PASSIVE SURFACE WATER FLUX METER
Julie C. Padowski
Chair: James W. Jawitz
Cochair: Kirk Hatfield
Major Department: Soil and Water Science
Current methods for determining pollutant loads typically involve collecting
separate instantaneous measurements of water velocities and solute concentrations at
discrete points in space and time. The data must be combined, interpolated and integrated
after collection in order to estimate local water and solute fluxes in streams. The
frequency with which these parameters are measured typically depends upon the
availability of resources (time, money, manpower, etc.) and often leads to under-
A method is presented here for direct measurement of surface water flux
(velocity) and solute mass flux using a Passive Surface Water Flux Meter (PSFM). The
PSFM integrates instantaneous measurements of water velocity and solute concentration
to yield a direct measurement of time-averaged water and solute mass fluxes. The shape
of the PSFM profile determines the velocity distribution around a submerged device,
creating a pressure gradient across two known points. A permeable, sorptive cartridge is
attached between these points, and fluxes are measured by simultaneously sorbing
contaminants while eluting a resident tracer from this cartridge. Analysis of the cartridge
reveals the relative amount of resident tracer lost and mass of contaminant sorbed, which
is then used to calculate water and solute fluxes.
Several combinations of PSFM bodies and cartridges were tested to validate the
theoretical equations for water and solute flux. Flume laboratory trials were performed
under steady state and transient flow conditions. A cylindrical and a hydrofoil-shaped
PSFM were tested with two different tracers: a visible, non-reactive dye and a non-visible
ethanol tracer. Experiments showed that both PSFM shapes produced a sufficient
pressure gradient for flow through the cartridge to occur. The non-reactive dye and the
ethanol tracers predicted water velocities within 8% and 26% of the true stream velocity
under steady-state conditions, respectively. True vs. ethanol-estimated solute fluxes
differed by 28% when measuring solute mass flux under the same conditions. Transient
solute flux experiments were inconclusive.
Results showed that the theoretical concepts of the PSFM were validated in steady-
state flow conditions using both the cylindrical and hydrofoil-shaped PSFMs. The PSFM
shape did not affect measurement accuracy under the velocity ranges tested, however, the
type of cartridge used did affect how precisely flux was measured in the flume. In
conclusion, this preliminary study suggests that with further cartridge development the
theoretical model of the PSFM could be used to accurately predict water and solute fluxes
and holds promise for measuring pollutant loads in natural systems.
The impact of human activities on local and regional ecosystems has become a
global concern over the last several decades. Although anthropogenic processes have
been affecting overall environmental quality for millennia, it is within the past century
that noticeable changes have taken place, especially in terms of water quality degradation
(Meybeck and Helmer, 1989). The health of watersheds across the globe have been
rapidly deteriorating under the pressures of expanding human populations, industrial and
agricultural growth and other socio-economic factors (Meybeck and Helmer, 1989;
Carpenter et. al, 1998; Parry, 1998). This deterioration is commonly referred to as
"pollution" and can have long-ranging effects, both through space and time.
Water quality monitoring programs were established by the late 1800's to identify
and quantify factors that lead to the degradation of water quality. Since then, a list of
over 100 descriptors have been recognized due to advances in sampling and analyzing
techniques, including dissolved oxygen, pH, temperature, total suspended solids (TSS),
bacteria, metals, nutrients, organic matter, inorganic compounds and radionuclides; many
of which are currently used in U.S. water quality monitoring (Meybeck and Helmer,
1989; United States Environmental Protection Agency [USEPA], 2002a). These
descriptors have allowed monitoring programs to identify, characterize and even attempt
to remediate those water bodies considered polluted.
In an attempt to deal with surface water degradation, the United States Congress
passed the Clean Water Act in 1972, requiring by law that states must identify and restore
all polluted waters within their borders to acceptable water quality levels (33 USC 1313).
The Clean Water Act has been largely successful at controlling and/or eliminating
pollution threats from point sources, such as wastewater treatment plants and industrial
discharges, but has been found to be deficient in its ability to manage non-point source
pollution, such as runoff from urban and agricultural lands (Cooter, 2004; United States
Geological Survey [USGS], 1996). These sources are more difficult to monitor and
adequate characterization of a water body often requires significant time and resources to
properly sample and process water quality data, making it impractical to monitor every
stream, river and lake (Cooter, 2004).
The Passive Surface Water Flux Meter (PSFM) is described here as a new method
for surface water quality analysis. The PSFM is a sister device of the Passive Flux Meter
(PFM), which is currently being used for groundwater water quality monitoring (Hatfield
et al., 2004; Annable et al., 2005; Clark et al., 2005). Both devices measure time-
averaged solute and water fluxes at discrete locations without the aid of electronic
hardware or external controls, but the PSFM does so in surface waters instead of
groundwater (Hatfield et al., 2002). The low cost and passive nature of the device makes
it practical for large-scale field applications, such as developing Total Maximum Daily
Loads (TMDLs), by requiring little expense or manpower to use and maintain the PSFM
device. The cumulative measurements ensure sampling has taken place throughout the
length of the deployment and directly integrates measurements of discharge and solute
concentration to yield total flux, thereby reducing the number of separate measurements
that must be made.
1.1.1 Total Maximum Daily Loads (TMDLs)
According to the Clean Water Act of 1972, each state, territory or tribe is required
to develop a list of impaired waters and establish ambient water quality standards for
each lake, river and stream based on their designated use(s). For all water bodies that do
not meet these standards, a TMDL must be developed and implemented.
The TMDL was established under the Act as a means for quantifying the maximum
amount of a pollutant that a water body can receive and still meet water quality standards
(33 USC 1313). A TMDL must be created for each pollutant and should represent the
sum of allowable loads for the contaminant including background levels, point and non-
point sources, and a margin of safety to account for uncertainties (33 USC 1313).
Total maximum daily load development takes years and relies on one or more load
assessment approaches to determine water quality standards for a particular water body.
Assessments are made on a geographical (i.e., watershed) basis and must recognize and
respond to a multitude of pollutants coming from not only surface waters, but
groundwater and atmospheric deposition (USEPA, 1991). The level of effort needed to
create a TMDL depends on the size of the geographical basis, the severity of impairment,
and contaminants present (USEPA, 1991). Logically, inclusive large-scale watershed
assessments require significantly more resources than small watershed or stream
assessments, but can not be disregarded because they are necessary for investigating
regional and national water quality.
Due to the intensive, long-term nature of the TMDL process, states often solicit
other agencies and firms to monitor water quality. Frequently, the United States
Geological Survey (USGS) is called upon because its involvement in national programs
such as the National Water-Quality Assessment (NAWQA), the National Stream Quality
Accounting Network (NAS-QAN), the National Water Information System (NWIS) and
a nationwide stream-flow gaging network; all programs that have been monitoring stream
data since the early 1900's (USGS, 2000). The information from these USGS programs
provides valuable background data useful for establishing water quality criteria, and help
state agencies develop monitoring plans that will optimize sampling frequency and
location, and aid in data analysis and interpretation (USGS, 2000).
1.1.2 Current Methods for Determining Contaminant Loads
Water quality indicators are typically measured in terms of contaminant
concentration Cm [M/L3], or contaminant load MQ [M/T], which is defined as the
maximum amount of contaminant a water body may receive in a given time period and
still meet water quality criteria (USEPA, 2002a; Florida Department of Environmental
Protection [FDEP], 2004a; FDEP, 2004b; FDEP, 2004c). Measurements of stream
discharge Q [L3/T] and contaminant concentration Cm [M/L3] are often collected
simultaneously as separate point discharge and point concentrations. Total loads are
determined from these point values by integrating with respect to time
M,= Q.Cdt. (1-1)
Water flow, or discharge, is typically calculated from instantaneous measurements
of stream stage and velocity across a channel transect over a given period of time. Stage
measurements are usually taken directly using float devices or a variety of instruments
that rely upon electrical, ultrasonic, or electromagnetic signals to detect and record
changes in water levels (John and Haberman, 1980; Gupta, 1989). Stage data can then be
used to calculate the cross-sectional area for the channel's width at this transect. This
area may then be combined with velocity data to determine Q as follows:
Q = Ach (1-2)
where Ach [L2] is the cross-sectional area of a channel transect and v [L/T] is the average
velocity. The same information can also be obtained from in-situ hydraulic devices, such
as weirs, orifices and flumes, all of which use specialized equations to quantify discharge
directly (John and Haberman, 1980).
Stream water velocity is measured either automatically or manually, usually as a
time-series of point measurements. Manual velocity measurements are typically taken
with a hand-held flow meter, usually either current or acoustic doppler, and are recorded
at several depths along each section of the channel profile to yield an average velocity for
each section (Gupta, 1989; Webb et al., 1999; USEPA, 2001). The most commonly used
methods are the two-point and six-tenths-depth methods, where velocity readings are
taken as the mean of 0.2 and 0.8 depths or at 0.6 depth, respectively. In cases where
channel walls are irregularly shaped, average velocity can not be expected to be at 0.6
depth. Therefore, anywhere between 3-10 velocity point measurements are taken at
several depths along the vertical profile.
The stream water contaminant concentrations can be measured by manual
collection, with autosamplers, or with in-situ probes designed to detect specific analytes
(USEPA, 2001; Wang et al., 2004). Special precautions must be taken when collecting
water samples to ensure that the sample is protected from degradation or outside
contamination (USEPA, 2001). The depth and location at which the sample is collected
must be taken into account. Just as flow velocity varies within a channel, concentrations
also may not be uniform. Therefore, methods similar to those used for water velocity
measurements are also used when collecting water samples to ensure that representative
average concentrations are taken (Webb et al., 1999; USEPA, 2001). Individual samples
are then combined with discharge data to calculate total loads using Eq. (1-1).
As with water flow measurements, solute samples must be collected on an
appropriate time-scale to accurately characterize the system being studied (Webb et al.,
1999; USEPA, 2001). This is particularly important when looking at non-point source
contamination. Heavy rainfall and poor drainage can move large quantities of
contaminants through a given system in a short period of time. Sampling schedules often
do not coincide with these storm events, and even autosamplers can be overloaded during
events that last for an extended period of time.
1.2 Study Rationale
When taking into account the error associated with current methods for measuring
water flow and the uncertainty associated with infrequent sampling; a method for
measuring continuous water and solute fluxes seems appropriate.
The PSFM is a low-cost, low-maintenance device that directly measures average
water flux (velocity) and average solute flux. Flux could be considered a more useful
method for quantifying contaminant transport in streams because neither water flux (v)
nor solute flux (J), requires knowledge of the cross-sectional area of the channel, a
variable often difficult to calculate with accuracy. Therefore, measuring contaminants in
terms of solute flux, or mass of contaminant per unit area-time, eliminates a source of
inaccuracy incurred by estimation of Ach in relatively steady flow systems.
Because the PSFM could produce a more accurate estimation of pollutants in
stream channels, it may be useful for the restoration of impaired water bodies. Almost
half of the waters reported to the EPA as impaired cited agriculture as the source of
pollution and one-fifth cited nutrients as their main pollutant (USEPA, 2002b). Most of
these nutrient-impaired waters list nitrogen (N) and phosphorus (P) as the main sources
Currently, eutrophication problems occur throughout the United States and have
recently become a major concern. Most eutrophication problems arise when an excess
input of N and/or P accumulates in a surface water body and starts to such a level that
aquatic ecosystems and water quality are affected. Many problems have been linked to
eutrophication including decreased dissolved oxygen contents, algal blooms and a
general decrease in biodiversity, many of which can affect a water body's usefulness for
human consumption and recreation (Carpenter et al., 1998; USEPA, 1999). Most nutrient
inputs occur via non-point source runoff from urban and agricultural activities, making it
difficult to quantify and regulate (Carpenter et al., 1998). In particular, Parry (1998)
reported that nutrient pollution from agricultural sources accounts for more than 60% of
impaired rivers and 50% of impaired lakes in the USA.
A great deal of water quality management in Florida is focused on problems
associated with eutrophication due to P enrichment. Excessive fertilizer application, the
destruction of small natural wetlands and the creation of drainage ditches have all
contributed to P problem on a local and state-wide scale (Krazner, 2005). In particular,
the Everglades National Park, a naturally oligotrophic wetland system, has been
adversely affected by increased P inputs from a combination of urban and agricultural
development (Perry, 2004).
The extent to which eutrophication occurs depends not only on the nutrient load,
but on the flow of the river or stream (USEPA, 1999). It is known that advective
transport is the primary transport mechanism of solutes in surface waters, and that flow
rates and water depths can have a considerable effect on their distribution (Reddy et al.,
1999). The flow regime within a channel most obviously affects water depth and
velocity, but also indirectly affects temperature and the residence time of nutrients and
gases. In general, higher flow rates are typically negatively correlated to P removal and
sequestration by sediments and vegetation (Reddy et al., 1999). Slower flows allow
nutrients to remain in one reach of a channel for a longer time period, increasing the
availability to plants and bacteria by prolonging the residence time. Flow also affects the
amount of turbulence and re-aeration that occurs within a channel, which indirectly may
alter nutrient uptake by plants (USEPA, 1999). In systems where phosphorus is a likely
pollutant, water quality monitoring may also be useful for determining the source of the P
and the expected maximum concentrations. Therefore, the low-maintenance PSFM could
be a useful tool for monitoring P transport in flowing surface water bodies since its
design would allow solute fluxes to be calculated along the stream velocity profile at any
number of locations over extended time periods.
In this study, phosphate (P043-), was chosen as the contaminant of interest because
of its frequent listing as a major pollutant impairing streams throughout the United States.
However, before the PSFM is applied to the field, the ability of this new device to
accurately measure v and Junder steady-state and transient conditions must be tested.
The validation of the conceptual model of the PSFM under these flow conditions will be
the first step towards using this low-cost, low-maintenance PSFM design as a
revolutionary new tool for surface water quality and TMDL monitoring.
The objectives of this study were to 1) design and build a physical prototype PSFM
2) experimentally validate the conceptual and mathematical PSFM model outlined in
Klammler et al. (2004), 3) test the hydraulic properties and functional velocity ranges for
two PSFMs with different shapes, and 4) characterize and verify the ability of the PSFM
to accurately measure point solute and water flux under steady-state and transient
THEORETICAL BACKGROUND AND DEVICE DESIGN
The conceptual model of the Passive Surface Water Flux Meter (PSFM) describes a
self-contained, passive unit capable of directly measuring water and solute fluxes at a
discrete location within a flowing surface water channel (Hatfield et al., 2002; Klammler
et al., 2004). The PSFM model consists of a permeable, sorptive cartridge that is
connected to openings in the surface of an impermeable, symmetric, streamlined body
(Figure 2-1). When submerged in a flowing system, the profile of the PSFM creates a
pressure gradient along the perimeter of the device inducing flow through the cartridge.
Each cartridge is packed with a sorptive porous media impregnated with a known mass of
a resident tracer. This tracer elutes from the cartridge at a rate proportional to the stream
flow, while at the same time sorbing and retaining solutes dissolved in the water. After
retrieval, the media is analyzed in the laboratory to determine the mass of solutes sorbed
and mass of tracer remaining, giving estimations of water and solute flux, respectively.
Openings Flow Direction
Figure 2-1 Schematics of PSFMs of different shapes: a) cylindrical, and b) hydrofoil.
Water enters via opening 1 and exits through opening 2.
Flow through the cartridge a function of the pressure gradient along the surface of
the device. It is well known that the velocity distribution on the surface of an
impermeable cylinder may be calculated if for a given flow field (Green, 1939). To find
the velocity distribution along a hydrofoil shaped device, conformal mapping must be
used to relate the velocity distribution of around the cylinder to that found around a
differently shaped object. Upon knowing this distribution, the pressure gradient along the
profile of the PSFM may be determined. A significant, stable pressure distribution
around the PSFM is necessary for measuring water and solute fluxes; therefore, several
initial conditions must be met. Different flow conditions may affect the boundary layers,
and therefore the pressure distribution, along the surface of an object submerged in a flow
field (Zdravkovich, 1997). To maintain a stable pressure gradient, wakes should be
prevented from forming behind the device to decrease the possibility of interference at
the boundary layer where the openings are located, a phenomenon that might interfere
with the flow field and thus with the calculation of water and solute fluxes (Klammler et
A steady pressure distribution also depends on the ratio of the channel's dimensions
to those of the PSFM. As detailed in Klammler et al. (2004), three-dimensional flow
around the PSFM should be avoided to minimize the occurrence of flow disturbances.
To meet these conditions, the PSFM should be inserted vertically into the stream flow
and be large enough to extend the entire depth of the channel. The width of the device
should be small enough that edge effects from the channel walls do not interfere with the
flow field around the PSFM. In this study, a channel-to-PSFM ratio of 2.5:1 was
estimated as acceptable, where the maximum diameter of the PSFM was five times
smaller than the width of the channel, allowing 2.5 times the diameter of the PSFM on
either side of the device.
2.1.1 Flow Field Determination
The conceptual model laid forth in Klammler et al. (2004) provides equations for
determining the pressure difference as well as the local velocity at two openings on the
PSFM. The pressure and velocity distributions around the submerged body can then be
used to estimate the velocity of the flow field around it.
Since the shape of the PSFM defines the flow field around it, conformal mapping
may be used to relate the complex potential of the flow field to the complex coordinates
of the device. The flow field around a cylinder was adapted to that of a Joukowsky
(hydrofoil) profile by Klammler et al. (2004) as follows:
z, = [z, -(1-b).a]+ -(2-1)
[z, (1- b) a]
where z~ = x, + iy, represents the complex coordinates of an impermeable circle with a
radius of a [L], z = xJ + iy, are the transformed complex coordinates of the Joukowsky
profile in the z,-plane, b is a dimensionless parameter defining the shape of the profile by
the chord-to-width ration, which ranges between 0 to 1, where b = 0 defines a circular
(blunt) profile and b = 1 defines a profile that is a straight line (slender).
Using these transformed coordinates, Klammler et al. (2004) used v, [L/T], the
complex conjugate of the flow velocity around the PSFM to find vo [L/T], the velocity of
the flow field
vC I 2 Vo (2-2)
[z (-b .a)]2
where x [-], is a defined a proportionality constant that depends on the shape of the
PSFM body and the locations of the two openings, points 1 and 2. Once the velocity at
these points is known, the static pressures at these points can be determined using
p + = -.(p 2) (V -v2 (2-3)
pg 2g pg 2g pg 2g
where p [M/LT2] is the static pressure, vl and v2 [L/T] are local velocities on the PSFM
surface at openings 1 and 2, respectively, p [M/L3] is the density of water, and g [L/T2] is
gravitational acceleration (Klammler et al., 2004). Eqs. (2-2) and (2-3) are then
combined to yield water flux (velocity) Vo [L/T]
o = 2- --2 (2-4)
0 2 X12
where vo is determined by the static pressure differences an location of points 1 and 2.
Based on this, a known head difference between any two points may be used to
calculate the stream velocity around the device if X is known. Using the equations above,
and known x values, the velocity may be estimated for different head differences
depending on the shape and location of the openings, as shown in Figure 2-2. This
figure displays estimated velocities for a range of head differences based on real x values
used for the PSFM devices tested in this study.
0 20 40 60 80
Figure 2-2: Estimated head-velocity relationships for two different PSFM devices.
Larger head differences are produced at a given velocity using the known X
values for a cylindrical PSFM.
2.1.2 Flux Determination Under Steady State Conditions
Under steady state conditions, it is assumed that v vo, and the solute
concentration in the water are constant throughout the time of the measurement. When
these parameters are held constant, water flow through the cartridge, Qc [L3/T], can be
measured by calculating the relative amount of resident tracer remaining on the cartridge.
If the assumptions of linear, instantaneous, reversible sorption of the tracer are met, then
Qc may be calculated as follows
Qc = -(2-5)
where M,r [-] is the relative mass of the tracer remaining after elution time t [T]
(determined by quantitative analysis of the cartridge) with respect to the initial mass of
the tracer, A [L2] is the cross-sectional area of the cartridge, O [-] is the porosity of the
media, Rt is the retardation factor of the tracer and L [L] is the length of the cartridge
(Klammler et al., 2004).
Hatfield et al. (2004) developed an alternate method where by non-linear tracer
elution functions may be used to calculate water flow, by super-imposing a linear
function over segments of the non-linear elution function. In this way, the elution may be
described by different linear equations for different ranges of relative mass remaining
(Figure 2-3). Using this technique, a new Rt may be calculated for each linear segment
of the elution curve
R, = (2-6)
Here, i (i=1, 2,..., p) denotes each segment of the linear approximation, (, (P,
represents the mass fraction eluted according to the linear segment and Rd, is the
retardation factor for i (i=1, 2,..., p). These parameters are all determined directly from
the plot of the linearized pieces, where Rd, is obtained from the terminating abscissa of
segment i and pi is the intercept of segment i on the y-axis (Hatfield et al., 2004). The Rt
values calculated in Eq. (2-6) may then used in Eq. (2-5) for the appropriate M,.
0 Rdl Su Rd3
Figure 2-3: Linearized segments fit to a general non-linear elution curve. Here, three
linear segments have been fit to the non-linear elution function G(r) where p,
and Rd, define each segment [Reprinted from Hatfield et al. 2004].
After Qc has been evaluated, the head difference across the cartridge is estimated
AH= -P = (1-M,). t (2-7)
pg pg) Kt
where K [L/T] is the hydraulic conductivity of the porous media within the cartridge.
Equation (2-7) may then be used to calculate Vo from Eq. (2-4) if the x values of the
openings are known.
While Qc and AH can be described by a linear relationship, Figure 2-2 shows that
the relationship between head difference and velocity is not linear. Therefore the
relationship between stream velocity and the flow rate through the PSFM cartridge will
not be directly proportional.
Solute mass flux J, [M/(L2T)] may be calculated with the equation
J, = vo C, (2-8)
where flux measurements rely on estimates of the stream velocity, found using Eqs. (2-7)
and (2-4), and on the mass yield of the solute of interest from the sorptive media within
the cartridge (Klammler et al., 2004). The total mass of the solute Ms [M] retained by the
sorptive media is determined by quantitative analysis and used to estimate the
concentration of the solute C, [M/L3] in the stream flow where
C -- (2-9)
It is important to note that C, will be calculated incorrectly if any solute exits the
cartridge. Therefore, it is important to replace the cartridges before the sorption capacity
is exceeded. The duration of deployment is estimated based on the retardation factor
associated with a particular solute and the expected flow rate though the column. High
flow rates may introduce non-equilibrium conditions, limiting sorption and therefore
affecting the final calculation of Cs.
2.1.3 Flux Determination Under Transient Conditions
Transient conditions more closely represent natural systems and therefore are
important to take into consideration. Under steady-state conditions, the area of the
stream channel is assumed to be constant. Natural channels often are irregularly shaped,
causing stream flows to change in cross-sectional area depending on their stage height.
Variations in both v, and Ach are problematic when trying to estimate water velocity with
a PSFM under transient conditions.
In general, the water velocity Vo can be described by
S= Q(t) (2-10)
where it can be seen that ifAch(t) is constant, Vo is directly proportional to Q. In cases
where Ach(t) varies with time, there is no direct way to calculate Q without the aid of
additional equipment. Therefore, in this study transient studies were conducted only by
varying solute flux while maintaining a constant water velocity.
When flow remains relatively steady, the water and solute fluxes can be calculated
using the equations developed for steady state conditions, only the solute flux will now
represent an average solute concentration in the stream channel over the duration of
deployment. This average concentration is a direct result of the measurement itself being
time-averaged, therefore no estimates on the concentration range over that time period
may be made.
2.2 Development of a Passive Surface Water Flux Meter
Physical prototypes were developed to verify the conceptual model as described in
Klammler et al. (2004) and to test the ability and accuracy of these devices for measuring
water and solute fluxes. The PSFM was developed in two parts; the impermeable body
and the permeable, sorptive cartridge.
2.2.1 PSFM Body Development
Since a wide range of flow conditions may be experienced in the field, this study
examined the plausibility of using differently shaped devices for different flow regimes.
In theory, PSFM shapes can be selected to best fit the expected flow regime. Blunt
shapes create larger pressure gradients than more slender shapes at slower velocities (as
shown in Figure 2-2), but also develops wake formations before their more slender
counterparts. Prototype bodies with two different shapes [b = 0.85 (hydrofoil) and b =
1.0 (cylinder)] were constructed to compare their ability to measure water and solute
fluxes under various channel velocities.
This study tested four devices: a cylindrical PSFM and three successive generations
of hydrofoil PSFMs. All four were constructed with a rigid outer body and openings
along a portion of the device. The openings were installed at several different heights
along the vertical profile of each device with several openings per "level" so that fluxes
could be measured at various heights and pressure gradients.
The first two generations of hydrofoil PSFMs were constructed in a similar
manner; both devices consisted of a vertical rod (spine) mounted on a horizontal base
with several wooden "ribs" positioned along the length of the spine. The wooden ribs
were cut to a chord/width ratio where b=0.85, and were spaced evenly along the spine.
The first hydrofoil (HF 1) was made of fiberglass with h = 35 cm and L = 19.5 cm (Figure
2-4a). The second generation (HF2) with h= 26.3 cm and L= 19.5 cm was fabricated by
molding plastic around wooden ribs identical to those used in the first generation (Figure
2-4b). The third generation hydrofoil (HF3) has an h=64.1 cm and L=51 cm and was
formed by molding plastic to into a shape of b=0.85, eliminating the wooden frame and
ensuring a better profile form (Figure 2-4c). The cylinder PSFM (h=40 cm, D= 11.4 cm)
was made from a section of PVC pipe. Four openings were drilled into one front quadrant
of the device at four different heights. The device was mounted vertically to a rod for
stability during experiments (Figure 2-4d).
Figure 2-4: Four PSFM devices used to calculate water and solute fluxes. a) first
prototype of the hydrofoil-shaped PSFM (HF 1) b) second hydrofoil-shaped
PSFM (HF2) c) third prototype of the hydrofoil-shaped PSFM (HF3) d)
2.2.2 PSFM Cartridge Development
Three different methods were evaluated for estimating water and solute flux. The
first investigated the use of a visual dye tracer as the resident tracer. The other two
approaches each used a form of non-visual, organic resident tracer. Each method was
tested using a different type of cartridge. The type of porous media associated with each
method/cartridge varied depending on the resident tracer used.
All cartridges were made using Kontes borosilicate glass columns with 500 |tm
polypropylene mesh filters (Spectrum Labs, Inc.) and a Kontes nylon three-way valve at
each end. The cartridges were connected to the body of the PSFM by Tygon tubing. The
direction of flow through the column was controlled by connecting the inlet of the
column to the port with the highest static pressure.
To pack the cartridge, a slurry of the porous media was made by submerging the
solids in their original supernatant, or in deionized (DI) water when no supernatant was
available (i.e. ion exchange resins). The slurry was poured into the bottom of the
cartridge and vibrations were applied to tightly pack the material and release any trapped
air bubbles. Once full, and the cartridge was sealed using the three-way valves to close
the cartridge either end until ready for use.
The first method used cartridges (L= 15 cm, I.D. = 2.5 cm) that were packed with
Dowex Marathon MSA anion exchange resin. This media acted both as a sorbent for the
solute of interest, phosphate, and as a non-reactive matrix through which a dye tracer
could travel. Dye was injected via a three way valve on the influent end of the cartridge,
and displacement through the column was tracked visually. Upon completion of the
experiment, the cartridge was dissected, sectioned, and analyzed for phosphate. No
chemical analysis of the dye tracer was necessary.
The second method used cartridges (L= 10 cm, I.D. = 1.5 cm) that were packed
with Lewatit S6328A anion exchange resin and an organic resident tracer made of 246
ppm 1-heptanol, 2-octanol, and 165 ppm 2-ethyl-l-hexanol and 1-octanol. Water and
solute flux measurements were made by analyzing the column contents. The resin was
removed and well-mixed to ensure homogeneity before being divided for analysis.
The third approach used cartridges (L= 10 cm, I.D. = 1.5 cm), packed with two
different kinds of porous media. The Lewatit S6328A anion exchange resin was packed
into the first 45% of the cartridge for the sole purpose of sorbing phosphate. The
remainder of the column was filled with activated carbon pre-equilibrated with a suite of
organic tracers: 1000 ppm each of methanol, ethanol, isopropyl alcohol (IPA), and tert-
butyl alcohol (TBA), and 500 ppm 2,4-dimethyl-3-pentanol (2,4 DMP). Analysis of
water and solute fluxes was performed by removing the resin from the activated carbon,
then analyzing the resin for Ms, the amount of P043- sorbed, and the activated carbon for
Mtr, the amount of tracer remaining.
2.3 PSFM Hydraulic Analysis
The conceptual model of the PSFM was first tested by verifying the existence and
range of the predicted pressure gradient along the surface of the device. Ancillary
experiments were performed to investigate the accuracy with which the PSFM could
estimate stream flow velocities.
2.3.1 Laboratory Experiments
Cartridges were first calibrated under known conditions before use in experimental
trials. Calibrations were done to find the hydraulic conductivity of the cartridge's porous
media by estimating the relationship between specific discharge and hydraulic gradient.
These studies were also used to estimate the proper duration of deployment.
Cartridges were fit on either end with manometers so that pressure differences
across the column could be accurately measured. The cartridge was attached to a
constant head reservoir consisting of a 10 L Marriott bottle filled with PO43-spiked DI
water. The height of the reservoir was adjusted to create known pressure gradients and
the water discharged from the column was collected and measured (Figure 2-5). Darcy's
Law was used to relate the specific discharge q [L/T] and the hydraulic gradient dH/dl [-]
q -= -K (2-11)
where A [L2] is the cross-sectional area of the column and K [L/T] is the hydraulic
conductivity of the porous media.
Constant head reservoir
Figure 2-5: Diagram of cartridge calibration apparatus. Changes in the elevation of the
constant head reservoir are directly proportional to the change in dH/dl.
2.3.2 Laboratory Results
Results from the visual tracer method showed the resin had a K = 9.8 cm/min.
The resin used in the second method had a K= 16.7 cm/min and the resin/activated
charcoal combination used in the third method had a combined K= 18.3 cm/min (Figure
y = 9.8301x
R2 = 0.9622
0 0.1 0.2
Hydraulic Gradient (dH/dl)
y = 16.727x
0 0.1 0.2 0.-
Hydraulic Gradient (dH/dl)
y = 18.336x *
Fe = 0.9727
0 0.1 0.2 0.3 0.
Hydraulic Gradient (dH/dl)
Figure 2-6: Calibrations for three different types of porous media. a) visual tracer
method b) organic tracer (resin-only) method c) organic tracer (resin and
activated carbon) method.
2.3.3 Flume Experiments
Flume experiments were performed in a rectangular (L = 18 m, w = 0.6 m),
recirculating flume at the Coastal Engineering Laboratory at the University of Florida.
Flume velocities were controlled using a crank ball valve and further adjusted with the
use of an underflow weir located at the end of the flume channel. Local tap water from
Gainesville Regional Utilities was used in all flume experiments.
The velocity of the water in the flume Vo [L/T] was estimated using the equation
o --h (2-12)
where Wch is the width of the flume channel and hw is the height of the water. The flume
discharge Qch was obtained from the underflow weir equations developed by Prathaba et
Q = Cd -ab(2gho)2 (2-13)
Cd= 0.611. a 0072 (2-14)
where Cd [-] is the discharge coefficient, a is the gate opening [L], b is the channel width
[L], g is gravitational acceleration [L/T2] and ho is the approach flow depth [L]. Water
velocity profiles in the flume were also measured in some experiments using an acoustic
doppler velocimeter, or ADV (SonTek FlowTracker Handheld ADV). The PSFM
devices were deployed in the center of the channel a sufficient distance from the
underflow weir to avoid any perturbations in velocity due to the gate opening. U-tube
manometers were made from the device tubing before being attached to the cartridges so
that static pressure differences could be measured. The cartridges were then attached to
the body of the PSFM with the influent end connected to the opening with the higher
static pressure and the effluent end to the opening with the lower static pressure.
2.3.4 Flume Results
The true water velocity within the flume was calculated using both an underflow
weir equation vweir (Eq. 2-13) and the acoustic doppler velocimeter VADV. Comparison of
these two measurement techniques revealed a 12% difference between the two
measurements (Figure 2-7). The strong correlation (R2=0.94) between the two
techniques suggested either technique could be used to measure true water velocity;
therefore, all experiments compared estimated water fluxes to those measured by the
ADV. The ADV was selected to represent true water flux since velocity profile
information could be obtained with this device. This is particularly advantageous since
local estimated vs. measured velocities may be compared along each sampling level of
y = 0.8769x
S40 RF = 0.9444
0 20 40 60
Figure 2-7: Comparison of"true" water velocities using a resin/tracer PSFM. A one-to-
one line shows the 12% difference in measurements made by the underflow
weir vs. the ADV.
Stream flow velocity vH [L/T] was calculated by measuring the difference
between the static pressures at two points on the PSFM body. The known AH
and X values were applied to Eq. (2-4) and compared to true velocities calculated using
the underflow weir Eq. (2-13) and those estimates made using an ADV (Figure 2-8).
S*Cyl. y=1.0195x F =0.9091
0 HF3 y= 1.0921x F2= 0.8656
) 20 40 60
0 20 40 60
b. vtrue (cm/s)
Figure 2-8: Comparison of estimated vs. true stream velocities based on differences in
head. a) true velocity Vtrue is measured by ADV, estimated velocities are
measured using the cylindrical and 3rd generation hydrofoil PSFMs b) true
velocity vre, is measured with the underflow weir and compared to velocities
estimated using three generations of hydrofoil PSFMs.
Results showed estimated velocities were within 2-10% of the true stream velocity
when compared against the ADV measurements (Figure 2-8a). The cylindrical PSFM
performed slightly better than the hydrofoil shaped device (HF3), but a lack of conclusive
data about the hydrofoil PSFM leaves the accuracy of this device unknown. Data from
Figure 2-8b shows estimates made using the hydrofoil PSFMs and the sluice-gate
equation yield a higher percent difference, but when true stream velocities are adjusted by
12%, or the difference between the ADV and sluice-gate velocity measurements, the
ability of the hydrofoil PSFMs to estimate true water velocity are on par with those made
using the ADV measurements. The only exception is the first generation hydrofoil PSFM
(HF ) who under-predicts the true velocity by -33%. The first prototype's information
may be neglected since all studies were performed with successive generations.
Therefore, it may be concluded that either device may be used when estimating velocities
in the range of 15-60 cm/s, the maximum range of the flume.
WATER FLUX EXPERIMENTS
Accurate water flux estimations are important for determining mass loads in
flowing surface water systems. Therefore, three combinations of porous media and
resident tracers were tested; a visual dye resident tracer on ion exchange resin, a suite of
organic alcohol tracers on ion exchange resin, and a suite of different organic alcohol
tracers on activated carbon. Water velocities were measured using three techniques:
1. Head Difference (VH)- denoted throughout the remainder of the manuscript as the
subscript "H," where the difference in static pressures at the ports was measured
before the cartridge attached. This difference (AH) was then used to predict water
velocity vH according to Eq. (2-4).
2. Dye Movement (vD)- denoted throughout the remainder of the manuscript as the
subscript "D" where the velocity of the dye was used to estimate the average flow
rate through the cartridge. The estimated flow rate yields a AH which may then be
used to predict water velocity vD according to Eq. (2-4).
3. Tracer Remaining (vr)- denoted throughout the remainder of the manuscript as the
subscript "T," where the relative mass of tracer remaining in the cartridge is used to
determine the average flow rate through the cartridge. The estimated flow rate
yields a AH which may then be used to predict water velocity VT according to Eq.
3.2 Porous Media Selection
The two types of porous media considered for use in PSFM cartridges were ion
exchange resins and activated carbon. Activated carbon was chosen because it has been
shown as an effective matrix for the elution of tracers in groundwater flux meters
(Hatfield et al., 2004; Annable et al., 2005). The ion exchange resin was chosen for its
high permeability, its ability to function under a wide pH range, and its known ability to
selectively sorb hydrophobic and charged solutes (i.e. PO43-). In addition, the light
surface color of the resin held potential for easy visual identification of dye tracer.
One silver-impregnated granular activated carbon (989 12_30: Barnebey Sutcliffe
Corp., Columbus OH) and eight different ion exchange resins were selected for testing
based on their availability and known physical characteristics (Table 3-1).
Table 3-1. Ion exchange resins tested for PSFM use.
Resin Name Type Structure Manufacturer Manufacturer's Comments
Dowex anionic, macroporous Dow well suited for use in high
Marathon strongly organic waters and has very
MSA basic good fouling resistance
Lewatit anionic, macroporous Sybron use with hydrophilic high
S6328A strongly Chemicals molecular weight anionic
basic Inc. substances and colorants
Macro-T anionic, macroporous Ionac Brand
Amberlite anionic, gel Rohm and high chemical resistance and
strongly Haas excellent resistance to fouling
Dowex anionic, gel Dow well suited for use in high
Marathon 11 strongly organic waters and has very
basic good fouling resistance
Supelite non- gel Supelco, Inc. good for environmental
DAX-8 ionic analysis, organic and
Cabsorb cationic, crushed GSA
zeolite mineral Resources,
3.3 Visual Tracers
Dye tracer studies are commonly used for tracing water flow paths both in the
laboratory and in the field. Due to their frequent use, some dyes, such as Rhodamine
WT, Brilliant Blue, and Acid Yellow 73, have been well characterized in terms of their
sorption properties and compatible media types (e.g., Smart and Laidlaw, 1977; Kasnavia
et al., 1999; Stem et al., 2001; Kasteel et al., 2002). Few, if any, studies have been done
using dyes as tracers on ion exchange resins. Resins and other sorbents (e.g., activated
carbon, fly ash, tree bark, rice husks) have been thoroughly tested for their ability to
remove dyes from wastewater (Faria et al., 2004; Karcher et al., 2001; Karcher et al.,
2002; Morais et al., 1993; Sun and Yang, 2003), but little characterization has been
recorded on dyes that were not strongly bound by sorbents. Therefore, dye/resin
combinations were chosen based on general physical and chemical characteristics rather
than cited literature.
3.3.1 Tracer Criterion
A visual tracer, such as a dye, was considered in this study as an alternative to a
non-visual tracer when quantifying water flux. Dye tracers would be advantageous since
the volume of water that had passed through the cartridge could be detected visually
without sacrificing the entire cartridge.
In order for a dye tracer to be successful, it must 1) be chemically stable and readily
detectable both in the laboratory and the field, 2) move along the porous media within the
cartridge in plug-flow fashion, or in such a way that distance traveled by the dye could be
easily related to the volume of water that has passed though the cartridge, 3) not interfere
with the sorption or detection of the solutes of interest, and 4) be able to provide
information during a range of time periods (i.e. hours, days, weeks) appropriate for field
use. With these criteria met, water flux could then be determined by visual inspection of
the cartridge, eliminating the need to analyze the tracer.
A transparent cartridge containing a suite of dyes, each with a different color and
retardation factor, was proposed for the PSFM. The suite of dyes would contain several
dyes expressing a range of retardation factors. One dye would represent the lifetime of
the cartridge, serving as an indicator for when the cartridge should be replaced. Twelve
dyes with a low toxicity and low cost (less than $2/g) were selected from Green (1991)
based on their polarity and solubility in water (Table 3-2).
Table 3-2. Dyes tested for PSFM use.
Dye Name CAS Number Solubility Ionic Charge
Basic Blue 3 33203-82-6 30 mg/mL Cationic
Basic Yellow 11 4208-80-4 good Cationic
Brilliant Green 633-03-4 40 mg/mL Cationic
Chrysoidin 532-82-1 20 mg/mL Cationic
Malachite Green 569-64-2 60 mg/mL Cationic
Rhodamine 6G 989-38-8 20 mg/mL Cationic
Amaranth 915-67-3 60 mg/mL Anionic
Direct Red 23 3441-14-3 40 mg/mL Anionic
Erioglaucine 3844-45-9 30 mg/mL Anionic
Fluroescein 518-47-8 40 mg/mL Anionic
Phloxine B 18472-87-2 90 mg/mL Anionic
Rhodamine WT 37299-86-8 good Anionic
3.3.2 Materials and Methods
Batch and column experiments were performed using different combinations of the
8 ion exchange resins and 12 dye tracers. Tracers were studied separately so that each
dye could be characterized individually.
Batch experiments were performed to provide preliminary sorption information,
and were analyzed by visual inspection only. Desired adsorption and desorption
characteristics were based on the relative removal of dye from solution and the relative
release of dye from the resin, respectively. Desorption studies were executed using tap
water, since the chemical composition would be most similar to conditions found in field.
Column studies were used to measure dye movement in exchange resin. These
experiments were performed by either packing a small layer of resin impregnated with
the dye tracer near the inlet of a resin-filled cartridge, or by direct injection of dye into
the cartridge. Tap water was allowed to flow through the column using constant head
reservoirs for a minimum of 100 pore volumes. Dye movement along the length of the
column was measured visually from the center of mass of the dye.
3.3.3 Results and Conclusions
A total of 4 dye/resin combinations were found to meet at least half of the criteria
listed in Section 3.3.1 (Table 3-3). These combinations were:
* Basic Blue 3/Dowex Marathon MSA
* Basic Yellow 11/Dowex Marathon MSA
* Brilliant Green /Dowex Marathon MSA
* Malachite Green/Dowex Marathon MSA
Of the four possibilities, Basic Yellow 11 and Brilliant Green combinations were rejected
because of continual desorption of dye into the supernatant, even after color appeared to
have been removed from the resin. This was undesirable since PO43- analysis was
conducted using colorimetric techniques and any dye in the supernatant could affect
results. The Malachite Green combination was rejected because of a possible Fenton-
reaction chemical breakdown, explained by Dutta et al. (2003) as a known chemical
reaction such that hydroxyl radicals in weakly acidic solutions degrade Malachite Green
dye. This reaction was witnessed in laboratory experiments where loss of green color
occurred within 48 hours.
Table 3-3: Tracer criteria met by dye/resin combinations. An "X" signifies that the
criterion was met.
Basic Blue 3
Basic Yellow 11
Basic Blue 3
Direct Red 23
Dowex Marathon MSA
Dowex Marathon MSA
Dowex Marathon MSA
Dowex Marathon MSA
Dowex Marathon MSA
Dowex Marathon 11
Dowex Marathon MSA
Dowex Marathon MSA
Dowex Marathon MSA
Dowex Marathon MSA
Dowex Marathon MSA
Dowex Marathon 11
Dowex Marathon MSA
Dowex Marathon MSA
The combination that met the most criteria, Basic Blue 3/Lewatit, was used in
flume experiments to visually determine stream velocity. Cartridges were first calibrated
in the laboratory using Darcy's Law (Eq. 2-11) to measure the hydraulic conductivity,
based on the ratio of the specific discharge q [L/T] and hydraulic gradient dH/dl [-]
(Figure 3-1). This calibration was then used to compare the differences in q as
determined by the volumetric measurement of effluent collected (qD) and the velocity of
the dye moving through the column (qv) at different head gradients using the equations
qv L (3-2)
where V [L3] is the volume of effluent collected, A is the cross-sectional area of the
cartridge, t [T] is the sample collection time, L [L] is the distance the dye has traveled in
time t and r [-] is the porosity of the media. As shown in Figure 3-1, the slopes of qv and
qD are approximately equal, showing that dye is traveling at the same velocity as the
water within the cartridge (R l). Based on these results, it was concluded that the dye
provided accurate representation of the pore water velocity within the column, fully
exiting the cartridge after one pore volume.
4 q y= 12.014x- 0.3145 F = 0.9959
2 qD y=11.961x-0.293 F= 0.9996
0 0.05 0.1 0.15 0.2
Hydraulic Gradient (dH/dl)
Figure 3-1: Comparison of specific discharge using two different methods. The slope of
each line represents the measured hydraulic conductivity of the cartridge
based on two different estimation techniques, qv and qD.
Based on these results, flume experiments were performed by using a three-way
valve on the influent end of the cartridge to inject Basic Blue 3 into a glass column (I.D.
= 2.5 cm, L= 24 cm) packed with Dowex Marathon MSA exchange resin. The dye travel
time was used to estimate the velocity of water though the porous media. The head
gradient across the column was estimated based on the calibration relationship in Figure
3-1. These measurements were then used in Eq. (2-4) to solve for Vo (Table 3-4).
According to the flume experiments, the estimated water velocities differed from
the true velocities by 0.8-12.5%. This difference was considered small and thus it was
concluded that Eq. (2-4) was a valid equation for estimating stream velocities with a dye
tracer over short time periods. Although this dye tracer showed promise in terms of it's
ability to accurately water flux measurement, all dye/resin combinations tested had either
an R 1 or an R>>1, both of which were considered unacceptable for the time scale being
studied here. Estimates made at higher velocities and/or over longer periods of time
would ideally require a dye tracer with a retardation factor ranging from 3-500.
Table 3-4: Water flux measurements using a cylindrical PSFM and a dye tracer.
% Depth from True velocity Estimated velocity Difference (%)
Bottom vo (cm/s) vD (cm/s)
43 32.1 35.7 10.9
43 35.4 35.7 0.8
43 35.4 35.7 0.8
43 35.4 35.7 0.8
62 36.4 39.1 7.2
62 36.1 39.1 8.0
62 35.5 39.1 9.7
81 35.3 39.1 10.2
81 34.5 39.1 12.5
81 35.1 39.1 10.8
3.4 Organic Tracers
Previous works have shown that alcohols have been used successfully as resident
tracers on activated carbon in groundwater studies using a passive flux meter (Hatfield et
al., 2005; Annable et al., 2005). Since the PSFM is similar in nature to the groundwater
passive flux meter, cartridges were tested using sorbents and resident tracers similar to
those described in Annable et al. (2005). In this study, an activated carbon matrix and
ion exchange resin matrix were both evaluated for potential use with an organic resident
3.4.1 Tracer Criterion
The organic tracers selected for study were chosen using criteria similar to those
listed for the dye tracers: 1) must be simple to remove and detect after deployment, 2) the
mass lost after a given time period must be proportional to water flow through the
column, 3) must not interfere with the sorption or recovery of the solutes of interest, and
4) must be able to provide information during a range of time periods appropriate for
3.4.2 Materials and Methods
Two suites of organic tracers were examined, each with a different porous media
matrix. Sorption/desorption properties of the two types of resin/tracer and activated
carbon/tracer cartridges were characterized using column elution studies. The
resin/tracer combination tested tracer elution properties at flow rates of 0.6 mL/min,
while the activated carbon/tracer cartridges used flow rates of 3.1, 5.1 and 7.1 mL/min.
Results from these experiments were used to predict tracer behavior during flume
Before deployment, the vertical velocity profile of the flume was characterized at
each velocity using the ADV. In every case, the velocity profile was found to be nearly
uniform, allowing cartridges attached to the PSFM at different water depths to serve as
replicates (Figure 3-2).
Both types of cartridges were tested at similar velocity ranges in the flume, and
velocities estimated using the static pressure difference (vH) and the organic tracer (vr)
were compared to true water velocities (vo) to determine how well the PSFM performed.
S 39 cm/s
2 30 A16cm/s
5 10 L
0 20 40 60
Figure 3-2: Velocity profiles for the range of velocities tested in the flume. The
maximum range of the flume was -15-60 cm/s.
Resin/tracers. The first combination tested used Lewatit S6328A as the sorbent
and the following four alcohols as the suite of resident tracers: 1-octanol (165 ppm), 2-
ethyl-1-hexanol (165 ppm), 2-octanol (246 ppm) and 1-heptanol (246 ppm). Before
loading the resin with tracer, the Lewatit was first washed in isopropyl alcohol (IPA), the
experimental extraction solvent, to remove impurities that were found to interfere with
the tracer peak detection. The resin was then drained and rinsed with DI water. After
rinsing, 1-L of the water-based tracer solution (listed above) was added to the resin and
placed on a shaker table to equilibrate for -24 hours. After equilibration, the resin was
drained and packed into glass columns (L = 14cm, I.D. = 1.5 cm) using techniques
described in Section 2.2.2.
After deployment, the cartridges were dissected, and the resin was homogenized by
thorough stirring before equilibration with IPA for tracer extraction. Samples were
allowed to equilibrate for -24 hours at a reciprocating speed of 25 rpm, then were
removed and placed in a refrigerator to settle for -12 hours. Tracer mass remaining was
analyzed by GC/FID.
Activated carbon/tracers. The second method tested a pre-made combination of a
suite of five alcohol tracers on activated carbon. The alcohol tracers sorbed to the
activated carbon were methanol (1200 ppm), ethanol (1200 ppm), IPA (2400 ppm), TBA
(2400 ppm), and 2,4 DMP (1200 ppm). The activated carbon, however, was not used as a
sorbent for PO43- since batch studies revealed that a substantial mass of P043- could be
extracted from the unused samples of the media. Instead, the PSFM cartridge contained a
section of activated carbon in series with the Lewatit S6328A, which was used to sorb
Cartridges (L = 14 cm, I.D. = 1.5 cm) were packed so the last 575% was filled
with the activated carbon/tracer combination; the remaining 435% was packed with
Lewatit resin to sorb P043- (Figure 3-3). Lewatit was placed up-gradient of the activated
carbon so that no tracers would sorb to the resin upon elution, which could interfere with
PO43- analysis. This combination of media types was packed into glass columns (L = 14
cm, I.D. = 1.5 cm) using techniques described in Section 2.2.2.
After deployment, the cartridges were put on ice until they could be dissected and
analyzed. The cartridges were dissected by carefully separating the resin from the
activated carbon. The small amount of media in the cartridges precluded use of internal
replicates, therefore no homogenization was necessary before equilibration with isobutyl
alcohol (IBA) for tracer extraction. Samples equilibrated for -24 hours at a reciprocating
speed of 25 rpm, then were removed and placed in a refrigerator to settle for -12 hours.
Tracer mass remaining was analyzed by GC/FID.
Figure 3-3: Cartridge packed with Lewatit S6328A anion exchange resin and activated
carbon loaded with a resident tracer.
3.4.3 Results and Conclusions
Elution experiments were performed in the laboratory on each cartridge type before
being tested in flume experiments. An elution study was used to both predict tracer
desorption characteristics and how long cartridges should be deployed in the flume, since
the tracer must not fully elute before analysis.
Resin/tracers. The results from the elution study for the Lewatit/tracer
combination exhibited an nearly linear decrease in relative mass of tracer remaining,
MR/MI [-] over time (Figure 3-4). This linear trend in relative mass lost shows that at low
flow rates, equilibrium conditions exist within the cartridge and the tracer desorbs at a
rate proportional to the flow. Differences between the slopes of the four tracers became
apparent after -50 pore volumes. Because of the poor differentiation between tracers
early in the study, these tracers may be better suited for longer characterizing longer
PSFM deployments (i.e. 100-300 pore volumes). This study also revealed that 2E1H and
1-octanol had similar slopes and thus eluted at a similar rate, essentially yielding the
same results. Since the purpose of the suite of tracers is to estimate total pore volumes
over different ranges, only one of these tracers would be required in further studies if
performed over a timescale similar to that used in this study.
0 100 200 300
Figure 3-4: Breakthrough curves for a resin/tracer elution study at a flow rate of 0.6
Flume experiments were used to test performance of the PSFM body and
cartridge under controlled stream flow conditions of 16 cm/s, 39 cm/s and 57 cm/s. The
difference in pressure between the two ports (AH) were measured before cartridges were
attached to the PSFM. These differences were used to estimate water velocity (vH) using
Eq. (2-4). The deployment time for the cartridges at each velocity was based on the
expected mass of tracer remaining on the resin upon removal. Cartridges were allowed to
flow for 20, 45 and 90 minutes, respectively. Upon analysis of the cartridge, the relative
mass of tracer recovered was used to estimate the water velocity (vr). These estimated
velocities, VH and VT, were then compared to the true velocity as measured by the ADV.
Results showed that velocities estimated using the static port pressures correlated
well to true water velocities, differing on average by only 5%. The strong correlation and
small percent difference between velocities suggests that Eq. (2-4) estimated water
velocity well and may be considered an appropriate equation for PSFM water flux
estimation (Figure 3-5).
y = 0.9558x
30 =o 0.9266
0 10 20 30 40
Figure 3-5: Comparison of true vs. pressure-based estimates of velocities using a
cylindrical PSFM with resin/tracer cartridges.
Cartridges were analyzed to find the mass of tracer remaining on the resin after
deployment. In each case, tracer analysis revealed a 100% mass recovery for all
cartridges. These results are substantially different from those predicted by the elution
study (Figure 3-5). Although the true reason for cartridge failure was not investigated
here, it is believed that these results may be explained by non-equilibrium conditions
within the cartridge. The earlier elution studies were performed at a flow rate -10 times
lower than that experienced in the flume and therefore may not properly characterize
elution at higher flow rates. High flow rates through the column are a direct result of
large pressure gradients across the cartridge and therefore subsequent experiments with
this combination (left for future work) should investigate tracer elution at flow rates
expected in the cartridge during flume experiments.
Activated carbon/tracers. Several elution studies were performed using
activated carbon/tracer cartridges in order to characterize tracer elution at flow rates that
may be expected under a range of stream velocities. Experiments simulating expected
flow rates through the cartridge (3.1, 5.1 and 7.1 mL/min) were performed using a
constant head reservoir. Results revealed that ethanol and IPA tracers returned the most
uniform elution curves for the range of flow rates tested. The elution curves of both
tracers were non-linear in nature (Figure 3-6); therefore, not meeting the assumption of
linear desorption required by Eq. (2-5). Therefore, these elution curves were fit with
linear segments following the techniques used in Hatfield et al. (2004). The linear
segments whose parameters are listed in Table 3-5, produced a linear elution function for
four ranges of MR/MI. The mass remaining, as determined by GC analysis, was then be
used to calculate Qc based on the parameters of the linear segment which described it.
The cartridge flow rate was in turn used to calculate the AH across the PSFM cartridge, a
parameter required for estimating water velocity.
S0.8 5. 1 .1L/rrin
A 7.1 rL/rrrin
0 20 40 60 80
Figure 3-6: Breakthrough curves for a activated carbon/tracer elution study. a) IPA
elution at three different flow rates with linear segments described by the
parameters (p and Rdi.. b) Ethanol elution at three different flow rates with
linear segments described by the parameters (p and Rdi.
+ 3.1 rr/in
0.8 m 5.1 rr-/rrin
A 7.1 r.L/rrin
T4 | ~~ ------
Rdl d2 Rd3 Rd4
0 10 20 30 40 50
b. Pore Volumes (-)
Figure 3.6 Continued.
Table 3-5: Parameters used for linear segments of IPA and ethanol elution functions.
(p Parameter Rd, Parameter
(Pi 1.00 Rdl 2.50
(P2 0.21 Rd2 10.00
(P3 0.10 Rd3 20.00
(P4 0.06 Rd4 30.00
Isopropyl Alcohol (IPA)
(p Parameter Rd, Parameter
(Pi 1.00 Rdl 27.00
(P2 0.65 Rd2 60.00
Flume experiments were used to test the performance of the PSFM at three
different velocities (24, 39, and 53 cm/s). The differences in static port pressures, AH,
were measured before cartridges were attached to the PSFM and were used to estimate
water velocity vH using Eq. (2-4). After AH measurements were recorded, the estimated
velocity, vT, was found by deploying both hydrofoil and cylindrical PSFMs. The duration
of deployment was recorded and the cartridges attached to either device were analyzed
for the mass of tracer remaining. The proper Rd was then found according to the relative
mass remaining and was then used to calculate the average flow rate through the cartridge
(Eq. 2-5). From the flow rate, the average AH was estimated using Eq. (2-7), which was
then used to solve for the estimated water velocity (Eq. 2-4). These estimated velocities
were then compared to the true velocity (vo) as measured by the ADV.
The velocities estimated using the static pressures at the port openings (vH) were
correlated to true water velocities and differed on average by 14%. The good correlation
and relatively small difference in estimated vs. true velocities verified that Eq. (2-4) is an
appropriate equation for estimating water velocity using this technique (Figure 3-7).
0 20 40 60 80
Figure 3-7: True vs. pressure-based estimate of water velocity.
Tracer-based estimates of water velocity were made using the ethanol tracer
(Figure 3-8). Although the IPA tracer better characterized the total pore volumes flushed
during the experiment, it could not be used for analysis. Interference from the desorption
of IPA from the Lewatit S6328A resin, which had been pre-treated in IPA to remove
impurities before packing obscured true tracer information in several cartridges. Instead,
the ethanol tracer was successfully used to calculate water flux, despite its short residence
time in the cartridge.
The water velocities estimated from organic tracer elution yielded results that were
relatively similar to the measured, true velocities. Although the correlation between
tracer-estimated and true velocities was lower than the estimates made using only the
static pressure differences, measurements predicted flux reasonably well, over-estimating
true water velocity by 25%.
y = 1.2528x +
60 R = 0.7617
0 20 40 60 80
Figure 3-8: True vs. tracer-based estimate of water velocity.
Variation in both sets of estimated velocities was noticeable at higher water
velocities. Over-estimation was possibly due to sorption of some of the resident tracer to
the ion exchange resin before or after deployment, since no inert buffering material was
used between the two medias. Variations in the estimated velocities could also be due to
the temperature at which these experiments took place. Laboratory elution studies were
performed under room-temperature conditions, whereas flume experiments were
performed using water averaging approximately 80-85 degrees Fahrenheit; a by-product
of recirculating water through a large pump. This higher temperature may have altered
the desorption properties of the resident tracers, adding error to the estimates made with
Although more variation was found in the activated carbon/tracer results (Figures
3-7 and 3-8) than in the resin/tracer experiment, the good elution data suggests that
problems with water flux measurements were a product of experimental error, rather than
chemical or design failures. Based on these data, it was concluded that this combination
was best-suited for measuring water fluxes with PSFMs needing only minor
modifications to reduce errors associated with deployment.
In conclusion, it was found that velocities estimated using the difference in static
pressures at two port openings most accurately predicted true water velocity. These
results validated the use of the theoretical PSFM equations for determining water flux
using Eq. (2-4). The dye tracer yielded close estimates of true water velocity, but the
short residence time of the dye made this method of water flux estimation inappropriate
for experiments that were longer in duration. Because no dye was able to accurately
characterize the time frames our experiments were performed in, dye tracers were
abandoned for organic tracers. Organic tracers were able to satisfy more tracer criteria
than the dye tracers, but were found to estimate water velocity with less accuracy. Of the
tracers tested, the resin/tracer combination elution experiment suggested these tracers
may be appropriate for longer PSFM deployments under slow flow conditions, but
require further study for applications at higher velocities. The activated carbon/tracer
combination yielded usable elution data and the preliminary results from flume studies
suggest that this method may be the best choice for future PSFM work.
SOLUTE FLUX EXPERIMENTS
Many studies have used ion exchange resins to measure contaminant fluxes
because they strongly sorb and retain a wide range of ions (Skogley and Dobermann,
1996). In the field, resins have been successfully used to collect information on nutrient
fluxes in snowmelt, soil solutions and throughfall (Susfalk and Johnson, 2002; Skogley
and Dobermann, 1996; Simkin et al., 2003). In general, resins are advantageous over
other types of sorbents because they are manufactured, therefore these synthetic sorbents
can be made with uniform physical and chemical properties, such as particle size, ion
selectivity and sorption strength (Skogley and Dobermann, 1996). For these reasons, ion
exchange resins were selected as the sorbent for phosphate measurements in the PSFM
Two kinds of ion exchange resins, Dowex Marathon MSA anion exchange resin
and Lewatit S6328A anion exchange resin, were tested to evaluate their ability to sorb
phosphate (PO43-). In order to be considered an acceptable sorbent for the PSFM, the
media had to meet certain criteria where 1) the sorbent must have a large capacity to
tightly store the contaminant of interest and 2) the contaminant must be easy to remove
from the sorbent and analyze. The resin was tested using a combination of batch, column
and flume tests to fully characterize each potential porous media.
4.2 Materials and Methods
Batch experiments were performed to characterize PO43- sorption and desorption
from both Dowex Marathon MSA and Lewatit S6328A resins. All solute flux samples
were analyzed using glassware washed in a 1N-HC1 acid bath to remove trace
contaminants. Phosphate adsorption was characterized by combining 3 g of resin with 35
mL of a solution of NaH2PO4 (Fisher Scientific BP329-1) dissolved in DI water. Samples
represented a range of concentrations and were allowed to equilibrate for 24 hours at a
reciprocating speed of 25 rpm. All samples were allowed to settle for approximately 24
hours before analysis. The amount of P043- adsorbed by the resin was then calculated
from the amount of P043- left in the supernatant after equilibrium. Adsorption
experiments were analyzed using the Hach Total Phosphorus Digestion Method #10127
(0-100 mg/L P043-) on a Hach DR/4000 Spectrophotometer (Hach, 2004).
Desorption studies were performed by replacing the supernatant from the samples
used in the adsorption study with 2M KC1. Samples were allowed to equilibrate with the
extraction solution for 24 hours under similar conditions. After allowing -12 hours for
settling, the supernatant of the sample was again analyzed using the same methods as
described in the adsorption study.
A column experiment was performed to predict solute movement along the
cartridge matrix and mass recovery percentages. The cartridge was packed using
standard procedures detailed in Section 2.2.2, and connected to a constant head reservoir
containing a known concentration of NaH2PO4 salt dissolved in DI water. The cartridges
were removed before the solute was detected in the eluent to ensure that all PO43- was
retained and the porous media was sectioned and extracted using 2M KC1 with
techniques similar to those used in the desorption experiments. Phosphate analysis was
performed using methods and instruments similar to those described in the
Following laboratory characterization, cartridges were considered ready for use in
flume experiments. After estimating the volume of water in the flume, sufficient mass of
NaH2PO4 salt was added to reach a desired solute concentration. Incidentally, the tap
water used in the flume was determined to have undetectable levels of P043- and therefore
was assumed to contribute no extra mass. In the flume, cartridges were connected to
submerged PSFM bodies submerged and dissolved PO43- was allowed to sorb to the
cartridges for a recorded amount of time.
The Lewatit/tracer combination was tested at three different velocities (16, 39 and
57 cm/s) at an initial concentration Co = 10 mg/L PO43-. The concentration range used
was high enough to ensure that sorbed PO43- mass would be above minimum detection
limits since limited access to the flume required short-duration experiments. Columns
were tested at two different depths on the cylinder PSFM, both of which were expected to
record the same solute and water fluxes since the velocity was found to be constant with
depth in the flume (Figure 3-2). Solute mass accumulated on the column was predicted
based on the initial P043- concentration Co and the flow rate through the column.
The Lewatit/activated carbon combination was tested at a v = 24, 39, 53 cm/s at a
Co = 13, 15, 10 mg/L, respectively. Columns were tested both on the hydrofoil and
cylindrical PSFMs at two and three depths, respectively. Since the velocity is held
constant with depth in the flume, measurements occurring at the same time but at
different depths were assumed to have recorded the same solute and water fluxes. Solute
mass retained was estimated in the same way P043- concentration was predicted in the
Upon removal from the flow field, the cartridges were put on ice until they could
be dissected and analyzed. All cartridge types were divided by their respective media
type and each thoroughly mixed to ensure homogeneity. All-resin cartridges were then
divided so that half the resin was analyzed for PO43-, while the other half was analyzed
for tracers. The activated carbon/resin cartridges were separated and the resin was
analyzed for P043-. All dilutions and standards were made using tap water from the
flume in order to account for any other factors that may have affected PO43- analysis. The
small amount of media in all the cartridges precluded use of internal replicates.
4.3 Results and Conclusions
Adsorptions isotherms were created based on sample concentrations ranging
between 0 and 30 mg/L PO43-. Isotherms were made for both Lewatit S6238A and
Dowex Marathon MSA resins and results revealed a linear trend for both resin types.
The Lewatit S6328A was found to have a Kd=1.06 L/g and an R=1800 (Figure 4-la) and
the Dowex Marathon MSA showed a Kd =1.34 L/g and an R=2200 (Figure 4-1b).
Sy= 1.0613x y=1.3367x
0.3 = 0.9833 >^C 0.6 = 0.9528
) 0 1 W 0 .2 *
S0 02E 0 Z
0 0.1 0.2 0.3 0.4 0 0.2 0.4 0.6 0.8
a. Supernatant C (mg/L) b. Supernatant C (mg/L)
Figure 4-1: Adsorption Isotherms for a) Lewatit S6328A and b) Dowex Marathon MSA.
The desorption studies performed on these resins revealed an extraction efficiency
of 84% for the Lewatit S6328A and 85% for the Dowex Marathon MSA (Figure 4-2).
Based on these data, both resins were considered to sorb and desorb PO43- equally well.
Q 0.6 y=0.8394x
y = 0.8479x 999
S0.3 = 0.9979
2E 01 1 1. Ia2 ^
0 0.1 0.2 0.3 0.4 0 02 04 0.6
a. Mass Adsorbed (mg/g) b. Mass Adsorbed (mg/g)
Figure 4-2: Desorption data for a) Lewatit S6328A and b) Dowex Marathon MSA resin.
4.3.1 Column Experiments
A column experiment was performed to predict PO43- movement under simulated
flow conditions. A cartridge packed with Dowex Marathon MSA (procedures found in
Section 2.2.2) was exposed to a solution containing 18 mg/L PO43- at a flow rate of
approximately 4 mL/min for two hours. Upon removal, the resin from the cartridge was
divided into 8 sections and PO43- was extracted from each section using 2M KC1.
Colorimetric analysis of P043- was performed using methods developed by Hach (2004)
and results revealed a 92% mass recovery. In addition, analysis of each section showed
that after two hours, the majority of the P043- traveled less than 2 cm into the cartridge
before sorbing to the Dowex Marathon MSA resin (Figure 4-3). No column study was
performed on the Lewatit resin, since the adsorption/desorption profiles were similar.
0 2 4 6 8 10 12 14
Column Length (L)
Figure 4-3: Mass sorbed with distance along a PSFM cartridge. A line is super-imposed
to connect data points.
4.3.2 Steady-State Flume Experiments
Flume solute flux experiments were performed simultaneously with water flux
experiments by dissolving a known mass of NaH2PO4 in the water and allowing the
solute to sorb to the resin in the cartridge. After deployment, the cartridges were
removed and the resin thoroughly mixed and extracted with 2M KC1.
Solute fluxes were calculated based on the water velocities from Chapter 3 and the
mass of P043- recovered after extraction. Two trials were performed to measure solute
flux; one using Lewatit with organic tracers, and one with activated carbon/resin
combination with organic resident tracers.
Resin/tracers. The Lewatit/organic tracer study (Table 4-1) revealed that when
analyzed at an 85% extraction efficiency, the flux meter over-estimated the true solute
concentration by only 3% at a v = 40 cm/s, but under-estimated the same concentration
during the other trials (v=15 and 60 cm/s) by 45 and 15%, respectively. The 45%
difference between actual and estimated mean solute concentrations was most likely
exaggerated due to the fact that only a small sample set was available for that velocity
range (n=2). Sample sizes this small are not able to compensate for significant variations
in the data. In addition, the estimated concentrations reflect the errors incurred when
computing the estimated water flux and the errors associated solute analysis. Small
errors in both estimates were possibly magnified since the estimated C is the product of
these two variables.
Table 4-1: Concentrations measured from Lewatit flux meters.
Actual C Estimated C
Sorbed Sorbed (using Vo) (using vH)
Column ID (mg/g) (mg/g) (mg/L) (mg/L)
Co= 10.56 mg/L PO43'- v=60 cm/s
1-1 0.56 0.66 13.55 15.94
1-2 0.36 0.42 8.82 10.38
1-3 0.59 0.70 14.46 17.02
1-4 0.55 0.65 9.24 10.87
1-5 0.47 0.56 8.04 9.46
1-6 0.47 0.55 7.94 9.34
Co= 9.07 mg/L P043'- v=40 cm/s
2-1 0.41 0.48 9.96 11.72
2-2 0.33 0.39 8.05 9.47
2-3 0.37 0.43 7.20 8.47
2-4 0.35 0.41 6.72 7.91
Co= 8.98 mg/L P043'- v=15 cm/s
3-1 0.08 0.09 1.98 2.33
3-2 0.16 0.18 7.72 9.08
mean (n=6) 12.17
Solute flux was determined using Eq. (2-8), where Vo was the water flux measured
with the ADV in Chapter 3. Because no tracer data was available (due to no net loss of
tracer), estimated solute flux could only be calculated using Qc data as determined by the
static pressure differences at the port openings. Estimated solute flux JH, was compared
to the "true" solute flux, the product of the true water velocity (vo) and the measured Co
[M/L3]. Figure 4-4 shows that JH is well correlated to the true solute flux differing by
only 4%. This small percent difference between true and estimated solute fluxes suggests
that this PSFM method may be used for estimating solute flux when the difference in port
pressures is used to solve Eqs. (2-4) and (2-8).
y = 0.9558x
30 F- = 0.9266
0 10 20 30 40
Figure 4-4: Comparison of estimated vs. true solute fluxes using a resin/tracer
Activated carbon/resin. The P043- analysis of the cartridges in the
Lewatit/activated carbon study (Table 4-2) revealed that the flux meter under-estimated
the solute concentration in the flume water by 16, 52 and 43% for the velocities v= 53, 39
and 24 cm/s, respectively. This data set follows a similar trend to that found in Table 4-1,
where high water velocities more accurately predicted true solute concentrations. Again,
reasons for this trend may stem from the fact that higher velocities had more accurate Vo
data and had a larger number of samples than either of the trials at lower velocities.
Therefore, conclusions for the data in Table 4-2 are similar in nature to those given for
results shown in Table 4-1. Poor estimates of Qc can affect estimates of C and v and
small sample sizes can alter the true mean of time-averaged data, but it not clear which
plays a large role in the variation of C.
Table 4-2: Concentrations measured from Lewatit/activated charcoal flux meters.
Corr. Corr. C. Corr. C.
Column ID Sorbed
v=50 cm/s, Co= 13.0
1-7 (H) 0.08
1-8 (H) 0.55
v=35 cm/s, Co= 15.0
v=25 cm/s, Co= 9.8 m
3-5 (H) 0.13
50 cm/s 35
mean (n=8) 11.09 (n=
SD 2.22 SD
CV 0.20 CV
% Diff 16 %
0.71 14.38 10.65
0.42 14.50 6.27
0.57 14.44 8.27
0.21 9.33 3.17
0.15 9.61 12.79
0.12 9.66 1.56
0.15 9.63 7.07
cm/s 25 cm/s
3) 8.40 (n=5) 6.15
2.20 SD 4.99
0.26 CV 0.81
3iff 52 % Diff 43
Solute flux was determined during flume experiments using estimated water
velocities VH and VT. Table 4-3 shows results of column performance as measured by
various estimation techniques. It can be seen that for Groups 1 and 2, JH and Jr did not
differ significantly from each other, both maintaining a percent difference of -25%.
Group 3 had a large variation between mean estimated solute fluxes, due in part to
cartridge malfunctions during deployment at this velocity.
Table 4-3: Differences between true and estimated solute fluxes (JH and Jr).
Jo Cest JH JT JH JT
mg/(cm2-min) (mg/L) mg/(cm2-min) mg/(cm2-min) % Diff. % Diff.
Group 1: Co= 13.0 mg/L P043-, V=50 cm/s
Mean 40.2 11.1 50.1 48.0 21.3 27.8
S.D. 0.7 2.2 5.7 11.6 10.5 17.5
C.V. 0.02 0.20 0.11 0.24 0.49 0.63
Group 2: Co= 15.0 mg/L P043-, v=39 cm/s
Mean 31.8 8.4 34.1 22.7 19.1 27.4
S.D. 5.6 2.2 3.4 6.4 18.2 21.5
C.V. 0.18 0.26 0.10 0.28 0.95 0.78
Group 3: Co= 9.8 mg/L P043-, v=24 cm/s
Mean 14.1 3.3 13.5 7.1 19.6 49.5
S.D. 0.1 2.6 3.4 3.4 8.2 24.4
C.V. 0.00 0.78 0.25 0.48 0.42 0.49
Figure 4-5 compares the estimated versus true solute fluxes graphically. Pressure-
based fluxes (Fig 4-5a) appeared to vary by only 17% whereas tracer-based fluxes (Fig 4-
5b) both over and under-estimated solute flux by -30% in either direction. The large
differences found between Jr and Jo represents the product of smaller errors associated
with estimates Vo and C. In general, J relies heavily on an accurate approximation of Qc,
since it is used to estimate both water velocity (Eq. 2-5) and water solute concentrations
(Eq. 2-9). Any small errors associated with Qc will be incorporated into these estimates
and magnified in Jwhen using Eq. (2-8).
Despite this possible propagation of error, general positive trends were observed
and suggest that there is a possibility that better correlations may be found with a larger
sample set and/or modifications to the PSFM cartridges to reduce experimental error.
S R2 = 0.8651 .
0 20 40 60
a. Jo (mg/cm2*min)
R2 = 0.6769
0 *- Cylinder
0 20 40 60
b. Jo (mg/cm2min)
Figure 4-5: Comparison of estimated vs. true solute fluxes using a resin/activated carbon
combination, a) correlation between the true and predicted velocities based on
Eq. (2-4). b) correlation between the true and predicted velocities based on
tracer elution data.
4.3.3 Transient Flume Experiments
Transient solute flux studies were performed for all three velocities in the
Lewatit/activated carbon study. In theory, solute flux as measured by the PSFM should
report an average of instantaneous discrete flux measurements over time; therefore,
measurements made over a range of C values should be quantified by the PSFM as the
average concentration over the time the cartridge was deployed.
Transient experiments were performed in the flume by attaching three cartridges
to the cylindrical PSFM at an initial concentration (C1) at to. At each sequential time-
step, stream solute concentrations were modified by adding a known amount ofNaHPO4
to the flume. Mixing was considered instantaneous based on the speed with which water
was recirculated in the flume. One cartridge was removed each time the solute
concentration increased, thus representing the average concentration during that time
step. Cartridges measured the average C of time periods to-ti, to-t2, and to-t3 and were
expected to show the average of the concentration changes experienced while deployed
W to ti t2 t3
Figure 4-6: Diagram of procedures for transient solute flux experiments. Ci-C3 represent
increasing solute concentrations, where solute is added in a step-wise fashion.
The x-axis represents the times at which cartridges are deployed and removed.
The flume experiment where v= 25 cm/s yielded transient solute flux data that did
not accurately predict true fluxes (Table 4-4). The difference in predicted vs. estimated
fluxes differed from 9-71%. These large differences in solute flux measurements are a
direct result of the large variations associated with the estimated solute concentrations
and water velocities. In some cases, such as during time-step ti-t2, these variations are
large enough to produce a solute flux measurement that appears to be accurate, but is
actually the product of equal over- and underestimations of water and solute data. In
addition, the fact that only one measurement was taken for each transient time/solute step
means inaccuracies in data measurement (and data variation) could not be accounted for
Table 4-4: Comparison of actual (JADV) VS. transient (Jr) solute fluxes and associated
Time JADV JT S.D. % Difference
min mg/(cm2.min) mg/(cm2.min) mg/(cm2.min)
to-ti 34.7 16.6 12.8 70.5
to-t2 46.0 41.9 2.9 9.2
to-t3 55.6 35.3 14.4 44.7
Time CADV CT S.D. % Difference
min mg/L mg/L mg/L
to-ti 15.00 6.27 6.2 82.2
to-t2 19.92 11.98 5.6 49.7
to-t3 23.52 16.92 4.7 32.7
Time VADV VT S.D. % Difference
min cm/s cm/s cm/s
to-ti 38.5 42.7 2.9 10.3
to-t2 38.5 54.5 11.3 34.4
to-t3 39.4 33.2 4.4 17.2
Although an accurate estimation of transient solute flux was obscured by a
collection of errors, distinct differences in the fluxes reported give some credibility to the
theory that transient solute fluxes may be determined using the PSFM. Further studies
aimed at reducing experimental error and increasing sample sizes would be needed to
confirm this theory and should be investigated in the future.
CONCLUSIONS AND IMPLICATIONS
The purpose of this research was to construct and validate the conceptual PSFM
model by characterizing the device at a fundamental level. Of most importance was the
ability of a physical model to validate the mathematical PSFM model as described by
Klammler et al. (2004).
Two differently-shaped PSFMs were created and tested under a range of controlled
stream flows to prove that the devices could generate the proper conditions required for
measurement of water velocity and solute flux. Results from a comparison of the
cylindrical and hydrofoil-shaped PSFMs showed that both designs were able to maintain
a head gradient sufficient to accurately estimate stream velocities, validating Eq. (2-4).
Although this experiment validated an important part of the PSFM theory, these
experiments did not include a velocity range wide enough to fully characterize the
hydraulic properties of the two devices. The range studied here was limited by the
maximum velocity the flume could produce.
Cartridge development examined several possibilities for measuring stream water
and solute fluxes by testing three combinations of resident tracers and porous media. The
visual dye tracer showed promise early in the investigations by accurately estimating
stream velocities, but failed to serve as an appropriate resident tracer for the PSFM due to
short residence time within the cartridge and its possible interference with the
colorimetric methods used to detect PO43-
Organic tracers had significantly longer residence times but were found to vary in
usefulness under experimental conditions. Tracers on resins performed poorly when
tested in the flume, but tracers on activated carbon estimated the difference between true
and predicted stream velocities with slightly less accuracy than the visual tracers,
according to the mathematics developed by Klammler et. al (2004). This reduction in
accuracy is most likely a product of errors associated with construction and analysis.
Although attempts were made to reduce the uncertainty found in these results,
variations in the most fundamental measurements (i.e. calculations of "true" velocities,
measured pressure differences) were unavoidable. Despite these uncertainties, water
velocity estimates made using resident tracer elution data and direct measurements of
head differences were considered accurate enough to suggest that many of the theoretical
assumptions regarding resident tracer elution and stream velocity were valid.
The accuracy solute flux estimates were less accurate than the water flux estimates,
but this was most likely due low sample populations and the general propagation of errors
through the analysis. Because both water and solute fluxes require an accurate estimate
of Qc, any error in the flow rate through the cartridge will be incorporated in water flux
estimates and magnified in solute fluxes, due to the use of Qc to predict both stream
solute concentrations and estimated solute fluxes. In general, estimations of Qc were
often similar to the actual Qc and solute and water flux results revealed that estimates
were relatively accurate, it may be reasonable to assume that the theoretical concepts
behind the device hold promise for making accurate predictions.
The transient conditions examined during solute flux experiments were at best
inconclusive, thus unable to validate any of the PSFM conceptual model regarding these
estimates. This was partially due to significant error associated with the calculations, and
due in part to a lack of sufficient data; one experiment with insufficient replication was
not able to properly characterize the ability of the PSFM to measure transient solute
This work opened up numerous avenues for future research. In particular, a more
in-depth, proper characterization of the PSFM cartridge combinations is needed.
Accurate solute and water flux ultimately relies upon the ability to correctly interpret
changes in resident tracer masses. A closer examination of the resin/tracer cartridge
would be extremely useful, since ideally, a PSFM cartridge should only contain one
porous media. Alternatively, the separation of the activated carbon and resin into two
cartridges (attached in series) may also be tested. This method may be particularly
useful, since the resin cartridge could be modified to sorb different contaminants without
affecting the media releasing the resident tracer. No matter the method chosen,
additional verification and replication of the experiments repeated at a broader range of
velocities than those detailed in this study would contribute significantly to the successful
validation of the PSFM model.
Ideas for immediate modifications to improve PSFM data include developing a
better casing for the PSFM cartridge, since luer lock valves often loosened, and fitting the
PSFMs with new tubing to reduce air leaks during experiments. In addition, modifying
the flume bed (increasing the roughness coefficient) to create a velocity profile more like
those found in field conditions would be useful, and possibly better prepare the PSFM for
field site testing.
LIST OF REFERENCES
Annable, M.D., Hatfield, K., Cho, J., Klammler, H., Parker B.L., Cherry J.A., Rao P.S.C.,
2005. Field-scale evaluation of the passive flux meter for simultaneous
measurement of groundwater and contaminant fluxes. Environ. Sci. Technol., 39
Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N., Smith,
V.H., 1998. Nonpoint pollution of surface waters with phosphorus and nitrogen.
Ecol. Appl., 8 (3), 559-568.
Clark, C.J., Hatfield, K., Annable, M.D., Gupta, P., Chirenje, T., 2005. Estimation of
arsenic contamination in groundwater by the passive flux meter. Environ.
Forensics, 6 (1), 77-82.
Cooter, W.S., 2004. Clean Water Act assessment processes in relation to changing U.S.
Environmental Protection Agency management strategies. Environ. Sci. Technol.,
38 (20), 5265-5273.
Dutta, K., Bhattacharjee, S., Chaudhuri, B., Mukhopadhyay, S., 2003. Oxidative
degradation of malachite green by Fenton generated hydroxyl radicals in aqueous
acidic media. J. Environ. Sci. Heal. A, 38 (7), 1311-1326.
Faria, P.C.C., Orfao, J.J.M., Pereira, M.F.R., 2004. Adsorption of anionic and cationic
dyes on activated carbons with different surface chemistries. Water Res., 38 (8),
Florida Department of Environmental Protection (FDEP) 2004a. TMDL Report: Nutrient
and Dissolved Oxygen TMDL for McKay Bay. Draft. WIBD: 1584B. Bureau of
Watershed Management, Tallahassee, FL.
Florida Department of Environmental Protection (FDEP) 2004b. TMDL Report: Nutrient
TMDL for Lower Sweetwater Creek. Draft. WIBD: 1570A. Bureau of Watershed
Management, Tallahassee, FL.
Florida Department of Environmental Protection (FDEP) 2004c. Total Maximum Daily
Load for Total Phosphorus in Lake Okeechobee, FL. Final Version. Bureau of
Watershed Management, Tallahassee, FL.
Green, S.L., 1939. Hydro-and Aero-dynamics. second edition. Sir Isaac Pitman & Sons,
LTD. London. pp. 60-74.
Green, F.J., 1991. The Sigma-Aldrich Handbook of Stains, Dyes and Indicators. Aldrich
Chemical Company, Inc. Milwaukee, WI.
Gupta, R.E., 1989. Hydrology and Hydraulic Systems. Prentice Hall, Englewood Cliffs,
NJ, pp. 227-273.
Hach, 2004. Method #10127- Molybdovanadate Method with Acid Persulfate Digestion.
Hach Company. Accessed October 2005 at
Hatfield, K., Rao, P.S.C., Annable, M.D., Campbell, T.J., 2002. Device and method for
measuring fluid and solute fluxes in flow systems. Patent US 6,402,547 B1.
Hatfield, K., Annable, M., Cho, J., Rao, P.S.C., Klammler, H., 2004. A direct passive
method for measuring water and contaminant fluxes in porous media. J. Contam.
Hydrol., 75 (3-4), 155-181.
John, J.E.A., Haberman, W.L., 1980. Introduction to Fluid Mechanics. Prentice-Hall,
Englewood Cliffs, NJ, pp. 493-496.
Karcher, S., Kornmuller, A., Jekel, M., 2001. Screening of commercial sorbents for the
removal of reactive dyes. Dyes Pigments, 51 (2-3), 111-125.
Karcher, S., Kornmuller, A., Jekel, M., 2002. Anion exchange resins for removal of
reactive dyes from textile wastewaters. Water Res., 36 (19), 4717-4724.
Kasnavia, T., Vu, D., Sabatini, D.A., 1999. Fluorescent dye and media properties
affecting sorption and tracer selection. Ground Water, 37 (3), 376-381.
Kasteel, R., Vogel, H.-J., Roth, K., 2002. Effect of non-linear adsorption on the transport
behavior of Brilliant Blue in a field soil. Eur. J. Soil Sci., 53 (2), 231-240.
Klammler, H., Hatfield, K., Annable, M., Jawitz, J., Padowski, J., 2004. The passive
surface water flux meter to measure cumulative water and solute mass fluxes. EOS
Trans., 85 (47), Fall Meet. Suppl., Abstract H13C-0422.
Krazner, K., 2005. The human context for Everglades restoration: the South Florida case
study. Yale F&ES Bulletin, 107, 25-59. Accessed September 2005 at
Meybeck, M., Helmer, R., 1989. The quality of rivers: from pristine stage to global
pollution. Global Planet. Change, 75 (4), 283-309.
Morais, L.C., Freitas, O.M., Goncalves, E.P., Vasconcelos, L.T., Beca, C.G.G., 1999.
Reactive dyes removal from wastewaters by adsorption on eucalyptus bark:
variables that define the process. Water Res., 33 (4), 979-988.
Prabhata, K.S., 1992. Sluice-gate discharge equations. J. Irrig. Drain. E-ASCE, 118 (1),
Parry, R., 1998. Agricultural phosphorus and water quality: A U.S. Environmental
Protection Agency perspective. J. Environ. Qual., 27 (2), 258-261.
Perry, W., 2004. Elements of South Florida's comprehensive Everglades restoration
plan. Ecotoxicology, 13 (3), 185-193.
Reddy, K.R., Kadlec, R.H., Flaig, E., 1999. Phosphorus retention in streams and
wetlands: a review. Crit. Rev. Env. Sci. Tec., 29 (1), 83-146.
Simkin, S.M., Lewis, D.N., Weathers, K.C., Lovett, G.M., Schwarz, K., 2004.
Determination of sulfate, nitrate, and chloride in throughfall using ion-exchange
resins. Water Air Soil Poll., 153 (1-4), 343-354.
Skogley, E.O., Dobermann, A., 1996. Synthetic ion-exchange resins: soil and
environmental studies. J. Environ. Qual., 25 (1), 13-24.
Smart, P.L., Laidlaw, I.M.S., 1977. An evaluation of some fluorescent dyes for water
tracing. Water Resour. Res. 13 (1), 15-33.
Stern, D.A., Khanbilvardi, R., Alair, J.C., Richardson, W., 2001. Description of flow
through a natural wetland using dye tracer tests. Ecol. Eng., 18 (2), 173-184.
Sun, Q.Y., Yang, L.Z., 2003. The adsorption of basic dyes from aqueous solution on
modified peat-resin particle. Water Res., 37 (7), 1535-1544.
Susfalk, R.B., Johnson, D.W., 2002. Ion exchange resin based soil solution lysimeters
and snowmelt solution collectors. Comm. Soil Sci. Plan., 33 (7-8), 1261-1275.
United States Environmental Protection Agency (USEPA), 1991. Guidance for water
quality-based decisions: the TMDL process. EPA 440-4-91-001. Office of Water,
United States Environmental Protection Agency (USEPA), 1999. Protocol for
developing nutrient TMDLs. EPA 841-B-99-007. Office of Water, Washington
United States Environmental Protection Agency (USEPA), 2001. Environmental
Investigations Standard Operating Procedures and Quality Assurance Manual.
Region 4. Athens, GA.
United States Environmental Protection Agency (USEPA), 2002a. National
recommended water quality criteria: 2002. EPA 822-R-02-047. Office of Water,
United States Environmental Protection Agency (USEPA), 2002b. National water
quality inventory: 2000 report. EPA 841R02001. Office of Water, Washington
United States Geological Survey (USGS), 1996. The strategy for improving water-
quality monitoring in the United States---final report of the intergovernmental task
force on monitoring water quality. Open-file report no. OF 95-0742. U.S.
Geological Survey, Reston VA. Accessed August 2005 at
United States Geological Survey (USGS), 2002. The USGS role in TMDL assessments.
USGS Fact Sheet FS-130-01. USGS Office of Water Quality, Reston, VA.
Wang, Q., Li, Y., Obreza, T., Munoz-Carpena, R., 2004. Monitoring stations for surface
water quality. Fact Sheet SL218. University of Florida. Gainesville, FL.
Webb, W.E., Radtke, D.B., Iwatsubo, R.T., eds. 1999. Collection of water samples: U.S.
Geological Survey Techniques of Water-Resources Investigations, book 9, chap.
A4, accessed 23 August 2005 at http://pubs.water.usgs.gov/twri9A4/
Zdravkovich, M.M., 1997. Flow Around Circular Cylinders Vol: 1- Fundamentals.
Oxford University Press, New York, NY, pp.18.
Julie C. Padowski was born in North Tonawanda, New York in 1981. After
graduating from high school, she attended the University of Rochester, NY. She
graduated with a degree in environmental science (B.S.) in May 2003. In an attempt to
escape the snowy North, she moved to Florida and received a Master of Science degree
from the Soil and Water Science Department in December 2005. After graduation, she
plans to start work in a water resources related field, hopefully remaining in a tropical or
sub-tropical climate. She also plans to broaden her horizons by spending some time
traveling outside the United States.