United States
Environmental Protection
Agency
Municipal Environmental Researc*
Laboratory
Cincinnati OH 45268
Research and Development
EPA-600/S2-81-195 Oct. 1981
Project Summary
Removal of Water Soluble
Hazardous Materials
Spills from Waterways by
Activated Carbon
George R. Schneider
A model for removing water soluble
organic materials from water by
carbon-filled, buoyant packets and
panels is described. Based on this
model, equations are derived for
removing dissolved organic com-
pounds from waterways by buoyant
packets that are either (a) cycled
through a water column, or (b) sus-
pended in the waterway by natural
turbulence and by panels mechanically
suspended in waterways. Computed
results are given for phenol spills. The
report considers the effects of tur-
bulence on the suspension of buoyant
packets; it also examines how tur-
bulent mixing and longitudinal dis-
persion of spills in waterways affect
the removal of water soluble hazardous
materials.
Buoyant packets are found to be
ineffective for removing spills from
waterways. The rapid dilution of spills
also renders panels ineffective unless
the spill is massive and the response is
rapid.
This Project Summary was devel-
oped by EPA's Municipal Environ-
mental Research Laboratory. Cincin-
nati, OH. to announce key findings of
the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
Activated carbon in one form or
another (packaged or loose, floating or
sinking in water, powdered, granular, or
fibrous) has been proposed for adsorbing
spilled, water soluble hazardous or-
ganics from waterways. Limited labora-
tory experiments and small field tests
have been conducted with carbon to
determine its spill cleanup potential.
The results of these programs have
been mixed, and it is not clear whether
the use of carbon-filled packets and
panels to clean up spills in large
nonimpoundable waterways is tech-
nically or economically feasible.
A technical feasibility study has been
conducted of the various in-situ
approaches to the removal of water
soluble hazardous organics from non-
impoundable waterways by means of
adsorption on carbon. The objective of
this study was to determine the removal
rate of dissolved pollutant and the
percent of dissolved pollutant likely to
be removed by using activated carbon in
packets and panels under the best
possible conditions. Related steps such
as the retrieval of the carbon-filled
packets and carbon regeneration were
not considered.
Analysis
Adsorption Model Basics
Packets or panels of carbon granules
immersed in a polluted waterway will be
considered to behave as fixed-bed
adsorbers with a small depth; the
carbon will adsorb dissolved organic
material only from fluid that is inside the
-------
packet (either stagnant or flowing
through). Packets of carbon that are
merely immersed in fluid with no flow
through them will adsorb organic
material only from the stagnant liquid in
the void space of the packet. Movement
of fluid through the porous packets is
induced by pressure differences across
the packets resulting from buoyancy
forces, turbulence, and velocity differ-
ences between the packets and the
surrounding fluid.
When the flowrate through the
carbon-filled packets or panels was
calculated, the flow resistance offered
by the fabric used to encase the
adsorbent was neglected, and the
packets were always assumed to be
oriented broadside to the external fluid
flow.
The overall resistance to mass transfer
from the bulk liquid to the solid includes
the resistive contribution of (a) the layer
of solution around the particles, (b)
diffusion of solute through the pores
within the particles, and (c) physical
adsorption at active sites on the solid.
The mass transfer coefficients from
the bulk liquid to the exterior surface
area of the carbon particles (step a
above) were computed using general
correlations obtained in packed beds at
low Rd numbers.
For steady-state mass transfer from
the water phase to the carbon, and for
piston flow through the packed bed (or
packet).
-UadC=kca(C-C,)dx.
(1)
The resistance offered by the pore
diffusion and physical adsorption steps
is assumed to be very small relative to
the resistance to mass transfer from the
bulk liquid to the exterior surface of the
carbon. Therefore, C, can be closely
approximated as the concentration of
the solute in equilibrium with the solid.
That is, the solute concentration C, is
assumed to be given by an experi-
mentally determined adsorption equi-
librium isotherm. The latter could be
used to relate C, to the bulk fluid
concentration, C, by means of a material
balance if the initial moles of solute and
the mass of carbon were known. Since
the carbon packets considered in this
study are fairly thin, and since the
packets moving through the liquid may
have fluid passing through them in
either direction (depending on which of
the two broad sides are facing the flow),
C, will be assumed to be constant for all
carbon granules and hence independent
of distance, x, through the packet. C, will
vary with time as the concentration of
organic material on the carbon in-
creases. The change in concentration
across a packet is given by integrating
Equation 1 from 0 to L and C,n to Cout to
yield after rearrangement
CourCm -
1 - exp I
c,
\-,
. kcaL ^
Ua ,
(2)
For fresh activated carbon, C, is
negligible for many organics, and the
fraction of pollutant removed from
water passing through a fresh carbon
packet is then given as
Cm - C0ut- i _ g™ I . kcal_ i
Cm ^ ~07~ I (3)
Pollutant Removal by
Buoyant Packets
Buoyant Packet Cycling
Some proponents have suggested
that waterways or lakes polluted by
spills of water soluble hazardous
organic compounds could be cleaned up
faster or more efficiently if buoyant
packets of activated carbon adsorbent
were used. These packets would be
fabricated in sizes small enough to be
mixed with water and injected at the
botttom of a polluted volume of water by
a solids-handling pump; they would be
buoyant enough to rise through the
pollutant water at some known terminal
velocity. The movement of the packet
upward through the water would
enhance mass transfer of the pollutant
to the carbon. Upon arrival at the water
surface, the packets would be automati-
cally gathered and pumped down to the
bottom of the water column again to
start another cycle.
A packet size of 10.2 by 10.2 cm (4 by
4 in.) with a thickness of either 1.27 cm
(0.5 in.) or of 0.64 cm (0.25 in.) was
chosen. Buoyancy was assumed to be
provided by adding foamed plastic
particles with a specific gravity of 0.2 to
the carbon granules used to fill the
packets. The foamed plastic would be
equal in size to the carbon granules
used. Calculations were performed for
carbon granules with a diameter of 0.12
and 0.06 cm (14 and 28 mesh).
For the purpose of this study, the
cycling of buoyant packets was assumed
to be used for the clean up of 3,786 m3
(106 gal) of water contaminated with
100 ppm of phenol. A carbon-to-phenol
ratio of 10 was assumed. The packets
were assumed to rise with the large
rectangular face normal to the flow
direction. The AP required to produce a
given terminal rise velocity, vt, for the
packet was calculated. This computed
AP was in turn used to calculate the
flowrate through the packet, Ua.
A packet rising through the water
column of depth D at velocity vt will treat
UaApD/Vt units of water. The fraction of
phenol removed from the water passing
through a packet is given by Equation 3.
Thus the amount of phenol removed
from the water by N packets of fresh
carbon after one pass through a depth D
at vt is given by the equation
VAC= P_UaApNC
vt
(4)
As the packets float on the water
surface before they are gathered and
returned to the river bottom, the carbon
will adsorb phenol from the fluid A
remaining in the void space of the \
packets. To account for this adsorption,
it has been assumed that for each cycle,
in addition to the pollutant adsorbed
while rising through the water, the
carbon in a packet will remove enough
phenol to attain equilibrium with a
volume of phenol-water mixture suffi-
cient to occupy the packet void space—
that is, (C-C,)NVP0. Thus the total
amount of phenol removed per cycle is
given by adding this quantity to Equation
4.
The total number of cycles, n, required
by fresh carbon packets to remove a
given fraction of the phenol (or any
pollutant) can be found from Equation 5,
1n
C _ Nn
vt
-UaAp
1 - expi
kcaL
Ua
j (5)
The total time required to reduce the
phenol concentration from C0 to C is
primarily a function of the packet
-------
retrieval time, the solids-handling pump
capacity, and the number of cycles, n.
According to Equation 5, if N should be
increased, the number of packet cycles
will be unaffected (N/V is determined by
the carbon to phenol ratio and the
amount of flotation matter required).
This effect applies to fresh carbon
packets only (the restrictions under
which Equation 5 was derived). Thus to
complete the desired cleanup job, the
same number of packets would have to
be retrieved and pumped to the bottom
of the water column. The volume of
fluid-solids handled by the recycling
pump (using a 15% maximum solids
loading as typical) is NnVp/0.15.
Numerical solutions of the mass
transfer equations for a carbon-phenol
ratio of 10 (using values of C, determined
from an adsorption isotherm for phenol
and Nuchar C-190 obtained for 80%
removal of phenol from a mixture with
an initial concentration of 100 ppm
phenol) are plotted in Figure 1.
Buoyant Packets Suspended in
Turbulent Rivers
If buoyant packets were injected into
a turbulent river, the random buffeting
of the packets by the eddies would
increase the immersion time. To remain
suspended in a river, a packet must
encounter a net downward force of
sufficient strength to counter the
steady, buoyant force. Suspension is
assumed to occur when the average slip
velocity is equal to the terminal rise
velocity of the packet. As before, it is
assu med that the packet wil I be oriented
broadside to the flow. Fluid entering
each packet is assumed to be well mixed
and hence representative of the river in
the surrounding vicinity. Fluid exiting
from a packet is assumed to mix very
rapidly with the surrounding fluid.
Figure 2 presents a simple schematic to
illustrate the adsorption model.
The differential equation for this case
is
-V
d£.=
dt
1 -exp
kcaL
(6)
or, separating variables and integrating,
1n C =. UaApNt
Co V
'b
12
11
10
9
8
7
6
5
4
3
2
1
Carbon/Phenol = 10
L - 1.27 cm, dp = 0.12 cm
L = 0.64 cm, dp = 0.12 cm
L = 1.27 cm. dp = 0.06 cm
10 15 20
Terminal Velocity, cm/sec
25
30
Figure 1.
Fluid pumping required to remove 80% phenol from 3,785 m3(106 gal)
with buoyant carbon packets (10.2 cm x 10.2 cm x L).
River
Well Stirred
Packets
Coul
Figure 2. Model buoyant packets suspended in turbulent rivers.
1 - expI
kcal
Ua
(7)
The V in the denominator of Equation 7
refers to the volume of water to be
treated, and t is the time the packets
remain in suspension.
Equation 6 was solved numericallyfor
a carbon-phenol ratio of 10 to obtain
values of C/C0 as a function of t. The
results of the numerical solutions for
the rate of phenol removal by packets
suspended in a turbulent river are
plotted in Figure 3. These results are
independent of spill size or of the
amount of water to be treated as long as
the water initially contains 100 ppm
phenol and the carbon-to-phenol ratio is
10. If, for example, the initial concentra-
tion of phenol were 250 ppm. Equation
7 indicates that the carbon packet
concentration, N/V, would be 2.5 times
greater than it would be for a 100-ppm
phenol-water mixture to maintain a
carbon-to-phenol ratio of 10. The
-------
information given in Figure 3 can be
easily converted to 250-ppm phenol-
water mixtures by merely dividing the
time axis by 2.5.
Pollutant Removal by Carbon-
Filled Panels Hanging in
Rivers
Spills of water soluble pollutants can
be removed from water courses by
panels packed with activated carbon
hanging in the water flow in some
homogeneous pattern downstream of
the spill and turned broadside to the
flow. The impact of the flowing water on
the panels will provide the AP for flow
through the porous panels and over the
activated carbon granules.
The system can be treated as a very
loosely packed bed of adsorbers consist-
ing of carbon-filled panels. Because the
panels would be located at various
distances downstream from a spill, the
amount of pollutant adsorbed in a panel
will depend on the location of the panel
along the watercourse. As in a packed
bed, the pollutant concentration will be
assumed to be constant over any cross-
section normal to the flow.
Assuming fresh carbon (i.e., C, is
negligible), the equation for the system
can be shown to be
1n
C _
1 - expI
kcaL
Ua
(8)
where Q is the volumetric discharge
rate relative to the carbon panels and Ac
is the total cross-sectional area of the
panels in the watercourse.
Calculated results obtained by using
Equation 8 are plotted in Figure 4 as the
percent of phenol removed per_ pass
over the carbon panels versus Q. The
spill size in pounds of phenol appears as
a parameter, and the carbon-to-phenol
ratio is 10. The solid curves in Figure 4
are for a slip velocity of 30.5 cm/sec (1
ft/sec). For a spill of 454 kg (1000 Ib) of
phenol, dashed curves are given for slip
velocities of 61.0 cm/sec and 91.4
cm/sec (2 and 3 ft/sec). Note that
increasing vs from 30.5 to 61.0 cm/sec
(1 to 2 ft/sec) (which aIso doubles Q) for
a 454-kg (1000-lb) spill, for example,
will result in a smaller percent of phenol
removal. For 1.27-cm-thick panels filled
with 0.12-cm carbon granules, the
percent of removal is maximized at a vs
V, = 3.81 cm/sec
22.9 15.2 7.62 7.62
L = 1.27 cm. dp = 0.12 cm
L = 0.64 cm. df = 0.12 cm
L = 1.27 cm, dv = 0.06 cm
Carbon/Phenol =10
Figure 3.
10 15
Time. Hours
Removal of phenol by suspended buoyant packets of activated carbon
(10.2 x 10.2 cm x L), C0 = WO ppm.
Carbon/Phenol =10
Vs = 30.5 cm/sec (1 ft/sec)
Vs = 61.0 cm/sec (2 ft/sec)
Vs = 91.4 cm/sec (3 ft/sec)
4536 kg (10.000 Ib) Phenol Spilled
072 kg (20,000
Panel Thickness = 1.27
(40,000 Ib)
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
20 40 60 80 100 200 400 600 WOO
Waterway Discharge Rate Relataive to Panels, Cubic Meters/Sec
Figure 4. Removal of spills from waterways by carbon-filled panels.
of about 15.2 cm/sec (0.5 ft/sec). If the
river velocity were 61.0 cm/sec (2
ft/sec), for example, and if the panels
could be allowed to drift down the river
at 45.7 cm/sec (1.5 ft/sec), the calcula-
tions indicate that the panels would
remove more phenol. But the polluted
stretch of water would take four times
longer to move past the panels, and
during this additional time, longitudinal
-------
dispersion will further dilute the pollu-
tant. The calculations ignore this
complicating factor.
Calculations in Figure 4, which are
based on Equation 8, are good for the
first portions of the spill-polluted water
to pass through the panel-festooned
region of the river. As more of the
contaminated water passes through the
panel region, the mass transfer equation
(a precursor of Equation 8) would have
to be numerically integrated (taking into
account a variation of adsorbed material
on the panel) with the axial distance of
the panel from the head of the panel
region. This is especially true for the
large spills and low values of Q. We
believe that the percent removal results
appearing in Figure 4 are higher than
can be attained in practice.
Turbulent Diffusion and
Dispersion in Waterways
Spills of water soluble materials into
waterways are rapidly diluted by eddy
diffusion and dispersion, making cleanup
attempts very difficult. Eddy diffusion
coefficients, though quite large com-
pared to molecular diffusion coefficients,
do not describe the primary dilution
action in watercourses, which is longi-
tudinal dispersion. Longitudinal dis-
persion involves the spreading out of a
water soluble substance along the
length of a waterway as a result of
variations in flow velocity across the
channel. Values of longitudinal disper-
sion coefficients measured for real
waterways are given in Table 1.
Profiles of mean concentration versus
distance for phenol spills of 18,140 kg
(40,000 Ib) and 1,814 kg (4,000 Ib) were
calculated for the Chicago Sanitary and
Ship Canal (CSSC) using data given in
Table 1. This waterway has the lowest
longitudinal dispersion (Table 1), and
the profiles plotted in Figure 5 are
thought to provide conservative values
of spill dilution that would occur in
rivers of comparable flowrate.
Discussion
The equations derived for the adsorp-
tion rate of pollutant from waterways by
buoyant carbon-filled packets (Equa-
tions 5 and 7) contain the total volume of
fluid to be treated, V, in the denominator.
As a result, an increase in V with time
resulting from turbulent mixing and
dispersion in waterways will increase
either the number of cycles or the
suspension time required to remove a
specified fraction of the pollutant,
Table 1. Typical Measured Dispersion Coefficients
Watercourse and Test Location
Chicago Sanitary and Ship Canal (Calumet Sag)
Missouri River (near Omaha, Nebraska)
Clinch River (near Clinchport. Virginia)
Copper Creek (near Gate City, Virginia)
Powell River (Sneedville, Tennessee)
Coachella Canal (Holtville, California)
Q
m3/sec
107.1
957.1
6.80
9.15
3.96
37.29
V
cm/sec
27.1
171.0
18.1
18.7
16.4
68.3
Ox
rrf/sec
3.56
1490
19.7
9.1
26.8
17.7
120
100
I
80
60
40
20
1 hr.
18.140 kg (40,000 Ib) Spill
1,814 kg (4,000 Ib) Spill
Flow = 107.1 m3/sec (3780 ft3/sec)
Ox = 3.56 rrf/sec
100 200 300 400 500 600 700
Distance from Maximum Concentration, Meters
800 900
Figure 5.
Longitudinal dispersion of water soluble spills in the Chicago Sanitary
and Ship Canal.
unless the number of carbon packets is
also increased with time so as to
maintain N/V constant.
Packet Cycling
Assume that response is required
within 1 to8hrtoan1,814-kg(4,000-lb)
phenol spill into the CSSC. With 18,140
kg (40,000 Ib) of activated granular
carbon and 8,645 kg (19,060 Ib) of
flotation material in 897,000 buoyant
packets, phenol would be removed from
a mixture in which the maximum
concentration would be between about
4 and 10.5 ppm by volume dissolved in
roughly 3 x 105 to 6.8 x 105 m3 (80 x 106
to 180 x 106 gal) of water spanning 800
to 1,800 m (2600 to 5900 ft) along the
channel. (Portions of the canal are
ignored where the calculated phenol
concentration is less than 0.5 ppm.) For
a response 1 hr after the spill, calcula-
tions in Equation 5 indicate that a 10%
reduction in phenol concentration
would require 155 cycles of the packets;
after 8 hr, 368 cycles would be needed
to effect a 10% reduction in concentra-
tion. With a factor of 15% solids by
volume (which can be handled by the
pumps), 6.07 x 104 m3 (16.05 pumped
(391.8 m3/cycle, or 103,500 gal/cycle).
If it were possible to reduce the packet
cycle time to 10 min, the time required
for 155 cycles (starting cleanup opera-
tions 1 hr after the spill) would be 61.3
hr. The solids-handling pumps would
have to circulate 39.4 m /min (10,400
gal/min). Of course, the phenol will
continue to spread longitudinally along
the channel during the cleanup process.
Buoyant Packets
If it were possible to keep the buoyant
packets suspended in the CSSC by
-------
turbulence (which appears to be highly
unlikely), a response to the 1,814-kg
(4,000 Ib) spill in 1 to 8 hr would produce
a situation requiring 25 to 60 hr to
remove 10% of the spilled phenol
(according to Equation 7), even if the
dispersion process were to cease. A
continuing dispersion process will
prevent the success of a carbon-filled-
packet approach to cleanup.
Carbon-Filled Panels
Unlike the equations derived for the
buoyant packets. Equation 8 (which is
derived for adsorption by panels hanging
in a waterway) does not contain the
volume of water to be treated. The
fraction of pollutant removed is directly
proportional to the panel area facing the
flow and inversely proportional to the
volumetric discharge rate relative to the
panels.
Carbon panels fixed in a watercourse
the size of the CSSC will remove about
4% of a 1,814-kg (4,000-lb) phenol spill
as the polluted fluid passes by the
panels, according to the model used to
compute Figure 5. Since the percent of
spill adsorbed is directly proportional to
the total panel area facing the flow, as
much as 33% of the phenol could be
removed if the carbon-to-phenol ratio
were 100 (i.e., if 181,400 kg (400,000
Ib) of carbon were used to clean up an
1,814-kg (4,000-lb) spill)).
This work assumes that equilibrium is
rapidly established at the activated
carbon surface. That is, the concentra-
tion at the fluid/exterior particle
interface is that value dictated by the
adsorption isotherm. With fluids con-
taining 100 ppm of dissolved pollutant,
the neglect of small deviations from
equilibrium at the fluid/particle inter-
face (perhaps 1 or 2 ppm) will cause
errors of a few percentage points in the
computations. But when the fluid
contains an average of less than 5 ppm
of pollutant, as it does after 8 hr of
dispersion of an 1,814-kg (4,000-lb)
spill into the CSSC, the neglect of
similar deviations from equilibrium can
result in substantial errors in the
computations.
Unless the spill is massive compared
with the total flow of the waterway, or
unless the spill response team begins
cleanup operations very shortly after
the spill, carbon-filled panels mechan-
ically suspended in the waterway will be
able to remove only a small percentage
of the pollutant.
Nomenclature
Ac Total cross-sectional area of
panels normal to the flow direction
in the watercourse
Ap Cross-sectional area of a packet
(10.2 x 10.2 cm for this study)
a Total exterior particle surface
area for mass transfer per unit
volume of packed space
ax Cross-sectional area of panels in
length dx
C Concentration of solute, moles/
volume, in the bulk fluid
C, Concentration of solute at the
exterior of the adsorbent particle
C,n Concentration of solute in fluid
entering a packet or panel
C0 Initial concentration of solute (at
time = 0)
C0ut Concentration of solute in fluid
exiting a panel or packet
D Depth of fluid
Dx Dispersion coefficient in x direction,
longitudinal dispersion (mVsec)
dp Diameter of particles in packets or
panels
kc Mass transfer coefficient (cm/sec)
L Thickness of carbon-filled packet
or panel
N Number of packets
n Number of cycles
Q Volumetric discharge rate relative
to carbon panels
R6 Reynolds number for a packed
bed, dpUap/Gufl)
t_ Time
Ua Superficial flow velocity, the
average linear velocity through a
bed computed on the basis of the
empty cross-sectional area
V Volume
Vp Volume of a packet
v Velocity, cross-sectional area
mean
vs Slip velocity, the velocity difference
between an object and the fluid in
which it is suspended
vt Terminal velocity
x Distance, as defined where used
0 Packed bed void fraction or void
space
V Viscosity of fluid
p Density of fluid
The full report was submitted in
partial fulfillment of Contract No. 68-
03-2648 (Task 8) by Rockwell Inter-
national, Environmental Monitoring &
Services Center, under sponsorship of
the U.S. Environmental Protection
Agency.
George R. Schneider is with Rockwell International, Environmental Monitoring
& Services Center, Newbury Park, CA 91320.
John E. Brugger is the EPA Project Officer (see below).
The complete report, entitled "Removal of Water Soluble Hazardous Materials
Spills from Waterways by Activated Carbon," (Order No. PB 82-103 813;
Cost: $8.00, subject to change) will be available only from:
National Technical Information Service
5285 'Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Oil and Hazardous Materials Spills Branch
Municipal Environmental Research Laboratory—Cincinnati
U.S. Environmental Protection Agency
Edison. NJ 08837
•ft-U.S. GOVERNMENT PRINTING OFFICE:1981--559-092/3320
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use $300
RETURN POSTAGE GUARANTEED
IL 60601
------- |