WATER POLLUTION CONTROL RESEARCH SERIES • ORD-17O2ODBAO3/7O
ULTRAFILTRATIVE DEWATERING
OF SPENT POWDERED CARBON
U.S. DEPARTMENT OF THE INTERIOR • FEDERAL WATER QUALITY ADMINISTRATION
-------
WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Reports describe
the results and progress in the control and abatement
of pollution in our Nation’s waters. They provide a
central source of information on the research, develop-
ment, and demonstration activities in the Federal Water
Quality Administration, in the U. S. Department of the
Interior, through inhouse research and grants and con-
tracts with Federal, State, and local agencies, research
institutions, and industrial organizations.
A triplicate abstract card sheet is included in the
report to facilitate information retrieval. Space is
provided on the card for the user’s accession number and
for additional uniterms.
Inquiries pertaining to Water Pollution Control Research
Reports should be directed to the Head, Project Reports
System, Planning and Resources Office, Office of Research
and Development, Department of the Interior, Federal Water
Quality Administration, Room 1108, Washington, D. C. 20242.
-------
ULTRAFILTRATIVE D*':'..,'........
SPENT POWDERED CARBON
by
C. W. Desaulniers
R. W. Hausslein
Amicon Corporation
Lexington, Massachusetts 02173
for the
FEDERAL WATER QUALITY ADMINISTRATION
DEPARTMEHKCDATHE INTERIOR
Program #17020 DBA
Contract #14-12-528
FWQA Project Officer, E. L. Berg
Advanced Waste Treatment Research Laboratory
Cincinnati, Ohio
March, 1970
For sale by the Superintendent of Documents, U.S. Government Priming Offh"
Washington, B.C. 20402 - Price 70 cents
-------
FWQA Review Notice
This report has been reviewed by the Federal
Water Quality Administration and approved for
publication. Approval does not signify that
the contents necessarily reflect the views
and policies of the Federal Water Quality
Administration, nor does mention of trade
names or commercial products constitute
endorsement or recommendation for use.
-------
ACKNOWLEDGEMENT
The authors wish to express their gratitude to
Mr. David Doucette of the Arnicon Corporation and
Mr. Timothy Murphy, Superintendent of the Brockton
Water Pollution Control Facility for their assistance
and cooperation in this study.
-------
TABLE OF CONTENTS
page
SUMMARY
INTRODUCTION AND BACKGROUND 1
MATERIALS AND EQUIPMENT 3
Spent Car:bon Concentrates 3
Equipment 7
Membranes 16
DEWATERING RUNS 21
Virgin Carbon 21
Comparison of Dewatering Cells Using
Lebanon Samples 23
Thin Channel Runs Using Brockton Samples
With the Gear Pump 28
Thin Channel Runs With the Moyno Pump 32
Long Term Dewatering Runs 34
Effect of Operating Conditions on Flux
Rates 40
RHEOLOGICAL STUDIES 46
Low Shear Rate Experiments 46
High Shear Rate Experiments 48
Effect of Dispersants 50
PROPOSED PLANT SCALE PROCESS 55
General Description 55
Economics 56
RECOMMENDATIONS 66
-------
SUMMARY
The feasibility of dewatering spent powdered carbon from
secondary effluent treatment by membrane ultrafiltration
has been demonstrated. Spent carbon slurries containing
between 5 and 10% carbon solids have been dewatered by
continuous membrane ultrafiltration in a laboratory size
thin channel cell (0.1 ft 2 membrane area) to solids levels
of 25 to 30%. Typical dewatering rates range between 50
and 100 gfd at transmembrane pressures of from 10 to 50
psi in runs of up to 9 days duration. The product water
from all runs was invariably free of any suspended carbon
solids.
The economics of dewatering carbon via ultrafiltration are
attractive if membrane replacement is infrequent. Using
estimation procedures recommended by the Office of Saline
Water, the cost of dewatering carbon from 10 to 20% solids
assuming membrane flux rates of only 50 gfd ranges from
$0.0015 per pound of carbon for a membrane lifetime of six
months to $0.0035 per pound of carbon for a membrane life-
time of 1 month.
-------
INTRODUCTION AND BACKGROUND
The efficacy of activated carbon as an adsorbent in
waste—water renovation is known to be enhanced if the
carbon is introduced into the water in finely divided
particulate form. A major limitation to the use of fine-
particulate carbon in this application is the difficulty
of recovering the solid from the liquid in sufficiently
high concentration to render its final dehydration and
thermal regeneration economic.
Efforts to develop means of recovering the carbon have
been directed toward addition of polymeric flocculants to
promote sedimentation. Such addition typically results in
a 5-10% carbon solids slurry; this slurry may then be
centrifuged to increase its solids content to between 20
and 25%. The concentrate is then thermally regenerated.
A less costly and more efficient method of dewatering the
slurry is highly desirable.
The recent development of high—flux, low-pressure ultra-
filtration membranes and devices which are quantitatively
retentive for colloidal materials and which are not
plugged or fouled by the solids they retain, provides a
very promising new approach to carbon-slurry concentration.
These systems have been successfully utilized, for example,
(1) To concentrate polymer latices to Ca. 60% polymer
solids without significant sacrifice in ultrafil-
tration rate, yielding concentrates which are stable
and of relatively law viscosity.
(2) To concentrate thixotropic, small particle size clay
suspensions at 6% solids to 12-18% solids also
without significant sacrifice in ultrafiltration
rate.
Such slurries, which become pseudoplastic and thixo—
tropic at high solids concentrations, can be effec-
tively dewatered by ultrafiltration because their
weak network structure is destroyed in the shear
field transversing the surface of the ultrafiltra-
tion membrane. While the particle-particle contacts
—1—
-------
are broken, dewatering takes place driven by the
hydraulic pressure drop across the membrane. In
addition, the membrane surface is also kept clear
y the high fluid velocities.
By analogy to latex and clay dewatering, the utility of
ultrafiltration for concentrating fine-particle carbon
slurries can be determined by subjecting representative
carbon slurries to membrane—ultrafiltration at pressures
between 10 and 50 psi, and determining the fractional
water removal possible without producing unmanageably
high consistency or thixotropy in the concentrate.
Although the addition of polymeric flocculants is neces-
sary for concentrating the carbon slurries by coagulation-
flocculation, it is possible that these flocculants will
impede the ultrafiltration process by reducing the effec-
tiveness of the shear field in separating the individual
carbon particles. Assuming such to be the case, the
effect of various chemical deflocculants added prior to
ultrafiltration should be evaluated. These deflocculants
perform their function by preferentially adsorbing on the
surface of the carbon particles, thus detaching the ad-
sorbed moieties of the polymeric flocculant. Typical de-
flocculants include the polyphosphates, silicates, poly—
carboxylates, and lignosulfonates.
—2—
-------
MATERIALS AND EQUIPMENT
Spent Carbon Concentrates
Sources
During the course of the program several samples of
spent carbon concentrate were obtained from the FWPCA’s
Advanced Waste Treatment Pilot Plant at Lebanon, Ohio.
These concentrates were prepared from secondary effluent
treated with 200 mg/l Aqua Nuchar A* activated carbon
powder and 2 mg/i Prirnafloc C_7**polymeric flocculant.
The time span between the preparation of these concen-
trates at Lebanon and their processing in the dewatering
apparatus in Lexington was typically between 1 and 2
weeks. Although the solids content of the Lebanon slur-
ries varied little from 10%, their fluidity and ease of
dewatering varied considerably, as noted in the descrip-
tion of each run.
Most of the spent carbon used in this program was pre-
pared from secondary effluent at the treatment plant in
Brockton, Massachusetts. The carbon concentrates from
Brockton were usually dewatered within 1 to 3 days of
preparation and tended to be more uniform in their de-
watering properties.
The Brockton Water Pollution Control Facility, is one
of the few relatively modern municipal treatment plants
in the Boston area with both primary and secondary
treatment. The plant was completed in 1964 and is
currently operating at 9.5 mgd out of a design capacity
of 12 mgd. Although the waste is largely domestic in
origin, a considerable proportion of industrial waste
(particularly from tanneries) is in the system from
time to time. Sampling during these periods was
avoided.
*
West Virginia Pulp & Paper Company
**
Rohm & Haas Company
—3—
-------
To prepare concentrates of spent carbon from fresh
secondary effluent, a 500 gal steel tank was installed
next to the Final Settling Tank. The secondary effluent
was pumped directly to the tank from the overflow of the
secondary settling tank. Aqua Nuchar A previously pre-
pared as a 10% concentrate in water was added with
stirring. The carbon was always added at the 200 mg/i
dosage level. A 2% concentrate of Primafloc C—7 at
3—5 mg/i dosage was added over a period of 30 minutes to
settle the carbon. After 2 hours, most of the carbon
floc had settled. The clear supernatant was decanted
by siphoning. The spent carbon concentrate plus some
unavoidable supernatant, was then drained from the bot-
tom of the tank. After overnight storage in the labora-
tory, the carbon usually occupied about 2 gallons, cor-
responding to a carbon solids content of about 5%,
which was used for the dewatering experiments.
Effects of Flocculant Dosage
A Primafloc C-7 concentration of 2 mg/i is employed at
Lebanon to obtain a spent carbon concentrate of approxi-
mately 10% solids. However, preliminary laboratory
flocculation experiments at Amicon indicated that a
2 mg/i flocculant dosage would require an impractically
long settling time in the 500 gallon “batch” tank in
Brockton.
In a typical experiment, a 600 ml sample of Brockton
secondary effluent (100 mg/i COD) was mixed with
200 mg/i carbon and after preliminary settling the
carbon was poured into a 250 ml graduated cylinder.
At 2.2 mg/i of C-7, greater than 4 hours were required
to obtain a well-defined sediment. After settling
overnight, the sediment was determined to contain 8.2%
solids. As shown in Table I and Figure 1, further in-
creasing the flocculant dosage reduced the settling
time and the solids concentration. A nominal 4 mg/i
C—7 dosage was therefore selected for use at Brockton.
This 4 mg/i dosage typically resulted in 5% solids
concentrates after 2 hours settling.
-4-
-------
DOSAGE OF PRIMAFLOC C-7 (mg/2)
00
*SEDIMENT OBTAINED FROM SECONDARY EFFLUENT (100 mg/i COD)
CONTAINING 200 mg/i AQUA NUCHAR A
FIGURE 1. SOLIDS CONCENTRATION OF SEDIMENT
VS. PRIMAFLOC C—7 DOSAGE
8
7
6
5
4
*
I-
2
LU
LU
U)
U-
0
z
0
I-
I-
2
LU
0
2
0
0
U)
-j
0
U)
3
2
2 4 6 8
10
—5—
-------
Table I
Effect of Flocculant Dosage on Settling Time and Sediment Concentration
Secondary Aqua Primafloc Appearance % Solids
Effluent Nuchar A C-7 Settling of in
COD, mg/l mg/i mg/i Time Supernatant Sediment
100 200 9 20 mm clear 2.5%
100 200 4.5 2-3 hrs clear 5.0%
*
100 200 2.2 overnight black 8.2%
*
Supernatant contained sufficient carbon to be black, opaque. Demarca-
tion between supernatant and sediment was consequently diffuse.
-------
Additional information on the effect of the flocculant
and carbon concentrations on the solids content of the
sediment was obtained later in the program on a poor
quality secondary effluent with an unusually high (410
mg/i) COD. This high COD content was due to a foaming
problem in the Brockton plant. The COD could be
reduced to 136 mg/i after filtration through a 6 inch
sand column. A combination of 4 mg/i C—7 and 200 mg/i
Aqua Nuchar were insufficient to give the desired rate
of settling and a clear supernatant for this effluent.
Flocculation tests showed that this material needed
higher carbon and floccuiant dosage levels. The results
are shown in Table II and Figures 2 and 3.
Equipment
Three types of cells were used to dewater carbon concen-
trates in this program. These were a stirred batch cell,
a flow-through, short path, medium channel cell and a
flow-through, long path, thin channel cell.
The Batch Cell
An Amicon Model 400 batch cell was used to determine
the ultimate level of dewatering possible with a virgin
carbon slurry and in some of the preliminary runs on
the Lebanon concentrate. This cell, shown in Figure 4,
has a capacity of about 400 ml. The feed may be pres-
surized at up to 50 psi. This cell is equipped with a
magnetic stirring bar located near the surface of the
membrane in order to minimize the local accumulation
of retained solids. Liquid that passes through the
membrane is collected in channels beneath the porous
support and exits through the ultrafiltrate port shown
in the figure.
The Short Channel Cell
A Dorr—Oliver short channel cell was used in some of
the preliminary runs with the Lebanon concentrate. In
this cell (See Figure 5) the feed stream is passed
along a channel about 4 in. long by 1 1/2 in. wide and
1/8 in. deep. The channel is bounded on one side by a
membrane, beneath which is located the porous membrane
—7—
-------
8
7
*
z
w
c
U i
Cl)
U-
0
z
04
I-
I-
z
U i
0
z
0
0
C /)
-j
0
C/)
O 10
DOSAGE OF PRIMAFLOC C-? (mg/fl
*OBTAINED FROM SECONDARY EFFLUENT (240 mg/I COD) TREATED WITH
500 mgI) AQUA NUCHAR A SETTLED 48 hours
FIGURE 2. SOLIDS CONCENTRATION OF SEDIMENT
VS. PRIMAFLOC C-7 DOSAGE
2 4 6 8
—8—
-------
30
DOSAGE OF PRIMAFLOC C-7 (mg/I)
*OBTAINED FROM 400 mI OF SECONDARY EFFLUENT (240 mg/I COD)
CONTAINING 500 mg/I AQUA NUCHAR A SETTLED 48 hours
FIGURE 3. SEDIMENT VOLUME VS. PRIMAFLOC C-7 DOSAGE
60
55
50
45
40
35
*
U
U
I —
z
LAJ
w
U)
0
U i
-J
0
>
0
2 4 6 8
10
—9—
-------
RELIEF
VALVE ____
Irili PRESSURE
AIR AIR
LAJ .A AJ.J A..A_A A AJ. AA .W.AI
L I Q U I D
MAGNETIC
MEM BRANE SI) RR ER
I ‘ ‘
ULTRAFILTRATE : :::‘- ---- COLLECTOR
_________________ ‘ CHANNELS
MEMBRANE SUPPORT
FIGURE ‘4. MODEL ‘400 CELL
—10—
-------
INLET OUTLET
MEMBRANE .JI
T /
> 1 -
:
M MBRANE COLI ECTOR
SUPPORT ULTRAFILTRATE CHANNELS
INTERNAL DIMENSIONS
4.0 in. x 1.5 in. x 0.12 5 in.
(VERTICAL SCALE EXAGGERATED)
FIGURE 5. SHORT PATH THIN CHANNEL CELL
—11--
-------
Table II
Laboratory Flocculation of High COD Brockton Secondary Effluent
% Solids
Secondary Aqua Primafloc Settled Recovery
Effluent Nuchar A C-7 Settling Volume in
COD, mg/i mg/i mg/i Time ml % Solids Sediment
240 500 3 8—10 hrs 34 6.6 87
240 500 5 36 6.2 88
240 500 7 42 5.4 89
I -a
240 500 9 20 mm 48 4.8 90
240 600 3 8—10 hrs 38 6.5 86
240 700 3 8—10 hrs 43 6.4 85
Settling trials were carried out in 600 ml beakers
containing 400 ml of effluent.
Settled volumes were measured after 48 hrs. of
quiescent standing from the time of C—7 addition.
-------
support and collector channels leading to the ultra-
filtrate port. The short channel cell is particularly
well suited for continuous dewatering of rapidly ultra-
filterable materials.
The Long Path Thin Channel Cell
An Amicon Model TC-l thin channel cell was used for
most of the dewatering runs in this program. An ex-
ploded view of this cell is shown in Figure 6 and the
principal of operation is shown schematically in
Figure 7. The number, length and thickness of the
channels are determined by the spacer plate shown in
Figure 6.
Most of the runs in this program were carried out using
a spacer plate with two parallel spiral channels each
of which was 28.8 in. long, 1/4 in. wide and 0.060 in.
deep. In general the feed stream enters the two ports
near the periphery of the spacer. The two parallel
spiral streams join just before exiting through the
center port. Note that only one channel is shown in
the schematic Figure 7.
As in the short channel cell, the channel itself is
bounded on one side by the membrane which is supported
by a porous sintered polyethylene disc. Liquid passes
through the membrane and the disc into collecting
channels and out the ultrafiltrate port.
The effective membrane area of the two parallel chan— 2
nels is (2) (28.8 in.) (0.25 in.) = 14.4 in. 2 = 0.10 ft
Under typical operating conditions, a high flow rate
along the channel is maintained in order to minimize
concentration polarization, i.e. the accumulation of
retained solids against the membrane’s surface which
causes reduction in flux. A high, but not unusual,
circulation rate of 1 gpm through two 0.060 in. x
0.25 in. channels corresponds to a linear velocity of
10.5 ft/sec.
—‘13—
-------
ULTRAFILTRATE OUTLET
KNOB (6)
SUPPORT DISC
CHANNEL SPACER PLATE
FIGURE 6.
EXPLODED VIEW OF AMICON TC-1 CELL
)P PLATE
RINGS
STREAM INLET
-14-
-------
ULTRAFI LTRATE
FROM
CIRCULATING.
PUMP I
FIGURE 7. LIQUID FLOW PATTERN IN TC-1 MEMBRANE MODULE
“SKIN” SIDE OF
ANISOTROPIC DIAFLO
MEMBRANE
CONTACTS PROCESS
STREAM IN CHANNEL
FLOW CHANNEL
SPACER
TO RECYCLE
RESERVOIR
—15--
-------
The circulation rate through the channel far exceeds
the amount of liquid passing through the membrane.
Under typical operating conditions 1/2 to 1% of the
circulating feed stream is dewatered per pass through
the cell. Thus, the volume of liquid entering and
leaving the upstream side of the cell is typically
50-100 times greater than that leaving as ultrafiltrate.
The apparatus used to circulate the feed, and monitor
flow rates and pressure drops is shown schematically
in Figure 10. The actual apparatus used for Runs 8B—l
to 9B-3 is shown in Figure 9.
After the completion of Run 9B-3, it was recognized
that the gear pump (which is part of the standard
apparatus shown in Figure 9) had inadequate pumping
power as well as poor wearing qualities for the high
concentration of abrasive slurries which were being
produced. A Moyno worm drive pump was used for the
remainder of the runs. The hardware system used with
the Moyno is shown schematically in Figure 10.
Membranes
Inasmuch as the particle size of the material to be
retained, Aqua Nuchar A, has a median size of llp,,
ultrafiltration media with relatively large pores could
be employed. The particular medium used for most of the
dewatering runs in this program was an Amicon XM-l00
membrane which has a nominal pore diameter of about
0.01 microns. A cross section of a typical ultrafiltra-
tion membrane is shown in Figure 11. Note that the
structure varies from a very dense “skin” side to a
more highly porous spongy side. The skin side, which
has the 100 A diameter pores (below the limit of reso-
lution of the photograph) is placed in contact with the
feed stream. No carbon particles in excess of 0.01
micron diameter or inacrornolecular material of greater
than 250,000 molecular weight can penetrate the membrane
skin. The XN—100 membrane has anominal distilled water
flux rate of 200-250 gfd at 10 psi transmernbrane pressure.
In several preliminary runs, a spun bonded polyolef in
sheet (Tyvek, E.I. DuPont de Nemours and Co., Inc.) was
used as the filtration medium.
—16—
-------
GAUGE
PROTECTOR
UPSTREAM
PRESSURE
GAUGE
FIGURE 8.
TC—1 ULTRAFILTRATION SYSTEM SCHEMATIC
ULTRA Fl LTRATE
4
PROCESS
FLOWMETER
N 2
SAMPLE
VALVE
SAMPLE
PORT
RESERVOIR
—17—
-------
ULTRAFILTRATION MODULE
UPSTREAM
PRESSURE GAUGE
-—SAMPLE
VALVE
—2O LITRE
RESERVOIR
CONTROL
F USES
SPEED
CONTROL
COVER
SCREW
GREASE
FITTING
PUMP
PUMP MOTOR
•1
PUMP SPEED CONTROL
GAUGE
PROTECTOR
OD
FLOWMETER
POWER
SWITCH
c T.
— -—- SAMPLE
PORT
Front View
LINE CORD— ’
Bock View
FIGURE 9. TC-l ULTRAFILTRATION SYSTEM
-------
ULTRAFI LTR ATE
ULTRAFILTRATION ________________
MODULE
MEMBRANE
0
RESERVOIR OUTLET iNLET J
GAUGE _____
GAUGE _____
MOYNO PUMP
VARI DRIVE
MOTOR
NO FLOWMETER USED, RECIRCULATION RATES DETERMINED BY DiRECT
MEASUREMENT OF RETURN TO RESERVOIR
FIGURE 10. TC-1 NOYNO ULTRAFILTRATIGN SYSTEM
—19—
-------
GEORGIA INSTITUTE OF TECHNOLOGY
EIIGINEERING EXPERIMENT STATION
ANALYTICAL INSTRUMENTATION LASS
1
‘ I L
PROJECT NO.
SAMPLE 10 cit4—’)--- L t %
MAGNIFICATION
.&T. _______
FIGURE 11.
CROSS SECTION OF SURFACE OF XM-100 MEMBRANE
—20-
-------
DEWATERING RUNS
Virgin Carbon
Initial dewatering runs were carried out with virgin car-
bon slurries in the TC-l unit (See Figure 9). Runs were
of short duration, terminated by carbon plugging the chan-
nel. Subsequent experiments demonstrated that the plugging
was due to the high efficiency of dewatering in this unit.
The combination of high membrane flux and the highly
efficient thin channel device in conjunction with the
easily dewatered virgin carbon gave rise to these plugging
problems. The results of the dewatering runs on virgin
carbon are summarized in Table III. It was concluded
from these runs that if it were possible to limit the de—
watering of the carbon to 20% solids in a single pass,
then runs of longer duration could be made in which the
entire contents of the reservoir were concentrated.
In an effort to determine independently the dewatering
rate and ultimate carbon solids achievable with virgin
carbon, a 10% Aqua Nuchar A slurry was dewatered in the
batch cell under conditions of no recirculation or stir-
ring. The resultant cake assayed 31% solids.
Inspection of the cake suggested that this 31% solids
level was a true upper limit and not an artifact due to
some thixotropic behavior caused by weak structuring of
the carbon. Thus it is likely that any mechanical means
of dewatering the carbon - ultrafiltration, centrifugation,
etc. - can at best only approach this 31% level as an
upper limit.
The average water flux rate through the membrane during
this dewatering was about 80 gfd at 2 psi transmembrane
pressure. Und.er identical conditions distilled water
fluxes at the rate of 150 gfd. The lower flux rate of
the carbon slurry indicates that the hydraulic permea-
bility of the carbon filter cake rather than the permea-
bility of the membrane partially controls flux rate.
—21--
-------
Table III
Dewptering of Virgin Carbon in the TC-1 Cell-Gear Pump
XM-l00 membrane Recirculation rate: 0.132 gpm (estimated)
Feed: 1.0 gal of 10% Aqua Nuchar A
Membrane area = 0.1 ft 2
* Duration Run Average %
Pressure of Run Terminated Flux Solids
( psi] ( mm) tJltrafiltrate by Rate(gfd) in Plug Comments
Enter ing/
Leaving
42/10 15 0.132 gal single 127 not 30 mu
clear plug in determined channel
channel 2 paths
30 long
20/0 11 0.125 gal 164 25%
clear
20/0 12 0.094 gal I’ 113 20%
clear
20/0 4 0.021 gal ‘I 75 21% inlet holes
clear in channel
doubled in
size
60/0 3 0.015 gal plug 72 28% 60 mu
clear filling channel
entire 4 paths
channel 18 long
*
Pressure on the ultrafiltrate side of the membrane is always
zero. The “average” transmembrane pressure is usually taken
as the mean of the inlet and outlet pressures, e.g., 42/10 psi
“averages” to 26 psi.
-------
Thus it would be of little value to use a membrane of
higher distilled water flux rate. Subsequent experiments,
using more open media than the XM—lOO membrane, corrobo-
rate this.
Comparison of Dewatering Cells Using Lebanon Samples
Two differing lots of spent carbon, Lebanon Nos. 2 and 3
were used in the following runs in which the dewatering
efficiencies of the three types of ultrafiltration cells
were compared. Lebanon No. 2 was completely free of odor,
similar in appearance and dewatering characteristics to
a virgin carbon slurry. In contrast, Lebanon No. 3 was
foul smelling, contained numerous microorganisms (visible
under the microscope) and was comparatively difficult to
dewater. In retrospect, it is quite likely that the
Lebanon No. 2 was a thermally regenerated carbon.
Table IV shows the results of a series of dewatering runs
made in the batch ultrafiltration cell. Lebanon No. 2 was
easily dewatered at a high flux rate using either an XM-100
membrane, a piece of Tyvek spun bonded polyolefin, or a
piece of Whatman No. 5 filter paper. The filtrate was
quite clear in all cases. A magnetic stirrer was used
in each case which stalled when the solids content reached
about 14%. In each case with Lebanon No. 2 the slurry
dewatered to the same 30% solids dry cake. The higher
fluxes using the Tyvek and filter paper are due to the
higher pressures employed.
In contrast, Lebanon No. 3 dewatered with much more dif-
ficulty in the batch cell. In spite of the higher pres-
sure employed its flux was only 20 gfd instead of the
110 gfd found with Lebanon No. 2. The maximum solids
attainable in the batch cell was only 19% in contrast to
the 30% attainable with Lebanon No. 2 and virgin carbon.
Dewatering in the Short Channel Cell
Table V shows the results of dewatering experiments in
the short channel cell. About one gallon of feed was
recirculated through this cell using the equipment shown
in Figure 9 with the replacement of the short channel cell
for the TC-1 shown in the figure. Using an XM—l00 membrane,
—23—
-------
TABLE IV
Dewatering Runs in the Model 400 Batch Ultrafiltration Cell
Volume of charge: 0.08 gallons
All filtrates were clear.
All runs were magnetically stirred.
% Solids Transmembrafle Average
Charge Initial Final Pressure Flux Filter Medium
Lebanon *
No. 2 9 ca 30 3 psi 110 gfd XM-100
*
No. 2 9 ca 30 3 initially 350 gfd Tyvek
30 Psi final
*
No. 2 9 ca 30 20 psi 475 gfd Whatman
No. 5
No. 3 11 19 25 psi 20 gfd XM—100
*
Stirrer stopped at 14% solids.
-------
TABLE V
Dewatering Runs in the Short Channel Cell-Gear Pump
XM—l00 membrane, circulation rate 0.132 gpm (estimated)
Pressure
Volume Enteriflg/
Circulated % Solids Leaving
Charge ( ga11on I Initial Final ( psi) Flux Comments
ui Lebanon
No. 2 0.950 9 18 6/0 iriiti l 90 gfd clear filtrate,
8/0 final 6 hr run
No. 3 0.950 11 11.5 10/0 26 gfd 1/16 in. slime
buildup on mem-
brane - 0.045
gal. of ultra-
filtrate in 1 hr
No.3 0.950 11 12 20/0 initial 30 gfd same 1/16 in.
75/0 final gel layer - 0.1
gal. of ultra—
filtrate in 1 hr
-------
the Lebanon No. 2 slurry (regenerated carbon) was de—
watered from 9 to 18% solids at the low transniembrane
pressure of 6—8 psi, at a 90 gfd flux rate. Although
the flux rate was comparable tc that obtained with
virgin carbon slurries, no plugging was observed in this
short channel cell.
In contrast, the dewatering of Lebanon No. 3 was much
more difficult. After one hour the feed had dewatered
from 11 to only 11.5% solids at a flux rate of 26 gfd.
The flux then dropped to a very 1cM value at which time
the run was discontinued. Upon disassembling the cell,
a 1/16 in. layer of slime or gel (which occupied half the
thickness of the 1/8 in. channel) was observed on the
meithrane - in all likelihood causing the reduced flux.
In an additional run with Lebanon No. 3 at higher pressure
(25 psi instead of 10) the flux rate increased by only l5%.
A flux rate that is only weakly dependent on pressure is
characteristic of ultrafiltration conditions in the gel
controlled region, a clearly undesirable operating condi—
t ion.
Dewatering in the Thin Channel Cell
The results of dewatering experiments in the spiral
module TC-1 is shown in Table VI. Using Lebanon No. 2,
the run terminated with a plugged channel after several
minutes of operation, as did virgin carbon in this cell.
Relative to the short channel cell, the TC-l cell is far
more efficient due to its longer channel length and
smaller channel height. The 60 psi pressure at the cell
inlet resulted from channel plugging.
In contrast to Lebanon No. 2, the Lebanon No. 3 was more
successfully dewatered in the thin channel unit. Since
the intrinsic flux rate of this material is lower than
that of Lebanon No. 2, no plugging was observed as the
slurry was concentrated from 11% to 19% solids. The
pressure rose gradually from 10 to 20 psi during the
course of the run. The 70 gfd flux is the highest ob-
served in any of the cells for Lebanon No. 3. Very little
slime buildup was noticed on the membrane.
—26—
-------
TABLE VI
Dewatering Runs in the Spiral Module TC-l-Gear Pump
XM-lOO membrane, circulation rate —‘0.132 gpm (estimated)
Linear velocity: 1.41 ft/sec
Channel length: 16 in. (each one)
Channel depth: 0.060 in.
Pressure
Volume Entering/
Circulated % Solids Leaving
Charge ( gallons) Initial Final ( psi) Flux Comments
Lebanon
No. 2 0.925 9 9.5 5/0 initial low plugged after
60/0 final (less passage of 0.015
than gallons of ultra—
20 gfd) filtrate
No. 3 0.925 11 19 10/0 initial
20/0 final
-------
The dewatering properties of Lebanon No.3 are probably
more typical of those of spent carbon slurries. For
such systems, the long thin channel device is most
suitable. The high shear conditions in the thin channel
keep the membrane surface clean, thus inhibiting the ac-
cumulation of solids. For such easily dewatered materials,
however, as Lebanon No. 2, highly efficient dewatering
devices can lead to plugging.
Any full scale ultrafiltrative dewatering device should
be able to accommodate both types of sludge and should
therefore be equipped with suitable control devices to
reduce the transinembrane pressure when easily dewaterable
sludges are pumped through the unit. (See later discussion.)
Thin Channel Runs Using Brockton Samples With the Gear
Pump
Inasmuch as the previous runs demonstrated that the most
efficient dewatering device for spent carbon slurries was
the thin channel spiral module, a second series of experi-
ments was undertaken to optimize the dewatering procedure
using spent carbon from Brockton. The first of these runs
was carried out using Tyvek as the filter medium, in order
to determine whether a rnicroporous ultrafiltration mem-
brane was in fact necessary in view of the relatively
large size of the carbon particles to be removed. A
spent carbon slurry from Brockton (See Table VII) was
circulated thrOugh the system using both 0.060 in. and
0.030 in. thick channels. Both runs exhibited rather low
flux rates especially considering the low carbon concen-
tration at which dewatering took place. The run with the
0.030 in. channel was terminated by plugging.
The low flux and tendency to plug were quite likely due
to the very high COD of the secondary effluent used to
prepare the carbon slurry; 410 mg/i of COD adsorbed by
200 mg/i of activated carbon tends to result in a
noticeably slimy slurry. Since the distilled water flux
for Tyvek at these transmembrane pressures is several
hundred gfd, the low fluxes observed are a strong mdi-
cation of gel layer formation.
—28—
-------
A subsequent batch of spent carbon prepared at Brockton,
8C, was derived from a secondary effluent much lower in
COD, 75 rng/l. Improved fluxes of 41 and 50 gfd were ob-
served at lower transrnembrane pressures. These fluxes
are comparable to the 70 gfd found for Lebanon No. 3
slurry shown in Table VI.
For the runs in Table VII, the circulation rate was about
0.32 to 0.37 gpm. Since the initial volume of spent car-
bon slurry charged to the reservoir is about 0.80 gallons,
the time for complete turnover, assuming good mixing of
the reservoir contents (not always the case), during the
initial portion of each run is about 2—2.5 minutes.
During the latter portion of these runs, when the slurry
has been concentrated about fourfold, the turnover time
is about once every half minute. The rate of ultrafiltra-
tion during the major portion of these runs is such that
approximately 1% of the total water is removed per minute,
or 0.5% per pass.
Inspection of the equipment after the completion of
Runs 8C-l and 8C-2 suggested that the inability to achieve
solids contents higher than about 16 or 17% was due to
the inability of the gear pump to suck up the locally con-
centrated (due to lack of stirring) slurry from the
reservoir.
Consequently during the series of runs beginning with 9B-l,
a stirrer was placed in the reservoir during the latter,
high viscosity portion of each run. As can be seen from
Table VIII, this procedure permitted higher final solids
levels to be obtained.
The 9B series of runs were also carried out under slightly
different conditions during the early portion of each run,
while the solids concentration was increasing up to 10%
with very little viscosity change. During this part of
the run, the circulation loop was sealed from the atmo-
sphere so that the system could be pressurized by apply-
ing 15 psi nitrogen to the reservoir. At a system pres-
sure of 15 psi, if the head pressure of the pump is 10 psi
(a function of the circulation rate) the actual inlet
pressure to the cell is 10 + 15 = 25 psi. Since the pres-
sure head loss takes place almost totally within the cell,
—29—
-------
TABLE VII
Dewateririg Runs in the Spiral Module TC-l Cell-Gear Pump
Circulation rate: 0.32—0.37 gallons/minute
Linear velocity: 3.7 ft/sec for 60 mu channel
Pressure
COD of Sec. Volume Entering/
Effluent Circulated % Solids Leaving
Charge Source mg/i ( gal) Initial Final ( psi) Flux Filter Comments
Br ockt on
8B-l 410 0.80 3.8 14.6 20/0 26 gfd Tyvek 60 mu
channel;
ge 1
buildup
8B-2 410 0.80 3.8 11.0 20/0 16 gfd Tyvek 30 mu
channel;
plugged
8C—1 75 0.80 4.0 16.7 between 41 gfd XM-100 60 mu
7/0 and channel
10/0 psi
up to 10%
solids
8C-2 75 0.77 4.0 16.0 between 50 gfd XM—100 60 mu
7/0 and channel
10/0 psi
up to 10%
solids
be twe en
10/0 and
22/0 psi
up to 16%
solids
-------
TABLE VIII
Dewatering Runs in the Spiral Module PC-i Cell-Gear Pump
XM-l00 membrane Circulated rate: 0.40 gpm (estimated)
Linear velocity: 4.3 ft/sec
Channel thickness: 0.060 inches
Channel length: 29 inches
Trans membrane
Pressure
COD of Sec. Volume Eritering/
Effluent Circulated % Solids Leaving Flux
Charge Source mg/i ( gailonsj Initial Final ( psi) ( gfd) Comments
w
H
Brockton * Step 1
9B—l 75 0.99 4.5 9.7 25/15 65 ultrafil—
trate COD
Step 2 10 mg/l
9.7 24.5 25/0 50
Step 1
9B—2 75 1.02 4.0 9.0 25/15 45
Step 2
9.0 22.0 25/0 55
9B—3 65 1.85 5.0 17.5 30/0 35 used for
viscosity
and dis-
* persant
The COD of the ultrafiltrate of Run 9B-l was 10 mg/l. studies
-------
the outlet pressure is 0 + 15 15 psi. The pressure on
the downstream or ultrafiltrate side of the membrane is
always zero (atmospheric). Thus, in the experiment above,
inlet pressure of 25 psi and an outlet pressure of 15 psi
give an average transmembrane pressure of 20 psi.
Thin Channel Runs With the Moyno Pump
Due to wear and inadequate capacity for pumping carbon
slurries a substitute for the gear pump was sought.
Based on the experience of the FWPCA in pumping carbon
slurries, the gear pump was replaced by a Moyno Pump.
The pumping arrangement was somewhat different with the
Moyno as shown in Figure 10. Note in particular that the
separate reservoir was eliminated; rather the exiting
feed stream returning from the ultrafiltration cell was
discharged directly into a hopper built directly over the
feed entrance of the Moyno.
The initial runs with the Moyno pump were carried out with
virgin carbon slurries. The results are shown as runs
9-4 and 9-5 in Table IX. It was immediately apparent that
operations with the Moyno pump substituted for the gear
pump were very much improved. A circulation rate of
0.92 pgm (over twice the maximum achievable with the gear
pump) was easily achieved which resulted in fluxes of 200
and 250 gfd. Compare this data with the virgin carbon run
data using the gear pump in Table III where the fluxes
were 75—150 gfd.
The first run with the Noyno pump on a spent carbon slurry
was performed on material prepared at Brockton, designated
9—6. In order to achieve the high dewatering level of
24.8%, it was necessary to increase the speed of the Noyno
pump as the run progressed. Increasing the speed in-
creased the pressure drop through the cell (along the
channel). In Run 9-6 this pressure drop increased from
10 psi at the beginning of the run to 80 psi by the end of
the run. A high flux rate was maintained throughout most
of the run, averaging 125 gfd, falling off somewhat as
the solids increased above the 20% level.
—32 -
-------
TABLE IX
Dewatering Runs in the Spiral Module TC-1 Cell-MoynO Punip
XM-100 membrane Circulation Rate: 0.80 gpm: Moyno Pump
Fluid velocity: 8.5 ft/sec
Channel depth: 0.060 in. Channel length: 29 in.
Press ure
COD of Sec. Volume Entering/Leaving
Effluent Circulated % Solids psi Flux
Charge Source mg/i ( gallons) Initiai Final Initial Final ( gfd) Comments
Virgin
none 0.92 10.0 19.2 unknown 200
carbon 9-4
Virgin
none 0.92 10.0 20.0 15/0 30/0 250
carbon 9-5
Brockton 80 1.98 5.8 24.8 10/0 80/0 125 Temp. rose to 42°C
W 10-1 Final flux at
( J
125 gfd.
Brockton COD of filtrate
65 1.85 4.1 24.0 20/0 40/0 140
10-2 56 mg/i. Temp.
rose to 50°C.
Final flux at
100 gfd.
unknown 1.50 10.2 31.0 20/0 80/0 70 Temp. rose to 65°C
Lebanon 5
10-3 Ultrafiltrate
turned pale green
prior to run term.
Lebanon 5
unknown 0.92 10.2 17.0 20/0 35/0 70
10-4
Brockton * Exten. running to
114/76 See Tables meas. flux, pres.,
10—5
* Total COD is 114 mg/i; after filtration, the COD is 76 m9/1 , viscosity para-
the difference representing the COD of the suspended solids. meters.
-------
The behavior of the carbon slurry on Run 10—1 was similar.
The temperature was monitored during this run. Starting
at room temperature, it began climbing during the latter
portion of the run (when the pump speed was increased)
resulting in a final temperature of 42°C. The increased
temperature decreased the viscosity of the slurry as was
observed on subsequent runs. Note that even at the high
final solids level of 24% the flux rate was maintained at
125 gfd.
Run 10-2 was similar to Run 10-1 except that a greater
volume was circulated. Thus the longer duration of the
run resulted in a greater temperature rise which in turn
permitted an extra 2% final dewatering.
Run 10-3 was performed on Lebanon No. 5. The material as
received- had a characteristic anaerobic sulfide odor not
observable in the freshly prepared Brockton slurries.
Lebanon No. 5 dewatered to an unusually high 31.0% solids.
Inasmuch as this high a solids level was not approached
with any other carbon slurry, it is suspected that this
sample contained an unusually high amount of organic
matter, possibly contained within the pores of the carbon
particles. The presence of organic matter would also be
consistent with the lower flux values of 70 gfd that were
found for this slurry. By the time this run was terminated
the temperature of the slurry had risen to 65°C from room
temperature; in addition the final half gallon of ultra-
filtrate had a pale green cast.
The same slurry was used in Run 10-4, but the dewatering
was terminated purposely after the solids concentration
had reached 17% in order to prepare a material for test-
ing the effects of dispersing agents on viscosity. The
viscosity experiments are reported in the section on
Rheological Studies.
Long Term Dewatering Runs
Having found the general operating conditions needed to
obtain high rates of dewatering and high final solids
concentrations, two long term dewatering runs were car-
ried out. The purpose was to obtain more quantitative
relationships between operating conditions and dewatering
-.34-
-------
rates and in particular to determine whether membrane
performance deteriorates with time. As will be shown
later in the Economics, the feasibility of membrane de—
watering is most highly dependent on membrane lifetime
and average flux throughout this lifetime.
The two long term dewatering runs were carried out for
6 and 9 days using spent carbon slurries prepared at
Brockton. As in all the previous experiments, no carbon
was found in the ultrafiltrate during these runs. Long
term flux rates of about 45 gfd were maintained at room
temperature operation (ca 27°C) . The cell was immersed
in a cooling bath for this portion of the experiment. At
somewhat elevated temperatures (42°C) , fluxes of the order
of 65 gfd were obtained, but no sustained runs at 42°C
were carried out. Although membrane flux rates do appear
to decrease slowly with time, no evidence of plugging or
gross destruction of the XM-l00 membrane was observed.
Some wear was observed in Run 11—1 near the entrance and
exit ports. The COD’s of the ultrafiltrates were typi-
cally in the range of 25-35 mg/i, indicating no breakthrough.
Details of these runs follow.
A long term run designated 11—1 was made as follows:
A 5.6% solids spent carbon slurry from Brockton was pre-
pared in the usual manner. An initial charge of 2.35 gal.
was dewatered to a 10.0% solids slurry in a period of
about 3 hours at an average flux rate of about 100 gfd.
A slight temperature increase from room temperature to
27°C was observed. The long term dewatering run was con-
sidered to commence with the resultant 1.32 gallons of
l0%solids slurry. This material was recirculated through
the 0.060 in. deep channel TC-1 cell using the Moyno pump
for 144 hours. The membrane was an XM-lOO. The ultra—
filtrate was recombined with the circulating slurry in
the reservoir above the Moyno pump hopper. During the
course of the run evaporative losses which took place
from the reservoir surface increased the solids level to
14% by the end of the run. In order to maintain iso-
thermal operating conditions during most of the run, the
cell was immersed in a water bath. When flux rates at
the elevated temperature conditions, resulting from the
pumping work, were to be determined, the cell was removed
—35--
-------
from the water bath (during the last 6 hours of the run).
The course of the run is shown in some detail in Figure 12.
During the first 24 hours of the run the circulation
velocity was maintained at the relatively low level of
5.1 ft/sec which resulted in an average transmembrane
pressure of 7 psi. Over the next few hours in a series
of two steps the circulation velocity was increased to
8.5 ft/sec causing the pressure at the cell entrance to
climb to 18 psi. Note however that the rise in transmem-
brane pressure from 7 to 9 psi caused more than a propor-
tional increase in the flux rate of from about 20 to 40 gfd.
During the next 120 hours no changes in operating condi-
tions were made. During this time the flux rate decreased
from 40 to 25 gfd. The extent to which this decrease in
flux is due to membrane compaction or is due to the slowly
increasing solids contents of the slurry from evaporation
is not known.
At the 142 hour mark, the cooling was removed resulting in
a temperature rise from 25 to 42°C over the next 4 hours.
The transmembrane pressure increased, but the flux rate
increased more than proportionally.
The circulation velocity was then increased to 11.3 ft/sec
and resulted in an average transmembrane pressure of
17 psi and the rather high flux value of 110 gfd.
The run was then terminated and the cell was disassembled
for inspection of the membrane. The membrane directly in
the channel areas appeared to be essentially unattacked.
Some erosion had taken place near the exit region; in this
area it appeared that some small carbon granules had
penetrated into the subskin region of the membrane.
During the entire course of the run the ultrafiltrate
samples were clear. COD determinations were also made on
the ultrafiltrate from time to time in order to have
advance warning of impending membrane failure. The COD
values are located along the top of Figure 13. These
values appear to show no consistent rise or decline with
time.
—36—
-------
120
TIME (hours)
TC-I CELL
MOYNO PUMP
XM-IOO MEMBRANE
CHANNEL HEIGHT 0.060
10% SOLIDS/5 liters
RUN STARTED AT 2.75 hours MEMBRANE LIFE
A P=7psi
B E P=6psi
C t P7.5psi
D P9 psi
E COOLING
F t P Ilpsi
G P I4 psi
H Pl7psi
V=5.I ft/sec
V=5.I ft/sec
V 6.8 ft/sec
V 8.5 ft/sec
REMOVED, t P 9psi T=25°C
V=lI.3 ft/sec T 42°C
V Il.3 ft/sec
V lI.3 ft/sec
T=48°C
in.
FIGURE 12. LONG TERM DEWATERING RUN 11-1
—37—
100
80
FLUX
(gfd)
60
40
20
0
0
50 100 150
200
-------
A second run designated 11—2 was then made as follows:
As in the previous run the spent carbon slurry was pre-
pared at Brockton in the usual way. The COD of the
secondary effluent from which the slurry was prepared was
100 mg/i of which 42 mg/i was removable by sand filtration.
The COD of the clear supernatant after the carbon floccu-
lation was 20 mg/i. Details of the run are shown in
Figure 13.
The initial slurry dewatering up to the 10% level is
shc ’in. At this level, the flux rate of 87 gfd is com-
parable to the 75 gfd obtained at the same solids level
commencing Run 11-1, although a higher initial fluid
velocity of 9.3 ft/sec was employed in this run. Again
the system was kept at 26-29°C by keeping the cell
immersed in a water bath.
The operating conditions were left unchanged for the next
44 hours during which time the flux appeared to asymptote
at the 60—65 gfd level. At this point the fluid velocity
was momentarily increased to 11.3 ft/sec to see if a
periodic “scouring ’ t would elevate the flux. Although the
flux did increase to 80 gfd during the scouring operation.
the flux returned to its previous 60-65 gfd value when
the original circulate velocity of 9.3 ft/sec was restored.
During the next 24 hours the flux further decayed to about
50 gfd at which time the circulation rate was briefly
increased, again with no permanent effect on the flux
rate.
Operating conditions were left unchanged during the next
3 days during which the flux rate was constant at 45 gfd.
At the 161 hour mark, the cooling bath was removed and the
fluid velocity was increased to 11.4 ft/sec arid resulted
in a substantial flux increase to 98 gfd at a temperature
of 43°C.
The fluid velocity was then restored to 9.3 ft/sec while
maintaining the temperature at the 43°C level. Under
these conditions of elevated temperature, the run was
continued through the 216 hour mark (9 days) at which
time it was terminated because of a leaking rear seal on
the Moyno pump.
-38-
-------
120
8.2 ft/sec
in.
V=9.3 ft/sec
T=28°C
TC-1 CELL
MOYNO PUMP
XM-I00 MEMBRANE
CHANNEL HEIGHT 0.060
I0% SOLIDS
A 5% SOLIDS) V 6.2 ft/sec
B DEWATERING,V= 7.6 ft/sec
C 10% SOLIDS,RUN BEGINS, P I2 psi
D t P l3psi V=9.3 ft/sec
E E P=l6psi V=lI.3 ft/sec
F t P I3 psi V DECREASED TO
G V IN CREASED TO 10.5 ft/sec
H t P=I7.5 psi V=I0.5 ft/sec
I t P l3 psi V8.2 ft/sec
J COOLING REMOVED, V 8.2 ft/sec
K V 9.3 ft/sec
L P= 20 psi V=II.3 ft/sec
M P= 14 psi V 7.6 ft/sec
N 1 P=I6 psi T=42°C
0 RUN ENDED, MOYNO LEAK
FIGURE 13. LONG TERM DEWATERING RUN 11-2
t3P= 16 psi
100
80
FLUX
(gfd)
60
40
20
0
0
50 100 150 200
TIME (hours)
250
—39—
-------
Note that operations at the elevated temperature and the
slightly decreased fluid velocity of 7.6 ft/sec re-
sulted in a flux rate of about 65 gfd.
As in the previous run the COD levels were monitored from
time to time and showed no consistent increase or decrease
throughout the run.
Effect of Operating Conditions on Flux Rates
Two sets of experiments were carried out to determine the
effect of operating variables such as pressure and circu-
lation rate (i.e. shear rate) on flux at various solids
levels. In brief, the results showed that at circulation
rates in excess of 0.5 gprn, a linear velocity of 5.2 ft/sec
in this experiment, the flux rates were directly propor-
tional to the average transmembrane pressure at solids
concentrations up to 13%; no runs above 13% were made.
At lower circulation rates in the range of 0.04 to 0.42
to 0.06 gpm, 0.42 ft/sec to 0.63 ft/sec in this system,
flux rates were found to be relatively independent of pres-
sure, butrnore depeade.nt on the fluid velocity. In addition,
at the lower circulation rates, permeabilities (flux rates
per unit pressure) were far less than at the higher circu—
lation rates. These results are consistent with the
behavior of other (non—carbon) slurries. Details of the
runs are given below.
Circulation Rates in Excess of 0.5 gpm (5.2 ft/sec )
A spent carbon was prepared by treating a secondary ef-
fluent with 200 mg/i Aqua Nuchar A and 4 mg/l Primafloc
C-7 in the usual manner. The COD of a secondary effluent
from which the slurry was derived was 114 mg/i of which
38 mg/i was removable after passage through a sand
column.
The resultant 5% slurry was run at three pump speed
settings at which the recirculation rate, the pressure
drop across the cell (from inlet to outlet on the up-
stream side) , and the transinernbrane flux rates were
measured. The results are shown in Table X. The slurry
was then dewatered to 9% solids and the above procedure
repeated. The slurry was finally dewatered to 13% solids
-40 -
-------
TABLE X
Dewatering as Function of Solids Content for TC-l/Moyno System
XM-l00 membrane, channel height: 0.060 in.
Spent carbon slurry prepared at Brockton
Transmembrane
Pressure Flux/Average
Circulation Fluid Entering/ Transmembrane
Carbon Moyno Rate Velocity Leaving Flux Pressure
% Solids Setting gpm ft/sec ( psi) gfd ( gfd/psi )
0 1 0.58 6.2 10/0 133 26.6
(tap water) 2 0.85 9.1 17/0 220 25.9
3 1.05 11.2 24/0 323 26.9
5 1 0.53 5.7 12/0 85 14.2
2 0.82 8.8 19/0 123 12.9
3 0.99 10.6 28/0 165 11.8
9 1 0.58 6.2 16/0 29 3.6
2 0.87 9.3 25/0 55 4.4
3 1.05 11.2 35/0 72 4.1
13 1 0.51 5.5 23/0 25 2.2
2 0.66 7.1 30/0 33 2.2
3 0.81 8.7 42/0 42 2.0
-------
and the same operating variables were measured. Tap
water had been previously run through the system in
order to obtain the 0% solids data.
As can be seen from the table the flux per average
transmembrane pressure (permeability) at any given
solids concentration is fairly constant, indicating
that the transrnembrane pressure is the variable that
controls flux rate at these recirculation rates. These
results are also displayed in Figure 14.
Circulation Rates Between 0.04 and 0.05 gpm (0.4-0.6 ft/sec)
A spent carbon slurry was prepared. The COD of the
secondary effluent from which the slurry was prepared
was 100 mg/i, 33 mg/i of which was removable by sand
filtration. The solids content of the settled slurry
was 5.6%. To obtain reduced circulation rates at trans—
membrane pressures comparable to those in the preceding
runs, the flow through the TC-l cell was throttled by
valves installed on the inlet and outlet sides of the
TC-l cell. Pressure gauges were installed at the inlet
and outlet ports of the TC-l cell, and the valves
adjusted to give a low pressure drop between the cell
inlet and outlet to achieve a low circulation rate.
Results of these runs are shown in Table XI. Note that
circulation rates were about 1/10 those in Table X,
while transmembrane pressure drops were comparable.
Flux rates per unit pressure however, fell off sharply.
Comparing the throttled results at 5.8% carbon solids
concentration to those at 5% solids in the preceding
run, it is immediately apparent that the flux rates per
unit pressure drop are about an order of magnitude
lower for the throttled runs. These results suggest that
in the throttled runs a carbon solids cake of low
hydraulic permeability formed on the membrane surface.
Ultrafiltration carried out under such conditions, where
the accumulation of ultrafiltered solids on the membrane
surface controls the flux rate, generally results in a
flux that is relatively insensitive to the transmernbrane
pressure. Such is the case observed here.
—42 -
-------
18
160
140
120
I00
FLUX
(gfd)
80
BROCKTON SPENT CARBON
TC-I CELL, MOYNO PUMP
XM-I00 MEMBRANE
0.060 in. CHANNEL HEIGHT
DATA TAKEN FROM TABLE 10
T JRE 14. EFFECT OF TRANSMEMBRANE PRESSURE ON FLUX
5% SOLIDS
60
9% SOLIDS
40
20
13% SOLIDS
00
5 10 15 20 25 30
AVERAGE TRANSMEMBRANE PRESSURE (psi)
—43-
-------
TABLE XI
Dewatering Using Throttled TC-l/MOYnO System
XM-l00 membrane, channel thickness: 0.06 in.
Spent carbon slurry prepared at Brocktofl,
5.6% Solids
Trans membrane
Pressure Flux/Average Flux
Moyno Throttled Fluid Enter ing/ Transmembrane Circulation
Speed Circulation Velocity Leaving Pressure Rate
Setting Rate (gpm) ft/sec ( psi) Flux ( gal/psi) ( gfd/gpm )
0.2 0.038 0.41 14/12 19 gfd 1.46 500
1.0 0.053 0.57 17/14 42 gfd 2.70 795
2.0 not measured 26/24 42 gfd 1.68
3.0 0.058 0.62 40/38 35 gfd 0.90 605
(5% solids data from Table X reproduced below
f or comparison)
1.0 0.53 5.7 12/0 85 gfd 14.2 160
2.0 0.82 8.8 19/0 123 gfd 12.9 150
3.0 0.99 10.6 28/0 165 gfd 11.8 167
-------
It can be shown on theoretical grounds 1 that a flux
rate independent of pressure is the condition to be
expected when a layer of retained solids of low hy-
draulic permeability covers the membrane. The condi-
tion is frequently observed when ultrafiltering macro-
molecular solutions and suspensions and is often
referred to as the “gel controlled” region of opera-
tion. If the shear rate across the membrane surface
is substantially increased, as is the case when the
circulation rate is increased (since the shear rate
is directly proportional to the circulation rate) then
“gel” control can be eliminated and the flux rate be-
comes directly proportional to the transmembrane pres-
sure as is the case here.
In operating any full scale unit, it is more economical
to operate outside of the gel controlled region, i.e.
at high circulation rates, since the pumping cost
necessary for high circulation rates is less than the
additional membrane area required at the low flux rates
characteristic of the gel controlled region.
Under the unlikely conditions where pumping costs to
circulate the carbon slurries are high, the throttled
system may be of some advantage. The flux rates per
unit circulation rates for the two modes of operation
may be compared by examining the final column in Table XI.
The flux per unit circulation rate, otherwise termed
dewatering per pass, is about 3 to 5 fold higher for
the throttled runs.
-45—
-------
RIIEOLOGIC?˝ L STUDIES
Rheological properties of spent carbon slurries are of
interest for two reasons: (1) proper design of plant
scale dewatering equipment, especially with regard to
pump sizes, channel sizes, etc. and (2) determination of
the likelihood that dispersing agents would decrease the
viscosity of the slurries. hereby enabling them to be
dewatered more rapidly or o higher ultimate solids levels.
Most of the rheological studies were carried out using a
Brookfield Viscorneter. Because the shear rates in the
TC—l cell and in full scale plant equipment are much
higher than those encountered with a Brookfield viscoineter,
some of the pressure/circulation rate data generated in
the thin channel cell course of other experiments is
analyzed here in rheological terms.
L Shear Rate Experiments
The effect of carbon solids concentration on Brookfield
(Model RVT, Spindle No. 3) viscosity for virgin carbon
slurries and a typical Brockton slurry are shown in
Figure 15. Note that both virgin and spent carbon vis-
cosities are very low until the 12% carbon concentration
is reached, at which point the viscosities climb rapidly.
These data Would seem to indicate that dewatering much
above the 20% level would be difficult because of the
viscosity limitation. It should also be noted however,
that the viscosity decreases with increasing shear rate;
Ł or example the viscosity of the 21% spent carbon sludge
decreased from 14,500 centipoise at 5 rpm spindle speed
to 4,400 centipoise at 20 rpm spindle speed (fourfold
higher shear rate).
Actual shear rate values in a Brookfield decrease from
the spindle out into the bulk of the liquid. For the RVT
Viscometer, shear rates at the spindle are of the order
of l0 - to 10 sec -.
-46-
-------
I I I I I
1000 -
00
CONCENTRATION (weight %)
0-200 mg/i AQUA NUCHAR A WITH 75 mg/i COD FLOCCULATED WITH PRIMAFLOC C-7
A-VIRGIN AQUA NUCHAR A FLOCCULATED WITH PRIMAFLOC C-7
BROOKFIELD RVT SPINDLE NO.3
FIGURE 15. BROO}(FIELD VISCOSITY VS. CONCENTRATION
OF CARBON SLURRY
4000
3000 -
5 rpm
U,
Q.
0
-J
U i
U-
2000
10
4 8 12 16 20
-47-
-------
High Shear Rate Experiments
We may use the data in Table X as a source of high shear
rate viscosity data.
The TC—l cell may be used as a viscometer insofar as the
channels are of simple rectangular geometry and of uniform
dimensions along their entire length.
For a fluid moving in laminar flc through a thin channel
the shear stress at the wall of the channel is given by:
hAP
T =
w 2L
where h and L are the channel height and length respec-
tively, and AP is the pressure drop along the channel.
If the liquid is Newtonian, the shear rate is given by:
6V
= 2
wh
where w is the channel width and V is the volumetric
flc i rate (circulation rate). The viscosity is the ratio
of the shear stress to the shear rate.
Inasmuch as carbon slurries are not truly Newtonian, the
viscosity values calculated are so--called “apparent vis—
cosities”. However the same error is implicit in the
Brookfield viscosity values. The apparent viscosity
calculations are summarized in Table XII.
it is apparent from the table that the viscosities cal-
culated from this data are far lower than those determined
with the Brookfield at the same solids levels. This is
a manifestation of the tendency for viscosities of sus-
pensions to decrease markedly with in creasing shear rate.
It should also be noted that at the high shear rates the
viscosities at any given solids concentration are rela-
tively insensitive to changes in shear rate. Since it is
known that the viscosities of suspensions decrease with
increasing shear rate to an asymptotic value (the asymp-
totic region often being referred to as the “Upper
Newtonian Regime”), the shear rates in the TC-l unit may
-48-
-------
TABLE XII
parent Viscosity in TC-1/Moyno System as a Function
of Solids Content and Circulation Rate
Shear Rate Shear Stress
Circulation Fluid Pressure . = . 3L = hAP Viscosity
% Rate (mi/sec) Velocity Drop wh TW 2L cp
Solids Per Channel ft/sec psi sec - dynes/cm 2 Tw/ ’(lOQ )
0 18.3 6.2 10 7500 690 9.2
(tap 26.7 9.1 17 10900 1170 10.1
water) 33.3 11.3 24 13600 1650 12.1
5 16.6 5.6 12 6800 825 12.1
25.8 8.8 19 10600 1310 12.3
31.3 10.6 28 12800 1930 15.0
9 18.3 6.2 16 7500 1100 14.7
26.7 9.1 25 10900 1720 15.8
33.3 11.3 35 13600 2410 17.6
13 16.0 5.4 23 6600 1585 24.0
20.8 7.1 30 8500 2070 24.3
23.7(est.) 8.0 42 9700(est.) 2890 28.8
All data were recorded at about 24°C.
TC-1 cell was in cooling bath to hold temperature constant.
V = volumetric flow (cm 3 /sec) = circulation rate
w = channel width, 0.635 cm (0.25 in.)
h = channel height, 0.152 cm (0.060 in.)
L = channel length, 76 cm (30.0 in.)
AP = pressure drop (dynes/cm 2 ) = 68947 pressure drop (psi)
-------
be sufficiently high to cause complete breakdown of the
aggregates into primary particles. If such is the case,
then it would be expected that dispersants would be of
little value in a TC-l or other high shear unit. On the
other hand,.it is conceivable that dispersants could cause
aggregate breakdown at even lower shear rates than were
used in these experiments.
The calculated viscosity values for water (0% solids) are
rather high since water is known to have a viscosity of
about 1.0 centipoise at this temperature (independent of
shear rate). Unusually high calculated values of vis—
cosities are usually attributable to entrance effects and
turbulence. Water moving through the channels at the
circulation rates shown, has Reynolds numbers of between
2,000 and 3,500. Such values are in the so—called transi-
tion region between laminar and turbulent flow. It is
unlikely however that this would account for the tenfold
higher viscosity values calculated. These values are more
likely due to entrance and exit effects since the fluid is
subjected to rapid changes in velocity moving in and out
of the cell. In any case, the viscosities of those slur-
ries containing carbon (i.e., 5, 9, and 13%) are suff 1-
ciently high such that the Reynolds numbers are in the
laminar region.
Also note that at any solids level the apparent viscosity
increases with the shear rate. This behavior is consistent
with both turbulent flow and entrance effects.
Effect of Dispersants
The effect of dispersant addition on the apparent vis-
cosity of spent carbon slurries was evaluated for two
dispersants at low shear rates.
A spent carbon slurry prepared from a 60 mg/l COD secon-
dary effluent was concentrated in the TC-1 unit to 17.5%
solids. Various amounts of Tamol 731 dispersant (manu-
factured by Rohm and Haas, described as a sodium salt of
a polymeric carboxylic acid) were added to the slurry and
the effect on viscosity determined. The results are shown
in Figure 17. Although some reduction in viscosity was
observed for the first per cent or so of Tamol, the
—50—
-------
% SOLIDS OF SLURRY
15
MEASURED IN TC-1 UNIT AT HIGH SHEAR (ca. 10,000 sec ) RATES
FIGURE 16. APPARENT VISCOSITY OF A CARBON SLURRY
AS A FUNCTION OF THE SOLIDS CONTENT
20
U,
0.
C)
>-
-J
U)
LL
0
>-
I-
U)
0
0
U)
>
I-
z
w
Q.
0 5 10
—51—
-------
3.5
% TAMOL 731 ON CARBON
PREPARED FROM 60 mg/i COD BROCKTON SECONDARY EFFLUENT
TREATED WITH 200 mg/i AQUA NUCHAR A, FLOCCULATED WITH 5 mg/i
PRIMAFLOC C-7, AND SUBSEQUENTLY CONCENTRATED TO I7.5% SOLIDS
BROOKF$ELD RVT SPINDLE NO. 3
FIGURE 17 BROOKFIELD VISCOSITY VS. TAMOL 731 DOSAGE
6000
U)
Q.
C.,
F-
‘I )
0
C -,
C ,,
>
0
-J
Li
0
0
2.5 RPM
0
10 RPM
0.5
1.0
1.5
2.0 2.5 3.0
—52—
-------
absolute reduction in viscosity was rather modest and no
further reduction was shown at higher Tamol dosages.
In another experiment various amounts of Daxad 11 dis-
persant (manufactured by W. R. Grace, described as a
polymerized sodium salt of a short chain alkyl naphthenate
sulfonic acid) were added to Lebanon sample No. 5, which
had been concentrated in the TC—1 unit from 10.2% solids
level to 17% solids. As shown in Figure 18, the Daxad
markedly reduced the viscosity of the slurries particularly
at the low shear rates. Although it may be misleading to
compare the effectiveness of the dispersants inasmuch as
they were evaluated on entirely different sludges (the
Brockton material being four times as viscous as the
Lebanon material), it would appear that the Daxad is bet-
ter suited for reducing the viscosity of carbon slurries at
low shear rates.
Note in Figure 18 that with increasing spindle speed
(shear rate) the effect of the dispersing agent is di-
minished. This is consistent with the comments in the
previous section to the effect that, as the particles are
disaggregated by the shear forces themselves, the disper-
sant becomes proportionately less effective.
—53—
-------
% DAXAD 11 ON CARBON
(CARBON IS ASSUMED TO HAVE BEEN TREATED WITH 1-2% PRIMAFLOC C-7
AND IS 2/3 OF THE TOTAL SOLIDS)
20
SPENT CARBON FROM LEBANON CONCENTRATED FROM 10.2 TO 17.0% SOLIDS
BROOKFIELD RVT SPINDLE NO. 3
FIGURE 18. BROOKFIELD VISCOSITY VS. DAXAD 11 DOSAGE
6000
5 RPM
U)
C.)
>-
U)
0
C -)
U)
>
0
-J
U i
U-
0
0
0
0
00 4 8 12 16
-54-
-------
PROPOSED PLANT SCALE PROCESS
General Description
The foregoing results permit the preliminary design of a
continuous process to dewater spent carbon slurries by
membrane ultrafiltration. In view of the wide change in
rheological properties that the carbon slurries undergo
with concentration, a two—stage dewatering process is
envisioned.
In the first stage, the slurry would be dewatered from its
original 5 or 10% level up to about 15 or 17% solids.
In this concentration region the viscosity of these slur-
ries is quite low as shown in Figure 16. Conventional
pumps and other fluid handling equipment would be adequate
to circulate the slurries through the dewatering unit.
In the second stage, the 15 to 17% slurry would be further
dewatered to the 20% or above level. In view of the
sharply increasing viscosity with concentration of the
slurries in this region, the design of the second stage
fluid handling equipment is more critical than that of
the first. It is likely that the second stage would
employ a recirculating ioop to permit greater control
over the dewatering by smoothing out the effects of varia-
tions in the feed slurry solids concentration. The second
state would probably have to be provided with control
devices to automatically adjust process conditions in
response to the changing character of the incoming slurry.
A multiple version of the thin channel spiral cell that was
used for most of the dewatering work in this report would
not be the proper piece of dewatering equipment for a
field or commercial sized unit. A more suitable thin
channel device would be that of two concentric tubes,
through the narrow annulus of which would flow the carbon
slurry. If the outside tube is porous and lined on its
inside with membrane, the annulus would be the analogous
flow path to the thin channel ultrafiltration cell used in
this work. Such tubular devices are currently undergoing
prototype evaluation prior to commercial production.
—55—
-------
Figure 19 shows the assembly of a tubular cartridge.
The sludge would flow longitudinally through the annuli
shown in the figure. The membrane surface faces inward
and the membrane is supported externally by a braided
structure. The solid extruded core has ridges the height
of which determines the thin channel space. The ultra-
filtrate is collected after having passed through the mem-
brane and braided support. Not shown is the outer cover
for the entire cartridge. Figure 20 shows a cross sec-
tional detail of a channel using two different sealing
configurations. Not shown in this figure is the braided
membrane support.
The XM—l00 type would be the recommended membrane since
this was the membrane used successfully in this work. The
pore sizes in this membrane however are far smaller than
necessary to retain arbon particles, the pore sizes being
of the order of 100 A diameter while the particles are
several microns in diameter. While it is quite likely
that membranes or other filter media with pores as large
as 1 micron or possibly larger may be perfectly adequate
to completely retain carbon particles, as the pore size of
a filter medium approach the size of the carbon particles,
the chances of irreversible membrane plugging increase.
The filter medium should also be anisotropic in structure,
i.e. should have its smallest pores at the carbon contact-
ing surface, in order to reduce the chance of internal
plugging of the material.
The experimental results indicate that success in the
dewatering of carbon slurries depends less on the charac-
teristics of the membrane than it does on the high shear
conditions of thin channel flow as shown in Table X and XI.
Economics
Cost Estimation
The projected costs for dewatering spent carbon via thin
channel ultrafiltration are outlined below following the
estimation procedures outlined in the Office of Saline
Water publication 2 .
—56--
-------
INLET HEADER
MEMBRANE
EDGE
PROTECTOR
BRANE
BRAIDING
POTTING
MATERIAL
FIGURE 19. DETAIL OF TUBULAR MEMBRANE CARTRIDGE ASSEMBLY
PROCESS STREAM INLET PORT
EXTRUDED CORE
MANIFOLD SEAL
“O’ RING
ER SEAL “0” RING
—57—
-------
REINFORCED BUTT (SHOWN)
OR LAP ADHESIVE LONGITUDINAL SEAL
MECHANICAL LONGITUDINAL SEAL
MEMBRANE
FLUID CHANNEL
U,
OD
CORE
-A- -B-
FIGURE 20. ULTRAFILTRATIVE TUBULAR MEMBRANE CONFIGURATION DETAIL
-------
The following figures are given for a ten million
gallon per day treatment plant. The dewatering costs
are calculated for 5 and 10% spent carbon slurry con-
centrations, both being dewatered to 20% solids prior
to feeding into the reaction furnace. Since this is
a small unit to be installed within existing premises,
such costs as land acquisition, building construction,
raw water supply and product water storage have been
omitted. In the Captial Costs a 5 year lifetime has
been selected for amortization purposes.
Assuming treatment at the 200 ppm (0.02%) level, a
10 mgd treatment plant would yield 40,000 gallons of
5% slurry or 20,000 gallons of 10% slurry. To concen-
trate a slurry to 20% carbon, the amount of water to be
removed would be 30,000 and 10,000 gallons respectively.
The daily weight of carbon to be recovered will be the
same in both cases, 16,700 lbs.
The membrane area requirements are estimated assuming
two flux rates: 50 and 100 gfd.
Thus:
TABLE 13
Feed Ultrafiltrate Membrane Area
Conc. Volume Flux Rate Required
( %) ( gal/day) ( gal/ft 2 day) ( ft 2 )
5 30,000 50 600
5 30,000 100 300
10 10,000 50 200
10 10,000 100 100
Costs are calculated for each of these four combinations
at membrane lifetimes of 1/2, 1, 2, 6, and 12 months.
—59—
-------
COSTS
Capital Costs
50
100
50
100
$ 4000 $ 2000 $ 1600 $ 1200
3000 1500 1000 500
1200 1200 1200 1200
2400 2400 1800 1800
2000 2000 2000 2000
(1200) (1200) (1200) (1200)
$ 1500 $ 1000 $ 800 $ 700
750 500 400 400
1500 1000 1000 1000
1000 1000 1000 1000
$17350 $12600 $10800 $ 9800
$16370 $ 8450 $ 5780 $ 3190
8120 4320 3020 1820
4588 2550 1840 1226
2225 1370 1050 830
1634 995 860 750
$33720 $21050 $16600 $13000
25470 16900 13800 11600
21940 15150 12650 11000
19575 13970 11850 10600
18985 13600 11670 10550
5
5
10
Spent Carbon Feed (%)
Membrane Flux (gfd)
A. Essential Plant Costs
1. Special Equipment
a. Membrane cell, filters, gauges
b. Membranes ft 2 at $5.00 ft 2
2. Standard Engineering Equipment
a. Air operated diaphragm pump
b. Compressor 2 (1 standby)
c. Moyno pump for second stage or
second diaphragm pump
B. Other Plant Costs
1. Installation
2. Instruments
3. Contingencies
4. Engineering
Total Plant Investment
C. Working Capital
Membrane life
2 weeks
1 month
2 months
6 months
1 year
Total Capital Costs
Membrane life
2 weeks
1 month
2 months
6 months
1 year
-60-
-------
Spent Carbon Feed (%)
Membrane Flux (gfd)
*
Capital Cost /lb of Carbon Processed
5
5
10
10
50
100
50
100
Membrane life
A. Essential Operating Costs
1. Electric Power (typical calculation)
7.5 HP compressor
7.5 x .747 x 24 x 2 = 269 KWH
269 x $0.007 = $1.88/day
Moyno pump
2 weeks
1 month
2 morths
6 months
1 year
*5 years x 330 days x 16700 lb/day
Operating Costs
$0 . 0012
0.0009
0.0008
0 .0007
0.0007
$0 . 0008
0.0006
0.0005
0.0005
0.0005
$0 . 0006
0 .0005
0.0005
0.0004
0.0004
$0 . 0005
0.0004
0.0004
0.0004
0.0004
38KWZday x .007 = $0.27/day
Total $2.10 day
2. Supplies and Maintenance Materials
depending on membrane life
$ 2.10 $ 2.10 $ 1.27 $ 1.27
2 weeks
1 month
2 months
6 months
1 year
39000/yr
18000/yr
9000/yr
3000/yr
1500/yr
3. Operating Labor
Membrane life
2 weeks
1 month
2 months
6 months
1 year
$239.00
112.00
57 .60
21.20
12.12
$ 12.54
6.19
3.47
1.65
1.19
$119.50
56.00
28.80
10.60
4.80
$ 6.45
3.28
1.92
1.01
0.72
$ 80.30
37 . 90
19.70
7.57
4.55
$ 4.41
2.29
1.38
0.77
0.62
$ 40.90
19.70
10.60
4.55
3.30
$ 2.42
1.36
0.91
0.60
0.54
—61—
-------
Total Essential Operating Costs
Membrane life
5
5
10
10
50
1ÔO
50
100
Spent Carbon Feed (%)
Membrane Flux (gfd)
4. Maintenance Labor
5. Payroll Extras
Membrane life
2 weeks
1 month
2 months
6 months
1 year
$ 0.25 $
$ 1.92 $
0 . 96
0.56
0.29
0.22
$255.81
121.50
63. 98
25.49
15.88
2 weeks
1 month
2 months
6 months
1 year
0.19 $ 0.16 $
1.00 $ 0.69 $
0.52 0.37
0.32 0.23
0.18 0.14
0.14 0.12
$129.24 $ 86.83
62.09 41.99
33.33 22.74
14.08 9.91
7.95 6.72
1.58
0.85
0.53
0.32
0.27
6.56
0.65
0 . 15
0.39
0.23
0.16
0.11
0. 10
$ 45.14
22.72
13. 10
6.69
5 . 37
$ 0.89
0.52
0.37
0.26
0.24
$ 6.23
$ 0.59
Other Operating Costs
6. General Overhead and Administra-
tive Overhead
Membrane life
2 weeks $ 4.41 $ 2.29 $
1 month 2.22 1.20
2 months 1.28 0.73
6 months 0.66 0.41
1 year 0.50 0.31
*
7. Amortization $ 9.62 $ 7.44 $
8. Taxes and Insurance $ 1.01 $ 0.76 $
.
*This equipment is being amortized ove?r a 5 year
period at 4% per annum, membranes are excluded
from amortization.
—62—
-------
Spent Carbon Feed (%)
Membrane Flux (gfd)
9. Interest on Working Capital
Membrane life
5
5
10
10
50
100
50
100
Membrane life
Total Cost (Capital Plus Operating)
per lb Carbon Processed
Membrane life
2 weeks
1 month
2 months
6 months
1 year
Total Operating Costs
Membrane life
2 weeks
1 month
2 months
6 months
1 year
Operating Cost/lb
of Carbon Processed
$ 1.96
0 . 97
0.55
0 .27
0.20
$272.84
135.35
76.47
37.08
27.24
$0 . 0163
0.0081
0.0045
0.0022
0.0016
$0 . 0175
0 .0090
0.0053
0.0029
0.002 3
2 weeks
1 month
2 months
6 months
1 year
$ 1.01
0 .52
0.31
0.16
0 . 12
$140.74
72.01
42.57
22.85
16.58
$0 . 0084
0 .0043
0.0025
0.0014
0.0009
$0 . 0092
0.0049
0.0030
0.0019
0.0015
0.69
0.36
0.22
0.13
0 . 10
96.31
50.41
30.70
17 .57
14.30
.0058
0.0030
0.0018
0.0011
0 . 0008E
$0 . 0064
0.0035
0.002 3
0.0015
0.0013
0.38
0.22
0.15
0.10
0.09
53.23
30.28
20.44
13.87
12.52
0.0032
0.0018
0.0012
0.0008
0.00075
$0 . 00 37
0.0022
0.0016
0.0012
0.0011
2 weeks
1 month
2 monthr
6 months
1 year
—63—
-------
Discussion
This cost estimation is summarized in Figure 21 where
the cost of concentrating the spent carbon slurry to the
20+ % level from the 5 and 10% level in cents per pound
carbon is expressed as a function of the membrane life-
time (frequency of membrane replacement) at two membrane
fluxes: 50 and 100 gfd. As can be seen from the curves,
for membrane lifetimes less than 2 or 3 months, the cost
of dewatering carbon bares a simple inverse propor-
tionality relationship to membrane flux rate and membrane
life.
Reliable estimation of membrane life is made difficult
by the unavailability of long term dewatering data.
The longest run was of 9 days duration after which the
membrane appeared to be in very serviceable condition.
Based on these results it can be said with confidence
that membrane lifetime is at least 2 weeks, probably
1 to 2 months and possibly 3 months to a half year.
Were more robust membranes to be evaluated, the time
scale for membrane lifetime might be extended.
The average flux rate most likely to be encountered is
probably 100 gfd. Although low solids sludges dewater
at higher rates, high solids sludges dewater at lower
rates; thus the 100 gfd is a good “averag& value.
Because membranes shc ’z a long term slcM loss of flux,
the 50 gfd curve may be more appropriate for long mem-
brane lifetimes (in excess of several months).
Thus taking a rather conservative set of conditions, one
month membrane life, 50 gfd and 10% solids sludge to be
dewatered we arrive at a cost of $0.0035/lb of carbon
dewatered.
-64-
-------
2
3
4
5
5
I0
I0
50
100
50
I00
600
300
200
100
MEMBRANE LIFE (months)
CURVE FEED (%) FLUX (gfd) AREA ft 2
FIGURE 21. ESTIMATED COST TO ULTRAFILTRATIVELY DEWATER
SPENT CARBON FROM A 10 MGD TREATMENT PLANT
U)
4-
C
C -)
z
0
C)
0
I-
U)
0
C)
I.8
1.6
‘.4
1.2
l.0
0.8
0.6
0.4
0.2
00
2 4 6 8 10 12
—65—
-------
RE CON1 J1ENDATI ONS
To further evaluate ultrafiltrative dewatering of spent
carbon slurries, a continuous pilot scale dewatering unit
should be designed and constructed. A tubular unit such
as that described in the section Proposed Plant Scale
Process — General Description of about one square foot
area would probably be most suitable. For comparison,
the laboratory unit used in this program has a membrane
area of about 0.1 square feet.
To properly design tubular and larger process units, the
following operational information needs yet to be obtained.
The items are listed in approximate priority of order:
Membrane Lifetime
A protracted run of at least 1 month s duration should
be carried out to determine whether existing polymeric
membranes can withstand the sustained scouring action
of the powdered carbon. If the lifetime of existing
polymeric membranes is inadequate, the feasibility of
using more wear resistant inorganic microporous
materials should also be evaluated. Candidate materials
are porous graphite, silica and sintered metals.
Single Pass Dewatering
Runs should be carried out in which the 5 or 10% carbon
slurry feed is dewatered up to the 15% level in a
single pass using various channel heights, pressure drops
and paths lengths. Similar runs should be carried out
between the 15% level up to the 20% and higher level.
Operation at Elevated Temperature
Existing results indicate that operations at elevated
temperatures such as 50 and 60°C promote significantly
more rapid dewatering. This should be further evaluated
in a systematic fashion in order to reduce membrane area
requirements.
—66—
-------
Effect of Dispersants
Dispersants have been shown to be effective in reducing
the viscosity of spent carbon slurries at low shear
rates such as those encountered in a Brookfield
Viscorneter. Their effectiveness in the high shear con-
ditions of the thin channel cell should be determined
as a possible method of reducing pumping costs and mem—
brane area requirements.
—67—
-------
LIST OF REFERENCES
1. Michaels, A.S., Chemical Engineering Progress, 64 (12)
31 (1968)
2. A Standardized Procedure for Estimating Costs of
Saline Water Conversion, March 1966 (OSW Publication).
—68-
* U. S. GOVERNMENT PRINTING OFFICE: 1970 0 - 408-308
------- |