EPA-660/2-75-025
JUNE 1975
Environmental Protection Technology Series
Chemical Coagulation/Mixed-Media
Filtration of Aerated Lagoon Effluent
National Environmental Research Center
Office of Research and Development
U.S. Environmental Protection Agency
Washington, O.C. 20460
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development,
U.S. Environmental Protection Agency, have been grouped into
five series. These five broad categories were established to
facilitate further development and application of environmental
technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in
related fields. The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY STUDIES series. This series describes research
performed to develop and demonstrate instrumentation, equipment
and methodology to repair or prevent environmental degradation from
point and non-point sources of pollution. This work provides the
new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
EPA REVIEW NOTICE
This report has been reviewed by the Office of Research and
Development, EPA, and approved for publication. Approval does
not signify that the contents necessarily reflect the views and
policies of the Environmental Protection Agency, nor does mention
of trade names or commercial products constitute endorsement or
recommendation for use.
-------
ABSTRACT
The applicability of chemical coagulation/mixed-media filtration as
tertiary treatment for additional removal of suspended solids, oil,
and biochemical oxygen demand following an oil refinery aerated
lagoon has been demonstrated in a full-scale commercial unit. Two
modes of operation were developed: 1) operation as a "scalping"
unit, and 2) optimized operation to achieve near potable water
clarity. Operation as a scalping unit in which chemical pretreat-
raent is not optimized yields a significant improvement in many
water quality parameters, particularly in the aesthetics of aerated
lagoon effluent appearance. Optimized operation involves deter-
mining the requirements for chemical destabilization of the col-
loidal system and then responding by providing proper pretreatment.
Operating problems and the effect of operating variables were in-
vestigated during full-scale plant operations in the scalping mode.
Influent suspended solids concentration and water temperature were
the most significant independent variables. Mechanical limitations
were studied, including a filter bed disturbance that necessitated
a total bed replacement. High, localized backwash velocity caused
the invisible disturbance which reduced turbidity removal from
about 80 percent to 50 percent. Diagnostic procedures, design
changes, and the costs of operation and maintenance are reported.
A cold weather study showed that a three-chemical destabilization
pretreatment system is required for filtration of biocolloids in
brackish water. Determination of the optimal three-chemical de-
stabilization system using zeta potentials required evaluation of
zeta potentials in a manner which sorted out the effect of double-
layer repression. The colloid destabilization mechanisms of charge
neutralization and bridging were required for optimal filter per-
formance. For colder water temperatures, even with optimal chemical
treatment, the filter hydraulic loading must be decreased. The
change in hydraulic loading with temperature related directly to
the water's viscosity.
This report was submitted in fulfillment of Project Number 12050
GQR, Grant Number S803026, by Amoco Oil Company, Yorktown, Virginia,
under the partial sponsorship of the Environmental Protection Agency.
Work was completed as of March 1974.
ii
-------
CONTENTS
Page
Abstract ii
List of Figures iv
List of Tables vli
Acknowledgements viii
Sections
I Conclusions 1
II Recommendations 3
III Introduction 5
IV Pilot Plant Studies 7
V Design and Construction 13
VI Operation 22
VII Tests and Observations 28
VIII Project Critique 47
IX Theory and Practical Application of Chemical 49
Pretreatment
X Operating and Maintenance Costs 64
XI References 67
XII Appendices 68
XIII Addendum 77
iii
-------
FIGURES
No. Page
1 Mixed Media Filter Pilot Plant 8
2 Mixed Media Filter Pilot Plant Bed Pressure Profiles 10
3 Mixed Media Filter Pilot Plant Pressure Drop 12
4 Mixed Media Filter Flow Schematic 14
5 Mixed Media Filter Bed and Support Design 16
6 Yorktown Filter Plan View 17
7 Yorktown Filter Section View A-A 18
8 Yorktown Filter Section View B-B 19
9 Yorktown Filter Section View C-C 20
10 Filter Backwash Cycle Timer Functions 23
11 Mixed Media Filter Bed and Support, Second Design 27
12 Porous Bed Pressure Probe 29
13 Effect of Surface Disturbance on Yorktown Mixed 31
Media Filter
14 Yorktown Filter Chemical Treatment Tests 39
15 Yorktown Filter Chemical Treatment Tests 40
16 Yorktown Filter Chemical Treatment Tests 41
17 Yorktown Filter Chemical Treatment Tests 42
18 Yorktown Filter Chemical Treatment Tests 43
19 Yorktown Filter Chemical Treatment Tests 44
20 Yorktown Filter Chemical Treatment Tests 45
21 Zeta Potential of Colloidal Iron Hydroxide Solution 51
Plotted as a Function of pH
iv
-------
FIGURES CONTINUED
No. Page
22 Zeta Potential - pH Plot for Aluminum Hydroxide 52
23 Solubility Curve for Aluminum 53
24 Equilibrium Compositions of Solutions in Contact 54
with A1(OH>3
25 Equilibrium Compositions of Solutions in Contact 55
with Fe(OH)3
26 Conversion of A1(H20)6 +3 to Al (OHH 57
27 Change in Zeta Potential of Refinery Process Water 59
with Increasing Coagulant Dosage at Various pH
Levels
28 Variation of Final Zeta Potential with pH when Treating 61
Refinery Effluent Having Negative Zeta Potential with
Three Levels of Coagulant Dosage
29 Zeta Potential Distribution of Biocolloids in Aerated 78
Lagoon Effluent
30 Concept of the Double Layer and Zeta Potential 79
31 Condition - Response Schematic for Chemical Treatment 82
of Waterborne Colloids
32 Sequential Formation of Hydrous Aluminum Oxide Polymers 83
33 Example of Complex Which May Exist in Precipitated 84
Hydrous Aluminum Oxide Polymers
34 Thickness of the Double Layer 85
35 Distribution of Ions and Potentials in the Double Layer 88
Surrounding Colloids
36 Variation of Charge on Minusil with Specific Conductance 89
37 Variation of Charge on Minusil with Various Titrants and 90
Two Levels of Specific Conductance for Each Titrant
-------
FIGURES CONTINUED
No. Page
38 The Variation in Zeta Potential of Particles in 92
Aerated Lagoon Effluent (Fresh Water) with
Addition of Cationic Polyelectrolytes
39 Variation of Charge on Particles in Fresh Water 95
Aerated Lagoon Effluent with Addition of Various
Polyelectrolytes and NaCl
40 The Variation in Zeta Potential of Particles in 96
Aerated Lagoon Effluent with Alum at pH of 6.0
and 8.5
41 Charge on Suspended Matter in Aerated Lagoon 98
Effluent as a Function of pH
42 The Variation of Charge on Aerated Lagoon Particles 99
with Specific Conductance after Treatment by
Various Chemical Combinations
43 Head Loss - Turbidity Breakthrough Curve Filtering 102
Chemically Treated Aerated Lagoon Effluent
44 Recommended Hydraulic Loading Rates as a Function 103
of Water Temperatures and Showing the Correlation
with Water Viscosity
vi
-------
TABLES
No. Page
1 Yorktown Water Filter Pilot Plant Operating and 9
Inspection Data
2 Yorktown Mixed Media Filter Average Operating 25
Data and Laboratory Analyses
3 Yorktown Water Filter Chemical Treatment Test Data 38
4 Yorktown Mixed Media Filter Expenditure Pattern 65
5 A Partial List of Plant Tests 92
vii
-------
ACKNOWLEDGEMENTS
This work was sponsored by Mr. R. C. Mallatt, Manager, Environ-
mental Conservation of Standard Oil Company (Indiana), whose
program suggestions and editorial assertions are acknowledged with
sincere thanks.
The design, construction and operation of the full-scale facil-
ities, the pilot plant studies, analytical work, and report pre-
paration were performed by a team from Amoco Oil Company consisting
of Messrs. D. B. Meyer, J. H. Mitchell, L. K. Halverson, A. W.
Peters, and J. F. Grutsch.
The aid of Mr. A. L. Conn, Director, Government Contracts, Amoco
Oil Company, is gratefully acknowledged.
The support of the project by the Water Quality Office, Environ-
mental Protection Agency, Mr. George Key, and the help provided by
Mr. Leon H. Myers, Grant Project Officer, is acknowledged with
sincere thanks.
viii
-------
SECTION I
CONCLUSIONS
1. Chemical coagulation/mixed-media filtration has been demon-
strated to be an effective tertiary water treatment for
accomplishing further reductions in suspended solids, oil
content and biochemical oxygen demand following a refinery
end-of-pipe treatment sequence consisting of an API separator
and aerated lagoon.
2. The most significant factors influencing the treatment ef-
fectiveness in the scalping mode operation are incident
contaminant concentration, the applied chemical treatment
level, and seasonal aerated lagoon conditions.
3. Backwashing with lagoon water will keep the filter media
satisfactorily free of oil and waste accumulation, and re-
turning the backwash waste to the lagoon can be handled in a
way to avoid unmanageable sludge buildup.
4. The filtration facilities require a minimum amount of operator
attention and generate no objectionable wastes or odors.
5. The unit operation can be demonstrated easily, appearance of
the effluent is greatly improved and the facility therefore has
considerable public relations appeal. Both the process and
the results are easily visible and comprehensible.
6. Unless adequate hydraulic controls are provided, the filter
media can be physically disturbed, rendering it less effective
than designed. Safeguards are necessary to prevent such
disturbances.
7. Optimized operation to achieve near potable water quality
clarity requires response to the required water chemistry for
destabilization of colloidal material. This response may
require taking any, or all, of the following into account: 1)
pH control; 2) two or three chemical destabilization systems,
and 3) reduced hydraulic loading at water temperatures less
than 15.50 C (60° F).
8. Brackish water has a major impact on the effectiveness of de-
stabilization chemicals.
-------
9. Brackish water interferes with direct usage of zeta potentials
for determining optimal chemical pretreatment because the re-
duction of zeta potential by double layer repression must be
detected. This is readily achieved by determining zeta po-
tentials on diluted samples.
10. Colloid entrapment and double layer repression are destabili-
zation mechanisms to be sorted out and avoided for direct
filtration.
11. Charge neutralization and bridging are required destabiliza-
tion mechanisms for optimal filter performance.
12. Weakly anionic polyelectrolytes are much more effective than
nonionics for filter aids in a three chemical system.
13. Even with optimized chemical pretreatment, filter loading must
be decreased with decreasing water temperatures.
14. Recommended filter hydraulic loadings are proportional to the
viscosity of water,
15. Incorporation of powdered activated carbon up to 150 mg/1 had
no favorable impact on the destabilization chemistry of bio-
colloids.
-------
SECTION II
RECOMMENDATIONS
1. Future effort on this unit should include investigation and
implementation of the chemical pretreatment system required to
optimize performance to yield near potable water clarity for
year-round operation.
2. The system variables considered in the chemical pretreatment
system study should include the following: pH, multi-chemical
destabilization system; zeta potential measurements; water
quality parameters, and water temperature-hydraulic loading
relationships.
3. An effort should be made to determine if there is enough
difference in properties of inert colloids in primary effluent
and biocolloids in aerated lagoon effluent to result in any
practical difference in the approach to destabilization by
chemical treatment.
4. Pilot plant data should be used cautiously if there is a
difference between the media support and that proposed for a
full-scale unit.
5. Year-round data on filterable contaminants and hydraulic
requirements should be used as a design basis to set minimum
design specifications if satisfactory year-round performance
is required.
6. Jar tests can be used to determine the type of chemical treat-
ment and the specific chemicals to which the waterborne
colloids are most responsive, but the results should not be
relied upon to predict the dosage requirements. Recognition
should be made of the fact that dosage requirements will
likely vary with the seasons even at the same contaminant
concentration.
7. Some type of adjustable, mechanical restriction should be pro-
vided to prevent excessive backwash rates in case of instru-
ment malfunction or operator error.
8. Design considerations involve trade-offs between hydraulic
loading rates and chemical pretreatment. The more responsive
the colloidal system is to chemical pretreatment the higher
are the hydraulic loadings achievable. Hydraulic loadings
-------
relate to water temperature. Recommended guideline values for
granular media filter hydraulic loadings preliminary to water
chemistry studies are as follows: 1.5° C (35° F) to 10° C
(50° F), an overflow rate of 6.1 m/hr (2.5 gpm/sq ft); 10° C
(50° F) to 15.5° C (60° F), an overflow rate of 8.6 m/hr (3.5
gpm/sq ft); 15.5° C (60° F) to 21° C (70° F), an overflow rate
of 11 m/hr (4.5 gpm/sq ft; and 21+° C (70+° F), an overflow
rate of 12.2 m/hr (5 gpm/sq ft).
9. Zeta potential measurements are a useful tool to determine the
optimal chemical pretreatment requirements for direct filtra-
tion. Zeta potentials must be used for brackish waters in a
manner that sorts out double layer repression mechanisms.
Charge neutralization and bridging are destabilization mechanisms
to be emphasized if optimal direct filtration results are to
be achieved.
10. Brackish water systems require special attention if optimal
chemical destabilization of biocolloids is to be achieved.
11. Brackish water biocolloid installations should have provision
for a three chemical pretreatment system.
-------
SECTION III
INTRODUCTION
The Amoco Oil Company Refinery at Yorktown, Virginia, has always
maintained a secondary waste water treatment system for process
waters and consistently discharged treated water of better quality
than regulations have required. The treatment system consists of a
two-cell API separator and about five acres of aerated lagoon. In
1968, permission was requested and obtained from the State Water
Control Board to deepen the lagoon and add mechanical aeration.
These improvements enhanced the effluent quality considerably, but
the aesthetic appearance of the water did not meet Amoco's standards.
The effluent was well stabilized, but contained suspended solids
which imparted a gray, turbid appearance and a very slight irides-
cence was sometimes present.
During the summer of 1970, a pilot plant was set up to investigate
the feasibility of improving the effluent appearance. In the pilot
unit, alum coagulation and mixed-media filtration were successful
in removing the objectionable turbidity and producing a clear ef-
fluent. In addition to the cosmetic effect, the pilot plant demon-
strated considerable real value in reducing effluent oil and BOD,.
levels.
Based on the success of the pilot plant, design was begun in
September, 1970, on a full-sized, mixed-media filter to treat the
entire lagoon effluent. The simplicity of the design and the over-
all effectiveness of the treatment seemed of general interest and
broad application, and Amoco applied for a Research and Development
Grant from the Environmental Protection Agency, Water Quality
Office. Amoco accepted a grant of not more than $73,815 on June 15,
1971, for a demonstration project, No. 12050 GQR. The stated
objective of the project was to demonstrate the simplicity, rugged-
ness, and adaptability of chemical coagulation mixed-media filtration
for producing a high quality effluent from a refinery aerated
lagoon.
Mixed-media filtration may be defined as clarification of a suspen-
sion by passage through a bed of porous media that separates, and
retains within the media, the solids constituting the suspension.
Historically, filtration was viewed as a polishing step following a
clarifier; however, today, mixed-media filtration in the petroleum
industry usually means "direct filtration." Direct filtration in-
volves injection of required colloid destabilization chemicals with
immediate transfer to the filter, i.e., there are no flash mix,
flocculation, or clarification process steps prior to filtration.
-------
The normal operation of mixed-media filters involves downward flow
through the filter until pressure drop due to clogging or break-
through of suspended matter increases to a predetermined limit.
The filter is then cleansed by reversed flow fluidization after
pretreatment by air scrubbing or a hydraulic surface wash.
Filter media include beds composed of sand; sand and coal; sand,
coal and garnet; and other minerals and synthetic materials.
Optimally, the suspension being filtered should pass through a bed
of increasingly smaller pore size.
This report has been prepared to disseminate the information from
the mixed-media filter study in a form acceptable to the Water
Quality Office, Environmental Protection Agency.
-------
SECTION IV
PILOT PLANT STUDIES
During the months of June, July, and August, 1970, pilot studies
were conducted on the Yorktown refinery waste lagoon effluent and
the API separator effluent to determine the applicability of chem-
ical coagulation/mixed-media filtration on these streams. The
pilot unit was a small, self-contained, commercial treatment system
(Neptune Micro FLOC Inc., Sweet Water Boy). It consisted of a
flash mix tank, tube settler, a 0.372 sq m (4 sq ft) mixed-
media filter and the appropriate pumps, and controls for automatic
operation. The unit had been used in extensive testing at Amoco1s
Whiting, Indiana, refinery and at the Texas City, Texas refinery
before being sent to Yorktown. Local tests were deemed advisable
to determine applicability of this unit on waters from a refinery
processing foreign crude oil. The results of the tests were then
logically extended to an application in a refinery at Cremona,
Italy.
The flash mix chamber and the tube settling section of the pilot
unit were not tested at Yorktown because of the cost-benefit data
developed at Whiting. Pilot plant testing of an iron salt floc-
culant and/or polymeric secondary flocculants was similarly ruled
out. The pilot unit is shown schematically in Figure 1. Operation
entailed alum flocculation in the chamber above the filter bed,
down flow filtration and clear water storage for use as backwash
water. The influent water and alum rates were set and the filter
level was controlled by pumping from the bottom of the unit.
Data from several of the pilot plant tests are given in Table 1.
It is axiomatic that the proportion of biological oxygen demand,
sulfide, phenolics, oil, or any other contaminant removed by a
phase separation, such as filtration, will depend primarily upon
the proportion of the specific contaminant that is insoluble in the
liquid. Turbidity measurements were used to give an indication of
insolubles in the pilot plant effluent and to monitor each test run
continually. Spot data taken throughout the test program confirmed
a rough correlation between turbidity improvement and improvement
in the levels of specific contaminants. About 50 percent of the
oil and biological oxygen demand were removed by the filter. The
test runs were all terminated by increased filter pressure drop
after 4 or 5 hours without any apparent loss of filter effectiveness.
Effluent turbidity remained essentially constant throughout each
test run.
Typical pressure drop data taken during one of the tests using pro-
cedures described later in this report are given in Figure 2.
These data indicate that the bed is indeed an "in-depth filter."
Unlike a typical sand filter, material was trapped throughout the
-------
CO
Alundum
Support
Plate
X
Coal
1.55 mm
Silica Sand
0.65 mm
Garnet Sand
0.43 mm
I I I I I I I I I I I I I I I ITTI
Alum Solution
Raw Water
Backwash
Waste
Mixed media filter pilot plant
Figure 1
1
Clear Water
Storage
\
Clear Water
X
-------
Table 1. YORKTOWN WATER FILTER PILOT PLANT OPERATING
AND INSPECTION DATA
Test no.
70-1
70-2
70-3
70-4
70-5
70-6
70-7
70-8
70-21
70-27
API separator
effluent
70-104
70-104a
70-105b
70-105
70-106
70-116
Water rate,
m/hr gpm/sq ft
16.5
17.7
17.7
12.5
12.2
12.2
6.48
5.87
12.2
12.2
12.2
12.2
9.3
12.2
12.2
12.2
6.75
7.25
7.25
5.1
5.0
5.0
2.65
2.4
5
5
5
5
3.8
5
5
5
Alum rate,
mg/1
0
5.4
9.0
0
5.0
10.0
5.5
11.3
10
10
10
15
20
15
15
15
Turbidity Avg oil,
JTU mg/1
In
11.3
10.3
10.3
9.1
10.9
12.0
8.2
9.6
7.1
13.0
28
29
29
25
22
21
Out In
8.9
4.6
1.9
7.9
4.6
1.4
4.4
1.5
1.5
1.3 77
8.6 117
4.5
3.9
6.3 34
4.8 12
4.6 119
Out
17
43
14
20
7
24
Analytical procedures are listed in Appendix A
-------
101.6
-------
filter bed. The Alundum bed support plate also developed an in-
crease In pressure drop as the test runs progressed (Figure 3).
The effect of the support plate on effluent quality was not determined.
At the end of a test run, the effluent was shut off and the filter
was backwashed at 36.7 m/hr (15 gpm/sq ft) for 6 minutes. The
rate was 12.2 m/hr (5 to 10 gpm/sq ft) less than recommended, but
no evidence could be found to indicate the need for a higher rate
or a greater duration. Sludge settled to 5 percent of the backwash
water volume in less than 2 hours and did not change volume within
the next 60 days. Based on a 5-hour filter cycle at 12.2 m/hr (5
gpm/sq ft) and 36.7 m/hr (15 gpm/sq ft) for a 6-minute back-
wash, the backwash waste volume equalled 6 percent of the filtered
water, and the settled sludge volume equalled 0.3 percent. Tests
using polymeric flocculants to reduce the settled sludge volumes
were inconclusive.
11
-------
IH
flj
•P
§
E
o
ft
o
fn
n
-------
SECTION V
DESIGN AND CONSTRUCTION
RESPONSIBILITY AND CRITERIA
The process design of the chemical coagulation/mixed-media filter
system was the responsibility of Neptune Micro FLOC Incorporated
under contract with Amoco Oil Company. They provided the automatic
control mechanism, the filter media, and supervision for the media's
installation. Piping and construction drawings were prepared by
the local Amoco staff.
The unit was designed to treat 341 cu m/hr (1500 gpm) lagoon ef-
fluent of 9 to 12 Jackson Turbidity Units (JTU) and to produce an
effluent of less than 3 JTU. Based on the pilot plant tests and
the lagoon effluent data available at the time, the following re-
ductions in contaminants were expected:
Filter Influent Filter Effluent
BOD5, mg/1 7-25 *NMT 15
Suspended Solids, mg/1 5-40 NMT 3
Volatile Suspended Solids, mg/1 ? NMT 1
Oil (CC14 Extracted), mg/1 10-20 NMT 10
Sulfides, mg/1 NMT 1 NMT 1
Phenolics, mg/1 0-2 20% Reduction
*Not More Than
FINAL DESIGN FEATURES
A simplified process flow schematic is shown in Figure 4. The
basic differences between the final design and the pilot unit were:
1. The larger unit provided 13 minutes of residence time above
the bed, compared to 2 minutes in the pilot unit.
2. Two rotary surface wash arms, using filtered water, were
provided. The pilot unit had a stationary surface wash but
it was inoperative.
3. A plastic pipe under-drain system and graded gravel support
were provided in lieu of the Alundum support plate.
13
-------
v\\\\\\\\\V
YYYYYYYXXXX
O O O O O
JHHSc
Alum
Tank
x
Fe<
Pun
• Jf X.'
d
P
B.
Pu
M
w. Aerated Lagoon
imp
Mixed media filter flow schematic
Figure 4
BacTcwash
Waste
Clear
Water
to River
-------
4. No provision was made to store filtered water for backwashing
the bed. Lagoon effluent water was deemed of adequate quality
to be used as backwash.
The unit was designed to operate automatically and unattended.
Automatic shutdown devices were provided to prevent an overflow in
the event of an equipment malfunction. A 27.9 sq m (300 sq ft)
filter bed was specified, utilizing a 17.8 cm (7 in.) depth of 0.7
mm ilmenite covered with 11 inches of 1.0 mm silica sand and
topped with 27.9 cm (22 in.) of 1.3 mm coal. Five pumps were
specified, one each for feed, effluent, backwash, surface wash, and
alum injection. Liquid alum was to be delivered by truck and
stored in a 3.05 m (10 ft) diameter by 3.66 m (12 ft) high fiber-
glass tank. A small, clear well was provided to store filtered
water for the surface wash and to allow some surge capacity for the
effluent pump.
The unique feature of this design is the filter bed itself. The
quality of the effluent expected from any filter design depends
largely upon bed depth and media particle size. Small particle
sizes produce a more clear effluent at the expense of a more rapid
pressure drop build-up. Greater filter bed depth could be expected
to give greater effluent quality, but the effective bed depth on a
single media filter is limited by surface blinding. That is, ma-
terial capable of being trapped in any specific media will be
trapped within the top few inches of the bed; the remainder of the
bed will then have only clear water flowing through it.
The mixed-media filter design uses three different sizes of material
arranged so that the water becomes progressively more clear as it
flows down through the bed. High quality effluent can be expected
because of the fine particle size in the bottom of the filter.
This design also gives an effective bed depth because each type of
media removes some of the contaminants, and the fine particle
portion of the bed is actually treating partially filtered water.
An upflow backwash would normally tend to reclassify particles of
the same specific gravity, with the smaller ones on top. The
specific gravities of the three bed materials in this design were
chosen to prevent such normal reclassification by particle size,
but enable reverse classification into 3 layers, with the small,
dense particles on the bottom. Ilmenite has a specific gravity of
about 5, while sand and coal are about 2.5 and 1.5, respectively.
The filter bed and support design are shown in Figure 5.
CONSTRUCTION
The unit is an open-top, reinforced concrete box, as depicted in
Figures 6-9. It was built beginning in February, 1971, by Amoco
15
-------
1.0 mm Silica Sand
0.7 mm Ilmenite
..95 * .48 (3/rx 3716) V
i:? 1i?^? &¥& °S
Gravel
Gravel
Gravej. <
5-1 X 2.5
(2 X 1) Gravel
Surface Wash
Arm
Dimensions in
centimeters (inches)
25 Underdrain Pipes, 30-5 (12 in.) o.c.
Mixed media filter bed and support design
Figure 5
16
-------
"f*r
1 I
B
TT
i i
o
B
f
5-8
ir
1.66
•3 (1)
P-
8.2 (27)
>3 (l)\f
Dimensions are in meters (feet)
Yorktown filter plan view
Figure 6
17
-------
•5.5 (216)
CD
Inlet Nozzle
Waste Nozzle-
v~
A
__ *•>
^\^t
^-^
^| I
"*^ , ^
V_'
...
Q
it
it
'i
Media
»
Support Gravel
.
Scale 1:1+8 (l/V = 1')
\
1.27
1
.8lf
>
J
1
• 53
L
(50)
(33)
\ ^-^
(
(^)
1
y
A
J21)
(168)
f
Dimensions in meters (inches)
Yorktown filter section viev A-A
(looking east)
Figure 7
-------
Vent with check valve
Effluent
ooooooooooooooooooooooooo
Plenum Chainber
Drain
Scale 1-M (lAM -I1)
Yorktown filter section view B-B
(looking south)
Figure 8
-------
o o o
/
Channel supports
r Swivel Joint
Media
Support
Scale 1:J»8 (1/4" « 1')
Dimensions are meters (feet)
Yorktown filter section view C-C
south)
U
O 0 O
l.JA (3-75)
All walls and
floor .3 (l)
Plgore 9
-------
forces and the in-plant construction and maintenance contractor
Catalytic Construction Company. The construction was straight
forward without any major complications or delays. On June 22,
1971, a representative from Neptune Micro FLOC began supervising
the 5-day media-placement procedure. By June 29, the unit was in a
usable condition with the major construction completed and only a
few instrumentation problems to be solved and the weather proofing
to be completed.
21
-------
SECTION VI
OPERATION
NORMAL FILTER CYCLE
The unit has been in essentially continuous operation since it
was started up on June 29, 1971. The control system has functioned
well, and the amount of operator attention required at the unit
has been minimal. An operator normally visits the unit once per
shift, primarily to inspect the equipment and to adjust the in-
fluent water rate in order to control the lagoon level. He also
records the pertinent operating data at that time, including
water and alum rates and effluent turbidity and pressure.
During a normal filtering cycle, water is pumped from the lagoon
through a flow controller and into the filter box. The 50 per-
cent alum solution is injected by a metering pump directly from
storage tank into the line upstream of the flow controller, thus
providing thorough mixing before the water enters the filter
tank. Coagulation, as described later, begins immediately, and
floe particles of varying sizes up to 5 mm can be seen in the
gentle swirling water above the filter bed. The water flows by
gravity through the media into the under-drain system and into a
clear well from which it is pumped to the York River. As the
voids in the media fill with material removed from the water, the
differential pressure measured across the filter bed increases.
Consequently, the pressure at the outlet falls. When it reaches
a preset level (usually .6 m (2 ft) of water head), a multi-
function timer that will automatically control a backwash is
started.
NORMAL BACKWASH CYCLE
The backwashing operation is illustrated by the sequence of func-
tions performed by the backwash timer and shown in Figure 10.
After the filter influent and effluent has been stopped, the
water is lowered to the level of the waste troughs by opening the
waste water valve. Surface washing is accomplished by water
jetting from nozzles on two arms rotating just above the bed
surface. The arms operate much like an inverted lawn sprinkler,
jetting about 1.2 cu m/hr (.5 gpm/sq ft) on the media surface
to break up any crust that may have formed. The recommended
primary backwash rate is a function of the water viscosity and
varies linearly from 39 m/hr (16 gpm/sq ft) at 4° C (40° F)
water temperature to 73 m/hr (30 gpm/sq ft) at 29.5° C (85° F)
for the original filter design. Backwashing at this rate causes
the filter bed to expand (covering the surface wash arms) and
22
-------
Cycle Time, minutes
Cam TiKier On
Influent Valve Open
Effluent Valve Open
Feed § Alum Pump On
u>
Effluent Pump On
Waste Valve Open
Surface Wash Pump On
S. Wash Valve Open
Backwash Pump On
Backwash Valve Open
Filter backwash cycle timer functions
Figure 10
-------
release the entrapped waste. The partially-fluidized, expanded
bed tends to be self-scouring. The timer begins another filter
cycle at the end of each backwash and then deactivates itself.
OPERATING DATA
Averages of operating and analytical data for the various periods
of this study are given in Table 2. Based on previous data, the
incident contaminant load was unexpectedly high and variable. On
several occasions, particularly during the March-May quarter, the
lagoon water became very milky and exhibited turbidities up to
300 JTU. Turbidity data from these periods have been excluded
from the averages as unreliable and nonrepresentative. The milky
condition is believed to have been caused by a colloid (possibly
sulfur) which resulted from the oxidation of a dissolved contami-
nant in the water. Filter effluent turbidity was virtually
impossible to measure during these periods because it changed
during the sampling and measuring procedures. Material that
passed the filter in a dissolved state became oxidized as the
samples were drawn and the resulting colloid produced a turbidity
that continued to increase for up to two hours.
Again, it is important to note that the removal of any contami-
nant from the water is dependent not only upon the efficiency of
flocculation and filtration as measured by suspended solids or
turbidity removal but also upon the extent to which that con-
taminant is dissolved in the water. For example, the gradual
deterioration of filter performance over the first four periods
is not evident from the sulfide removal data. The increase in
dissolved oxygen can be attributed to the exposure of water from
the bottom of the lagoon to the air in the shallow, open-filter
tank.
Significant data from the summer of 1972 are not available be-
cause the grant period had expired. The filter bed was obviously
displaced, however, the unit was in continuous operation during
the period and effecting a beneficial change on the lagoon
effluent, but laboratory documentation was not obtained.
CHRONOLOGY
A chronological listing of events pertinent to the unit opera-
tions, modifications, and evaluation is given in Appendix B.
CHANGES FROM ORIGINAL DESIGN
No significant changes have been necessary in the original mode
of operation. The backwash flow rate correlation was changed
24
-------
Table 2. YORKTOWN MIXED MEDIA FILTER AVERAGE OPERATING DATA AND LABORATORY ANALYSES
N9
bn
Quarter ending
Data from appendix
Water rate, cu tn/hr
(gpm)
Alum rate, wppm
Water temperature, °C
(°F)
Turbidity, JTU
Oil, mg/1
Fhenollcs, mg/1
Sulfide, mg/1
PH
Suspended solids, mg/1
Volatile sus. solids, mg/1
Ammonia, mg/1
Dissolved oxygen, mg/1
BOD5, mg/1
Sept. 1, 1971
C
%
Inlet Removed
255
(1125)
20.0
27
(81)
20
7
1.1
0.7
7.1
18
35
84
73
73
14
73
54
Dec. 1, 1971
D
%
Inlet Removed
331
(1456)
24.4
20.5
(69)
32
16
0.7
3.1
7.0
43
35
54
1.4
19
79
49
34
26
68
71
1
-39
44
March
Inlet
332
(1461)
29.4
10.5
(51)
65
16
0.5
0.9
7.4
68
50
55
2.6
18
1, 1972
E
%
Removed
63
38
20
33
43
42
0
-62
39
June 1, 1972
F
%
Inlet Removed
296
(1304)
31.6
15.5
(60)
68
17
0.9
1.0
7.5
69
50
52
0.7
34
47
22
14
50
45
40
3
-65
12
Dec. 6, 1972
G
%
Inlet Removed
343
(1510)
21.2
18.5
(65)
33
9
0.6
0.9
7.3
90
18
67
36
29
54
53
29
-------
when an error was discovered in the original instructions, and
again when the media was changed, but the basic method of oper-
ation has been satisfactory.
Several minor equipment changes have been necessary. Thin red-
wood strips have been bolted to the edges of two of the backwash
troughs to compensate for level differences between troughs and
equalize the waste flow distribution into the gullet. A .84 sq
m (9 sq ft) redwood Impingement baffle was installed in the
gullet to prevent displacement of the top media layer as des-
cribed later. A secondary flocculent nozzle was mounted on the
impingement baffle pointing into the center of the influent
nozzle. The continuous recording turbidimeter which was orig-
inally mounted inside the control cabinet was remounted on the
control building wall to provide easier access and prevent
moisture from affecting electrical circuitry.
The most significant modification of the original design was a
complete changeout of the filter media and gravel support. The
replacement was necessary because the original bed had become
severely disturbed. Support gravel (2.5 cm) was found in the
sand layer and most of the ilmenite and some of the sand had
migrated through the support and under-drain laterals into the
under-drain plenum. The vendor indicated that only four or five
such failures had been observed in four to five thousand units.
The new bed was an advanced design as shown in Figure 11. It
provided a finer media for the bottom layer to improve effluent
quality without additional head loss because the upper layers of
media were redesigned to collect a larger percentage of the
contaminant. The new design had the added advantage of requiring
a lower backwash rate. Thus, it eliminated the need to replace
an undersized waste water effluent nozzle. Responsibility for
the bed replacement was shared, the manufacturer supplying the
material and technical assistance, and Amoco supplying the labor.
The approximate costs were $4,000 for material, $2,000 for freight,
and $1,250 for five days technical help. Labor costs included
three laborers, a foreman and a crane operator for ten days.
Five days were required to remove the old bed and clean the areas
between the under-drain laterals, and five days were required to
lay the new bed.
26
-------
33.8 (33)
365-8 ( 144)
30.5 (12)
127 (»)
131.6 (40) 2Q_3 (-
51 (20)
22.9
(1..5)
53
NOTE:
Dimensions in cm (in.)
1.45 mm COAL
1.3 MM COAL
SALVAGE FROM OLD BED
0.33 mm SILICA SAND
9,2? nn
'i wm-iru*.i-r.i
•fLW.m-'AW- 7.\'H
_L
SUHFACC
ARM
1.3 mm GARNET
.6 mm GARNET
2 X 1 GRAVEL
Q Q
25 UNOERORAIN PIPES, 30.5 (12) O.C.
Scale
(l/U in. = 1 ft)
Mixed media filter bed and support, second design
Figure 11
27
-------
SECTION VII
TESTS AND OBSERVATIONS
MEDIA SURFACE DISTURBANCES
On several occasions when the media was purposely exposed, a bed
disturbance was noted that involved about one-third of the bed
surface. Coal from the west end of the south wall had been
deposited In a mound over the west surface wash area leaving a
hole 51 cm (20 in.) below the original bed surface (Figures 6 and
7). This coal displacement was determined to have been caused by
the influent water during normal operation. Water entered the
filter from the east, traveled 7.6 m (25 ft) and impinged on the
west wall. From there the water moved over the gullet wall and
4.6 m (15 ft) on the south wall, and then down about 1.5 m (5
ft), impinging on the bed surface and displacing the coal. The
surface was completely restored by a normal backwash so that the
disturbance was only evident if the unit was drained before
backwashing (not usually the case).
The effect of the coal displacement on filter effluent quality
could not be documented, but pressure profiles indicated that the
areas of the bed where the hole and mound would be formed did
about equal work immediately after a backwash, before the coal
had been completely displaced. The hole in the bed did provide a
minimum resistance path for the backwash water. Evidence of high
velocity could be seen on the water surface in the southwest
corner during the first minutes of a backwash. A redwood splash
baffle directly in front of the inlet nozzle completely elimi-
nated the bed disturbances.
MEDIA PRESSURE PROFILES
Media pressure profiles were taken from different locations in
the bed with a probe, shown in Figure 12. One leg of a manometer
was connected by transparent flexible tubing to the small tube in
the probe. Tubing from the other leg was weighted and allowed to
hang below the water surface in the filter. Then, by inserting
the probe into the bed and positioning the piston opposite the
probe holes at the desired depth, the pressure drop from the
surface to that depth could be read directly on the manometer.
The total head loss from the filter surface to the pressure
switch on the filter effluent was never accounted for by probe
measurements of the media. The head loss unaccounted for, about
.46 m (1.5 ft) of water at 341 cu m/hr (1500 gpm), remained
essentially constant. It was assumed to have been partially due
to pressure drop through the support under-drain system and
28
-------
i 6.4 cm
9 cm . O.D. Brass
exchanger tube **
J_
^ *
. y J
^•K/
^_ i\p^^
:t. "A- A"
1
s
o
3
}
L
At
*
1
i
s
rH O
WO
* O*
I
GO LT^
0 cm
to
rH
WO
0 5
•
&
• B
I
ta
3
cr
0
o
n
01
0)
r-H
O
t-l
1-9
1
<
M_-
r
i
<
>
-j-<
\
\
\
cm
1 '
1
\
\
r\
*~\*\
i
>
i
im V
/
/
r
_*
v
i
r
6.k cm ,
/
Score tubing with
to aid positioning
piston at elevation
of holes in probe
•
.64 •» Stainless
•teel tubing
a
j
3
n
Machine J Cm
0-ring grooves
top & bottom
V-s
^••{3
.36 cm drill-P^
recess .. Tt"
.16 Cm ^ crossdrill -r
^^•«
Ji
HEI
0
ss
•
f
•
4
0
t
f
t
»^^
Adapt top o£.'
tubing to hook
to manometer
tube
E
^
m
E— H M
1 C O<
rrj °
i w
to
1.9 cm
ube
Elevation
Porous bed pressure probe
Figure 12
29
-------
partially due to measurement error Induced by a wall effect on
the probe itself. The success of the media design at "in depth"
filtration can be inferred from profiles as shown in Figure 13.
Head loss developed at all levels of the media as filtration
progressed.
BACKWASH TIME AND WATER RATE
The recommended backwash flow versus water temperature relation-
ship that was supplied with the unit was in error. As a con-
sequence, all the backwashes for the first four and one-half
months of operation were about 65 percent of the intended rate.
Effluent quality during this period was poorer than expected from
the pilot plant but the percentage of contaminant removal was
only slightly less than expected. There was no significant
improvement in contaminant removal after the backwash rate was
increased. Head recovery was increased about 50 percent to 1 m
(3 ft) of water by the more vigorous backwashing, thus filter
run lengths were increased 50 percent.
The filter bed had expanded, raising the surface about 5 cm (2
in.) within the first two months of operation. The bed surface
subsequently dropped 17.8 cm (7 in.) to 12.7 cm (5 in.) below the
original level after the higher backwash rate was put into prac-
tice. On the erroneous assumption that coal was being backwashed
out of the bed, the surface wash cycle was set to terminate
before the backwash valve had reached its maximum opening. The
quantity of coal that could be seen flowing into the waste troughs
(less than a handful per backwash) was not affected by the change
in surface wash. The manufacturer reported that a small loss of
coal was to be expected with the higher rate.
Visual observation of the waste water flowing into the backwash
troughs indicated that most of the material trapped in the bed
was released between one-half and five minutes after the backwash
valve was opened. The minimum acceptable backwash time was
determined by observing the head recovery on backwash for two
successive filter cycles. The first backwash observed for this
test was the shorter of the two and was of the same duration that
had been set up as standard operation a week earlier. The second
cycle was given the designed seven minutes. Head recovery after
a five-minute backwash was within 6 cm (2.4 in.) water of that
after a seven-minute wash.
Because the Yorktown filter uses lagoon effluent water for back-
wash and returns the waste to an upstream portion of the lagoon
for settling, there is only a minimum penalty for extending the
backwash time. It has been left at seven minutes. If, however,
filtered water were being used for backwash water, the reprocessing
would consume valuable filter capacity and the backwash time
would have been lowered to about five minutes.
30
-------
0)
•
I
-p
£
tt>
t-l
-------
OIL CONTAMINATION
During October, 1971, a torrential rain flooded the API separator
and produced an oil slick on the lagoons. Free oil was pumped
onto the filter during normal operation and up through the filter
during backwash. Oil from the backwash apparently coated the
media and was subsequently partially flushed out by the filtered
water to form a slick in the effluent sump for the next several
hours. Oil removal efficiency of the unit remained good in spite
of the oil soaked condition of the media. Routine samples taken
during the next week had 62 and 87 mg/1 oil in the influent water
and 9 and 23 mg/1 in the effluent. Media samples taken ten days
after the initial contamination showed coal with 200 and 330 wppm
oil and silica sand with 60 wppm. Forty days after the incident,
no oil could be found on the coal.
During the week following the rainstorm, and at other times when
the filter influent oil level was high, the backwash waste con-
tained free oil. Oil that was invisible in the influent water
was apparently agglomerated into floe particles that subsequently
broke up and released larger oil droplets that were visible on
the backwash waste water.
MEDIA SAMPLING
Attempts at core sampling with plastic pipe proved unsuccessful
with 2.5 cm (1 in.) and 5 cm (2 in.) pipes held either vertically
or on an angle. In general, if the media was fluid enough to
enter the sampler without distortion, it would fall out the free
end as the sampler was withdrawn from the bed. All media samples
were obtained by digging through the coal to the sand. Then,
while standing on the sand, the investigator obtained samples of
the lower two layers of media by using a clam-shall type posthole
digger. Extreme care was necessary to prevent disturbing the top
support layer.
MEDIA SUPPORT DISTURBANCES
About four months after the proper backwash rate was initiated,
four holes were dug in the bed to confirm the extent of media
loss. In each case, the coal layer was about 53 cm (21 in.) deep
(originally 56 cm (22 in.)) and had a sharp line of demarcation
between it and the sand layer. The sand and fine garnet layers
(originally 28 cm (11 in.) and 17.8 cm (7 in.), respectively)
were not distinguishable as separate layers, but appeared as
40.6, 48.3, 38.1, and 43.2 centimeters (16, 19, 15, and 17 in.)
of sand in the four respective holes. The top two support layers
7.6 cm (3 in.) coarse garnet and 7.6 cm (3 in.) of .95 x .48 cm
(3/8 by 3/16 in. rock) were mixed in a layer from zero to 12.7 cm
(5 in.) thick. A small coal loss to the backwash waste had been
expected, but the 12.7 cm (5 in.) drop in the bed surface seemed
to have come from the top support layers.
32
-------
Three tests were conducted to determine the extent of the support
loss. A .64 cm (.25 in.) diameter rod was pushed into the bed at
each intersection of a .3m (1 ft) by .3m (1 ft) grid while the
filter was operating in a normal manner. The rod was forced down
until the support layer was felt, and the length of the rod below
the water level was noted. The readings obtained in this manner
are somewhat subjective but were found to be reproducible within
about 5 cm (2 in.). The surface of the support was not flat and
did not remain constant. In the first test the maximum reading
was 20 cm (8 in.) above the original level and the minimum was 20
cm (8 in.) below with the average of 312 readings being 5 cm (2
in.) below the original level. Three weeks later the extremes
were 22.9 cm (9 in.) above and 20 cm (8 in.) below, with an aver-
age of 7 cm (2.75 in.) below. The high spots and the low spots
did not necessarily appear in the same places during the two
tests. The maximum difference between two adjacent readings (,3m
(1 ft) apart) was 30.5 cm (12 in.).
At the suggestion of the filter manufacturer, a 12.7 cm (5 in.) by
15.2 cm (6 in.) plate was welded to the end of a rod and the
apparatus used as a probe to determine the level of the support
layer during a backwash. The extreme readings were 38 cm (15 in.)
above and 20.3 cm (8 in.) below the original, with the average
about 2 cm. (.75 in.) below original. The maximum variation
between adjacent locations was 51 cm (20 in.). At several of the
locations where the support was recorded more than 30.5 cm (12
in.) above original level, the probe felt as if it were resting
upon a steep mound of mud. No physical evidence of these "mud
hills" could be found in the two locations opened by coring at
that time. In fact, no mud was ever found in the bed during the
approximately one dozen samplings, or during any other investigations.
A possible explanation for the phenomenon of feeling "mud hills"
when none were there was suggested during a visit of the manu-
facturer's representative, but it was never proved. The under-
drain system was designed to have a 4 m (13 ft) water head loss
at 1840 cu m/hr (8100 gpm) backwash rate. It actually exhibited a
5 m (16.5 ft) water loss at only 1567 cu m/hr (6900 gpm). This
indicated a flow restriction that was postulated to have caused
localized high velocities similar to a subsurface fountain. These
fountains or jets might have had enough force to tend to support
the probe if it were restrained from "sliding off the hill". The
under-drain system was later proved to be unrestricted, but part
of the support gravel was packed with finer gravel and sand there-
by restricting flow through the support.
Another probe was constructed using .64 cm (.25 in.) mesh screen
instead of the steel plate. It was unavailable to test the water
jet theory, but later proved unsatisfactory in probing the replacement
33
-------
filter bed. The bars that were used to stiffen the screen formed
a basket and trapped support and media on top of the screen. Its
use was discontinued to prevent a possible mixing of the support
media in the new bed.
The downward migration of the media and the disturbance of the
original support were determined to be critical and irreparable,
and the entire bed had to be replaced. The reason for the problem
is believed to have been localized, high backwash velocity.
Several causes for high velocity in the bed were postulated but
none could be proved. The probability of air in the backwash was
slight because the air vent had been functional since the startup.
The possibility of poor distribution because of the broken under-
drain pipe was eliminated by Inspection during the bed changeout.
Another theory is that mud balls were formed during the early
weeks of operation when the surface wash arms were not functioning,
that the mud caused poor backwash flow distribution and gravel
displacement, and that the mud was broken up and eliminated by the
surface wash when it was put into service. This theory seems
unlikely. A more plausible explanation would be that the backwash
rate which was set by the manufacturer and limited by an under-
sized backwash waste nozzle was inadequate to clean the bed. The
large surface depression created by the inlet water turbulence
before the baffle was installed afforded a low resistance channel
to pass an inordinate portion of the backwash water. This, then,
reduced the cleansing in the remainder of the bed and the situation
compounded itself until the localized velocity was sufficient to
carry gravel up into the filter media. With the gravel support
disturbed, the finer media and support then migrated down into the
coarser layers and eventually through the under-drain laterals.
About 1.53 cu m (2 cu yds) of fine media were found in the plenum
during the bed changeout.
OIL DISPERSANT TEST
Laboratory analyses had indicated that oil dispersants or sur-
factants that at one time were used to wash down equipment would
cause a milky appearance in the API separator water. The milky
condition in this case was attributed to a fine dispersion of oil
in the water. A test of the filter on such a dispersion was
performed by dripping Jansolv directly into the feed pump suction
screen. About 12.2 m/hr (5 gpm/sq ft) of 52 JTU, 12° C (54° F)
water was treated with 21 wppm alum for the test. The rate of
head loss was .088 m/hr (.29 ft/hr) with no Jansolv and .107
m/hr (.35 ft/hr) with 11 wppm of the dispersant. The dif-
ference between the average effluent turbidities was even less
significant, going from 16 JTU with no dispersant to 17 JTU with
it.
34
-------
JAR TESTS
The bench scale screening tests of secondary flocculants were
done with a gang stirrer. This device could be used to stir
from two to six samples at the same speed simultaneously. The
stirring speed could be varied from 15 to 100 rpm. Typically,
four, 8-ml beakers of lagoon water would be stirred at about
20 rpm to simulate the movement of water in the filter, above
the bed. Different flocculant-alum combinations were put into
each sample and the floe formation was observed. The procedures
were quantitative, but the results could only be expressed as
"this is better than that" and were extremely subjective. The
following conclusions were drawn from the jar test when no pH
adjustments were made:
1. Alum and polymeric flocculant together treat Yorktown waste
water better than either can alone.
2. Alum should be added before the polymer.
3. Cationic polymers work best for Yorktown waste water.
4. The turbulence produced by the filter feed water pump does
not affect the treatability of the water.
5. A sample of water from a 7° C (45° F) lagoon does not respond
to 30 wppm alum differently than the same water after it
has been heated to 25° C (77° F).
The lack of observable temperature effect on chemical treatment
response was somewhat different than observed for physical
phenomena. Forty 2-hour quiescent tests indicated considerable
temperature effect on settling. Samples of 53 JTU water settled
after chemical treatment to yield a supernatent of 18 JTU at
40.50 c (105° F), 18 JTU at 24° C (75° F), 21 JTU at 13° C (56° F),
and 27 at 4° C (39° F). Apparently, the increase in water vis-
cosity at lower temperatures is sufficient to impede quiescent
settling, but not clarification by flocculation in moving water.
MEDIA STERILIZATION
Although repeated sampling had indicated no media contamination
or oil buildup problem, the condition of the support gravel was
not certain. It seemed possible that a biological culture might
have begun to grow on the support and that it could contribute
detrimentally to the filter effluent. The unit was backwashed,
shutdown, and allowed to stand full of a copper solution for 30
minutes in an attempt to loosen any biomass that might have
accumulated. Sufficient copper sulfate was used to give 10 mg/1
35
-------
copper in the water but the actual solution strength was some-
what less. Some of the ionic copper was reduced to elemental
copper; it plated out on the metal dissolving apparatus and formed
a colloid imparting a red-brown color to the water. No evidence
of a blomass could be observed in the solution as it was drained.
Time requirements for ingestion and toxication by copper are
probably greater than 30 minutes, but it seems likely that copper
was deposited in the unit to give a residual effect. No evi-
dence of a biomass was ever seen and the unit performance did
not improve after the treatment; however, it was performing
excellently at the time and any improvement in performance would
have been difficult to detect.
pH EFFECT
Yorktown lagoon effluent has exhibited extremes of pH from 6.0
to 8.9 during this study, but by far the majority of observations
have been in the range of 6.8 to 7.5. The number of the pH
observations that were considered above the range for good alum
coagulation is listed below as a fraction of the total number
of observations for the respective periods.
Quarter Ending 9-1-71 12-1-71 3-1-72 6-1-72 12-1-72
pH greater than 7.5 0/5 3/11 11/26 8/35 2/27
pH greater than 8.0 0/5 1/11 4/26 3/35 0/27
Repeated instances of high pH during early 1972 caused no decline
in filter performance that could be attributed to pH alone.
These high readings were the result of upstream conditions and
practices (that have since been changed) and were always accom-
panied by high, filter inlet turbidities. Treatment effectiveness
was not noticeably different from other periods of high turbidity
at a pH range of 7 to 8.5. No facilities for pH adjustment were
deemed justified.
CONTAMINANT LOAD, TEMPERATURES, AND THROUGHPUT EFFECTS
Correlations were developed between turbidity removal and incident
turbidity, temperature, and throughput by plotting log sheet data.
Scatter of these data was extreme but plotting median percent
turbidity removal against 20 JTU increments of inlet turbidity,
2.8° C (5° F) increments of temperature or 11.4 cu m/hr (50 gpm)
increments of throughput indicated the following linear relation-
ships :
R • 100 - 3.82 (TP-7)
R - 1.14 T + 0.8
R - 90 - 0.357 I
36
-------
Where: R = percent turbidity removal
TP = hundreds of gpm throughput from 10 to 17.5
T = temperature from 4.5° C (40° F) to 29.5° C (85° F)
expressed in degrees F.
I = incident turbidity from 10 to 140 JTU
These relationships were not intended to indicate unit capability,
only to give a history of first-year, overall performance including
operator error and equipment malfunction and to serve as a basis
for evaluation of later improvements. The majority of high inlet
turbidities occurred during the winter when the hydraulic load was
the greatest and the water temperature the lowest, thus lending a
bias to the data.
CHEMICAL TREATMENT TESTS
Definition of optimum chemical treatment for any specific waste
water certainly would have only limited applicability. Simple
chemical pretreatment using primary coagulants likewise has only
limited applicability to most waste waters since it typically re-
sults in a "scalping" operation that does not achieve essentially
complete removal of suspended material by filtration. Optimum
chemical pretreatment to achieve consistent near-potable, water
clarity usually requires more response to water chemistry than
just the use of primary coagulants. This aspect of pretreatment
is discussed in a later section comparing results at other re-
fineries to results at Yorktown.
The results of Yorktown1s preliminary treatment tests are shown in
Table 3 and Figures 14 through 20 only to indicate basic equipment
responses. The date of each test has been coded into a test
number using one digit for the year, two for the month, two for
the day, and one for the test number. Each test was begun after a
normal backwash during which the polymer addition rate was con-
firmed at the injection point. The filter-effluent turbidity and
pressure were recorded as the only criteria for a successful
chemical treatment. The head loss rate after the first 20 minutes
was essentially constant for the 2 1/2-to-3-hour tests. In this
open top unit, a decrease in pressure (head loss) at the effluent
is equivalent to an increase in differential pressure as often
reported in the literature. Effluent turbidity did not remain
constant or increase at a constant rate during most normal opera-
tions. An arithmetic average of the turbidity at 20, 50, and 120
minutes after the test started was arbitrarily chosen to represent
the effluent turbidity as a single number. The polymeric flocculant
in use at the time was Betz 1175.
The effects of contaminant load and temperature on effluent quality
are deomonstrated in Figure 14. Polymer addition at a constant
alum dosage improved effluent quality as shown in Figure 15.
37
-------
Table 3. YORKTOWN WATER FILTER CHEMICAL TREATMENT TEST DATA
Test #
201121
201122
202081
202091
202091
202101
202111
202251
202291
203011
203071
203281
203282
204051
204062
204063
204064
205301
206011
206012
206021
206022
206031
206032
305021
305031
306141
Water rate,
m/hr gpm/sq ft
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
6.5
7.3
9.8
13.7
6.5
6.5
9.8
12.2
12.6
13
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(5.
(2.
(3.
(4.
(5.
(2.
(2.
(4.
(5.
(5.
(5.
0)
0)
0)
0)
0)
0)
0)
0)
0)
0)
0)
0)
0)
0)
0)
0)
0)
67)
0)
0)
60)
67)
67)
0)
0)
17)
33)
JTU
Inlet
55
55
80
65
65
54
48
52
33
28
80
52
52
170
155
155
155
100
50
63
51
50
44
43
22
16
23
Outlet
11
14
18
22
15
9
9
15
9
6
21
16
17
110
110
39
48
17
12
21
25
17
4
11
3
2
6
Temp.
0C (Op)
14.5
14.5
5.5
6.5
6.5
5.5
6
10
11.5
13
11.5
11
13.5
13.5
13.5
13.5
13.5
23
21.5
22
21.5
23
22
23
-
-
29
(58)
(58)
(42)
(44)
(44)
(42)
(43)
(50)
(53)
(55)
(53)
(52)
(56)
(56)
(56)
(56)
(56)
(73)
(71)
(72)
(71)
(73)
(72)
(73)
-
-
(84)
Alum
wppm
30
33
45
30
30
15
7.5
22
30
30
9
21
21
27
3.8
27
27
52
10
10
10.4
9.8
22.2
14.9
28
27
20
Polymer
wppm
0.0
0.0
3.2
0.0
2.3
1.0
1.1
0.34
0.0
*
1.9
0.0
0.0
0.0
5.0
5.0
11.8
0.0
5.3
0.27
0.20
0.40
0.40
0.36
0.0
0.0
0.0
Head
m(ft)
.18
.12
.23
.13
.16
.12
.08
.1
.1
.16
.11
.09
.11
.04
.003
.06
.09
.13
.07
.08
.12
.05
.07
.12
.43
.54
.32
loss,
H20/hr
(0.60)
(0.38)
(0.74)
(0.44)
(0.54)
(0.39)
(0.27)
(0.33)
(0- 34)
(0.51)
(0.36)
(0.29)
(0. 35)
(0. 13)
(0.01)
(0.19)
(0.29)
(0.44)
(0.23)
(0. 26)
(0. 39)
(0. 16)
(0.24)
(0. 38)
(1.42)
(1.76)
(1.05)
NOTES: Test no. 203022 followed the injection of polymer in the backwash.
Test no. 203282 water contained 11 wppm oil dlspersant.
Tests no. 30502, 30503, and 30614 were with the replacement bed.
38
-------
40
OJ
fc
•H
3
I
30
20
10
12.2 m/hr (5 gpm/sq ft)
30 wppn» alur«
65 JTU inlet
6.5° C (W F)
.13 m/hr (.M ft/hr) head loss
33 JTU inlet
11.5° C (53° P)
.1 m/hr ('.33 ft/hr) head loss
20
40
60
100
80
Time, minutes
Yorktown filter chemical treatment tests
Figure 14
120
140
-------
40
•C-
o
30
•H
1
1->
m
20
10
12.2 m/hr (5 gpm/sq ft)
65 JTU inlet
30 v.ppm alum
6.5° C (W F)
0.0 v/ppir polyrner
.13 m/hr (.4U ft/hr) hea
0 1-
2.3 wppi? polymer
.16 m/hr (.5!)- ft/hr) head loss
20
40
60
100
80
Time, minutes
Yorktown filter chemical treatment tests
Figure 15
120
140
-------
40
12.2 m/hr (5 gpm/sq ft)
6°C
30
I
20
10
55 JTU inlet
15 wppm alum
1.0 wppm polymer
.12 m/hr (.39 ft/hr) head loss
i(8 JTU inlet
7-5 wppm alum
1.1 wppm polymer
.08 m/hr (.27 ft/hr) head loss
...U-.J...U, . . « »
20
40
60 80 100
Time, minutes
Yorktown filter chemical treatment tests
Figure 16
120
140
-------
40
ro
fr
•H
•H
•e
£
+->
w
30
20
10
ft)
12.2 m/hr (5
80 JTU inlet
9 v/ppm alum
11.5° C (53° F)
1.9 wppm polymer
.11 m/hr (.36 ft/hr) head
45 ppir alum
5.5° C (^2° F)
3-2 wppm polymer
.23 m/hr (.74 ft/hr} head loss
20
40
60
100
80
Time, minutes
Yorktown filter chemical treatment tests
Figure 17
120
140
-------
40
30
12.2 m/hr (5 gpm/sq ft)
12° C (5V F)
30 wppm alum
20
10
33 JTU inlet
0.0 \vppm polyrer
.1 m/hr (-33 ft/hr) head loss
27 JTU inlet
Anionic polymer in previous
.16 m/hr (.51 ft/hr) head loss
l 1 «
20
40 60 80 100
Time, minutes
Yorktovn filter chemical treatment tests
120
140
Figure 18
-------
80
60
40
I
w 20
0
6.5 m/hr (2.67 gpm/sq ft)
100 JTU inlet
52 vppm alum
0.0 vppm polymer
22.5° C (73° F)
.13 m/hr (.kk ft/hr) head loss
Effluent JTU data
ad j us ted to 12.2 m/hr (5 "gpm/s q ft)
on equal throughput^,
basis.
\
20
40
60
100
80
Time, minutes
Yorktown filter chemical treatment tests
120
140
Figure 19
-------
40
•B
20
m
w 10
0
13 m/hr (5-33 gpm/sq ft)
23 JTU inlet
20 wppm alum
0.0 wppm polymer
29° C (84° F)
.32 m/hr (1.05 ft/hr) head loss
20
40
60
100
80
Time, minutes
Yorktown filter chemical treatment tests
120
140
Figure 20
-------
Additional effluent improvement could usually be demonstrated by
reducing the alum feed when a secondary flocculant was being used.
Figure 16 is an indication of the response to 1/2 and 1/4 the alum
dosage reported for Figure 15 with about the same and double the
polymer alum ratio. Even after recognizing that the effluent
should be better with a lower influent turbidity, it still appears
that the lower treatment levels produced a better effluent and a
lower head loss rate. One result of over-treatment can be seen in
Figure 17. The relatively low minimum turbidity, high head-loss
rate, and poor sustained performance all indicate that the filter
was filling up with unnecessary chemical floe. Figure 18 was
developed from a test using a high molecular weight, weakly anionic
polymer in the backwash in an attempt to "make the bed sticky".
Forty-five grams of solid were dissolved and injected into the
backwash line during a 90-second upflow, following a normal back-
wash. The increased head-loss rate and the improvement in the
effluent quality indicate some success with the technique. This
"flypaper type" entrapment is not the normal use of filter media
and there is considerable doubt that backwashing would be effective
after prolonged use of this technique.
The curves in Figure 19 represent data from a test run at 6.53
m/hr (2.67 gpm/sq ft) with the solid line plotted as run. The
dashed line represents the same effluent turbidity versus cumula-
tive throughput data, but as plotted as if it had been run at 12.2
m/hr (5.0 gpm/sq ft). The similarity between the dashed line of
Figure 19 and the high alum curve in Figure 17 suggests that
throughput rate is not the predominant independent variable. Head
loss, when similarly adjusted from 6.5 to 12.2 m/hr, was .25 m/hr
(0.82 ft/hr) for the dashed line. In other words, contaminant
removal from a given quantity of water is nearly independent of
the filtering rate between 6.53 and 12.2 m/hr.
Test data from the replacement bed are represented in Figure 20.
This curve most nearly corresponds to test 202291 (the upper curve
in Figure 18), but the 40-percent lower-inlet turbidity during the
later test makes the magnitude of the improvement associated with
the new bed questionable. One of the stated purposes of the
newly-designed bed, no increase in head-loss rate, was not achieved.
For a comparable 75-percent contaminant removal, the new bed had
three times the head-loss rate of the damaged bed and twice that
demonstrated in the pilot unit.
46
-------
SECTION VIII
PROJECT CRITIQUE
AESTHETICS
The Yorktown mixed media filter has been a valuable addition to
the refinery waste water treatment system. It has removed more
contaminants from the water than expected and has exhibited con-
siderable public relations appeal. Visitors have been impressed
with the visible differences between the influent and effluent
water ana with the contaminant removal evidenced by waste water
flowing from the troughs into the gullet during a backwash. The
only disappointment associated with the project has been the
effluent turbidity level. An average of less than three had been
expected using only primary coagulants. Longer term operating
experience demonstrated that the brackish water biocolloid system
requires a more sophisticated chemical treatment than simple
primary coagulants, and further definition to this will be given
during the 1973-1974 winter, the most demanding season. Infor-
mation on hydraulic and contaminant loads that was collected to
provide a design basis for the unit was accurate, but quite in-
adequate. No consideration was given to possible differences
between summer and winter lagoon operation that might change the
filter influent conditions. Before startup of the plant filter, a
number of decisions were made that changed the properties of the
water to be filtered. For example, a change in company policy to
prevent partial deballasting of product tankers at sea caused an
unanticipated increase in throughput requirement, contaminant
loading and salinity. A second decision impacting on filter
operations involved backwashing the filter to the second cell of
the four cell lagoon. The objective of this operation was to
eliminate the substantial cost of a backwash surge pond and
desludging pit. However, this operation resulted in recycling
solids to the filter thereby increasing the solids burden, and
seems to have contributed to more pronounced spring and fall
turnovers which also increased the loading on the filter. Sub-
sequent to this experience, a backwash sludge pit has been added
to the system to effect removal of backwash solids from the
system.
Several years experience has shown that the Yorktown refinery
lagoon effluent lends itself to filtration without causing media
contamination problems. Traces of oil do not accumulate in the
bed and no mud balls were formed at the chemical dosages used
during this project.
47
-------
PROCESS LIMITATIONS
Pump sizes limit the unit throughput and treatment dosage to about
397 cu m/hr (1750 gpm) and 25 wppm alum. No permanent secondary
flocculant pump has yet been installed. Given the above con-
straints, the minimum effluent turbidity is limited to some un-
known extent by the untreated lagoon water left in the unit after
a backwash. Over compensation for this effect might cause an alum
over-treatment and shortened run lengths. Run lengths are deter-
mined by the head loss rate, which is a function of incident
turbidity and chemical treatment. The backwash switch has usually
been set to terminate a run after .6 to 1.2 m (2 to 4 ft) of
water total head loss. Run length then automatically shortens if
the influent turbidity rises.
FILTERED WATER FOR BACKWASH
Consideration has been given to adding filtered water storage and
a pump to allow backwashing with cleaner water. Present unit
capacity makes this nearly impossible since only 160 minutes of
filtering time are presently available in a 3-hour cycle. The
influent pump is down for 9 minutes and 11 minutes are required to
fill the unit after a backwash. Therefore, the influent rate must
be 386 cu m/hr (1,700 gpm) to average 343 cu m/hr (1,510 gpm) for
an extended period such as the last project quarter. The summer-
time backwash rate is about 1363 cu m/hr (6,000 gpm) and with a
minimum acceptable duration of five minutes the reprocessing load
would be 113.6 cu m (30,000 gallons). A rate 429 cu m/hr (1,890
gpm) would be required to maintain the same 343 cu m/hr (1,510
gpm) average. This is above the feed pump capacity and in an
untested region as far as filter efficiency is concerned.
48
-------
SECTION IX
THEORY AND PRACTICAL APPLICATION OF CHEMICAL PRETREATMENT
FILTRATION REQUIRES DESTABILIZATION OF COLLOIDS
Refinery waste water suspensions that are the target of filtration
applications usually are essentially all colloidal or a combination
of colloidal and very slightly flocculated suspensions. Work at
other Amoco refineries indicates that the optimal application of
filtration to these waters relates primarily to recognizing and
responding to the required water chemistry for destabilizing the
colloidal material and only secondarily to the design of the
filter.
Mechanisms for retention of solids within the pores of filter
media may be separated into two principal processes: a transport
step and an attachment step. The transport step involves movement
of the dispersed phase material to the vicinity of the filter
media surface, and the attachment step involves attachment of the
particles on the media surface. The transport mechanisms may
involve diffusion, interception, sedimentation and hydrodynamic
actions. The attachment mechanisms may involve van der Waal
forces, electrical double layer interaction, mutual adsorption or
hydrogen bonding (1). Experimental comparisons of the filtration
of colloidal, ferric hydroxide suspensions with the filtration of
well flocculated suspensions of ferric hydroxide, both performed
under identical conditions of filtration, demonstrate a much
higher filtration efficiency for the floe. Studies (2) have shown
that removal of colloidal suspensions by filtration appears to be
possible only when the colloidal particles to be filtered carry an
electro-kinetic charge opposite to the charge of the filter media
used. The removal of the colloid is by electro-kinetic sorption
of the colloidal particles on the surface of the filter media.
This kind of "filtration" phenomena seems to have limited appli-
cation to the filtration of industrial waste water because of
complications encountered in application, the complexity of which
are beyond the scope of this report.
For example, however, at a high ratio of the amount of colloidal
particles to the amount of filter media, the filter media soon
accepts the surface charge of the colloidal particles as a result
of sorption of them and the charge of the colloidal particles does
not change appreciably. As a result sorption ceases. If, in con-
trast, the amount of colloidal material is small in proportion to
the amount of filter material, the colloid is removed rapidly
initially but ceases when the media changes charge due to sorption.
49
-------
A dilution or filtration effect (sorption/desorption of "soluble"
surface ionic species) may cause the charge of the remaining
colloids in suspension to change and sorption is again initiated
until sorption changes the charge on the media once again. Thus,
by this mechanism erratic performance may ensue.
COLLOIDAL DESTABILIZATION
In most cases, the application of filtration to industrial waste
waters will not be optimized unless the colloidal material in
suspension is destabilized by essentially neutralizing the surface
charge on the colloids. The surface charges (measured as zeta
potential) of suspended matter in refinery process water effluents
have been found to range from -35 to -55 mV. The neutralization
of the charge using conventional primary coagulants is simple, but
must recognize the condition of the water. Overdosing with alum,
for example, can be a most deleterious response to make to a
direct filtration system that is not performing satisfactorily.
Destabilization of colloidal suspensions using salts of iron and
aluminum as primary coagulants must recognize the properties of
these primary coagulants. In Figure 21 the zeta potential of
colloidal iron hydroxide solutions is plotted as -a function of pH.
The zeta potential decreases in positive charge as the pH in-
creases until the isoelectric point is reached at a pH of 8.3 at
which the charge reverses. In the vicinity of the isoelectric
point, the charge may vary as indicated. Alum has similar zeta
potential-pH relationships as shown in Figure 22. The zeta
potential may be negative or positive over the pH range of 7.0 to
7.8.
The amphoteric property of alum shown in Figure 23 is an addi-
tional factor that must be considered. Other properties of water
bear importantly on alum solubility in alkaline solutions that may
sharply limit the operable pH range.
Stumm and O'Melia (3) describe the equilibrium composition of
solutions in contact with precipitated primary coagulants in the
interesting manner shown in Figures 24 and 25. These diagrams are
calculated using equilibrium constants for solubility and hydrolysis
equilibria. The shaded areas A and B in each figure are approx-
imate operation regions for air flotation and clarifiers (region
A) and direct filtration (region B). Both regions are assumed to
cover a pH range of 6.0 to 8.5. The coagulant dosage ranges from
33 to 200 mg/1 in region A and 3.3 to 20 mg/1 in region B. These
figures are useful in the interpretation of some of our filtration
results.
50
-------
+UO
+30
+20
+10
Data Scatter
Q>
I
a
-p
a
-10
-20
J L
10
Isoelectric Point pH = 8.3
-30
-40
Zeta potential of colloidal iron hydroxide
solution plotted as a function of pH
Figure 21
51
-------
+10
03
1)
I
01
IS!
-10
6
pH
10
Zeta potential - j»H plot for aluminum hydroxide
Figure 22
-------
•§
r-H
O
CO
Ul
Soft H20
Hard E20
pH
Solubility curve for aluminum
Figure 23
-------
-2 —
-k —
VJl
-6
-8 —
-10 —
-12
\ \
A1(OH)3 (s)
\ \ \
A - Operating Region for Air
Flotation and Clarif iers
B - Operating Region for
Direct Filtration
C - A1(OH)J
D -
• • v
\\ \
\\\ \
Equilibrium conpositlons of solutions In contact with
Figure 2k
-------
-2
_]|.
\
CO
-12
A - Operating Region for Air Flotation and Clarifiers
B - Operating Region for Direct Filtration
C - F
F - Fe~
\\VA\\
Equilibrium compositions of solutions in contact with
Figure 25
-------
With reference to Figure 25, the isoelectric point for ferric
hydroxide coincides with the region of minimum solubility, and the
operating regions for water treating (destabilization) always
yield a hydrolyzed primary coagulant with a desirable, positive,
zeta potential.
In many refinery situations, however, it is difficult to use this
desirable condition because the presence of sulfides and strongly
reducing conditions cause the reduction of ferric to ferrous iron
and the formation of mixed iron sulfides with no coagulation
powers. In fact, in some refinery waters the use of iron coagu-
lants at modest dosages may contribute to stabilizing solids
rather than destabilizing them.
While alum has no redox or sulfide chemistry comparable to iron,
its amphoterism and solubility pose definite limitations on alum
usage. With reference to Figure 24, a substantial portion of
operating region B lies in the area where alum is soluble and the
predominant equilibrium species is negative, A1(OH)4. In the
more acidic part of region B, however, the concentration of
equilibrium ionic species is very much lower and much less nega-
tive. Considering these data, it is not unexpected that investi-
gators consistently report optimal coagulation/flocculation results
with alum at a pH of 5-6.
With inspection of Figure 24, one may question why alum is ef-
fective at all for neutralizing negatively charged colloids in the
indicated operating regions. One approach to explaining observed
performance requires understanding that the data in Figure 24 are
equilibrium data; but before equilibrium is reached, substantially
different conditions exist.
Alum very readily hydrolyzes to form polymers in a complex manner
not well defined. The hydrolytic pathway and reaction rates are
affected by pH, temperature, other ions, etc. One hypothesized
route which includes different aluminum hydrolysis products known
to exist is outlined in Figure 26. When alum is added to water in
amounts which exceed the solubility limits, sequential kinetic
reactions occur until the ultimate precipitate is formed and the
ionic species appropriate to the pH equilibrate with the pre-
cipitate. The hydrolytic reactions are not instantaneous, and as
they proceed, positively charged hydroxo polymers are formed which
are available for colloid adsorption. The hydrolyzed species have
enhanced adsorption capabilities possibly due to larger size, less
hydration, and the presence of coordinated hydroxide groups (3).
In solutions more alkaline than the isoelectric point, the posi-
tively charged polymers are transient and, at equilibrium, anionic
polymers prevail. In modestly alkaline solutions, the transient,
56
-------
OH
A1(H20)5(OH)
A1(OH)3(H20)3 (s)
OH
A12(OH)5
+ o
Conversion of AlfHgO^ to
Figure 26
57
-------
positively charged polymers appear to contribute to destabilization
of colloids. On the other hand, in solutions more acidic than the
isoelectric point, the positively charged polymers prevail at
equilibrium and destabilization of colloids is achieved at sig-
nificantly lower coagulant treatment levels.
While coagulation/flocculation of refinery waste water can fre-
quently be achieved over a wide pH range using primary coagulants
supplemented with polyelectrolytes, typically there is a very
definite optimal pH that may simplify, or even be required by,
direct filtration. Operating at the optimal pH yields benefits of
maximum removal of discontinuous phase material with minimal
coagulant requirement and sludge generation.
In Figure 21 (page 51), the zeta potential-pH relationship is
shown for a primary coagulant. The use of this primary coagulant
to treat a refinery process water containing negatively charged
colloids is shown in Figure 27 at three pH levels.
At a pH of 6, a minimal dosage of coagulant rapidly changes the
zeta potential of solids in the system to zero thereby destabil-
izing the suspension adequately for direct filtration.
Frequently, refinery waste waters only require 5 to 10 mg/1 of
primary coagulant in this pH range to achieve destabilization for
filtration. When this level of chemical treatment is observed in
jar tests, the coagulant dosage may be much below that required
for generation of floe; however, excellent filtration results are
achieved. The destabilized particles penetrate into the pores of
the filter media, the media aids the flocculation process, and the
particles are effectively trapped with high solids loadings. Too
much primary coagulant can be added at a pH of 6 and some suspended
matter restabilized. To much coagulant puts an unnecessarily
heavy solids burden on the filter and the result is rapidly in-
creasing pressure drops, early breakthrough and excess backwash
sludge to handle.
At a pH of 8, substantially more primary coagulant may be needed
to neutralize the zeta potential. The dosage may be 30 mg/1 or
more with the generation of a substantial floe size instead of the
almost invisible pin-point floe of suspensions successfully fil-
tered at a pH of 6. The more voluminous floe does not penetrate
into the filter media as well as does the pin-point floe and lower
solids loading, higher pressure drops, and more chemical usage,
backwash and sludge volumes result.
At a pH of 9, the zeta potential of the system may never be neu-
tralized and, of course, is not optimal as such for direct fil-
tration. With iron salts, chemical coagulation/flocculation at a
58
-------
VO
Increasing Iron Salt Addition
Change in zeta potential of refinery process
water with increasing coagulant
dosage at various pH levels
Figure 27
-------
pH of 9 or more can yield superb results with some waters, but a
different clarification mechanism is involved. Comparatively
large dosages of coagulant are used which generate large, voluminous
floe particles that trap and enmesh the suspended matter in the
water being treated and retain the solids in the sludge blanket.
This approach is not optimal for direct filtration but requires
pretreatment in a clarifier to handle the voluminous sludge volume
generated. The clarified water is then polished by filtration
using polyelectrolytes to destabilize the residual suspended
matter.
Figure 28 illustrates the effect of treating a refinery effluent
water with fixed levels of overdosage, underdosage, and optimal
dosage of primary coagulant at various pH levels. For this ex-
ample, the optimal dosage is defined as the coagulant required to
bring the zeta potential of the system into the target zeta po-
tential range of + 5 inV necessary for destabilization of the
suspended matter preparatory to direct filtration. To emphasize
the importance of pH for these systems of over, under, and optimal
treating with coagulant, there is a pH range where the system is
destabilized and receptive to clarification by direct filtration.
For the optimal dosage in curve A, the system is destabilized over
the widest pH range of 5.5 to 8.4. Underdosing (curve C) provides
a destabilized system for filtration only at a pH of about 6.5 or
less. Overdosing (curve B) results in the narrowest pH range for
effective destabilization of about 0.6 pH units, i.e., 7.65 to
8.25.
Industrial effluent waters may vary rapidly in pH over a short
time span which makes it desirable to determine the optimal
coagulant-pH balance. Frequently, poor or no pH control is
available. In these cases, the destabilization of suspended
matter frequently can be "desensitized" to pH somewhat by using
coagulant/polyelectrolyte combinations. The preferred operating
procedure that frequently makes the filtration system essentially
free from upset is a two chemical system: a primary coagulant
supplemented by a polyelectrolyte. Two cases seem appropriate:
1) For the operating range of 6-9 pH, an underdosage of primary
coagulant is used and is supplemented by addition of a cationic
polyelectrolyte. This combination of chemicals broadens the
effective pH range of destabilization of curve C in Figure 25,
and, additionally, the chemical properties of the polyelectrolyte
binds and makes the destabilized particles more adherent to the
filter media. 2) For the operating range of 5-7 pH, the poly-
electrolyte used to supplement the primary coagulant may be a non-
ionic or very weakly anionic. The non-ionic or very weakly anionic
polyelectrolyte is useful in systems that tend to modest over-
dosage situations with primary coagulant. Massive overdosing with
60
-------
+1*0
+30
+20
13 +10
I °
o
Ou
5 -10
-20
-30
B
\
A
B
C
D
Optimal Iron Salt Dosage
Overdose with Iron
Under Dosage with Iron
Target Zeta Potential Range
PH
Variation of final zeta potential with pH when treating
refinery effluent having negative zeta potential
with three levels of coagulant dosage
Figure 28
-------
primary coagulant can be compensated for somewhat by non-ionic or
very weakly anionic polyelectrolytes but rapidly blind the filter.
These chemical treatments are also useful when high hydraulic
loadings are applied to the filter or rapid rate increases occur,
both of which may tend to redisperse trapped solids and cause
premature breakthrough.
Strong, or very strong, anionic polymers have always performed
decidedly poorest in our tests comparing cationic, non-ionic,
very weakly anionic, weakly anionic, fairly strong anionic,
strongly anionic, and very strongly anionic polyelectrolytes. In
one instance, we found treatment with only a cationic polyelec-
trolyte at a less than 1 ppm to outperform all other combinations
of chemical treatment for destabilization of suspended matter
preparatory to filtration.
Another distinction that may be important is the nature of the
colloids to be destabilized: inert materials including clays,
silica, coke fines, oil particles, etc., or biocolloids consisting
chiefly of bacteria. Amoco's work on colloid destabilization at
other refineries has been largely on inerts, or a mixture of
colloids predominately inerts with some biocolloids. The work of
McLellon et al. (4) and Stumm and Morgan (5) indicates that where
principally biocolloids are concerned, stability is less dependent
on repulsive electrostatic interactions and relates primarily to
the interaction of the hydrophllic surface of the biocolloid with
the aqueous solvent. The source of colloid surface charge is
through acid-base interactions of ionogenic functional groups.
The authors cited (A, 5) conclude that aggregation of biocolloids
is by chemical interaction of the hydroxy metallic polymers with
the ionogenic groups of the colloid followed by chemical bridging.
If the destabilization mechanism for biocolloids is such a special
case, the chemical pretreatment may be optimized by emphasizing
the use of organic polymers known to form chemical bridges. The
optimal chemical pretreatment would then be expected to be a
multichemical system:
1. Two Chemical Systems
a) Optimize pH.
b) Use a primary coagulant plus an organic cationic poly-
electrolyte, or
c) Use a primary coagulant plus an organic non-ionic or
weakly anionic polyelectrolyte.
62
-------
2. Three Chemical Systems
a) Optimize pH.
b) Use a primary coagulant/organic cationic polyelectrolyte/
weakly anionic polyelectrolyte treatment sequence.
63
-------
SECTION X
OPERATING AND MAINTENANCE COSTS
Routine unit operation has required only about 10 percent of an
operator's time on an around-the-clock basis. Abnormal conditions
such as high contaminant concentrations with changes in chemical
treatability of the water have required non-routine laboratory
analyses and technical assistance. The proportionate cost of
these expenses has declined with increasing operator experience
and with increased emphasis on control of the upstream facilities.
Because of the remote location of the unit, power costs are the
only utility expense. Instrument air is furnished by an onsite
compressor and all the heat tracing is electrical. The costs
shown below are based on $0.0091 per KWH and an average cycle
time of three hours.
$/Day $/3785 cu m
(10& gal)
Power:
Filtering 62 HP, 23 hr 9.69
Backwashing 215 HP, 1 hr 1.46
Light and Heat 0.5KW 0.11
11.26 5.22
Chemicals:
Alum $39.63 cu m (15c/gal)
Polymer $1.00/Kg (45c/lb)
23.12 10.75
Labor:
Operator 10%, 24 hr/da,
$5.20/hr
Supervision & Technical 75%
Overhead 62%
35.36 16.50
Maintenance:
Estimate 1.1% of cost 3.80 1.76
73.54 34.23
64
-------
Chemical cost data have been estimated on the basis of test data
because actual long-term experience with polymeric flocculant is
not yet available. An average of 341 cu m/hr (1500 gpm) through-
put has been assumed with a 20 wppm alum dosage in the summer and
40 wppm alum plus 2 wppm polymeric flocculant during the winter.
Maintenance costs for the first year resulted largely from repairs
that were necessary after a shutdown caused by freezing. Addi-
tional heat tracing and new bonded valve liners were installed at
that time. These should not be recurring expenses. The continuing
maintenance expenses consist of instrument maintenance, weather-
proofing repair, and pump repair.
Amoco's contribution to the filter bed replacement was not con-
sidered normal maintenance costs and was included with the unit
construction costs during the last two project quarters. The ex-
penditure pattern for this project is shown in Table 4.
65
-------
Table 4. YORKTOWN MIXED MEDIA FILTER EXPENDITURE PATTERN
Engineering and construction
Quarter
ending
11-30-70
2-28-71
5-31-71
8-31-71
11-30-71
2-29-72
5-31-72
8-31-72
11-30-72
Operation
Quarter
ending
8-31-71
11-30-71
2-29-72
5-31-72
8-31-72
11-30-72
Engineering
1,304
2,128
1,038
576
$5,046
Maintenance and
operation
8,014
5,931
6,670
5,635
5,219
5,395
$36,864
Material and
equipment rental
3,515
17,019
58,727
4,724
1,391
47
$85,423
Post construction
studies & reports
22,874
11,413
9,793
9,226
5,605
5,411
$64,322
Labor
1,206
22,827
10,069
922
707
706
$36,437
Totals
30,888
17,344
16,463
14,861
10,824
10,806
$101,186
Total
4,819
20,353
82,592
15,369
2,313
47
707
706
$126,906
66
-------
SECTION XI
REFERENCES
1. Ives, K.J., "Depth Filtration of Liquids." Filtration and
Separation, pp. 700-703 (Nov./Dec., 1970).
2. Heertjes, P.M., and C.F. Lark, "The Functioning of Deep Bed
Filters Part I: The Filtration of Colloidal Solutions." Trans.
Instn. Chem. Engrs.. 45:T139 (1967).
3. Stumm, W., and C.R. O'Melia, "Stoichiometry of Coagulation,"
Jour. AWWA. 60:514 (1968).
4. McLellon, W.M., and T.M. Keinath, and C. Chao, "Coagulation
of Colloidal and Solution Phase Impurities in Trickling Filter
Effluent." Jour. WPCF. 60(1):77-91 (1972).
5. Stumm, W., and J.J. Morgan, "Chemical Aspects of Coagulation,"
Jour. AWWA. 54:971 (1962).
6. Riddick, T.M., Control of Colloid Stability through Zeta Potential.
Volume I. Livingston Publishing Company, Wynnewood, Penna.,
pp. 14-248 (1968).
7. Overbeek, J.T.G., and J. Lijklema, "Electric Potentials in
Colloidal Systems in Electrophoresis," In: Electrophoresis
(Theory, Methods, Applications), Volume 1 (M. Bier, ed.).
Academic Press, New York, pp. 1-33 (1959).
67
-------
SECTION XII
APPENDIX A
ANALYTICAL PROCEDURES USED IN PROGRAM
Parameter
Oil
Phenolics (Amino-antipyrine)
Suspended Solids, Total
Suspended Solids, Volatile
Ammonia
Turbidity
Sulfides
Dissolved Oxygen
BOD5
COD
Procedure Reference
AOM #68
API 716-53C1)
Standard Methods 224C
Standard Methods 224D
Standard Methods 132A & B
Turbidimeter-HACH Model 1860A
API Method 713-53W
Standard Methods 218B
Standard Methods 219 4ii
Standard Methods 220
(1) API Method for Sampling and Analysis of Refinery Wastes
68
-------
APPENDIX B
CHRONOLOGICAL OBSERVATIONS PERTINENT TO THE
OPERATION AND STUDY OF THE YORKTOWN MIXED MEDIA FILTER
June 29, 1971
The unit was started up and operated successfully at 6.6 m/hr (2.7
gpm/sq ft) and about .3 m/hr (1 ft/hr) head loss. The inlet
turbidity of 13 was 86 percent removed.
August 9, 1971
The unit was partially drained and the filter bed was found to
have expanded and was preventing the surface wash arms from rotating.
It was later learned that the backwash rate was inadequate and the
bed expansion could logically have resulted from a material build-
up in the lower layers of media. The surface wash system was
deactivated and subsequent backwashes had no surface wash.
September 9, 1971
The surface wash arms were raised about 5 cm (2 in.) to clear the
bed and put into routine service.
October 5. 1971
A flooded API separator caused free oil to be carried through the
filter during both filtering and backwashing cycles.
October 12. 1971
The bed was drained to obtain media samples. Some of the coal
from the southwest corner of the bed had been displaced, leaving
an 45.7 cm (18 in.) deep hole in the corner and a mound of coal
over the west surface wash arm pivot. About one-third of the bed
surface was involved. No sand or ilmenite were visible.
November 29. 1971
New recommendations were received for the backwash rate; the new
rate being about 50 percent greater than the originally specified.
December 9, 1971
A temporary baffle was placed in front of the inlet nozzle in an
attempt to solve the persistent problem of having coal shifted
from the southwest corner to a mound over the west surface wash
arm.
69
-------
December 30. 1971
Adjustments were made to the backwash troughs to equalize flow
distribution between the troughs. Two of the troughs had been
carrying about 75 percent of the flow.
January 24. 1972
The temporary inlet baffle was replaced by a permanent one which
proved to be very effective in stopping the coal displacement.
February 18. 1972
A dedication ceremony was held with enforcement officials from the
various agencies and representatives from the news media in
attendance.
February 22. 1972
An adjustment was made in the backwash timer to stop the surface
wash before the backwash flow started.
April 7, 1972
Several holes were dug in the bed to inspect the media and confirm
the media losses.
May 18, 1972
The media loss seemed to have stopped and the surface wash arms
were lowered about 20.3 cm (8 in.) so that they were 5 cm (2 in.)
above the bed surface as originally specified.
May 31, 1972
The original grant period expired.
June 30. 1972
The bed was probed and sampled by a manufacturer's representative.
July 11. 1972
A 3-month grant extension was requested in order to include data
from the bed repair in the final report.
August 21, 1972
The unit was taken out of service to replace the bed.
70
-------
September 6. 1972
A new, redesigned, bed had been installed and the unit was put
into service.
November 22. 1972
A 6-month grant extension was approved.
April 15. 1973
The unit was drained for inspection of the bed and installation of
a permanent secondary flocculant nozzle. The surface of the media
was about 16.5 cm (6.5 in.) below the surface wash arms. Some
fine garnet was observed on top of the surface wash support beam.
May 18, 1973
Work on the final report was initiated in order to make project
results to date generally available.
Approval was received for a supplemental program aimed at optim-
izing chemical preconditioning of filter feed during winter oper-
ating conditions to be carried out during the period December,
1973 - March, 1974. A supplement to the project report will be
issued.
71
-------
APPENDIX C
YORKTOWN WATER FILTER ROUTINE OPERATING AND INSPECTION DATA
June 29, 1971 to September 1, 1971
Influent Effluent
Min. Med. Max. Avg. Mln. Med. Max. Avg.
Water rate, cu m/hr 204 273 397 255
Alum rate, wppm 20.0 20.0 20.0 20.0
Water temperature, °C 26.5 27 29.5 27
17 3.2
14 1.9
1 0.3
2 0.6
7.3 6.9
20 4.8
Turbidity, JTU
Oil, mg/1
Phenolics, mg/1
Sulfide, mg/1
pH
Suspended solids, mg/1
Volatile sus. solids, mg/1
Ammonia, mg/1
Dissolved Oxygen, mg/1
BOD5, mg/1
8
0
0.1
0
6.9
2
0.6
0
19
5
0.7
0
7.0
20
2.9
16
28
54
6
4
7.5
45
8
182
20
7
1.1
0.7
7.1
18
3.6
35
0.9
0
0
0
6.8
2
0.5
0
3.4
0
0.2
0
6.9
5
1.8
18
6 2.5
35 16
Note: The dissolved oxygen, pH, and BOD5 data from this period were based
on only five sets of samples.
72
-------
APPENDIX D
YORKTOWN WATER FILTER ROUTINE OPERATING AND INSPECTION DATA
September 1, 1971 to December 1, 1971
Influent Effluent
Min. Med. Max. Avg. Min. Med. Max. Ave.
Water rate, cu m/hr 250 341 386 330
Alum rate, wppm 20 24.4
Water temperature, °C 10 18 27 20.5
1.2 3.7 63 6.6
4 84 8.3
0.1 0.3 2.0 0.45
1 34 2.3
6.5 6.8 7.5 6.9
6 30 13.6
4 27 10.1
53 87 53
0.4 1.6 3.8 1.9
7 21 11
Turbidity, JTU
Oil, mg/1
Phenolics, mg/1
Sulfide, mg/1
pH
Suspended solids, mg/1
Volatile sus. solids, mg/1
Ammonia, mg/1
Dissolved oxygen, mg/1
BOD5, mg/1
8
0
0.2
0
6.7
8
4
29
0
3
25
7
0.7
1
7.0
38
20
54
0.6
15
135
105
2.4
38
8.1
79
72
90
6
43
32
16
0.7
3.1
7.0
43
35
54
1.4
19
1.
0
0.
0
6
2
0
28
0
2
73
-------
APPENDIX E
YORKTOWN WATER FILTER ROUTINE OPERATING AND INSPECTION DATA
December 1, 1971 to March 1, 1972
Influent
Water rate, cu m/hr
Alum rate, wppm
Water temperature, °C
Turbidity,
Oil, mg/1
Phenolics ,
JTU
mg/1
Sulfide, mg/1
PH
Susp ended
solids, mg/1
Volatile sus. solids, mg/1
Ammonia, mg/1
Dissolved
BOD5, mg/1
oxygen, mg/1
Min.
238
20
5.5
23
2
0.1
0
6.0
33
25
24
0
5
Med.
341
10
48
13
0.3
1
7.4
70
54
54
2.1
10
Max.
397
16
260
87
1.7
2
8.8
104
78
119
10.3
61
Avg.
332
29.4
10.5
65
16
0.5
0.9
7.4
68
50
55
2.6
18
Min.
1
1
0.1
0
6.0
15
8
25
0
4
Effluent
Med.
16
7
0.2
1
7.1
36
25
53
5.2
7
Max.
100
67
1.4
1
8.2
96
66
126
10. 0
37
Avg.
24
10
0.
0.
7.
39
29
55
4.
11
4
6
2
2
74
-------
APPENDIX F
YORKTOWN WATER FILTER ROUTINE OPERATING AND INSPECTION DATA
March 1, 1972 to June 1, 1972
Influent
Water rate, cu m/hr
Alum rate, wppm
Water temperature, °C
Turbidity, JTU
Oil, mg/1
Phenolics, mg/1
Min.
182
20
10
14
2
0.2
Med.
295
14.5
58
13
0.5
Max.
386
23
220
68
18.7
Effluent
Avg. Min. Med. Max. Avg.
296
31.6
15.5
68 1.5 26 200 36
17 0 7 44 13
0.9 0 0.3 16.7 0.8
Sulfide, mg/1 01 10 1.0 0 0 9 0.5
pH 6.9 7.3 8.9 7.5 6.6 7.1 8.8 7.2
Suspended solids, mg/1 18 56 145 69 11 31 110 38
Volatile sus. solids, mg/1 6 42 94 50 1 28 88 30
Ammonia, mg/1 25 53 79 52 25 52 74 50
Dissolved oxygen, mg/1 0 0 2.5 0.7 0 1.7 5.9 2.0
BOD5, mg/1 1 23 171 34 1 13 171 30
75
-------
APPENDIX G
YORKTOWN WATER FILTER ROUTINE OPERATING AND INSPECTION DATA
September 6, 1972 to December 1, 1972
Influent Effluent
Min. Med. Max. Avg. Min. Med. Max.Ave.
Water rate, cu m/hr 204 341 386 343
Alum rate, wppm 20 21.2
Water temperature, °C 10.5 18.5 26.5 18.5
Turbidity, JTU 21 34 50 33 1.6 9 28 11
Oil, mg/1 0 8 54 9 0 4 42 6
Phenolics, mg/1 0.2 0.4 2.0 0.6 0.2 0.3 1.0 0.4
Sulflde, mg/1 0 0 80 0.9 0 0 60 0.4
pH 7.0 7.3 7.7 7.3 6.8 7.1 7.8 7.1
Suspended solids, mg/1 57 90 141 90 17 36 83 42
Volatile sus. solids, mg/1
Ammonia, mg/1
Dissolved oxygen, mg/1
BOD5, mg/1 9 18 27 18 7 13 16 13
76
-------
SECTION XIII
ADDENDUM
Review of initial operational data indicated the chief causes of
periodically poor filter performance were related to seasonal
changes in aerated lagoon performance, colder water temperatures,
and other unidentified changes in water quality. Facilities
available were not adequate to respond to changed water quality as
required to properly chemically condition the water in preparation
for filtration. Seasonal deterioration in filter performance
occurs during winter; therefore, a winter program was developed
emphasizing study of the water chemistry needed to prepare the
aerated lagoon effluent for direct filtration, and the plant
facilities were operated to quantitatively demonstrate the need
for proper water pretreatment. The general program developed was
as follows:
1. Determine the effect of primary coagulant over the pH range
6-8.5.
2. Determine the effect of primary coagulant and cationic polymer
over the pH range 6-8.5.
3. Determine the effect of primary coagulant and nonionic and
weakly anionic polymers over the pH range of 6-8.5.
4. After determination of the most effective chemical pretreatment
per 1, 2, and 3, determine the effect of adding various
dosages of powdered carbon at the optimized chemical pre-
treatment conditions.
5. Determine if the use of powdered carbon is a viable means to
"desensitize" the chemical pretreatment such that the charged
particles do not have to be so carefully neutralized.
ELECTRICAL CHARGE ON SUSPENDED MATTER IN YORKTOWN AERATED LAGOON
EFFLUENT
As the first step in a series of backgrounding studies the zeta
potential (ZP) of the suspended matter in the aerated lagoon ef-
fluent was determined according to the equipment and method of
Riddick (6). A probability plot of these data in Figure 29 indi-
cates the expected average ZP to be about 11.4 mV with a range of
8.5 to 15. The concept of zeta potential is illustrated in Figure
30. In initiating the study using electrical techniques an imme-
diate problem was encountered involving the brackish nature of the
effluent. High salt content (8,000-10,000 ppm) resulted in very
77
-------
.512 5 10 20 30 iiO 50 60 TO 80 9095 9099 99«5 99-o
Probability of Zeta Potential Being Bqual to, or LOTS Hum, Indicated Value
Zeta potential distribution of biocolloids In aerated lajgoon effluent
Figure 29
-------
-j-
A = Rigid Layer Migrating with Particle
as Single Kinetic Unit
B = Diffuse Layer of Counterions
C « Electric Potential Surrounding
Particle
D « Zeta Potential
Concept of the double layer and zeta potential (after Riddick, 6)
Figure 30
-------
high specific conductivities, requiring use of ancillary automatic
sampling equipment with the zeta meter to facilitate measurements
of ZP. The brackish nature of the water not only affected the
measurement of the ZP but also the interpretation of the measured
ZP and the water chemistry for pretreatment. These latter subjects
will be discussed in later parts of this section.
SCREENING OF PRIMARY COAGULANTS AND CANDIDATE CATIONIC, NONIONIC,
AND WEAKLY ANIONIC POLYELECTROLYTES
Broad water chemistry experience at other Amoco refineries pro-
vided background for candidate treatment chemicals and combin-
ations, and screening tests were conducted at the Yorktown refinery
using conventional jar test techniques. The results of these
initial jar tests were not consistent with typical experience at
other Amoco refineries. To focus more accurately on the impact of
salinity and temperature, backgrounding tests were conducted at
another Amoco refinery having a fresh water aerated lagoon. The
effects of temperature and salinity were investigated by adjusting
1) the temperature of fresh sample of aerated lagoon effluent to
the desired test temperature and 2) the salinity to various specific
conductivities by adding solid NaCl. The jar tests results re-
flected operating observations at Yorktown; i.e., 1) at warm
summertime water temperatures, 26.5+° C (80+° F), salinity effects
were minor, but 2) the combination of colder and highly saline
water had a more observable effect on the chemical pretreatment
requirement for colloidal destabilization. Jar tests indicated
that in warm brackish water a two-chemical system at low concen-
trations was satisfactory whereas for cold brackish water a three-
chemical system at higher concentrations was required (see page 62).
Polyelectrolytes screen varied widely in effectiveness, and
optimizing chemical pretreatment involved testing a series of
commercial polyelectrolytes at various combinations until an
optimal combination was found. For defining chemical pretreatment
requirements for colloidal destabilization and direct filtration,
jar tests were useful only for gross screening tests. This was
because the results of jar tests were more applicable if "sweep
floe" coagulation procedures were used rather than charge neutral-
ization mechanisms. (In a later section, the replacement of jar
test techniques by zeta potential measurements to accurately
quantify the chemical treatment required will be described.)
WATER CHEMISTRY
This study pointed out the care with which destabilization chemistry
must be approached, identified, and semantics or word usage de-
fined. These points were illustrated by the fact that chemical
pretreatment guidelines for direct filtration from fresh water ex-
perience had to be further developed for brackish water situations.
80
-------
Various colloid destabilization mechanisms were illustrated by
results of this study. Their identification and description can
be discussed in relation to the condition-response schematic for
chemical treatment of waterborne colloids outlined in Figure 31.
Destabilization of the waterbourne suspended solids may involve
four mechanisms: 1) colloid entrapment or removal via the sweep
floe mechanism, 2) reduction in surface charge by double layer
repression, 3) bridging by polymers, and 4) charge neutralization
by adsorption.
Colloid entrapment involves chemical treatment with comparatively
massive amounts of primary coagulants. The amount of coagulant
used is typically so great in relation to the amount of colloidal
matter that the nature of the colloidal material is not relevant.
The amount of primary coagulant used may be 5 to 40 times as much
as is used for charge neutralization by adsorption. The rate at
which the primary coagulants form hydrous metal oxide polymers
(Figure 32) is relatively slow and depends upon water temperature
and pH, and coupled with the concentrations used initially, exposes
all negatively charged colloidal material to charge neutralization
by the transient cationic species. The polymer matrix is 3-di-
mensional and voluminous, providing for entrapment of solids as
shown by Figure 33. As the polymer contracts, freeing solvent
water molecules, and settles, the suspended solids remain enmeshed
in the settling floe and appear to be swept from the water, hence
the description of the process as a "sweep floe" mechanism. This
destabilization mechanism is not the subject of this report be-
cause it is not applicable to direct filtration. The compara-
tively massive chemical treatment rapidly blinds the filter causing
very short runs.
Reduction in surface charge by double layer repression is caused
by the presence of an indifferent electrolyte, which in the case
investigated of brackish water treatment was sodium chloride.
Riddick (6) summarizes the effect of electrolytes on thickness of
the double layer in the data reproduced in Figure 34. For water
and monovalent electrolytes, the thickness of the double layer is
approximately 10 Angstroms (A°) for 0.1M, 100 A° for 0.001M, and
1000 A° for 0.00001M. For double layer repression of colloid
surface charge in brackish waters, the sodium ions of the indif-
ferent electrolyte which surrounds the colloid particles in order
to electrically balance their negatively charged surfaces have
less tendency to diffuse away from the colloid surface as the
salinity increases. Some salt concentration may eventually be
reached such that the thickness of the double layer may be small
enough that two colloids may approach each other close enough that
van der Waals forces cause aggregation. An important aspect
81
-------
Waterborne
Suspended Solids
i
Charge Neutralization
Bridging
"Sweep Floe" Mechanism
Double Layer Repression
Inert Solids
Cold Water
1
Biological
Cell Material
>
' >
Fresh Water
t
Brackish Water
1
Hot Water
Fresh Water
Brackish Water
Condition - response schematic for chemical
treatment of waterborne colloids
Figure 31
82
-------
1 OH"
[AI(H?O)JOE)]
X OH"
[(H?0)3(OH) Al
CD
I
1
/OHN/ ^OB
[(H20)2(OH)2 Al^f [>1^ ^ A1(OH)3(H20)J ^
OH
OH
[(H20)(OH)3 Al'
1OF"
(For solid phase see isometric in following figure.)
[(H20)i,
Al
OH
OH"
[(H£0)3(OH)
\
Sequential formation of hydrous alunlmm oxide polymers
Figure 32
-------
Example of complex which may exist In precipitated
hydrous aluminum oxide polymers
Figure 33
-------
00
1*000
100
a
a
I
10
Type of Electrolyte
Thickness of
Double Layer
0.1 1 10 100
Electrolyte Concentration In MUllnols/Liter
I I i
10 100 1,000
Nad, ng/1 for Reference
Thickness of the double layer (after Rlddlck 6)
Figure 3^
i i 11
10,000
1,000
-------
of double layer repression is that the quantity of colloidal
charge is not significantly reduced but just the extent to which
it extends out from the colloid surface (Figure 30). This relates
to the nature of the destabilizing chemical (salt) and its mode of
action; i.e., the sodium ions remain free in the solvent and cause
rapid dissipation of the charge as the distance from the colloid
surface increases. Double layer repression can improve solids
removal by direct filtration, but this mechanism does not achieve
the best results and can conceal definition of optimal chemical
pretreatment to achieve best filtration results if the interference
of this destabilization mechanism is not recognized.
Bridging by polymers describes the destabilization mechanism where
the molecules of the added chemical attach onto two or more colloids
causing aggregation. Weakly anionic organic polymers are negatively
charged; however, they are especially useful for aggregating and
binding together aggregates into particles that resist redispersion.
Thus, attractive forces of a chemical nature seem to overcome
electrostatic repulsion forces due to like charges. Bridging by
polymers proved to be an important destabilization mechanism for
application to direct filtration. Charge neutralization by ad-
sorption of the destabilizing chemical to the colloid is a key
mechanism for optimizing removal of waterborne solids from brackish
waters by direct filtration. Perhaps adsorption is a poor word
choice here, since the mechanism may not be different from bridging
previously discussed. While the mechanism may be the same, the
end results are different in that the colloidal charge may not
only be reduced to zero but beyond zero, i.e., reversed. Charge
neutralization by adsorption infers that the colloid-water Inter-
face is changed and thereby its physical chemical properties.
This destabilization mechanism can explain those cases where
optimal chemical dosages were found and overdosing resulted in a
deterioration in, or failure of, direct filtration. Charge
neutralization correlated with plant performance as the optimal
destabilization mechanism. For plant control of direct filtration,
charge neutralization was the key test parameter correlating with
performance of the refinery filter when filtering brackish, aerated
lagoon effluent. Brackish water required that charge neutralization
be measured after dilution with distilled water to separate the
effects of double layer repression and charge neutralization;
i.e., under plant conditions of high salinity, the addition of
destabilization chemicals could reduce the ZP to approximately
zero by a range of chemical treatments. However, when double
layer repression was the cause of reduced ZP, reduced filter run
lengths and performance were observed. Reducing the ZP to approx-
imately zero, as measured by means responsive to charge neutralization,
86
-------
pointed out more definitively the required destabilization treat-
ment and resulted in optimal filter performance. Some examples of
charge neutralization mechanisms are shown schematically in Figure
35.
Waterborne colloids subject to chemical destabilization and fil-
tration fall into two general categories: inorganic materials
such as clays and sand and organic materials such as bacteria and
organic colloids. Both categories of colloidal matter may be
stabilized because they are charged and/or are highly hydrophilic.
Both categories of colloidal matter also may vary in response to
treatment by destabilization chemicals, and within each category,
the state of subdivision seems to require additional consideration;
i.e., extremely small colloidal particles are sometimes more
difficult to aggregate for removal by filtration.
With reference to Figure 31, direct filtration is not concerned
with the "sweep floe" or colloid entrapment mechanism for de-
stabilization, but rather charge neutralization and bridging. As
will be illustrated later, double layer repression must be care-
fully identified as a destabilization mechanism or optimal filtration
results will not be achieved.
The nature and size distribution of the colloidal material must be
addressed next. Inorganic colloids such as clays, silicas, metal
oxides, etc. are generally much easier to handle than biocolloids
in terms of amount and complexity of chemical treatment required
for destabilization. Typically a one-chemical, and occasionally a
two-chemical system is required for destabilization of inorganic
colloids (see discussion pages 62 and 63).
Double layer repression of charge is illustrated in Figure 36
where the variation in ZP with specific conductance is shown for
colloidal silica. The specific conductance and ZP of an 84 mg/1
Minusil and salt solution were measured with results of 25,000 and
nil, respectively. Successive dilutions with distilled water were
made followed by ZP and conductance measurements for Curve A. To
a second sample of 8.4 mg/1 Minusil, 2.5 mg/1 of alum was added,
which resulted in a charge reversal on the colloid from about -30
to +30 ZP by adsorption. After alum addition to the second sample,
the specific conductance was increased in increments by adding
NaCl and measuring ZP and conductance. In both cases, a maximum
colloidal charge was observed at a specific conductance of 2,000-
3,000. At higher conductivities, the ZP decreased sharply due to
double layer repression. In the case of silica, high sodium
counterion concentrations eventually achieved almost total re-
duction in ZP. In the case where the charge on the silica was
87
-------
No Specific Adsorption
'*;
-i
Specific Adsorption of
Cations
9
'&
@
Strong Specific Adsorption
of Cation*
j
Specific Adsorption of
Anioni
A Rigid Layer Migrating vith Particle aa Single Kinetic Unit
B Slipping or Trietional Boundary Layer of Ions
X Distance fron An Arbitrary Point Inside Solid Phase
$ Electrical Potential
( Zeta or ELectrophoretlc Potential
Distribution of lone and potentials in the double layer
surrounding colloid* (after Overbeek ?)
Figure 35
68
-------
+ko i—
Minus11 + 2.5 ppm Alum
+20
+10
0
•H
s o
o
-p
-10
-30
O
_L
123"*5 10 15 20 2
Specific Conductance, Mlcromhoe X 10"^
Variation of charge on Minusil with specific conductance
Figure 36
89
-------
Tltration of 8.k ppm Minus11 at Indicated
Specific Conductance and Titrant
Curve Specific Cond.
A
B
C
D
E
F
G
H
2,850
12,000
8.5
15,500
8.0
15,500
7.5
15,000
Titrant
Alum
Alum
Catlonlc PE E-32
Cationic PE E-32
Cationic PE 573
Cationic PE 573
Cationic PE E-5
Cationic PE E-5
+60
+50
-»IK3
1 *3°
PV
3 +20
£ +10
00
-10
-20
2 3
Titrant, ng/1
J
8
Variation of charge on Minusil with various tilranis and two
levels of specific conductance for each titrant
Figure 37
90
-------
reversed by adding alum, the colloid-water interface was obviously
changed in physico-chemical properties and/or the chloride counter-
ion was not as effective for double layer repression as was sodium.
Further ZP measurements using silica and determining the effect of
specific conductance on the performance of various organic cationic
polyelectrolytes and alum are shown in Figure 37. Again, specific
conductance was changed by adding solid NaCl.
Curves A and B (Figure 37) indicate titration with alum at two
levels of specific conductance, 2,850 and 12,000, respectively.
Even 2,850 is high since refinery effluents with only fresh water
inputs are well under 1,000. At the higher specific conductance,
the increased electrolyte impacts on the intensity of charge
reversal achievable by alum and on the stoichiometry to achieve
zero ZP.
The results suggest anion penetration of the alum polymer by
chloride. Anion penetration, the replacement of a coordinated
group such as aquo, hydroxo, or another anion, can be visualized
with reference to Figures 32 and 33. With reference to Figure 32,
penetration by chloride ion may impair the olation of alum to
polymers and reduce the formation, or charge, of the transient
cationic alum polymers important to colloid charge reduction.
Other data on Figure 37 are ZP-titration curves using three com-
mercially available organic cationic polyelectrolytes at two
conductivities. On polyelectrolytes 573 and E-5 the salt did not
impact on their charge reversal properties. For polyelectrolyte
E-32 a salt effect was observed, but the extent of charge reversal
was greater in both cases than achieved by 573 and E-5.
For contrast with the data for destabilization of inorganic colloids,
a series of tests were run to determine the effectiveness of de-
stabilization chemicals on biocolloids in fresh and brackish
water. ZP-titration curves for refinery, fresh water (specific
conductivity 610), aerated lagoon biocolloids are shown in Figure
38. One immediately interesting observation was that E-32, which
reversed the charge of the inorganic colloid to the highest ZP
(Figure 37), was comparatively ineffective for reducing and re-
versing the charge on biocolloids. On the other hand, other
organic polyelectrolytes (573 and 522) could effect charge re-
versal of biocolloids in fresh water. The inferiority of E-32 for
destabilization of biocolloids for filtration was ^verified in
plant tests (Table 5, runs 26 and 27). '
91
-------
+10
4>
VO
to
Specific Conductivity 610
15 20 25
Polyelectrolyte Added, •«/!
The variation in zeta potential of particles in aerated lagoon
effluent (fresh vater) with addition of cafionic polyeiectrolytes
Figure 38
-------
Table 5. A PARTIAL LIST OF PLANT TESTS
Run
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
m/hr
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
8.6
8.6
8.6
8.6
8.6
8.6
8.6
8.6
8.6
8.6
8.6
8.6
8.6
8.6
13.9
Chemicals:
gpm/
sq ft
(5)
(5)
(5)
(5)
(5)
(5)
(5)
(5)
(5)
(5)
(5)
(5)
(5)
(5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(3.5)
(5.67)
522
573
E-32
905
2700
1110
Concentration, mg/1
Nonlonlc
Alum Cat Ionic or Anionlc
21 8 (522) 0
10 8 (522) 0
0 8 (522) 0
20 0 0
20 0 .1 (905)
10 0 .1 (905)
20 0 .1 (905)
20 4 (573) 0
15 4 (573) .007 (905)
20 15 (522) .1 (2700)
20 4.5 (522) .1 (2700)
20 0 0
20 0 .11 (2700)
20 0 .1 (2700)
20 0 0
20 0 .17 (2700)
20 9 (522) .17 (2700)
20 9 (573) .14 (2700)
35 3 (573) .17 (2700)
10 15 (573) .157 (2700)
20 16 (573) .157 (2700)
0 15 (522) .16 (2700)
0 17 (522) .17 (2700)
20 0 0
20 0 .3 (2700)
20 10 (E-32) 0
20 10 (E-32) .3 (1110)
20 0 .3 (1110)
20 15 (522) .3 (1110)
- Betz DK522 (Cationic)
• American Cyanamld (Cationlc)
- Armak (Cationlc)
• American Cyanamld (Nonlonlc)
- Calgon WT 2700 (Weakly Anionlc)
- Betz 1110 (Weakly Anionic)
Early breakthrough
Early breakthrough
Very poor
Poor
Poor
Poor
Poor
Poor, early breakthrough
Poor
Good , 1.3 hours to
Fair, 1.0 hours to
Poor, .2-. 3 hours
.65 hours
.65 hours
1.5 hours, fair at
Fair, 2 hours
Good, 3.5 hours
2 . 5 hours
Fair, 1.5 hours
2.75 hours
3.0 hours
1.25 hours
2.25 hours
Very poor
Poor
Very poor
Poor
0.9 hours
0.9 hours, better
breakthrough
breakthrough
this loading
at this loading
93
-------
The Impact of specific conductance (salinity) on biocolloid de-
stabilization is shown in Figure 39. Curve E shows the impact of
increasing salinity on the ZP of biocolloids in fresh water,
aerated lagoon effluent. The initial specific conductance of 610
was increased by adding solid NaCl and measuring the ZP. This
double layer repression of ZP yields a curve similar to E-32 in
Figure 38, which may yield a clue to the biocolloid destabilization
mechanism of E-32 and why it is not effective. Curves A and B in
Figure 39 are replots of fresh water data in Figure 38 for the
purpose of comparing to brackish water data in curves C and D.
Curves C and D were developed by taking a fresh water, aerated
lagoon sample, adding solid NaCl to give a specific conductance of
12,000 and immediately running a ZP-titration curve. The initial
ZP of the biocolloids was only very slightly negative due to
double layer repression by the salt. Two very interesting and
revealing observations from curves C and D are: 1) for curve C,
the salt approximately halved the effectiveness of cationic poly-
electrolyte 573 for reversing the ZP of the biocolloids to about
+3 ZP, and 2) for curve D, the salt completely destroyed the
effectiveness of cationic polyelectrolyte 522 for reversing the ZP
of the biocolloids. In fresh water, 522 was very effective for
reversing the ZP of biocolloids (Figure 38).
For fresh water colloidal systems, jar tests and plant experience
at refineries have demonstrated the importance of pH for the
evaluation of chemical destabilization systems. Jar tests on the
brackish aerated lagoon effluent at the Yorktown refinery did not
reflect the pH response that fresh water systems did; therefore,
this aspect of the destabilization chemistry was investigated
further.
ZP-alum titration curves on biocolloids in brackish aerated lagoon
effluent were run at a constant pH of 6.0 and 8.5. These data
are plotted in Figure 40 and show that 1) at a pH of 8.5, alum has
an insignificant impact on the ZP of biocolloids in brackish
water, and 2) in contrast to fresh water experience, alum does not
appreciably affect the ZP at a pH of 6.0. The data for brackish
water at 6 pH again suggest that in brackish water, anion pene-
tration of the alum polymer severely interferes with olation of
alum and its ability to reverse the ZP. Qualitatively, more
acidic conditions resulted in more reduction of ZP with alum
addition, but quantitatively the effect was not as pronounced as
in fresh water.
The nature of biocolloids in aerated lagoon effluent reflect their
environment. The ZP of the brackish aerated lagoon effluent
94
-------
A = Addition of Catlonlc 573 at Specific Conductance of 6lO
B = Addition of Catlonlc 522 at Specific Conductance of 6lO
C = Addition of Catlonlc 573 at Specific Conductance of 12,000
D = Addition of Catlonlc 522 at Specific Conductance of 12,000
E • Addition of NaCl(s) to Give Indicated Specific Conductance
?
3
-10
Polyelectrolyte Added, «g/l
1,000 3,000 5,000 7,000 9,000 11,000
Specific Conductance, Micromhos
Variation of charge on particles In fresh water a«rated lagoon
effluent with addition of various polyelectrolytes and NaCl
Figure 39
95
-------
-10
I
I
-15
-20
pH Constant at 6.0
Specific Conductance 11,000
pH Constant at 8.5
Specific Conductance 11,000
Charge on Particles In Distlllec1. Water
1
I
10
20
30 k>
Alum Added, ng/1
60
TO
Th« variation In seta potential of particles in aerated lagoon
effluent with alum at _pH of 6.0 and B.-5
Figure 1»0
-------
averaged -11.4 mV and ranged from -8.5 to -15 as shown in Figure
29. When biocolloids were removed from brackish water by milli-
pore filtration and redispersed in distilled water, the ZP inten-
sified to -23. The ZP of brackish aerated lagoon biocolloids
responded to pH as shown in Figure 41, a curve characteristic of
biocolloids. Biocolloids in fresh water aerated lagoons had a ZP
distribution similar to brackish water biocolloids; however, when
their environment was changed by adding salt, the ZP could be
decreased to almost zero (Figure 39). Plant-scale filter tests
conducted concurrent with laboratory investigations indicated that
early breakthrough of solids was characterized by having a sig-
nificant negative ZP. Pretreatment with a three-chemical de-
stabilizing system could achieve essentially zero ZP, but differences
in filtration performance would occur between essentially the same
colloidal system treated to zero ZP with different chemical
treatments.
Differences in plant performance of chemical destabilization
systems relate to the destabilization mechanism and the requirements
of the colloidal system. Shown in Figure 42 are data that permit
these differences to be identified and the optimum chemical de-
stabilization system determined. These curves and interpretations
were developed by measuring the ZP of brackish feed to the plant
filter with an initial specific conductivity of 15,000 and various
chemical treatments and by following plant filter performance.
Each sample was successively diluted with distilled water and the
ZP and conductivity determined with each dilution; i.e., the
curves are read from right to left. Curve A shows the response to
salinity of the untreated biocolloids. As the salinity decreases,
the repression of the double layer decreases and the charge of the
ZP intensifies to less than -15 mV. Curve B shows the response of
the biocolloid system to dilution when treated by 10 mg/1 alum, 15
mg/1 cationic PE 73, and 0.16 mg/1 anionic PE 2700. Initially,
the ZP was approximately zero at a specific conductance of 15,000.
Upon dilution, however, a negative ZP rapidly developed. This
suggests that some of the charge reduction to achieve satisfactory
plant operations had not been achieved. Curve C shows that in-
creasing the concentration of the alum component of the three-
chemical system for Curve B increases significantly the amount of
destabilization by charge neutralization. Increasing the alum
component further and decreasing the amount of cationic poly-
electrolyte in the three-chemical system resulted in a completely
ineffective treatment for destabilization (Curve D).
Thus, for biocolloids in brackish waters, using ZP to determine
optimal chemical treatment requires that ZP be measured at low,
97
-------
+10 ,—
+5 ~
00
ti
fi
-5
-10
-15 i
Initial Specific Conductance 11,000
k
PH
Charge on suspended matter In aerated lagoon
effluent as a function of. pH
Figure kl
-------
0 _
Curve Alum 373 2700
3^5678 9 10 11
Specific Conductance, Micromhos X 10"
12 13
The variation of charge on aerated lagoon particles with specific
conductance after treatment by various chemical combinations
Figure 42
99
-------
specific conductance (by dilution) to insure that charge neu-
tralization is the predominate destabilization mechanism. Charge
neutralization results in the most stable, filterable aggregates
of suspended matter most resistant to redispersion by hydraulic
forces in the filter. These interpretations of ZP data and evalu-
ation of optimal chemical treatment were confirmed by simultaneous
plant test data.
The destabilization of biocoHolds in brackish water therefore
requires a carefully determined three-chemical treatment pre-
paratory to direct filtration if optimal results are to be achieved;
a one, two, or non-optimal three-chemical system results in sig-
nificantly poorer plant filter results. These data indicate that
the colloid system has at least three components: one responsive
to alum; a second responsive to certain cationic polyelectrolytes;
and a third responsive to both. The total chemical treatment
requirement can be outlined as follows:
Total Chemical = (Alum)(X) + (cationic PE)(Y) + (Alum + Cationic PE)
(Z) + Anionic (k)
Where X * that component responsive to Alum,
Y = that component responsive to cationic PE,
Z = that component responsive to both, and
k = the overall requirement of the destabilized colloids
for a weakly anionic PE to resist redispersive hy-
draulic forces in the filter.
The functional mechanism for destabilization of biocolloids in
brackish water by three-chemical systems is not obvious. Data in
Figure 40 indicate that under these conditions alum is a very poor
prospect for chemical destabilization; yet, alum is required as
part of the chemical treatment system for optimal results. The
role of alum is not clear. Perhaps it reacts with the biocolloid
fraction not responsive to cationic PE and changes the physico-
chemical properties of the biocolloid surface such that this fraction
is then responsive.
The third chemical component of the required treatment is also
somewhat unusual in that the conventional nonionic polyelectrolyte
widely used as a filter aid is not effective; weakly anionic poly-
electrolytes are required.
100
-------
Laboratory ZP data for various chemical treatment systems were
supplemented with plant performance data to aid in interpreting
laboratory observations. Typical plant data accumulated for each
test period are illustrated in Figure 43. Run length comparisons
were determined as the time when the filter effluent reached 10
JTU. A partial list of plant runs is shown in Table 5, page 93.
Guidelines for the filtration of biocolloids in brackish water can
be summarized as follows:
1. Warm water conditions, 29.5+° C (85+° F), require relatively
simple chemical pretreatment systems and permit higher
hydraulic loadings.
2. The colder the water conditions the more important and com-
plex the chemical pretreatment.
3. Even with optimized chemical pretreatment, in colder water
the filter hydraulic loading must be decreased.
4. The filter hydraulic loading as a function of water tem-
perature recommended in Figure 44 also relates to water
viscosity.
Powdered activated carbon (PAC) was incorporated into the opti-
mized chemical treatment system at levels of 50, 100, and 150
mg/1. No "desensitizing" of the biocolloid treatment system was
observed; indeed, if an effect existed at these levels of treatment,
it would appear to be one of interference. PAC samples from four
vendors were screened and similar effects observed. All of the
PAC used had a substantial negative zeta potential which may have
increased the destabilization chemical requirement.
APPLICATION OF RESULTS
The method for determining the optimum chemical treatment for de-
stabilization of colloidal systems was applied at other Amoco re-
fineries. Using temporary, make-shift chemical addition facilities
plant tests were run. In all cases, including the use of both
filters and dissolved air flotation units for phase removal, the
improvement in unit operation was excellent. As a result of these
favorable plant tests, the Yorktown filter is being retrofitted with
the required chemical addition facilities as are new facilities
under construction at Amoco1s Texas City refinery. Startup of
both these facilities is anticipated to be in April, 1975. The
continuous, long-term operation of these facilities is needed to
provide a measure of performance which can be sustained in the
treatment of industrial effluents when the water chemistry require-
ments are met.
101
-------
2.1 (71—
8.56 m/hr (3-5 gpm/sq ft)
•P
In
, 1.2 (4) —
n
n
Hydraulic Loading:
Alum: 20 mg/1
Cationic Polyelectrolyte (DK-522): 9
Anionic Polyelectrolyte (WT-2700) : 01
.6 (2)
•3 (D-
•o
•H
123
Filter Run Length, Hours
Head loss - turbidity breakthrough curve
filtering chemically treated aerated lagoon effluent
Figure 1*3
102
-------
.7 (6)
12.2 (5)
i 9-8 (M
7-3 (3)
(2)
Recommended Water /K
Temperature Hy- *£•
draulic Loading/£|
—Envelop
_: .07 (.8)
Temperature-Viscosity
Relationship for Water
I
10 (50) 21 (70) 32 (90)
Temperature, °C (CP)
o
«
a
1 (1.10)
to
00
.13 (i.to)
.16 (1.70)
CD
O
O
o
•P
I
Recommended hydraulic loading rates as a function of
water temperatures and shoving the
correlation with water viscosity
Figure kk
103
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-660/2-75-025
2.
3. RECIPIENT'S ACCESSIOf»NO.
4. TITLE ANDSUBTITLE
Chemical Coagulation/Mixed Media Filtration of
Aerated Lagoon Effluent
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Grutsch, J. F.
R. C. Mallatt, A. W. Peters
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
American Oil Company
Yorktown, Virginia
10. PROGRAM ELEMENT NO.
1BB036
11. CONTRACT/GRANT NO.
S803026-01
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Robert S. Kerr Environmental Research Laboratory
P.O. Box 1198
Ada, Oklahoma 74820
13. TYPE OF REPORT AND PERIOD COVERED
Demo - 6/71-3/74
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES __^__
Referenced by EPA Effluent Guidelines Division. Prepared in cooperation with the
Petroleum-Organic Chemicals Wastes Section, RSKERL, Ada, Oklahoma 74820
16. ABS1
Operating problems and the effect of operating variables were investigated during full
scale plant operations in the scalping mode. Influent suspended solids concentration
and water temperature were the most significant independent variables. Mechanical
limitations were studied, including a filter bed disturbance that necessitated a
total bed replacement. High, localized backwash velocity caused the invisible dis-
turbance which reduced turbidity removal from about 80 percent to 50 percent. Diag-
nostic procedures, design changes, and the costs of operation and maintenance are
reported.
A cold weather study showed that a three-chemical destablization pretreatment system i;
required for filtration of biocolloids in brackish water. Determination of the
optimal three-chemical destabilization system using zeta potentials required evaluatior
of zeta potentials in a manner which sorted out the effect of double-layer repression.
The colloid destabilization mechanisms of charge neutralization and bridging were
required for optimal filter performance. For colder water temperatures, even with
optimal chemical treatment, the filter hydraulic loading must be decreased. The
change in hydraulic loading with temperature related directly to the water's viscosity.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
Filtration, Particle size, Chemical
coagulation, Construction, Costs, Aerated
Lagoon.
8. DISTRIBUTION STATEMENT
Release Unlimited
EPA Form 2220-1 (9-73)
19. SECURITY CLASS {This Report)
21. NO. OF PAGES
111
20. SECURITY CLASS (Thispage)
22. PRICE
$1900
* U.S. GOVERNMENT PRINTING OFFICE: 1975-698-986 /I I REGION
------- |