United States
Environmental Protection
Agency
Municipal Environmental Research
Laboratory
Cincinnati OH 45268
EPA 600 2-78-209
December 1978
Research and Development
Treatment of
Combined Sewer
Overflows by High
Gradient Magnetic
Separation
On-Site Testing With
Mobile Pilot Plant
Trailer
-------
EPA-600/2-78-209
December 1978
TREATMENT.OF COMBINED SEWER OVERFLOWS
BY HIGH GRADIENT MAGNETIC SEPARATION
On-site Testing With Mobile
Pilot Plant Trailer
by
David M. Allen
Sala Magnetics, Inc.
Cambridge, Massachusetts 02142
Contract No. 68-03-2218
Project Officers
Hugh Masters
and
Richard P. Traver
Storm and Combined Sewer Section
Wastewater Research Division
Municipal Environmental Research Laboratory (Cincinnati)
Edison, New Jersey 08817
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Municipal Environmental Research
Laboratory, U.S. Environmental Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
ii
-------
FOREWORD
The Environmental Protection Agency was created because of increasing
public and government concern about the dangers of pollution to the health
and welfare of the American people. Noxious air, foul water, and spoiled
land are tragic testimony to the deterioration of our natural environment.
The complexity of that environment and the interplay between its components
require a concentrated and integrated attack on the problem.
Research and development is that necessary first step in problem solu-
tion and it involves defining the problem, measuring its impact, and search-
ing for solutions. The Municipal Environmental Research Laboratory develops
new and improved technology and systems for the hazardous water pollutant
discharges from municipal and community sources, for the preservation and
treatment of public drinking water supplies, and to minimize the adverse
economic, social, health, and aesthetic effects of pollution. This publica-
tion is one of the products of that research; a most vital communications
link between the researcher and the user community.
Combined sewer overflows cause substantial pollution of major waterways
and coastal areas in the U.S. The associated large surge volumetric flow
rates in combination with the exponentially increasing solids loadings which
often occur during a "first flush" storm situation are at best difficult for
even the most advanced current integrated wet and dry weather sewage treat-
ment facilities to handle. It is also very costly using conventional tech-
nologies to have on hand reserve capacity of this magnitude for such in-
frequent events. The use of magnetic filters and a seeding technique,
however, offer an alternative to reserve capacity inherent in the magnetic
nature of the separation itself. Increased storm solids can be handled simply
by increasing chemical and seed dosages, while increased flows can be accom-
modated either by bringing on additional separators and/or by increasing the
magnetic field to compensate for higher surface loadings. Such adaptability
is attractive in the treatment system both from a capital cost and an opera-
tion and maintenance point of view.
This report describes a preliminary on-site stage of testing of this
technology in which specific problems were addressed and overall feasibility
was studied.
Francis T. Mayo
Director
Municipal Environmental
Research Laboratory
iii
-------
ABSTRACT
Seeded water treatment using a SALA-HGMF® high gradient magnetic separa-
tor pilot plant system was conducted on CSO and raw sewage at Sala Magnetics,
Inc. in Cambridge, Massachusetts and at on-site locations in the Boston area.
These tests built further upon the data base collected during the first phase
of this contract (see EPA Publication 600/2-77-015) with special emphasis on
specific design and operational parameters, long-term durability and mainte-
nance problems, and system adaptability to integrated wet and dry weather flow
conditions. The on-site results reported, although not equaling those ob-
tained on uniform batch samples in-house, were nevertheless good and proved
that SALA magnetic filters are effective on fresh CSO and raw sewage dynamic
solids loading and flow rate conditions typically associated with storm water
and integrated wet and dry treatment systems, even where tidal salinity in-
trusion is experienced.
This report was submitted in- fulfillment of Contract #68-03-2218 by Sala
Magnetics, Inc. under the sponsorship of the U.S. Environmental Protection
Agency. The work covers a period from August 1976 to December 1977.
IV
-------
CONTENTS
Page
Foreword ±±±
Abstract ±v
List of Figures vii
List of Tables ix
Acknowledgments x
Section
I Conclusions 1
II Recommendations 3
III Introduction 5
Overview of High Gradient 5
Magnetic Filtration
Principles of High Gradient 6
Magnetic Filtration
Magnetic Seeded Water Treatment 10
as Applied to CSO and Raw Sewage
Previous Work 10
Present Contract Goals 10
IV Description of the Pilot Plant Trailer 12
Physical Layout 12
Flow Diagram and Description of 14
Pilot Plant System
V Testing Sites 20
Tests at Sala Magnetics 20
Cottage Farm Work 20
Columbia Park Testing 22
v
-------
Section Page
VI Discussion of Results 24
Summary of Results 24
Overview of Test Conditions 24
Backflush Tests 25
Chemical Addition Techniques 33
Matrix Loading Tests 41
Limits of Flow Velocity on 48
Magnetic Filtration
Sludge Evaluation 48
Static Mixer versus Flash Mixer 51
Storm Profile Tests 54
VII Cost Estimates for a High Gradient Magnetic 66
Filter Based 25 MGD CSO/Sewage Treatment
Facility
Capital Costs 66
Operation and Maintenance Costs 70
Design Considerations 70
References 78
Glossary 80
VI
-------
LIST OF FIGURES
Number Page
III-l Cut-Away View of High Gradient Magnetic Separator 7
III-2 Matrix Materials Used in High Gradient 9
Magnetic Separation
IV-1 Magnetic Filtration Pilot Plant Trailer 13
IV-2 Simplified Pilot Plant Floor Plan and 13
Flowsheet
IV-3 Detailed Floor Plan of Mobile Trailer 15
IV-4 Pilot Plant Trailer Interior 16
IV-5 Process Flowsheet for Seeded Water Treatment 17
Pilot Plant
V-l Cottage Farm Chlorination and Detention Facility, 21
Cambridge, Massachusetts
VI-1 Backflush Volume versus Hydrotank Pressure 28
VI-2 Backflush Effectiveness 29
VI-3 Effect of Residence Time on Turbidity 40
VI-4 Effect of pH on Turbidity 42
VI—5 Suspended Solids versus Time into Cycle for 44
Various Seed-to-Solids Ratios
VI-6 Seed-to-Solids Ratio: Suspended Solids 45
Variation in Cycle
VI-7 Seed-to-Solids Ratio: Turbidity Variation 45
in Cycle
VI-8 Suspended Solids versus Time into Cycle for Feed 47
and Treated Samples Using Different Matrices
VI-9 Matrix Loading Curves for Two Matrices 47
VI-10 Flow Flux Rate versus Suspended Solids 49
VI-11 Flow Flux Rate versus Turbidity 49
VI-12 Storm of 9/20/77: Physical Parameters 56
Vll
-------
Number Page
VI-13 Storm of 9/20/77: Suspended Solids and Turbidity 57
of Feed and Treated Samples
VI-14 Storm of 9/26/77: Physical Parameters 58
VI-15 Storm of 9/26/77: Suspended Solids and Turbidity 59
of Feed and Treated Samples
VI-16 Storm of 10/14/77: Physical Parameters 60
VI-17 Storm of 10/14/77: Suspended Solids and Turbidity 61
of Feed and Treated Samples
VI-18 Probability Curve for Suspended Solids on 10/14/77 Run 64
VI-19 Probability Curve for Turbidity on 10/14/77 Run 65
V1I-1 Proposed Design for High Gradient Magnetic Filter 71
Based 25 MGD CSO/Sewage Treatment Plant
VII-2 Top and Side Views of Alum Residence/Flash Mix/ 77
Flocculator Chain for 25 MGD Seeded Water
Treatment System
viii
-------
LIST OF TABLES
Number Page
VI-1 Backflush Technique Comparison 26
VI-2 Backflush Time Tests 27
VI-3A Backflush Pressure and Duration Data 30
VI-3B Backflush Pressure and Duration Data 31
VI-4 Chemical Addition Sequence 33
VI-5 Polyelectrolyte Comparison 37
VI-6 Residence Time 39
VI-7A Flush Water Solids Concentrations 52
VI-7B Flush Water Solids Concentrations 53
VII-1 Capital Cost Estimates for a High Gradient 67
Magnetic Filter Based 25 MGD CSO/Sewage
Treatment Facility
VII-2 Operation and Maintenance Cost Estimates for 73
a High Gradient Magnetic Filter Based 25 MGD
CSO/Sewage Treatment Facility
VII-3 Power Consumption for a 25 MGD Facility 74
VII-4 Assumed Parameter Values for a 25 MGD System 75
IX
-------
ACKNOWLEDGEMENTS
We are thankful for the cooperation of Mr. Andrew Fisicelli of the
Metropolitan District Commission (MDC) for allowing the operation of our
mobile trailer at the Cottage Farm and Columbia Park locations. Special
thanks to all employees at those locations who were so helpful in so many
ways throughout the testing.
Also deserving of thanks are EPA Project Officers Hugh Masters and
Richard Traver and Chief Richard Field, all of the Storm and Combined
Sewer Section of the USEPA. They have provided guidance and support
throughout the project.
This project was conducted under the supervision of Dr. John A.
Oberteuffer of Sala Magnetics, Inc. In-house and on-site pilot plant test-
ing and overall project organization was carried out by David M. Allen,
Project Engineer. Richard L. Sargent contributed in the section on cost
estimation and was active in trailer design. Assembly of the mobile unit
was directed by installation engineer, Robert Vinton.
-------
SECTION I
CONCLUSIONS
The test results reported here on magnetic filtration of CSO and raw
sewage have given valuable information on system performance. A mobile
trailer pilot plant capable of processing 6 gpm at 100 gpm/ft2 was constructed
and used successfully for the bulk of the testing performed, including the
actual profiling of three individual storm flows at an on-site location.
Although initial problems were encountered in obtaining adequate stabili-
zation of the flocculation train due to tidally induced salinity fluctuations
in the incoming waste water, it was found that these problems could be over-
come by the use of additional magnetite, a less salinity-sensitive flocculant,
and some other minor changes in the process stream and operation procedure.
By incorporation of these refinements in monitoring and technique,
essentially all process problems were overcome and the technology was demon-
strated as one capable of handling even the most adverse of the dynamic surge
flow circumstances. This represents a major step in the proving of seeded
water treatment as a viable process for the treatment of CSO.
The best practical backflush technique for optimal solids removal and
concentration was found to be a combination of air-water flush with indepen-
dent air and water sources (50 psi constant air and 35 psi initial charge on
the hydrotank water source). This flushing technique was shown to prevent
any significant build-up within the matrix from occurring in the rather short
term test runs performed. To ensure a clean matrix over long intervals, a
periodic chemical rinse is necessary, and extremely effective. Water back-
flush pressure was found to be optimal when kept to > 25 psi initial charge
on the hydrotank. A 5 second backflush time was shown to be sufficient in all
cases tried for complete solids removal, and even shorter flushes (of down
to 2 seconds) gave consistently excellent results.
Chemical flocculation techniques were studied to determine the best chemi-
cal addition sequence, the optimal flocculant and starting pH, and the minimum
residence time necessary for best floe formation and clarification of the
waste water studied. It was found that although it made no noticeable differ-
ence when in the sequence magnetite was added, alum had to be added prior to
polyelectrolyte addition (^ 2 minutes), with a total residence mixing time of
more than 4 minutes. Prolonged residence times of up to 20 minutes were found
to enhance clarification only slightly (<5%) while decreasing the average floe
size (unless much slower mixing rates were used). Optimum starting pH (as in
the previous EPA-funded study) was found to be at natural pH (6.74) for the
flocculation procedure used.
1
-------
A number of different Hercofloc and Betz flocculants were tested on the
same effluent in a comparative test. It was found that the most effective in
removing turbidity was an anionic polymer, Hercofloc 831. Betz 1160 (used for
all testing in this report) and Hercofloc 849 were also found to give relative-
ly good results.
Matrix loading tests were used to help determine the minimum seed-to-
sblids ratio necessary for good magnetic separation. It was found that this
ratio should be at least 2.5:1 for unsieved, commercial grade magnetite in
order to achieve adequate separation efficiency and cycle length. This pro-
portion perhaps could be cut in half by grinding the commercial magnetite to
a finer size. A matrix design comparison using a matrix loading test was
unable to delineate differences between two similar designs due to fluctuations
in incoming solids loadings created by rainy weather.
Flow velocity tests showed that seeded water treatment efficiency de-
creases rather sharply above 200 gpm/ft^, although significant deterioration
in results occurs between 90 and 200 gpm/ft^ as well. At 90 gpm/ft , control-
led results approached 95% removal in a batch-type continuous pilot plant run
indicating that slower process rates could not possibly improve results enough
to compensate for the resulting loss in capacity in most applications.
A limited evaluation of sludge character showed the thickener underflow
to have a solids concentration of 28% (with a 3:1 seed-to-solids ratio) and
a density of 1.13 g/cc. The backflush water as it came into the thickener
ranged from 0.8 to 14 g/£, depending on the solids loading and flushing tech-
nique used.
-------
SECTION II
RECOMMENDATIONS
In order to adequately demonstrate the feasibility of magnetic separation
techniques for the treatment of CSO and raw sewage, more and larger scale
testing must be undertaken. The following outline gives a simplified version
of our recommendations for future work.
Step I: Additional Testing with SALA Mobile Trailer at EPA Location
A. Modification of 6 gpm Mobile Trailer.
1. Make suitable for continuous unattended runs of
12 hours.
2. Improve analytical and monitoring facilities.
3. Upgrade sludge handling system.
B. On-Site Pilot Testing of 6 gpm Mobile Trailer at EPA Location.
1. Establish performance on specific waste water.
2. Obtain design criteria for a 1 MGD system.
3. Evaluate sludge characteristics.
4. Initiate seed recycle evaluation.
5. Study long-term separator matrix maintenance
requirements.
6. Profile storms to evaluate system adaptability and
performance on CSO.
Step II: Design, Construction and Testing of a 1 MGD Magnetic
Filter Pilot Plant
A. Design of the 1 MGD, skid-mounted system.
-------
B. Construction and Installation of the 1 MGD Pilot
Plant.
1. Assembly of major subsystems at Sala Magnetics, Inc.
2. Transport and final assembly at EPA location
C. Testing of 1 MGD Pilot Plant.
1. Demonstrate system performance.
2. Treat sludge with traditional existing facilities.
3. Finalize design and implement equipment for the following
studies:
a. Magnetite seed disinfection and recycle.
b. Sludge densification and disposal.
Step III: Design and Construction of a 25 MGD Permanent
Demonstration Seeded Wastewater Treatment Pilot Plant
at a High-Visibility Location
Step IV: Full-Scale Municipal Integrated Wet and Dry Weather
Treatment Facility
-------
SECTION III
INTRODUCTION
OVERVIEW OF HIGH GRADIENT MAGNETIC FILTRATION
The use of magnets to separate substances of varying character is not
new. Magnetic separation techniques have been used since the nineteenth
century to remove tramp iron and to concentrate iron ores. A variety of
conventional magnetic separation devices is in wide use today. These
devices generally separate relatively coarse particles of highly magnetic
material containing large amounts of iron from nonmagnetic media (direct
filtration).
In recent years magnetic devices have been developed which are capable
of separating even weakly magnetic materials of micron size at inherently
high flow rates. These so-called "high gradient magnetic separators" have
been designed to maximize the magnetic forces on fine, paramagnetic mate-
rials. They are capable of efficient separation of even weakly magnetic
suspended solids or precipitates for which conventional magnetic separation
techniques are ineffective. This capability is the result of the development
of a filamentary ferromagnetic matrix and a large volume, high-field magnet.
The combination of an efficient magnet and high gradient matrix permits the
economical production of strong magnetic forces over a large surface area
in the active volume of the separator. The separations may be carried out
economically, and at process rates of up to several hundred gpm/ft^-
For normally nonmagnetic colloidal material in polluted water, the ad-
dition of magnetic iron oxide (magnetite) along with a flocculating agent can
render these colloids sufficiently magnetic to be removed by high gradient
magnetic separators (indirect filtration). The machines provide the rapid
filtration of many pollutants from water with a small expenditure of energy.
Removal is much more efficient than with sedimentation because the magnetic
forces on fine particles may be many times greater than gravitational forces.
This technology has a high potential for use in water pollution control.
High gradient magnetic separation is used in the kaolin clay industry to
remove weakly magnetic impurities of less than 2 micron size from clay. In-
dustrial-size high gradient magnetic separators treat up to 60 tons per hour
of dry clay, as a 30 percent slurry.
Other proven applications for HGMS magnetic separators include iron ore,
feldspar, and many other types of mineral beneficiation. Waste reclamation
and recycling, ultra purification of chemical refractories and powders,
-------
removal of smoke stack particulates, cleaning of refueling pool waters at
nuclear power plants, steam purification and other thermal power applications,
and steel mill waste water purification are some of the recent problems that
HGMS magnetic separators are or will soon be handling. All are direct appli-
cations and do not require the addition of a seed or flocculant to be effec-
tive.
Besides CSO and raw sewage, high gradient magnetic separation is appli-
cable to numerous nonmagnetic waste waters such as paper mill wastes, elec-
troplating waters, secondary effluent polishing, potable water processing, on
board ship treatment of gray and black water, and almost any polluted stream
in which the goal is to remove all solids from the water portion.
PRINCIPLES OF HIGH GRADIENT MAGNETIC FILTRATION
Magnetic and Competing Forces
High gradient magnetic separators, like all magnetic separators, utilize
the interaction of magnetic and competing forces on a mixture of magnetic and
nonmagnetic particles to provide separation based on the magnetic suscepti-
bilities of the particles. The magnetic forces of attraction in a high gra-
dient magnetic separator hold the magnetic particles to the edges of the ma-
trix fibers while the competing hydrodynamic forces carry the fluid and non-
magnetic particles through the separator. For small particles the forces of
hydrodynamic drag are larger than gravitational forces, and increase with
slurry velocity in the separator. The magnetic forces necessary to trap
these particles must therefore be large.
Maximizing the Magnetic Forces
High gradient magnetic separators effectively maximize the magnetic
force on even weakly magnetic particles. The magnetic force (F ) on a
particle is given by the following expression:
Fm = vM grad H
where v is the volume of the particle, M is its magnetization, and grad H is
the magnetic field gradient that acts on the particle. The magnetic field
gradient appears in the expression for magnetic force for the following
reason. Placed in a magnetic field, all particles develop north and south
poles at either end as shown in Figure III-l. In a uniform field the net
force on a particle will be zero, since the field exerts an equal and oppo-
site force on either end of the particle. In a gradient magnetic field,
however, the force exerted by the stronger field at one end of the particle
will produce a net force on the particle. Therefore, the larger the change
in field across the particle (magnetic field gradient), the greater the force
on the particle.
The magnetization of ferromagnetic fibers, like those in the high gradi-
ent magnetic separator matrix, produces extremely high magnetic field gradi-
ents. It turns out that the greatest force is produced on the particles when
-------
MAGNETIC FORCE
grad H
L magnetic field
gradient
Lparticle magnetization
L-particle volume
COMPETING FORCE
hydrodynamic drag
FIGURE III-l CUT-AWAY VIEW OF HIGH
GRADIENT MAGNETIC
SEPARATOR
L slurry velocity
particle diameter
-slurry viscosity
-------
the diameter of the magnetized wire is approximately the same size as the
particle to be trapped. This matching of the fiber diameter to the particle
size is utilized in high gradient magnetic separators to produce extremely
large forces on even weakly magnetic particles.
Ferromagnetic Matrix
In order to produce strong magnetic forces over a practical surface area,
a filamentary ferromagnetic matrix is used, magnetized by a strong applied mag-
netic field. The effective trapping volume of this type of matrix is many
times larger than that achieved by the use of tacks, balls, or other small
ferromagnetic objects which also produce large field gradients when magnetized.
As shown schematically in Figure III-l, the introduction of the ferromagnetic
matrix into a uniform magnetic field produces a multitude of high gradient,
strong forces within the volume of the separator. A strong applied magnetic
field is required to magnetize these fibers, and when the field is turned off
the residual magnetization of the fibers is very low. For this reason, even
strongly magnetic particles are easily washed out when the applied field is
reduced to zero. Typical matrix materials are shown in Figure III-2.
Production of Strong Magnetic Fields
The ferromagnetic matrix is a relatively difficult magnetic structure to
magnetize; that is, a strong applied magnetic field is required to produce the
high field gradients along the matrix filament edges. The magnetic fields
in conventional magnetic separation devices are not sufficient to magnetize
the ferromagnetic matrix. Therefore, practical realization of high gradient
magnetic separation depends as much on the production of an economical, in-
tense magnetic field as it does on the ferromagnetic matrix. The SALA design
accomplishes this goal by placing the electromagnetic coil directly around the
working volume of the canister, and a steel magnetic flux return frame around
the coil. This not only enhances the field created, but also increases the
efficiency of the magnet and lowers its material cost.
Operating Variables of the Magnetic Separator
The efficiency of magnetic particle trapping in a high gradient magnetic
separator depends heavily upon the operating variables of the separator as
well as the size and magnetic susceptibility of the particles. The basic
operating variables in direct filtration are: the strength of the applied
magnetic field, the velocity of the feed passing through the matrix, the type
and configuration of the matrix, and the accumulated feed solids in the work-
ing volume of the separator (matrix loading). The recovery of the magnetic
particles increases with an increasing magnetic field because the magnetic
forces are stronger. The recovery of the magnetic particles decreases with
increasing feed flow velocity because the competing hydrodynamic drag forces
are larger. The recovery of magnetic particles decreases with increased
matrix loading since the high gradient trapping sites within the matrix be-
come filled, leaving fewer and weaker available trapping sites. For indirect
filtration applications (seeded water treatment), the additional factors of
seed and flocculant type and concentration, residence time and mixing rate,
pH, and other physical and chemical characteristics of the substances being
-------
»"
£>"
Expanded Metal
Stainless Steel Wool
FIGURE III-2 MATRIX MATERIALS USED IN HIGH GRADIENT MAGNETIC SEPARATORS
-------
separated are important operational variables in addition to those cited above.
Thus, seeded water treatment is dependent upon the maintenance of a delicate
chemical balance in order to achieve an effective union of suspended solids
and seed particles before their magnetic removal.
MAGNETIC SEEDED WATER TREATMENT AS
APPLIED TO CSO AND RAW SEWAGE
The seeded water treatment (mag-seed) process is a unique application of
high gradient magnetic separation to the removal of nonmagnetic suspended and
colloidal-sized particles suspended in a liquid medium (usually water). It
has considerable potential in a large number of effluent waste water cases
where certain standards must be met before disposal, as well as in some closed
loop operations where corrosion products or contamination- may result in degrad-
ation of liquid quality within the system. The system is of particular
interest for its possible application to CSO and raw sewage and a number of
other areas. Calculations for effectiveness of separation, economics of
capital investment and operating costs, land requirements, dependability,
process flow rates, and detention times, etc., have so far been favorable
in comparison with presently available technology.
PREVIOUS WORK
This report is a continuation of Report #600/2-77-015 (March 1977)
entitled, "Treatment of Combined Storm Overflows by High Gradient Magnetic
Separation." In that portion of the study, full descriptions and references
are provided for the physics and concepts involved in magnetic filtration.
In completing that work, both bench and continuous pilot plant runs were per-
formed at Sala Magnetics, Inc. in Cambridge on CSO and raw sewage trucked in
from the Cottage Farm Chlorination and Detention Facility (Cambridge) and the
Deer Island Sewage treatment Plant (Boston) . These tests showed clearly that
the seeded water treatment process could effectively and efficiently treat
these waste water samples. However, limitations in the pilot plant system
and lack of freshness in the sample volume suggested that an on-site test with
a slightly larger and more flexible system would be necessary before jumping
to demonstration size. A mobile system also would allow the performance of
on-site testing with several different effluent situations in order to provide
a maximum amount of design and cost estimating input. Whereas in th% previous
study CSO had been slightly aged and relatively static within the test period,
with a mobile trailer on location it would be possible to profile an actual
storm event, as it occurred, in order to study in detail the possible problems
and solutions unique to combined storm overflows (e.g., first flush loadings,
multiple separator storm function, required influent monitoring systems, etc.).
PRESENT CONTRACT GOALS
The present effort is designed to demonstrate the pilot-scale effective-
ness of SALA-HGMF® magnetic filter treatment of CSO, and to use this informa-
tion as a basis for further larger scale tests. Various design criteria and
10
-------
basic operating characteristics of the separator system were studied in some
detail in order that accurate costing projections could be made for full-size
integrated wet and dry weather treatment systems.
The extension of EPA Contract #68-03-2218 was performed in two parts:
Effort I extended the data base of the previous work on the then-existing
1 gpm pilot plant with several specified tests, and upgraded that original
pilot plant by means of a dual-magnet system, with more advanced controls, in-
stalled in a mobile unit; Effort II included on-site testing of the mobile
unit, including several storm flows, as well as completion of the necessary
backflushing, cleaning and sludge evaluations. With the information gained
from these efforts and from previous testing using the seeded water technique,
basic design and operating characteristics could be developed as a basis for
the generation of costs and criteria for a demonstration-scale system.
11
-------
SECTION IV
DESCRIPTION OF THE PILOT PLANT TRAILER
PHYSICAL LAYOUT
The EPA Seeded Water Treatment Mobile Pilot Plant Trailer is shown in
Figure IV-1. This 28 foot vehicle was constructed by Relco Corporation of
Billerica, Massachusetts and was furnished with interior pilot plant and lab
facilities by SALA. A simplified floor plan and pilot plant flow sheet are
shown in Figure IV-2. General specifications and requirements are listed
below.
General Dimensions
Trailer Dimensions: 8'W x 28'L x ll'H (2.4m x 8.5m x 3.5m)
Laboratory-Office: 8'W x 13'L (2.4m x 4m)
Gross Weight: 7000 Ib (3000 kg)
Pilot Plant Specifications
Flow Capacity: 10 gpm (0.6 I/sec)
Sludge Production: 1 Ib/hour (0.5 kg/hour)
Utilities Requirements
Electric Power: 8.8-22 kW
Tap Water: 5-25 gpm (exclusive of feed)
Compressed Air: 1-3 cfm at 40-60 psi
Appropriate Drainage
Typical Chemical Requirements
Magnetite: 0.6 Ib/hour (0.25 kg/hour)
Alum: 0.25 Ib/hour (0.1 kg/hour)
Polyelectrolyte: 0.04 oz/hour (0.001 kg/hour)
For further descriptions and detailed schematics refer to the operating
manual for the Sala Magnetic Filtration Pilot Plant.
12
-------
Magnetic.Filtration
pilot P'anT
FIGURE IV-1 MAGNETIC FILTRATION
PILOT PLANT TRAILER
Trailer Floor Plan
B c
M, D
M,
COUNTER TOP-
CABINETS UNDER
COUNTER TOP-CABINETS UNDER
INSTRUMENTS
FEED
V. KEY A - Wedgewlre Screen
B-Flash Mixer
o
C - Flocculallon Tank TREATED
0-Turbldlmeler WATER
E-Sludge Thickener I
M, - Small Magnet
M,-Large Magnet *
Pilot Plant Flowsheet
FIGURE IV-2 SIMPLIFIED PILOT PLANT FLOOR
PLAN AND FLOWSHEET
13
-------
As can be seen in Figure IV-2, the trailer is divided into two sections,
the forward portion being for laboratory analysis and office space and the
larger, rear portion containing the SALA-HGMF® Magnetic Separator pilot plant
system, sink and jar testing and millipore filter apparatus set-ups. Other
SALA laboratory equipment includes pH and conductivity meters, spectropho-
tometer, incubator, oven, balance, and magnetic stirrer. These facilities
allow for immediate processing of test samples giving the experimenter quick
feedback on the process (through turbidity measurements) and assuring that
post-collection contamination or aging does not take place. Other relevant
tests that can be performed easily with the presently available equipment are
total suspended solids, color, and fecal coliform.
Figure IV-3 shows a more detailed floor plan of the EPA Trailer, includ-
ing the major process components of the pilot plant. Figure IV-4 shows a
portion of the mobile trailer interior containing the majority of this equip-
ment.
The mobile unit is fully insulated and may be heated by electric base-
board heaters located in each of the two rooms. Running water is supplied for
pilot plant cooling and backflush water as well as for laboratory uses. A
floor drainage system leads to an external sump pump with a capacity of about
35 gpm. Ceiling fans provide ventilation for each room and numerous cabinets
and cupboards supply generous storage space for chemicals, glassware, hardware,
tools, spare parts, and other necessary equipment. Once in place, the trailer
is supported by four Simplex jack stands set on heavy steel blocks. The tires
should also be blocked for added safety.
The trailer is protected against burglary in several ways. Double locks
have been installed on each of the two doors and lexan storm panels have been
added to both of the front windows to prevent possible breakage and for extra
insulation. Inside, an ultrasonic alarm system capable of running off AC
(normal) or DC (emergency battery) adds a further deterrent to break-ins. A
first-aid station and fire extinguisher are wall mounted in the pilot plant
section.
FLOW DIAGRAM AND DESCRIPTION OF PILOT PLANT SYSTEM
Figure IV-5 shows the process flow sheet for the seeded water treatment
system. Untreated waste water is pumped into the trailer system by ^ submers-
ible pump (P-A) or any facility sampling pump. As it enters it passes through
a wedge wire screening device (where it is screened to -35 mesh size) and into
a 55 gallon drum with low- and high-level float switches to maintain a sample
reservoir for treatment. A variable speed mixer prevents settling within the
55 gallon barrel.
The feed pump (PI) draws sample from the drum and pumps it through a
rotameter (FMI) for process flow rate monitoring. The rate of pumping is de-
termined directly by the operator who sets the master control knob on control
panel #2. A float switch in the flocculator tank also shuts PI off when the
filtering throughput is unable to keep up with feed pumping rate.
14
-------
FLOCCU.ATOR
STATIC MIX UNIT
FLUSH WX UNIT
THICKENER
- POWER SUPPLIES
,—TURSIDIMETER
MAOHCTITE STOAAGE
weaai mm fcraiN
AUTOMATIC CONTROL AND FEED SACK PROCESSINO
FEED MIXER
CHBCCM. FLUSH PUMP
CHEMICAL MErCMNO PUMPS
ALUM STOHAOE
fH METERS .TumniMETeR OUTPUTS
POLTELECTROLm
CHEMICAL FLUSH WATER STORAGE
TE STORAGE
-CUPBOARD
-COUNrER TOP AND STORAGE AREA
FIGURE IV-3 DETAILED FLOOR PLAN OF
MOBILE TRAILER
-------
FIGURE IV-4 PILOT PLANT TRAILER INTERIOR
16
-------
CAUSTIC ALUM MAGNETITE POLYCLECTROLYTE
8TOHAGE STORAGE STORAGE STORAGE
^ tOLIMOID
y< ILICT. VALVi
Q
y< AIR VALVI
CHICK VALVI
FIGURE IV-5 PROCESS FLOW SHEET FOR SEEDED WATER TREATMENT PILOT PLANT
-------
Process water passes through one of two alternative chemical mixing units
before entering the flocculator. The first, a conventional flash mixer, con-
sists of a large (8 inch diameter polysulfate) cylinder mounted on a slant and
compartmentalized into four sections by three, toothed weirs. The water pass-
ing through each compartment is agitated by a variable speed paddle mixer
which disperses the chemical(s) being added in that section. In general, the
first compartment is reserved for pH monitoring and adjustment (if necessary),
the second is for alum addition, the third for magnetite, and the fourth for
polyelectrolyte. The alternative system is a static mix unit (1 inch diameter)
made by KOMAX. This 24-element unit disperses the chemicals (which are added
in the same order as above through ports in the statix mixer) by repeated
division and rejoining of fluid flow throughout the 18 inch length. In this
way, no moving parts are needed, and considerable space is saved in the system.
This unit is designed for flow rates on the order of 1 gpm and is less effect-
ive for substantially faster process rates.
Both chemical mixing units dump into the flocculator which is a large
(30 gallon) polyethylene tank agitated by a variable speed mixer. This pro-
vides contact time for floes to form before magnetic filtration. Also located
in this chamber are the flocculator feedback control unit and the feed pump
shut-off float. The feedback control unit provides an indirect contact between
the manually set feed pump (PI) and chemical metering pumps and the filter
pumps for systems I and II (pumps P2 and P3 respectively). This is done by a
mechanical linkage between a level float in the flocculator and a series of
microswitches. The microswitches, responding to flocculator water level,
control which system (I or II) is on and how fast that system is filtering.
Thus, when the incoming stream remains constant, the filtering rate will seek
that rate and hold it, as well as track any changes (surge simulation) imposed
on the feed pump and associated chemical metering pumps. The rate of flow
through the filters can therefore be controlled directly by flocculator level
which in turn is a function of the (manually controlled) feed pump rate. A
lag time is present due to the plumbing (flash mixer, etc.) preceding the
flocculator. A small cyclical fluctuation in level occurs due to the cyclic
operation of the magnetic filter (filter and backflush modes alternate). An
alternate manual control of filter pump operation (P2 and P3) is also provided
which is independent of the feed pump (PI) rate.
The pre-treated waste water then passes up through the magnetic filter(s)
via the suction action of the filter pump(s) (P2 and/or P3) and opening of air
valves VI and V2 and/or V3 and V4. These variable speed pumps have practical
capacities of 1.5 gpm (P2) and 5.5 gpm (P3), yielding flow flux rates of 100
gpm/ft2 for each, and pump through their respective rotameters before dis-
charge into the drain at the sample station. A small portion of this flow may
be tapped for continuous monitoring of turbidity through a Hach Model 1720
Low Range Turbidimeter when good separation is being achieved, and the turbid-
ity recorded on a Rustrak chart recorder.
Cooling water from the tap is provided to magnetic separators I and II and
to their power supplies through control and solenoid valves. Water flow
through these valves should be approximately 0.2 gpm for V20 (system I) and
0.4 gpm for V21 (system II), or not hotter than 60°C at the drain.
18
-------
Backflush is accomplished by a pressurized vessel (20 gallons) about
one-third filled with water and with the remaining volume air-pressurized to
^ 35 psi when fully charged. The water refill is normally accomplished by tap
water addition (via solenoid valve V13) but the system can be altered to allow
treated process water to be pumped (by P5) into the pressure vessel for normal
cycling (preferably at high process rates). A float switch inside the vessel
controls water level.
After water addition, the tank is automatically pressurized (via solenoid
valve V14) to a pre-set pressure as gauged by a Honeywell air switch. Should
the air-water recharge described above not be completed before the filter cycle
times out, the next step will be delayed until recharge is accomplished to
avoid inadequate matrix flushing. The tank may be manually vented by opening
solenoid valve V15 and a manual ball valve near the top of the pressure vessel.
This operation is necessary before shutting down the system in order to release
the pressure in the tank before the normally open valves become de-energized.
Backflush through the separators occurs when air valves V5 and V6 or V7
and V8 open with all other valves closed. This allows the pressurized water
to excape (via the demagnetized magnetic separator matrix) to the vented surge
tank which integrates the flow of the sludge-filled backwash water to the
thickener. Air is mixed with the backflush water as it leaves the pressure
vessel from one of two alternative sources. Solenoid valves V16 or V17 are
opened allowing an amount of air to enter the backflush stream. V16 allows
air to escape from the pressurized vessel itself, and thus the relative ratio
of air to water over the backflush interval is constant. When V17 is used,
the air pressure used to aerate the backflush water is independent of the pres-
sure in the vessel and the relative proportion of air in the water can be reg-
ulated. Thus the air-to-water ratio will increase over the backflush interval
for this operational mode.
The backflush water in the 5 gallon surge tank drains slowly into the
15 inch thickener where the sludge settles out and the backflush water over-
flows the weirs to be pumped (by P6) back into the process stream at the flash
mix unit or drained out of the system. Sludge in the thickener is raked to-
wards the cone shaped bottom of the tank where it must be periodically manually
discharged into a collecting vessel. (A sludge pump would, in a full-scale
system, deliver this material to a vacuum filter, incinerator, or other means
of disposal and/or recycle).
The third set of air valves associated with each magnetic separator are
the chemical backflush valves (V9 and V10 for HGMF I and Vll and V12 for SALA-
HGM!®II). These valves open only during chemical backflush which occurs once
every (x) cycles as determined by the manually set counter (chemical backflush
occurs only on the final step of the counter prior to its resetting, i.e.,
step 0). The (Parcolene) rinse used for chemical backflush is circulated in
a loop by pump P4 through the demagnetized matrix as with normal backflush,
but with much lower velocity and for a longer interval. Normal backflush
follows the chemical rinse before the next filter cycle is again initiated.
19
-------
SECTION V
TESTING SITES
During the span of the present modifications (4 and 5) of EPA Contract
#68-03-2218 testing has taken place at several different sites. Effort I was
spent at SALA using a slightly modified version of the 1 gpm continuous pilot
plant described in EPA Report 600/2-77-015, while Effort II was completed by
work done with the newly constructed mobile pilot plant trailer on-site at the
Cambridge Cottage Farm Chlorination and Detention Facility and at the MDC
Columbia Park Headworks in Dorchester. All testing was completed between
October of 1976 and the same month of 1977, with mobile trailer construction
and debugging taking a large portion of this time span, from late December to
mid June.
TESTS AT SALA MAGNETICS
Effort I testing was essentially completed with the 1 gpm continuous
pilot plant upon the collection and running of two storm activations from
Cottage Farm. In the first, on October 20, 1976, 400 gallons of unscreened
CSO was trucked to SALA in 55 gallon barrels and tests were performed which
included the evaluation of the maximum permissible matrix loading over a range
of seed-to-solids ratios and a preliminary static versus flash mixer compara-
tive run.
Dry weather prevented further activations until December 7, 1976 when
300 gallons of CSO was once again carried to SALA for pilot plant evaluation.
Further seed-to-solids runs were carried out on this water, as well as tests
designed to determine the maximum practical flow flux rate at a given condi-
tion of solids loading and seed addition rate.
Due to the delay caused by infrequent storms, and the expectation that
collection and running of storm water would be more efficient and relevant on-
site than after aging with transport and set up times, it was decided that
further Effort I specified testing be deferred until mobile trailer unit oper-
ation under Effort II.
COTTAGE FARM WORK
After many months of design and construction, the trailer-mounted pilot
plant was moved to the Cottage Farm facility on May 19, 1977 for testing
and debugging. (See Figure V-l.) This was accomplished soon thereafter at
20
-------
FIGURE V-I COTTAGE FARM CHLORINATION AND DETENTION
FACILITY, CAMBRIDGE, MASSACHUSETTS
-------
which time the system was made ready to accept storm water for Effort II test-
ing. Once again the weather did not cooperate and a storm testing was not
possible until July 11, 1977. By this time, the trailer had already been
scheduled to be moved to Columbia Park Headworks where raw sewage is continu-
ously available as well as CSO during rain storms. This storm provided an
opportunity to test the workability and capabilities of the pilot plant system
under real storm water running conditions.
The above static flow rate CSO run showed that suspended solids and tur-
bidity were significantly lowered (^ 75-80%) with this on-site process, using
an admittedly roughly controlled operating procedure. However, this first
run was a necessary prerequisite to the tests that followed as a system was
developed during the run for monitoring and checking all pilot plant parameters
and physical components.
COLUMBIA PARK TESTING
On July 18, 1977, the pilot plant trailer was moved to a site at the
Columbia Park Headworks in Dorchester. This was done so that some of the less
critical tests could be completed, using raw sewage, in the intervals between
storm flows. The new site was chosen on the basis of the expected storm water
dilution factor as compared with the Cottage Farm activation character, the
adaptability of our system to their available facilities, and the relative
security of the location.
Unfortunately, essentially all of the MDC-run Boston sewerage system is
beset by the problem of tidal inundation, and the resulting periodic fluctu-
ating salinity of the incoming waste water. This problem, originally thought
to be of minor significance, turned out to be a major obstacle to a systematic
comparative study of parameters related to seeded water treatment of CSO and
raw sewage, as the system is basically one of chemical coagulation, a technique
sensitive to such changes in water salinity. Fluctuations in raw sewage con-
ductivity of between 2000 and 20,000+ ymhos/cm occur regularly with the tides,
causing significant changes not only in the dosages required, but also in the
very nature of the reaction which takes place in destabilizing the suspended
matter. Because flocculation is, under normal conditions, dependent in a com-
plex way upon water characteristics which are difficult to monitor or adjust
for, the additional imposed variable of a dynamic salinity scheme, dependent
upon relative tidal height and raw sewage and storm water volumes, Makes pre-
cise system regulation and maintenance of optimum running conditions nearly
impossible. For this reason, tests that are best performed using a stable
waste water input for comparative analysis of various pilot plant parameters
were difficult to interpret in light of the constantly changing conductivity
measured in this case. Even storm water runs were affected by this overlayed
factor (although to a somewhat lessened extent) due to the large CSO dilution
volumes involved.
At Columbia Park, all remaining tests were completed and several addi-
tional experiments were performed. Testing included optimal backflush tech-
nique, backflush velocity, time and pressure requirements, matrix cleaning
22
-------
and sludge evaluations, minimum residence time needed, optimum chemical
addition sequence, and the pilot plant profiling of three storm water
flows. A further flash versus static mixer comparative test, pH jar tests,
and the trial of a variety of different flocculants were also performed during
this stage at Columbia Park Headworks. Most of the above testing (excluding
the three storms profiled) was performed using raw sewage.
23
-------
SECTION VI
DISCUSSION OF RESULTS
SUMMARY OF RESULTS
The overall results obtained in this study were highly satisfactory and
shed much light on the advantages of high gradient magnetic separation tech-
niques as applied to CSO. Several minor difficulties were overcome, and the
work led to improvements in both technique and system design to better suit
this particular application. The feasibility of the seeded water treatment
system for handling dynamic situations such as are encountered in integrated
wet and dry weather installations was clearly demonstrated.
The practicality of using seeded water treatment variable flows and mul-
tiple separators to handle storm surges while maintaining process effective-
ness was shown by our storm profile tests. The importance of adequate seed
in the system was made clear in these and other matrix loading-type tests.
Best chemical addition techniques were reconfirmed as well as the assertion
that pH adjustment is not necessary for raw sewage or CSO with this treatment
method. Backflush and matrix cleaning requirements were explored and optimal
techniques proposed. Preliminary examination of the generated sludge tends
to indicate that it is easily handled, dried or recycled for possible seed
recovery and reuse. The static mixer shows some promise in the system as an
energy saving and maintenance-free chemical disperser. The effects of in-
creased flow velocity at a certain set of conditions were determined to be
rather mild up to a flow flux rate of about 200 gpm/ft^, higher than was ex-
pected. These results and others have helped us to better project costs and
optimal designs for larger size systems.
OVERVIEW OF TEST CONDITIONS
In the previous study; the seeded water treatment technique was shown to
be highly effective in removing on the order of 90+% of all suspended solids,
turbidity, BOD, fecal coliforms, etc. from CSO and raw sewage. These tests
were carried out after the collection and storage of the waste water (usually
a lapse of at least one day) by drawing samples from large, well-mixed ves-
sels all of exactly the same age. In some instances, sample alteration
(e.g., bacterial growth due to warm temperatures inside the building, etc.)
did take place (for example, when it was necessary to run for more than one
day with the same sample). These changes usually involved a straightforward
conversion of available BOD to biomass, and only minor changes were noticed
in the chemical balance of the water (i.e., relative doses of chemicals needed
to achieve proper flocculation did not change and little self-destabilization
was noticed over these short durations). The result of all this was that
24
-------
rather controlled conditions existed in the waste water stream entering the
pilot plant allowing optimization to be precise. Further, the time lapse
necessary before processing of the sample undoubtedly had some effect on the
chemical interactions taking place in the system.
On-site tests, however, proved to be much different. CSO, by nature, is
a highly dynamic waste water. The "first-flush" phenomenon, where the sewer
system is flushed of sediment laid down during dry weather by the initial
gush of rain water, can result in an order of magnitude jump in suspended
solids passing through the system in a very short time. Depending upon the
relative rate and source (i.e., rain, snow, melt-off) of flow, the time of
year and time of day (relative raw sewage flow), the duration of the storm,
and the frequency of recent precipitation, the CSO in a given sewer system
will vary tremendously. Local inputs of industrial wastes can also cause
periodic fluctuations in waste water chemistry serious enough to upset both
biological- and chemical-clarification-based treatment systems. Finally, CSO
varies with the individual sewer system. Each has its own unique character
depending on numerous factors among which are: the nature of the drainage
area included, the type and relative proportion of industry, population
density, climate, and sewer pipe age and construction. All this means that
most on-site tests tend to be more rigorous than batch sample tests, and this
situation is even further accentuated with the dynamic combined storm over-
flow waters. The fact that the SALA on-site tests were able to deal success-
fully with this situation is a major step in demonstration of the applicabil-
ity of the technology.
The majority of the on-site testing under this contract was done at
Columbia Park where the additional complication of constantly changing con-
ductivity of the waste water overlayed the above-mentioned variables. This
problem, along with other operational troubles, made testing at this site
difficult, necessitating the repetition of most tests. The outcome of this
was that progressive results showed considerable refinement as well as im-
provement in separation effectiveness. By careful monitoring of the system
and proper chemical dosing, essentially all process problems were overcome
and the technology was demonstrated as one capable of handling even the most
adverse of the dynamic surge flow circumstances.
BACKFLUSH TESTS
Optimal Backflush Technique
The object of the magnet backflush is to purge all solids trapped during
the filter cycle from the filamentary magnetic matrix to free up sites for
further magnetic separation and prevent buildups from occurring which could
lead to filter clogging. Several possible techniques were chosen for com-
parison including single-surge, air-water flush from the pressure vessel;
single-surge, water-only flush; pulsed air-water backflush from the pressure
vessel; single-surge, air-water backflush using independent (constant) air
source; and single-surge, air-water flush from the pressure vessel with a
drained matrix. Previous tests had indicated that air-water flushing tech-
niques could solve matrix build-up problems associated with oily waste waters,
25
-------
The tests performed here attempted to determine whether the optimum backflush
technique for CSO is similar to that for other applications and to further de-
fine the critical parameters involved. The tests were performed on raw sewage
using the following pilot plant running conditions:
Flow Rate: 5.3 £/min, 115 gpm/ft2 (SALA-HGMF®I)
Magnetic Field: 1.9 kG
Residence Time: 6 minutes
Backflush: 2.5 seconds, 1800 gpm/ft2 @ 35 psi
Chemical addition rates were as follows:
Alum: 125 mg/£
Polyelectrolyte: 1 mg/£ (Betz 1150)
Magnetite: 260 mg/£
The raw sewage had a suspended solids concentration of 110 mg/£, a turbidity
of 97 FTU, and a conductivity of 18,600 ymhos/cm at the start of these tests.
The pH was 6.8.
TABLE VI-1 BACKFLUSH TECHNIQUE COMPARISON
Solids
Concentration in Backflush Backflush
Backflush Technique Backflush Volume Solids
(g/D (I) (g)
1. Air-water backflush from 2.8 3.2 8.9
pressure vessel, decreasing
air pressure*
2. Water backflush only 2.3 3.7 8.4
3. Air-water backflush in pulses 4.9 2.5 12.3
from pressure vessel
4. Air-water backflush with inde- 7.8 1.6 12.4
pendent constant air supply
(50 psi)
•
5. Air-water backflush from 2.0 3.5 7.0
pressure vessel after
matrix drain**
Air-water flush with decreasing pressure was accomplished by re-combination
of air from the top of the pressure vessel with water from the bottom at a
point in the backflush line before the matrix.
*
Some solids lost in test.
26
-------
Table VI-1 shows the solids concentration in the backflush water, total
solids collected in the backflush and the total volume for each backflush tried.
It can be seen from this that of those methods tried the air-water flush with
constant 50 psi air pressure appears to have removed the most solids at the
highest concentration.
Backflush Velocity, Time and Pressure
Several tests were performed in order to determine minimum backflush
requirements for the system. The first of these was done using a simple 35
psi pressure vessel air-water backflush while varying the backflush time (and
therefore the volume of backflush water used). The backflush times used on
this treated raw sewage were 10, 5, 2.5, and 1 second with the 2.5 second run
repeated as a check on technique. Each test run was followed by an unsampled
run with a 10 second backflush to avoid residual build-up between tests.
Other pilot plant and chemical addition statistics were identical to those
previously found for the optimal backflush technique tests.
Table VI-2, below, shows the results of this test.
TABLE VI-2 BACKFLUSH TIME TESTS
Backflush Time
(seconds)
10.0
5.0
2.5
1.0
2.5
Solids Concentration
in Backflush
(gAO
0.94
1.6
3.0
14.0
2.7
Backflush
Vo lume
(£)
15.0
9.4
3.6
0.83
3.9
Backflush
Solids
(g)
14.1
14.8
10.8
11.6
10.4
As is to be expected, and can be visually observed in the pilot plant,
the vast majority of solids wash freely from the matrix in the very first
slug of backflush water. However, from the above table, it seems evident
that a significant amount of suspended solids in addition to the initial
slug are removed with increased backflush duration (at least with the back-
flush technique being used). The difference between the backflush solids
recovered in the 10 second and 5 second flushes is essentially unchanged
while the 2.5 and 1 second flushes seem to have been somewhat less effective.
Due to the lack of control samples, however, these tests are not conclusive
as differences observed could conceivably be attributable to changes in the
raw sewage itself. Also, an even shorter backflush may result in superior
flushing effects when the air-water flush with constant pressure is used.
A second, more involved series of tests was run to further try to pin-
point the minimum backflush parameter values. In this experiment, the air-
water flush with constant 50 psi air pressure (optimal technique) was used
27
-------
CO
20
16
w
I12
EC
CO
u
-------
1.5h
NJ
VO
o
I—I
H
Si. ol
o
C/3
w
1-1
pq
W
H
.51
p
W
PS
CO
£ 7 second flush
0 5 second flush
• 2.5 second flush
X 1 second flush
10
15
20 25
PRESSURE VESSEL (psl)
i.
35
FIGURE VI-2 BACKFLUSH EFFECTIVENESS
-------
TABLE VI-3A BACKFLUSH PRESSURE AND DURATION DATA
u>
o
Sample
Type///
Feed 1
Treated 1
Floe Tank 1-1
1-2
1-3
1-4
Flush 1-1
" 1-2
" 1-3
" 1-4
Treated 2
Floe Tank 2
Flush 2
Feed 2
Treated 3
Floe Tank 3-1
3-2
" 3-3
3-4
Flush 3-1
11 3-2
" 3-3
" 3-4
Treated 4
Floe Tank 4
Flush 4
Flush
Condition
psi/seconds
35/7
35/2.5
35/1
35/5
30/5
25/7
25/2.5
25/1
25/5
20/5 *
Concen-
tration
(mg/£)
92
61
83
153
230
337
829
1900
6010
1320
46
206
1640
135
80
263
258
1240
3120
5430
1450
80
300
2850
Total Average Floe Treated Average Flushed/
Flushed Tank Solids Solids Lost Filtered Filtered
Solids (g) (g/cycle) (g/cycle) Solids Solids Ratio
(g/cycle)
-
6.4
-
21.1
14.9
16.3
17.4
18.0
4.8
21.6
18.5
i
8.4
I 27.3
J
17.0
18.7
15.5
15.4
8.4 :
26.2* 37.8
14.7
1.0
1.1
1.2
1.2
16.8
1.1
18.9
.90
1.0
.82
.81
29.4
.89
Total flushed solids after 6 minutes filter cycle instead of 5 minutes
-------
TABLE VI-3B BACKFLUSH PRESSURE AND DURATION DATA
Sample
Type///
Feed "3
Treated 5
Floe Tank 5-1
5-2
5-3
5-4
Flush 5-1
" 5-2
" 5-3
" 5-4
Treated 6
Floe Tank 6
Flush 6
Feed 4
Treated 7
Floe Tank 7-1
7-2
7-3
7-4
Flush 7-1
" 7-2
" 7-3
" 7-4
Feed 5
Treated 8
Floe Tank 8
Flush
Condition
psi/seconds
15/7
15/2.5
15/1
15/5
10/5
30/7
30/2.5
30/1
30/5
Concen-
tration
(mg/£)
71
56
367
337
330
340
2220
5180
2330
3230
60
347
3510
56
66
245
280
276
1050
2880
5490
68
67
200
Total Average Floe Treated Average Flushed/
Flushed Tank Solids Solids Lost Filtered Filtered
Solids (g) (g/cycle) (g/cycle) Solids Solids Ratio
(g/cycle)
•
23.3
20.7
3.1
22.6
19.3
'
16.8
19.6
15.5
5.9
36.1
-
6.3
36.4
•
6.9
28.0
:
7.0
21.0
30.2
.77
.69
.10
.75
30.1
.64
21.1
.79
.93
.73
14.0
-------
throughout, while varying the pressure vessel pressure and the total flushing
time. This was done in a systematic manner, using pressure vessel pressures
of 35, 30, 25, 20, 15, and 10 psi and backflush times of 7, 5. 2.5, and 1
second. The 20 and 10 psi runs were done with a 5 second flush only, while
all the others were repeated for each of the above flush times. An uncollected
7 second flush was performed in between each test run to ensure maintenance
of a clean matrix. Not only were backflush waters collected but feed water,
flocculator water, and treated water were all sampled and analyzed periodically
Figure VI-1 gives the running conditions used and shows the relation of back-
flush volume to backflush pressure for different backflush durations.
Table VI-3A and 3B show the results obtained in these tests. From these
data, the average flushed suspended solids to filtered suspended solids ratio
(backflush effectiveness) has been calculated and is plotted in Figure VI-2.
This figure shows a clear trend of increasing backflush effectiveness with
increased air charge on the pressure vessel over the interval tested.
Matrix Cleaning Requirements
The cleaning requirements for the matrix (in addition to normal back-
flush) were investigated to determine what if any special attention needs
to be given to this possible problem. CSO and raw sewage have been shown to
be relatively free of the grease and oil which have caused some problems in
other magnetic separation applications. Preliminary results have indicated
that although minor build-ups may occur, these easily can be eliminated by
better backflushing techniques and periodic chemical (caustic) rinsing. In
tests done on a similar application, a number of cleaning solutions were
compared for overall effectiveness. Several of these (one in particular) were
found to be extremely effective in maintaining a "like new" appearing matrix
with rather infrequent, low-velocity rinsing. As prolonged testing over a
period of months has not as yet been done, conclusions about a permanent
solution for matrix maintenance can not presently be made with certainty.
The two matrix examinations performed during this study to determine cleaning
requirements showed that build-ups were practically nonexistent when compared
to those observed in the past in other magnetic separation applications.
Experience has shown that these problems can be eliminated by the use of a
periodic chemical rinse in addition to an air-water back and forward flushing
technique. The chemical rinse frees up the trapped particles by dissolving
the greasy slime surrounding them and the reversal of flush direction midway
through provides equally high shear forces at both ends of the matrix* This
combination has been shown to maintain the matrix in a very clean state
virtually indefinitely. Further tests should be done, however, using the
present pilot plant (as well as the next, scaled-up version with modified
plumbing for reverse backflush) to determine frequency and duration require-
ments of such flushing techniques on CSO effluent. The chemical flush system
is already incorporated within the present system and needs only to be
tested over extended time intervals for detailed evaluation of its effective-
ness on CSO.
32
-------
CHEMICAL ADDITION TECHNIQUES
A number of jar test-type tests were performed to investigate the effects
of residence time on flocculation, to determine the optimal chemical addition
sequence to promote good flocculation, and to compare previous test results on
optimal pH for clarification in seeded water treatment with newly generated
data done on site. These tests were all done with fresh raw sewage or CSO
using a standard jar test procedure. A turbidity measurement after 5 minutes
of settling provided all quantitative results, while visual observation of the
flocculation as it occurred gave added input to the analyses. Although the
quantitative measurements cannot be translated directly into what we can expect
from magnetic separation, these tests do give us a relative baseline from which
the most likely to succeed techniques can be determined.
Chemical Addition Sequence
Table VI-4 gives the variety of chemical addition sequences tested and
results obtained. The test procedure was designed to give all samples a total
of 4 minutes of mixing before clarification was begun. This mixing was broken
up into 21/2 minutes at 110 RPM and 11/2 minutes at 60 RPM, simulating the
flash mix-flocculator situation to some extent. All samples were of one liter
size.
TABLE VI-4 CHEMICAL ADDITION SEQUENCE
Test #
Addition Sequence
1 Magnetite + Alum - Polyelectrolyte
2 Alum + Polyelectrolyte + Magnetite
3 Polyelectrolyte - Alum + Magnetite
4 Alum + Polyelectrolyte - Magnetite
5 Polyelectrolyte + Magnetite - Alum
6 Alum - Polyelectrolyte + Magnetite
Floe Characteristic
large
medium-fine
stringy
very fine
stringy
large
Turbidity(FTU)
4
5
18
4
15
4
33
-------
In test #1, magnetite and alum were added together and mixed for two minutes
before polyelectrolyte was added. After 30 seconds more, the mixing rate was
turned down to 60 RPM. In test #2, all chemicals were added simultaneously,
while test #3 was the inverse of test #1. Similarly, in tests #4 and #5,
polyelectrolyte was added simultaneously with alum and magnetite, respectively,
and mixed for two minutes before the addition of the remaining chemical.
Finally, in test #6, alum was added first followed two minutes later by poly-
electrolyte and magnetite.
These results confirm what has been known intuitively and demonstrated
scientifically, and what experience has indicated in the past: Firstly, when
polyelectrolyte is added before the alum, floes are stringy, weak and incom-
pletely formed (tests #3 and #5). Secondly, alum must be added to the waste
water sufficiently in advance to permit the coagulation process to be completed
(destabilization) before polyelectrolyte bridging is initiated. In tests #1
and #6, this optimum sequence was present, while in tests #2 and #4, the floes
were finer and took longer to settle as polyelectrolyte and alum were added
simultaneously.
The theories of coagulation and flocculation are complex and do not fully
explain what may happen in any specific instance or with a particular type of
water. Basically, coagulation is a process in which particles suspended in
water are destabilized, enabling their transport to one another where they
may attach if the right conditions exist. The term "flocculation" is usually
used to describe the transport and eventual bridging of destabilized particles
by a polymer.
Destabilization can occur in waste water by adsorption of a coagulent
species or, more rarely, by enmeshment within hydroxide or carbonate precipi-
tants. In the case of adsorption, as in seeded water treatment, polymers are
the chief agents involved in the adsorption process. They may be present
naturally in the waste water (i.e., biopolymers), they may be added to the
water (i.e.., synthetic polyelectrolytes) , or they may be generated within the
process from salts added to the waste water (i.e., alum). A combination of the
latter two conditions plays a major role in the chemical addition technique
used with the magnetic separation system.
The addition of alum (Al2(S04>3 x 18 H20) to water results in hydration
of the complex and the creation of acidic aquocomplexes such as Al(H20)g^+j
chemically known as ligands. This can then be further hydrated to hy,droxo-
metal polymers depending on how close their concentration in the water is to
its solubility limit. (This limit is a function of pH.) These polymers are
then responsible for precipitation of colloidal particles in waste water in
the form of metal hydroxides. The necessary alum concentration is dependent
on the amount of colloid present in the waste water (stoichiometric dependence),
Two mechanisms have been proposed for destabilization using alum. When
the pH of the water is below the isoelectric point (the pH at which positively
and negatively charged particles are equal) for alum (^ pH 6), positively
charged polymers are absorbed by charge neutralization with the dominant •
34
-------
negatively charged colloids found in waste waters. Above the isoelectric point,
as is normal with seeded water treatment, anionic polymers are thought to func-
tion by adsorption and interparticle bridge formation. Characteristic of alum
destabilization is its dependency on colloid concentration. When this concen-
tration is low, even though alum dosage is less, detention time for complete
destabilization increases significantly. The reverse is true for high colloid
concentration. Detention times required are low but necessary alum doses are
increased. In seeded water treatment, the addition of magnetic seed tends to
act as a coagulant aid, especially in low solids situations, thus increasing
the reaction rate and decreasing the necessary detention time.
Besides pH and colloid concentration, alkalinity of waste water is also
of importance in water treatment because, when present, it acts as a buffer
preventing pH change with alum dosing (alum is an acid). Depending on the col-
loid concentration present, the relative degree of alkalinity may be an advan-
tage or a hindrance to destabilization. In general, when a high solids loading
is present, high alkalinity is a disadvantage to alum destabilization as it
holds the pH near neutral (for sewage and CSO) where hydroxometal polymers are
not so highly charged and therefore higher alum doses are needed. When low
colloid concentrations are present the situation is reversed, as high alkalin-
ity is advantageous for destabilization of low colloid suspensions. In this
case, the alkalinity holds the pH at a near neutral point where high alum doses
result in the enmeshment of particles in an aluminum hydroxide precipitate
known as a "sweep floe." When both alkalinity and colloid concentration are
low, pH depression due to the alum can inhibit this phenomenon. In this case,
either additional colloid (to speed reaction rate) or alkalinity (to allow
sweep floe formation) must be added. In CSO and sewage treatment, conditions
intermediate to those described above are the most common and (especially with
CSO) the water may move from one extreme to the other in a short time.
The above discussion considers alum destabilization only, and does not
take into consideration the modification of final water chemistry with synthet-
ic polyelectrolytes. Soluble phosphorus concentration in the waste water is
a further consideration in alum dosing as alum reacts preferentially with this
substance, limiting the amount of metal available for solids destabilization.
This is not a minor consideration, as domestic sewage generally requires about
60 mg/£ just to precipitate out the phosphorus present.
Depending on the charge on the ionizable group of a polyelectrolyte, it
can be classified as being anionic (negatively charged), nonionic (no ionizable
group present) or cationic (positively charged). This quality, along with
others such as its molecular weight and the degree of branching, affect the
polyelectrolyte's qualities and usefulness in any particular application.
Water characteristics affecting polyelectrolyte performance include pH, (the
net charge changes with pH) and the concentration of divalent cations in solu-
tion. Polyelectrolytes are generally thought to act as bridges, binding to-
gether destabilized particles or groups of particles (e.g., alum floes) with
their long chain structure. When used alone they may, by their own nature,
result in destabilization, but in seeded water treatment they are much more
effective after alum destabilization has already occurred as they facilitate
generation of large, uniform and strongly bound agglomerations amenable to
separation. Precautions must be taken when using certain polyelectrolytes
35
-------
as some will tend to restabilize (as, for example, with a high dosage of cat-
ionic polyelectrolytes on a system lacking divalent cations), while some are of
limited use when their molecular weight is less than 1 million (i.e., low mo-
lecular weight anionic polyelectrolytes may be unable to overcome the potential
energy barrier between negatively charged particles), and most polyelectrolyte
floes will tend to rupture if mixed too strenuously.
The collision and final attachment of colloidal particles (particle trans-
port or flocculation) is successful only after destabilization has occurred.
The destabilization stage takes only up to a few seconds to be completed, pro-
vided conditions are favorable. While vigorous agitation promotes uniform
coagulent dosage, water pH, and alum adsorption, the resulting break up of
colloidal particles in the sewage or other waste tends to increase the surface
area exposed, thus demanding higher alum concentrations. Likewise, synthetic
polymers rupture easily and have a slower rate of adsorption due to their large
size, making moderate mixing and longer retention times more effective in the
system. It should be noted that stirring will only increase the aggregation
rate of particles greater then 1 y size* with the exception of viruses which
in magnetic separation have been shown to actively seek and adsorb to the mag-
netite seed surfaces, ensuring their efficient removal.
Flocculant Evaluation
In previous studies on the use of high gradient magnetic separation seeded
water technique for treating CSO and raw sewage, and for most of the present
study, flocculant type has not been considered as a possible system variable.
Alum is a well proven coagulant and although ferric chloride, ferric sulfate,
ferrous chloride, ferrous sulfate, and lime are also used in similar applica-
tions, preliminary subjective jar tests have shown alum likely to be the most
universally applicable and cost-effective coagulant for seeded water treatment
of CSO and raw sewage. Further jar test-type studies are needed for every
different waste stream, along with consideration of chemical availability and
cost. This applies even more strongly to the optimal polyelectrolyte chosen
for a particular seeded water treatment application.
A jar test comparison of synthetic polyelectrolytes was attempted on a
single batch of raw sewage at the close of the present study in order to try
to determine what (if any) polyelectrolyte type might prove most effective in
this instance (normally high, varying conductivity due to salt water inunda-
tion) . Nine different Hercules polyelectrolytes were compared along with the
Betz 1160 used for most testing reported herein. The polyelectrolytes tested,
their ionic natures, and the summarized results (subjective floe descriptions
and turbidity measurements of the supernatants) are given in Table VI-5. In
these jar tests, the conductivity of the sewage was 6000 ymhos/cm. All 1
liter samples were treated with 100 mg/£ alum, '^.25 g/£ magnetite, and after
two minutes of mixing at 120 RPM, 1 mg/£ of polyelectrolyte. After 30 seconds,
the mixing rate was reduced to 60 RPM for the remaining 1 1/2 minutes before
the settling period of 5 minutes.
For particles smaller than 1 y, Brownian motion is the dominant force.
36
-------
TABLE VI-5 POLYELECTROLYTE COMPARISON
Polyelectrolyte Ionic Floe Turbidity
Test # Flocculant Character Size (FTU)
1 Betz 1160 Cationic Medium 4
2 Hereofloc 849 Med. High Cationic Medium 4
3 Hereofloc 855 Med. High Cationic Medium-fine 3
4 Hereofloc 874 Med. Cationic Extremely fine 4
5 Hereofloc 815 Med. Cationic Extremely fine 3
6 Hercofloc 812 Low Cationic Extremely fine 3
7 Hercofloc 827 Nonionic Very fine 3
8 Hercofloc 818 Low Anionic Huge 4
9 Hercofloc 831 Med. Anionic Very large 2
10 Hercofloc 821 High Anionic Medium-fine 5
It was immediately apparent from these jar tests that the majority of the
flocculants tried were not suitable (at least at the concentration tried) for
seeded water treatment. Most formed only tiny floes late in the mixing pro-
cess and settled very slowly. Experience has shown that for optimal high
gradient magnetic filtration of coagulated suspended solids, floes should be
on the order of from 1 to 4 millimeters in size and of compact, uniform shape.
This large size tends to lower magnetite demand by increasing the probability
that some seed is incorporated into every floe agglomeration. When the floes
get any larger than this, their tendency to rupture in the flocculator as well
as in the matrix makes them less easily handled. Also, extremely large floes
(such as were experienced with Herfloc 818) tend to drop out of suspension so
rapidly that not all turbidity is removed from the water with the bulk of the
solids and magnetic seed, making the small amount of remaining solids less
likely to contact a seed particle. Because of the mixing rate limit imposed
by the large-sized floes, the short contact time caused by the rapid settling
of most of the water volume cannot be extended.
In the tests described above, Betz 1160 and Hercofloc 849 seemed to work
very similarly while Hercofloc 831 was shown to be superior to any tried on
this effluent sample in rapidly removing the most turbidity. Turbidity dif-
ferences measured after 5-10 minutes of settling did not clearly differentiate
between the various polyelectrolytes' effectiveness, as all were able to
settle by the end of this time interval. However, the rate of floe formation
and the size and shape of these floes were clearly distinguishable and gave
the experimenter a good indication of how effective they might be. A bench
test magnet set-up might be more appropriate for obtaining a "supernatant"
37
-------
for turbidity measurement as this technique would take into account the seed-
to-solid distribution in the floes and the resistance to shear imposed by flow
past the matrix fibers, as well as relative clarification of the waste water
by the flocculant. Low color is also of importance in treated water but may
not necessarily coincide with best turbidity removal by a flocculant. Also,
before any one polyelectrolyte is completely discounted for a particular waste
water it should be tested over a wide range of concentrations (preferably at
least four orders of magnitude), since overdosing often can cause restabiliza-
tion of the colloidal particles. It is important to emphasize the necessity
of flocculant evaluation before other test work is attempted so that the best
possible magnetic separation conditions can be achieved. Unfortunately, this
was not done at the start of the present study, and apparently less-suitable
polyelectrolytes (Betz 1150 and 1160) were used throughout. Although the
preparation of a large number of stock solutions of different polymers is some-
what time consuming, the jar tests are easily run and the results may save
many hours of work later on in a test program. An extensive evaluation is, of
course, necessary for any location considering a permanent installation. Other
tests, such as zeta potential, may further help to determine (or maintain)
optimal flocculants and concentrations for the waste water stream.
Effects of Residence Time
As mentioned earlier, the detention time needed to achieve complete de-
stabilization is dependent on a number of factors including waste water colloid
concentration and alkalinity. Further, the total turnover time necessary for
magnetic separation can be expected to be somewhat less than that for chemical
clarification, as the final separation is accomplished magnetically rather than
by a passive settling of the flocculated particles. Contact detention times
for coagulent/flocculent/seed additives to the waste water are, however, very
similar to those encountered in chemical clarification flocculating chambers.
The use of a rapid mix unit to disperse chemicals followed by a somewhat larger
flocculator to provide contact time sufficient to produce aggregates of suit-
able size for removal is common to both processes, although the desired result
may be slightly different. While clarification demands the largest and fastest
settling floe, magnetic separation techniques strive to obtain a floe of medium
size with good shear resistance and magnetite distribution.
Some simple jar tests were performed to try to determine the minimum res-
idence time necessary for good floe formation in the raw sewage used. The
procedure used was as follows. Magnetite (^ .25 g) was first added tg the
1 liter samples while mixing with a paddle stirrer at 100 RPM. Alum (100 mg/£)
was then added and the timing begun. After a predetermined mixing time (shown
below) Betz 1160 (1 mg/£) was mixed in before turning the stirrer down to
60 RPM for the remaining time interval. These test conditions are given in
Table VI-6, following.
38
-------
TABLE VI-6 RESIDENCE TIME
Total Residence Flash Mix (Alum) Flocculator (Polyelectrolyte)
Test # Mixing Time Mixing Time (100 RPM) Mixing Time (60 RPM)
1 15 sec 15 sec -0-
2 30 sec 30 sec -0-
3 1 min 30 sec 30 sec
4 2 min 1 min 1 min
5 4 min 2 min 2 min
6 8 min 4 min 4 min
7 12^2 min 6^ min 6 min
8 20 min 11 min 9 min
For the first two tests shown, alum and polyelectrolyte were added simul-
taneously and the mixing rate was left at 100 RPM for the entire span. In all
others, polyelectrolyte was added approximately 15 seconds prior to turning
down the stirrer to 60 RPM.
Turbidity was taken on the supernatant after a settling period of about
10 minutes, and Figure VI-3 shows the resulting curve. These results seem to
indicate that for this raw sewage a minimum mixing detention time on the order
of from 3 to 5 minutes is necessary to facilitate the destabilization and
transport of a majority of the colloidal particles. After this initial con-
tact period, water quality improved only gradually with mixing time, and it
can be assumed that in most cases the additional capacities required for such
minor possible benefits would not be deemed desirable.
Observations of the on-going flocculation give some further insight into
the significance, as well as the limitations, of this test. For tests #1 and
#2, no floes were apparent. In tests #3, #4 and #5, respectively, the floes
observed were progressively larger, reaching optimal size (and the largest in
this experiment) when the mixing was done over a period of 4 minutes. This
represents a 2 minute flash mix period for alum destabilization, and 2 minutes
for polymer bridging and final agglomeration to take place. In tests #6, #1
and #8, the floes became progressively finer so that in #8 they were barely
visible. This is because shear forces at the higher mixing rate immediately
ruptured the largest floes formed after alum destabilization (i.e., after
2 minutes of mixing) and these floe fragments were then restabilized by the
secondary adsorption of the polymer to particle surfaces resulting in smaller
agglomerations. In other words, the longer the shearing, the smaller the
average floe is likely to be. However, turbidity obtained after settling con-
tinues to improve even with increased shearing because particle contact time
also increases, and the the shear created in this case is not violent enough
to cause break-up and restabilization of individual colloid particles, which by
39
-------
-e-
o
60
54
48
42
-ir
-36
32
£24
18
12
Jar Tests
10/19/77
Conductivity 6000
10 12 14 16
RESIDENCE TIME (min)
18 20
Alum 100 mg/l
Magnetite 250 mg/£
Betz 1160 1 mg/£
22
FIGURE VI-3 EFFECT OF RESIDENCE TIME ON TURBIDITY
-------
definition are too small to settle out. It is expected, however, that these
smaller floes would be less effectively removed by magnetic separation than
the larger agglomerations. In the trailer-mounted pilot plant, as it now ex-
ists, the residence time possible in the flash mix chamber at full-flow capac-
ity for the system is about 1 minute, and about 5 1/2 minutes for the floccu-
lator system. These times correspond well with the test results just outlined
although alum contact time may be borderline for some effluents. More complex
calculations can be made for sizing-up to larger scale and are available in
the literature.*
Effects of pH on Clarification
Jar tests were conducted on a sample of raw sewage at Columbia Park to
compare the effects of pH on the seeded water treatment of a waste water with
a high salt content to that determined on CSO in earlier studies. The conduc-
tivity of the sewage at the time of these tests was 16,300 ymhos/cm, near to
levels generally reached at high tide.
For these tests, 1 liter samples were taken and their pH altered to a
known (measured) value. Although alkalinity measurements were not available,
it is likely that substantial amounts existed as some buffering effects were
evident. Figure VI-4 shows the turbidity of the supernatant after a standard
jar test procedure for the pH-adjusted sewage samples. It can be seen that,
as in previous tests, the optimal clarification with an alum/magnetite/poly-
electrolyte flocculation technique was at natural pH (6.74), with floccula-
tion dropping off rapidly below pH 6 and only gradually deteriorating with
waters of a more basic pH. At natural pH, floes were large and distinct and
they settled very rapidly compared to the tests where the pH was above 7,
showing finer and slower settling agglomerates as pH increased. Tests with
a pH below 6.5 were progressively more and more cloudy.
MATRIX LOADING TESTS
A matrix loading-type test proved valuable in helping to delineate physi-
cal phenomena occurring within the seeded water treatment process stream.
Such tests are done by sampling a single cycle at regular intervals to observe
the relative rapidity with which magnetic separation efficiency drops off
*Camp, T. R. and P- C. Stein. Velocity Gradients and Internal Work in Fluid
Motion. J. Boston Soc. Civ. Eng., 30: 219-237, 1943.
Camp, T. R. Flocculation and Flocculation Basins. Trans. Amer. Soc. of Civ.
Eng., 120: 1-16, 1955.
Hudson, H. E., Jr. Physical Aspects of Coagulation. J. Amer. Water Works
Assoc., 57: 885-892, 1965.
Hudson, H. E., Jr. and J. P. Wolfner. Design of Mixing and Flocculation
Basins. J. Amer. Water Works Assoc., 57: 1257-1267, 1967.
41
-------
H
M
P
M
Pi
H
30
27
24
21
18
15
12
Alum 100 mg/£
Betz 1160 1 mg/£
cloudy
Jar Tests
10/5/77
Conductivity 16,300
large
and clear
fine floes
10
pH
FIGURE VI-4 EFFECT OF pH ON TURBIDITY
-------
under the given conditions. Thus, by varying such things as seed concentra-
tion, flow velocity, matrix design, or magnetic field, limits can be deter-
mined by observing achievable cycle times and breakthrough characteristics
for a given set of circumstances.
Seed-to-Solids Ratio
During Effort I of this study, tests were performed on the 1 gpm continu-
ous pilot plant in order to attempt to determine the effects of varied seed-
to-solids ratios on the duty cycle, and the practical matrix loading before
breakthrough. In the first test series, seed concentrations of 100 mg/£,
250 mg/£ and 500 mg/£ were used on a CSO having a suspended solids concentra-
tion of approximately 100 mg/£. These yielded seed-to-solids ratios of 1:1,
2.5:1 and 5:1 for runs of 20 minutes each shown in Figure VI-5 as a plot of
suspended solids versus time into the cycle. This figure shows that for a low
seed-to-solids ratio (i.e., 1:1), breakthrough occurred almost immediately
while for runs with progressively more seed added, stable separation continued
for longer periods before any deterioration in filtration efficiency was
noticeable. Further, in the run with a 1:1 seed-to-solids ratio, even initial
separation was not up to par with that accomplished in the runs where more
seed was used, hinting that perhaps the magnetite itself was acting as a
"filter aid" in those instances.
The 2.5:1 seed-to-solids run had an effective duty cycle of ^15 minutes
or, assuming a 5-10 seconds total for backflushing operations, a net 99%
filter cycle function. No outer limit is shown for the 5:1 ratio as samples
were not collected after 20 minutes of filter time in that test.
A second CSO batch was run using seed concentrations of 135 mg/£, 200 mg/£
and 270 mg/£ yielding seed-to-solids ratios of 1.3:1, 1.5:1 and 2.8:1, re-
spectively. Figures VI-6 and VI-7 show these results plotted as percent
removals of suspended solids and turbidity versus time into the filter cycle.
The similarities between these curves and Figure VI-5 are immediately evident.
Once again, the run in which the seed-to-solids was lowest (1.3:1 in this case)
showed a deterioration in filtration efficiency relatively early in the cycle,
while those runs which had higher magnetite loadings maintained excellent re-
sults for progressively longer times. Thus, it has been clearly demonstrated
that relative magnetite concentration is a critical limiting factor in deter-
mining effective duty cycle. In addition, especially where feed solids are
low, magnetite may actually aid in flocculation by increasing the frequency
of collision of destabilized particles.
At first glance it may seem somewhat contradictory that cycle length is
increased with increased total matrix load (high seed-to-solids ratios), but
this is a logical result of the magnetic forces involved at the site of cap-
ture. If the materials being separated were 100% highly magnetic in nature,
essentially the entire canister volume surrounding the matrix material would
be available for their retention. This is because when the particles being
trapped on the surface of the magnetized matrix are themselves highly magnetic
throughout, the net gradient present on the new surface formed by their attach-
ment decreases only gradually with increased layer thickness (and buildup can
occur effectively). In this case, each new layer of magnetic material
43
-------
60
0
cn
T3
•H
O
CO
OJ
ft
W
3
CO
BO
7O
SO
SO
3O
SO
1 O
a.s:"u
Time into Cycle (rain)
FIGURE VI-5 SUSPENDED SOLIDS VERSUS TIME INTO CYCLE
FOR VARIOUS SEED-TO-SOLIDS RATIOS
-------
•a
•H
I-H
o
u
3
H
100 T
90
80
70
60
50
40
30
20
10
135 mg/J, magnetite
6 mg/J. solids
(1-3:1)
FIGURE VI-6 SEED-TO-SOLIDS RATIO:
SUSPENDED SOLIDS\VARIATION
IN CYCLE
0
100
8
12
16
90 -
80
70
60
50 •
40
30
20
10 -
270 mg/J.
magnetite
k!33 mg/£ solids \96 mg/J. solids
(1.5:1) \ (2.8:1)
20
135 mg/J! magnetite
106 mg/£ solids
(1.3:1)
157 FTU Turbidity
200 mg/£ magnetite
133 mg/J, solids
(1.5:1)
162 FTU Turbidity
^270 mg/£ magnetite
96 mg/£ solids
(2.8:1)
155 FTU
Turbidity
FIGURE VI-7 SEED-TO-SOLIDS RATIO:
TURBIDITY VARIATION
IN CYCLE
0
12 16 20 24
Time in Single Cycles (minutes)
28
45
-------
concentrates the magnetic flux nearly as efficiently as the matrix fibers in a
clean matrix, being limited only by shearing forces caused by flow or irregu-
larities in the layers themselves.
However, in seeded water treatment, nonmagnetic solids make up a large
proportion of the total volume to be held within the matrix. Here the floe
as a unit, although containing some magnetite, is considerably less magnet-
ically susceptible and therefore is not only held in the matrix with much less
force but also fails to maintain the strong gradients present on the matrix
fiber surfaces. Thus, the canister can no longer fill to volumetric capacity
but is limited by the relative proportion of magnetite in the floes as to the
layer thickness it can successfully hold before shear forces exceed magnetic
forces. Flow velocity, magnetic field, and floe shear resistance also can
enter into the picture as this limit is approached, suggesting a need for
multiple compensation schemes to Increase duty cycle and magnet efficiency.
Matrix Design
Matrix loading tests also were used to attempt to compare two slightly
different matrix designs. The two matrices used differed only in the relative
proportion of .045 inch and .156 inch spacers between their randomly cut,
aligned standard metal disks. In matrix #1, 4% inches of the length was
packed with .156 inch spacers and the remaining 3 inches with .045 inch spacers
while in matrix //2 the situation was reversed, with 3 inches of .156 inch
spacers followed by 4% inches of .045 inch spacers.
Figures VI-8 and VI-9 show plots of pilot plant effluent suspended solids
versus actual time into the filter cycles and versus matrix loading (calcula-
ted weight of sludge accumulated in the matrix) for tests run on the two
matrices. The dashed lines in Figure VI-8 show solids levels in the treated
stream after a single flush. Figure VI-8 also shows incoming feed suspended
solids on samples taken during these runs, and in fact documents the first
flush phenomenon that occurred during this storm run. Matrix #2 was run
first, and the feed suspended solids started out at '^ 90 mg/£, a normal dry
weather flow. But, close to 20 minutes into that test run cycle, the effects
of the earlier drizzle and the following rain began to be felt as the
suspended solids in the feed rose to ^ 200 mg/£ by the end of the run. After
changing matrices (^1 hour), the feed solids seemed to have leveled off at
'^ 200 mg/£, and the second matrix loading test was begun. As in the earlier
run, the incoming feed suspended solids reacted to increased CSO flofir right
in the middle of the run, this time jumping to over 550 mg/£ as the main rain
water flow cleaned the streets and sewers. Conductivity changes due to tidal
influences were also present (varying between 8000 and 16,500 ymhos/cm),
although somewhat buffered by the rain volume.
Because of the varying input of the feed during these runs, and perhaps
because the two matrices tested were so similar in construction, the plots
shown in Figures VI-8 and VI-9 do not show any significant differences
attributable to matrix design. The fact that matrix #2 lasted longer before
breakthrough (Figure VI-8) than did matrix #1 is due simply to the difference
in rate of accumulation reflected by the storm water loading. Figure VI-9
shows this, as no significantly different trends in matrix loading are evident.
46
-------
Q
60
CO
Q
i-J
O
CO
Q
W
O
w
PH
CO
CO
700
600
500
400
300
200
100
10/26/77
P3 10 £/min
Alum 100 mg/£
Magnetite 300 mg/£
Hereofloc 831 1.5 mg/l
X Feed (Matrix 1)
• Feed (Matrix 2)
0 Treated (Matrix 1)
• Treated (Matrix 2)
FIGURE VI-8
SUSPENDED SOLIDS VS. TIME INTO
CYCLE FOR FEED AND TREATED
SAMPLES USING DIFFERENT
MATRICES
2 46 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
TIME IN CYCLE (min)
100
if 80
CO
O
Q
pa
p
&
w
60
40
20
^ X Matrix 1
, Matrix 2
FIGURE VI-9 MATRIX LOADING CURVES FOR
TWO MATRICES
**s
^x
*•
NC^\H,
50
100 150
MATRIX LOADING (grams)
200
250
47
-------
LIMITS OF FLOW VELOCITY ON MAGNETIC FILTRATION
Figures VI-10 and VT-11 show results obtained on the 1 gpm pilot plant
at Sala Magnetics, Inc. in tests meant to determine the practical flow flux
rate limits on a particular batch of CSO from the Cottage Farm facility. The
solids loading of this storm water was '^ 130 mg/£ and a magnetite concentra-
tion of 200 mg/£ was used (a seed-to-solids ratio of 1.5:1). Samples were
taken at 2 and 4 minutes into the filter cycle for each flow rate tested in
order to indicate the relative rapidity of loading effects likely to occur.
The tests were done using four specially prepared canister/matrix assem-
blies with identical matrix construction and length but with varied diame-
ters. These were interchanged during a single pilot plant run in which the
system was kept under a strictly controlled equilibrium. Since the process
rate was kept constant, the flow flux rate through each of the canisters was
proportional to its volume (which of course depends on the diameter). This
gave flow flux rates at the process flow chosen of 86, 124, 194, and 343
gpm/ft^ of matrix fiber. In this way not only were all recalibrations of
feed, filter and metering pumps avoided, but other system parameters such as
total mixing and residence time, and shear factors introduced by the plumbing
were kept completely constant.
Figures VI-10 and VI-11 show clearly that as flow rate is increased fil-
tration efficiency is decreased (slope of curves) and effective cycle time
is shortened (divergence of curves). These results are predicted and ex-
plained in the earlier discussion concerning matrix loading of (nonmagnetic)
floes containing magnetite, and their effect of decreasing the magnetic gra-
dient with increased layer thickness (build-up) around a matrix fiber. When
a flow rate is reached in which the shear forces at the capture surface ex-
ceed the magnetic holding force for most floes, they will simply be dragged
through. The faster the flow rate, the more rapidly this point will be
reached and the earlier in the cycle breakthrough will be evident.
For practical design considerations, a 100-120 gpm/ft^ limit generally
has been used for seeded water treatment systems. Although lower flow rates
do result in some increased efficiency under most conditions, these improve-
ments generally are not large enough to justify the extra capital expendi-
tures necessary for such systems. Thus, depending upon the discharge water
quality requirements, frequency of peak flow, character of the feed, and
economics of the individual situation, the best flow rate can be chftsen.
SLUDGE EVALUATION
In the past, SALA's efforts under the present contract have concentrated
on the evaluation of treatment effectiveness and process applicability to
the purification of storm water using the seeded water process. Prior to
undertaking this contract, in-house proprietary studies were conducted by
SALA on the subjects of seed and chemical recovery and regeneration. Because
of the small scale of the pilot plant existing at that time, sludge genera-
tion was inadequate for evaluation of the flow sheets devised.
48
-------
100
90
80
m
3 70
60
50
40
1 30
0.'
20
10
0
100
90
80
70
60
at 2 minutes
FIGURE VI-10 FLOW FLUX RATE VS.
SUSPENDED SOLIDS
H
y-l
o
0)
a;
50
40
30
20
10
100
200
300
at 2 minutes
FIGURE VI-11
FLOW FLUX RATE VS.
TURBIDITY
100
200
Flow Flux Rate
(gpm/ft2)
300
49
-------
A detailed investigation of sludge handling and seed recycle options is
beyond the scope of the present study, but should be included in the next
scale-up of the system where proper equipment and significant sludge volumes
can accurately simulate relevant conditions. Deferment of this work was in-
dicated as neither the required flushing design nor a proper seed-to-solids
ratio had been established at the outset for CSO. These factors are most
important for determining the character of the sludge to be processed. The
data presented below and in other sections of this report constitute a first
step towards a determination of the best method for sludge disposal with a
concommitant recovery of its economically valuable components.
Nevertheless, a number of measurements were taken during this study on
backflush waters and sludges generated during runs on CSO and raw sewage.
Backflush solids concentration depends on a number of factors related to the
backflush procedure used (flush volume and effectiveness), the concentration
of solids, and flow velocity. In other words, only that amount of solids
trapped in the matrix during a filter cycle can be concentrated in the flush
water, and the water used for backflush, in turn, depends on a number of
parameters such as backflush time and velocity, relative amount of air, etc.
Tables VI-7A and 7B display data that was collected over several different
test runs showing flush water solids concentrations (in g/£) tabulated against
the corresponding flush parameters used to obtain them. Also shown are the
seed-to-solids ratios used and the actual number of grams of sludge and liters
of backflush water collected for each flush recorded. These figures are given
in brackets below each flush water solids concentration datum in the tables.
Several relatively obvious things can be noted upon examination of this
collection of data. Although no attempt was made to maintain uniform running
conditions while generating these figures, basic correlations are still quite
evident when moving across a row or down a column, especially in Table VI-7B.
For a constant backflush tank starting pressure, backflush solids concentra-
tion increases with decreasing flush time as a result of the decreased dilut-
ing volume. The only exception to this is for the 1 second flush at 15 psi.
This is explained by the fact that in the present system the plumbing between
the magnet and the collection site is always pre-primed with clean water from
the end of the previous flush; for such a short, low-pressure backflush surge
the solids from the matrix never reach the surge tank where they can be
measured. Naturally, when only the initial solids "slug" at the front end of
the flush is sampled, solids concentration is even higher than the highest
shown in the table (14 g/£), but backflush effectiveness in maintaining a
clean matrix may suffer with such abbreviated flushes, thus making them
impractical.
In addition, it can be seen that solids concentrations generally increase
as one moves across a row from left to right (decreasing backflush tank start-
ing pressure) for a given flush time. This is once again due to the decreas-
ing diluting volume generated for the backflush by this lowered charge pres-
sure, and thus the same exceptional circumstance is evident for the row of
1 second flushes as with the vertical column mentioned previously. Other
minor inconsistencies are due to the variety of uncontrolled conditions present.
50
-------
A sample of the thickener underflow sludge also was taken, and the solids
concentration was determined to be approximately 28% (280 g/£). This high
value is, of course, mainly a result of the large proportion of magnetite in-
corporated. A seed-to-solids ratio of 3:1 was used in this case, giving an
expected sludge solids concentration of a.7%. The density of this sludge in
wet form (supernatant water decanted) was found to be 1.13 g/cc, and after
drying at 200°C it increased to 1.73 g/cc. Incineration at 500°C of a sludge
generated at a seed-to-solids ratio of 1.5:1 reduced the mass significantly
(on the order of 30% of the dry solids volatilized). However, neither of the
dried sludge samples tested was found to be able to sustain a flame on its
own (autogenous incineration not detected) .
As mentioned earlier, no detailed flowsheet for seed recovery and/or
sludge treatment and disposal has yet been defined. Until more information
is known about the character of magnetite sludge it is of little value to
propose flow sheets or work out material balances for such possible systems.
Measurements of this sludge's caloric value, its bacterial and viral content
(especially associated with the magnetite surfaces), its specific resistence
to filtration, its settling rate and particle size distribution, its compress-
ability, its degree of hydration, and other parameters critical to its hand-
ling and dewatering characteristics are necessary in order to be able to
properly formulate an experimental framework leading to the identification of
the most compatible process options. It seems likely that in order for seeded
water treatment to play any major role in the future, recycling of the magne-
tite must be accomplished within the process, and done so economically. The
seed recycle step itself is in its infancy, and the process must be fully
refined before sludge treatment methods are even worth considering.
STATIC MIXER VERSUS FLASH MIXER
Several attempts were made to try to establish the applicability of a
static mixer to the seeded water treatment system as a replacement for the
mechanical flash mixer now used for chemical dispersion. Because of the
changing salinity and other dynamic water parameters encountered at Columbia
Park, this test, like most of the others, proved difficult. Further, because
of the time intervals necessary to make the changeover complete, poor results
were difficult to distinguish from changed water character necessitating
chemical dosage adjustments. Also, the back pressure caused by the static
mixer's resistance to flow made actual chemical addition rates difficult to
determine. This combination of factors may have influenced the results nega-
tively to some extent.
In all tests attempted where no adjustments were made in the flocculator,
the static mixer produced generally poorer floes, and on the order of 5 to 15%
poorer separation was achieved as compared to the flash mixer under those
running conditions. As mentioned above, although these runs were repeated
on three different occasions, factors other than the type of mixer used could
have had a significant influence on these data, and no final conclusion can be
reached at this time.
51
-------
TABLE VI-7A FLUSH WATER SOLIDS CONCENTRATIONS
Flush Time
(sec)
TYPE OF BACKFLUSH USED
(35 psi Pressure Vessel)
Air-Water
from Pressure
Vessel
Constant
Air Pressure
(50 psi)
Pulsed
Air-Water
from Pressure
Vessel
Water Only
Ol
ro
10
7
5
3
2.5
.94
(2.6
14/15)
1.6
(2.6
15/9.4)
2.9
(2.6
11/3.8, 9/3.2)
14.0
(2.6
12/.83)
7.8
(2.6
12/1.6)
4.9
(2.6
12/2.5)
2.3
(2.6
8.4/3.7)
Key: g/£ fAish solids concentration
(seed-to-solids ratio used
total g/total volume in back
flush)
-------
TABLE VI-7B FLUSH WATER SOLIDS CONCENTRATIONS
BACKFLUSH TANK PRESSURE (psi)
Air-Water Flush, 50 psi Constant Air Pressure
Ul
U)
Flush Time
(sec) 35 32 30
10
7 .83
(.16
15/18)
5 1.3 1.6
(1.6 (1.6
18/14) 19/11)
3 3.5
(.88
27/7.6)
2.5 1.9
(1.6
16/8.6)
1 6.0
(1.6
17/2.9)
28 25 24
1.2
(1.1
17/14)
1.5
(1.1
15/10)
3.6 5.4
(1.7 (2.2
26/7.2) 37/6.9)
3.1
(1.1
19/6)
5.4
(1.1
16/2.9)
20
2.3
(1.1
20/9.2)
5.8
(2.4
29/4.9)
18 15 10
2.2
(2.1
23/11)
3.2 3.5
(2.1 (2.1
23/7) 19/5.5)
5.3
(2.5
26/4.9)
5.2
(2.1
21/4)
2.3
(2.1
3.1/1.3)
flush solids concentration
(seed-to-solids ratio used
total g/total volume in back
flush)
-------
When the flocculator mixing rate was increased (20-30%), the static mixer
results were comparable to those achieved with the conventional system at a
slower flocculator mixing rate. This seems to indicate that, by itself, the
static mixer (KOMAX) unit is not dispersing the chemicals adequately. The
question then becomes one of economics. This is not a new application for
static mixers, and there is little doubt that they will work if designed and
sized properly for the system. But there is some question at this time as to
whether or not these units will actually save on power and system cost for
the seeded water process. For some systems, the savings in system size and/or
maintenance costs may dictate a static mixer even if the overall economics are
not positive.
STOKM PROFILE TESTS
Several storms were actively run on the mobile pilot plant trailer at the
Cottage Farm and Columbia Park facilities in order to observe system perfor-
mance over the duration of a storm event. The runs were designed to profile
the storm by monitoring water flow through the MDC facility, and to use the op-
portunity to make appropriate changes in the process rate of the scaled-down
system. In this way, studies could be made on what types of monitoring are
needed, what dosage changes are necessary, ana what results can be obtained
throughout the duration of a storm—especially in the first flush stage.
Analyses for turbidity, pH, conductivity, and total suspended solids were
made on site, immediately after collection, during three storm runs. The
following sections discuss the results of storm profiles made at the Columbia
Park location in the fall of 1977.
Storm of 9/20/77
The first CSO storm profile run was for a gentle rain which occurred in
the Boston area on September 9, 1977. At the outset of this storm, flow
channel //I in which the submersible pump was located was in the process of
being de-gritted, with the result that pilot plant start-up was delayed until
about 11:00 AM, just as the first storm flow was beginning to come through.
High tide on that day was at 5:40 PM, occurring during the latter portion of
the run, which lasted from 11:00 AM until 7:00 PM. No samples were collected
between 2:30 PM and 4:00 PM due to an interruption by visitors.
Figure VI-12 and Figure VI-13 show overlaid plots for all pertinent
monitored storm water parameters profiled during this pilot plant run, includ-
ing actual storm water flow through the Columbia Park headworks, pilot plant
flow rates used during the storm, feed water suspended solids, turbidity and
conductivity, and treated water suspended solids and turbidity.
For this run, the small (1 1/2 inch diameter) magnetic separator was used
with a range of process flow rates from .85 to 1.42 gpm, paralleling the
observed storm flow of 78 to 180 MGD in a step-type fashion. Jar tests
were done regularly (one per hour) in order to assure proper alum dosages,
and visual observation of flocculation proved to be most important in
detection of disequilibrium in the system. Optimal alum dosages started out
at 100 mg/£ at 11:00 AM, but decreased steadily to 30 mg/£ by 2:30 PM.
54
-------
By 6:00 PM, the optimal dosage was back to 60 mg/£. Polyelectrolyte concen-
tration was not varied from the 1 mg/£ of Betz 1160 normally used. Initially,
magnetite concentration was maintained at 85 mg/£, but was increased to 150
mg/£ after 2:00 PM as separation seemed to be suffering from seed deficiency.
A significantly higher seed concentration later was shown to be the single
most critical factor in controlling effluent solids concentration during the
first flush situation apparent in many storms. Note that the treated water's
solids and turbidity were brought back into control only after the magnetite
concentration was raised in the latter part of the storm simultaneous with a
lowered storm water solids loading, demonstrating that a minimum seed-to-
solids ratio must be maintained at all times for proper magnetic separation.
Figure VI-12 also shows how the rain water accentuated the tidally
induced salinity variation during the early part of this storm by diluting
the "low-tide CSO" to a conductivity of ^ 500 ymhos/cm, substantially below
the normal low of about 2500 ymhos/cm in this system at low tide. The two
to three hour time delay for the Boston sewer system is evidenced by the
conductivity plot given, as low tide occurred at 11:21 AM and minimum conduc-
tivity was not reached until after 2:00 PM. The salinity effects of the high
tide in the late afternoon were also affected by the storm's diluting waters,
with the normal conductivity of 20,000+ ymhos/cm at peak being significantly
suppressed.
The overall separation achieved during this storm was not particularly
good as a result of a number of factors, including lack of magnetite and per-
tinent feedback regarding storm water character. In successive storm runs,
experience proved valuable in the visual evaluation of CSO before and after
treatment, permitting more sensitive system adjustments to be made based on
changing conditions. Continuous monitoring of turbidity, alkalinity, pH, etc.
may be necessary in full-scale systems for CSO in order to maintain proper
flocculation.
Storm of 9/26/77
Figure VI-13 shows the monitored parameters plotted for the second storm
water profile run performed with the mobile pilot plant on September 26, 1977.
Rain started gradually early in the morning and then came down hard for a
brief time shortly after 12:30 PM before tapering off in midafternoon. In
this instance, the flow rate curve follows the entire storm from start to
finish, with actual pilot plant running profiling all but the initial start-up
of the storm water flow. Both magnetic separators were used covering a flow
rate of from 1.9 to 6.1 gpm.
A definite first flush situation took place after the heavy rain (at
about 1:30 PM) as feed solids jumped from a, 90 mg/£ to over 200 mg/£ in just
a few minutes. Once again, the sudden solids loading was not detected in time
to compensate with added magnetite sufficient to maintain efficient separation,
and the effluent solids, therefore, tended to parallel feed solids due to an
insufficient seed-to-solids ratio. Magnetite concentration was left at 150
mg/£ throughout the run, while alum was varied based on jar tests from 100 up
to 140 and back to 80 mg/£ during the duration of the storm. Polyelectrolyte
was also changed over a range of .87 to 1.5 mg/£ during the run.
55
-------
MGD
W
QJ
o
H
fn
O
4J
•H
PM
aao
soo
X Pxlot Plant Flow Rate
o Feed Conductivity
Time of Day (hr)
FIGURE VI-12 STORM OF 9/20/77: PHYSICAL PARAMETERS
CO
o
TO
O
0)
OJ
•H
>
•H
4J
O
fi
O
-------
2 cm
1 SO
CO
T3
•H
iH
O
TD
0)
T)
c
HI
ex
CD
Time of Day (hr)
FIGURE VI-13 STORM OF 9/20/77: SUSPENDED SOLIDS AND
TURBIDITY OF FEED AND TREATED SAMPLES
soo
ISO
16O
1-4O
1SO
Feed Suspended Solids
Feed Turbidity
Treated Suspended Solids
Treated Turbidity
100
•H
TJ
•H
PM
-------
MGD
gpm
On
CO
cn
0)
13
o
O
4-J
CO
P-i
JJ
O
23O
SOO
1SO
0 Storm Flow Rate
XPilot Plant Flow Rate
OFeed Conductivity
I
Time of Day (hr)
FIGURE VI-14 STORM OF 9/26/77: PHYSICAL PARAMETERS
-------
Ul
00
to
13
O
CO
C
0)
ft
en
3
CO
r
220|
200
1BO
16O
1<4O
ISO
10O
& Feed Suspended Solids
+ Feed Turbidity
• Treated Suspended Solids
Q Treated Turbidity
22O
2OO
ISO
TBO
J140
12O
10O
•H
T3
•H
J2
5-I
3
H
Time of Day (hr)
FIGURE VI-15 STORM OF 9/26/77: SUSPENDED SOLIDS AND
TURBIDITY OF FEED AND TREATED SAMPLES
-------
to
0)
O
rH
o
4-1
OT
13
C
-------
&0
CO
T3
o
CO
0)
CO
3
CO
k*
1 SO
16O
14O
A Feed Suspended Solids
+• Feed Turbidity
• Treated Suspended Solids
D Treated Turbidity
00
•H
T3
•H
(-1
3
Time of Day (hr)
FIGURE VI-17 STORM OF 10/14/77: SUSPENDED SOLIDS AND
TURBIDITY OF FEED AND TREATED SAMPLES
-------
From these data, a correlation is evident between high conductivity and
poor separation due to flocculation difficulties. High tide was at 11:09 AM
on that day, and because the storm was a rather small one, little dilution of
the salinity was noticed (the conductivity was 20,000+ ymhos/cm for a large
portion of the running time). When the conductivity subsided to< 8000 ymhos/cm
later in the day, flocculation, and the resulting magnetic separation, improved
significantly. This relationship was even more obvious in the dry weather flow
tests described earlier. Thus, lack of magnetite, high feed solids loading,
and high conductivity all helped to decrease magnetic separation effectiveness
during much of this run.
In the first two storms profiled, magnetic separation water treatment was
characterized by a highly unstable system unable to maintain good equilibrium
for either pilot plant process equipment or flocculation. This was mainly a
result of minor inadequacies in the equipment, the number of operators and
lab technicians, and the fact that running a storm is much more chaotic than
running under dry weather conditions due to the time limits imposed. Results
suffered from this lack of proficiency, but careful evaluation of these
results and the problems encountered led to improvements in systems and
techniques, and ultimately resulted in a substantial improvement in both
system functioning and quantitative results. In the third and final storm,
despite similar non-optimal conditions (salinity fluctuation, first flush
solids loading, non-optimization of flocculant, lack of monitoring feedback,
etc.), seeded water treatment was able to produce impressive results, and a
smoothly running operation was maintained for the duration of the run.
Storm of 10/14/77
The third storm event profiled in this study was done on October 14, 1977
on a typical, moderate rainfall which began as a drizzle shortly after 8:00 AM.
High tide this day was at 12:50 PM, but because of the relatively large volume
of rain, and a low high tide, conductivity did not play as large a role (maximum
conductivity reached was only 12,000 ymhos/cm), and jar tests showed that alum
concentration remained sufficient during the storm, being kept at 100 mg/£
throughout the run. However, magnetite concentration was increased to 400
mg/£ in anticipation of the high solids loading associated with the storm
water flows normally encountered, and this level was maintained for the
duration of the run. (If turbidity monitors for the feed were to be available
continuously, magnetite concentration could be varied continuously in response.)
Betz 1160 was kept at 1 mg/£. •
(fi)
The 3 inch SALA-HGMFW magnetic separator was used in this run, with flow
rates ranging from 1.4 to 3.3 gpm, corresponding to actual flows of 32 to 107
MGD. The lower flow rates through Columbia Park on this date were caused by
a diversion of part of the sewer flow due to a break in one of the upstream
lines.
Figures VT-16 and VI-17 show the data collected during this last storm.
It is immediately apparent from these curves that the system was much more
stable in this case than in the previous two cases. This is attributable
mainly to the added experience of the investigators, the increased magnetite
concentration used, and the more uniform salinity present during the storm.
62
-------
Magnetic separation was able to remove, with good consistency, about 80% of
the suspended solids and turbidity during this storm. Note that effluent
levels achieved did not parallel first flush solids loadings as they did in
the previous storms where magnetite was in short supply. By using more magne-
tite during the first flush, a consistently good effluent can be achieved. It
also can be seen from this figure that monitoring the storm flow rate with a
simple channel float level meter gives a fairly accurate indication of the
actual turbidity and suspended solids concentration of the storm water (at
least in this particular case). These curves tend to indicate that a rather
simple relation of flow rate to turbidity exists for a storm situation, and
a combination of these easily monitored parameters might prove useful in help-
ing to maintain optimal magnetite concentration and flocculation in lieu of
more expensive and sophisticated monitoring and feedback systems.
This final storm profile, then, is shown to be successful in meeting
essentially all of its objectives. Pilot plant operation was smooth, with no
technical difficulties. Equilibrium was maintained in the flocculation train
without any chemical concentration adjustment over the full 7^ straight hours
of running time in which samples were taken.
Cumulative probability curves for all the magnetically treated total
suspended solids and turbidity samples taken during this storm are given in
Figures VI-18 and VI-19. The geometric means for these curves are 25.6 mg/£
and 22.5 FTU, respectively, and they both have nearly identical spread factors
of -\j 1.38. The curves are useful in helping to determine how often a parti-
cular limit was (or is expected to be) exceeded during the course of opera-
tion. These data, although somewhat higher in absolute value than might be
possible with an optimized floaculant and a better regulated system, show good
consistency and a relatively small spread, demonstrating that seeded water
treatment is capable of handling the dynamic combined storm water flows typical
of the sewerage systems of a large number of cities in the United States.
63
-------
SUSPENDED SOLIDS (MG/L)
>0 OJ £-
0 0 0 $
10
0
99
—
—
99.9 99.8
--
—
_;:
: -
.. _.
- -
. __
.. _
_
... _
—
-
99 98
95
-
FIGURE \
PROBABILITY CURVE FOV
ON 10/14/7
-
—
- - - - —
:-.:
_.
H
-
-
-
-
/
-
9O 80 70 60 50 40 30 20 10 5
'118
, . .
-- ---
... , '
j r - -• - -
--J;! 2
._ . . . ... . __ _
\
„ ._ _ ,,.... . . _ _ . .
i; ---
?-- .... .. _ _ .. ._
£
^_^;;:::;:-:::::^:;:;-
.... ...
- • - - • - -
... . . ,!.. .
,. _' . . .
r - - - -
.... J 1 . _ .-._.
..... . . .._
. ... _ -
- • •
• MEAN - 25.6
....
-
--
-
_-
_
---
-"
_—
-
-
f
-
Z."
-
~
—
—
2 1
—
—
0.5
- -•
. _ .
- -
0.2
SPREAD FACTOR
1 — IUU-1'i" ! ' i — r
-- (---
- -
— . —
- - -•
....-._ .
1.0 O.OD 0.0
—
. _
r
r
r
r
t'
r
r
- \ \-
~ —
- —
..
-----
..__ _
'__ ^^
.
- . .
0.01 0.05 0.1 0.2 0.5 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99
PROBABILITY OF ACHIEVING LESS THAN "Y" MG/L IN A MAGNETICALLY TREATED CSO
99.8 99.9
99.99
-------
30 20 10 5 2 1 0.5 0 2 1.0 0.05 0.01
99 98 S5 90
FIGURE VI-19
PROBABILITY CURVE FOR TURBIDITY
ON 10/14/77 RUN
MEAN =22.5
SLOPE = 1.387 = SPREAD FACTOR
0.01 0.05 0.1 0.2 0.5 1 2 5 10 20 30 40 50 6O 70 80 90 95 98 99
PROBABILITY OF ACHIEVING LESS THAN "Y" FTU IN A MAGNETICALLY TREATED CSO
99.8 99.9
99.99
-------
SECTION VII
COST ESTIMATES FOR A HIGH GRADIENT MAGNETIC FILTER BASED
25 MGD CSO/SEWAGE TREATMENT FACILITY
The data generated in this and previous studies of seeded water treatment
have provided considerable information useful in the formulation of design and
size requirements for larger scale systems. In this section of the report are
presented the estimated capital, operation and maintenance-related costs for
a proposed integrated wet and dry weather flow treatment plant capable of pro-
cessing 25 million gallons per day at peak flow. The design of this treatment
facility has been upgraded considerably from the one outlined in the earlier
report (EPA-600/2-77-015), and includes high-quality system components through-
out, with reserve capacity and/or spare units as integral parts of the design.
CAPITAL COSTS
Capital costs for the 25 MGD combined wet and dry weather flow treatment
plant are summarized by sub-system in Table VTI-1. The total estimated capital
cost for this seeded water treatment system is just under 5.2 million dollars.
Not included in the capital cost estimates are systems for seed recycling
and alum recovery. These systems are contemplated for a full-scale facility,
but at this time too little process information is available for a determina-
tion of what should be included. It is hoped that a demonstration size plant
will be able to generate sufficient quantities of seeded sludge, in the near
future, for proper evaluation of various process alternatives in the recovery
and reuse of the magnetite seed and/or the aluminum sulfate incorporated
therein. The cost of land has not been estimated because of its site-specific
nature.
Installation costs were based on estimates from equipment suppliers, the
Chemical Engineer's Handbook (Perry), and the Richardson Rapid System. Capital
costs are represented at an Engineering News Record (ENR) index of 2700.
Labor was figured at $13/hour, based on an ENR of March 9, 1978.
No provisions have been made for the equipment and operational costs
associated with the disposal of either the dried sludge cakes or the grit
removed from the wastewater.
66
-------
TABLE VII-1 CAPITAL COST ESTIMATES FOR A HIGH GRADIENT MAGNETIC FILTER
BASED 25 MGD CSO/SEWAGE TREATMENT FACILITY
Part 1
Subsystem Estimated Installed Cost
De-gritting and Pre-screening $ 394,900
Feed Pumping 297,600
Flocculation Train & Chemical Feed 510,500
SALA-HGMF®High Gradient Magnetic Filters 990,000
Backflushing 141,300
Filter Piping and Valving 527,400
Thickening and De-watering 265,600
Disinfection 86,900
Electrical 137,100
Automatic Process Control & Monitoring 258,700
Physical Plant Construction 490,000
Subtotal: $ 4,100,000
Construction Contingency @ 10% 410,000
Subtotal: $ 4,510,000
Engineering and Administration @ 15% 677,000
TOTAL ESTIMATED COSTS: $ 5,187,000
Part 2
A breakdown of the subsystem components on which the above cost estimates
were made is given below.
De-gritting and Pre-screening System
. influent flow meter (Parshall flume);
. coarse bar screen;
. rotary wedge wire screens (6);
. entry pipe to plant;
. by-pass piping and valving;
. concrete troughs and pits;
. influent flow distributors;
. conveyor system; and
. grit hopper.
67
-------
TABLE VII-1, Part 2 (continued)
Feed Pumping System
. variable speed pumps;
. single speed pumps;
. support frames;
. piping for feed flow; and
. sump well.
Flocculation Train and Chemical
Feed System
. flocculators (3);
. flash mixers (6);
. static mixers (3);
. alum storage and delivery system;
. polyelectrolyte make-up and delivery system;
. magnetite slurry and delivery system; and
. support structures.
Magnetic Filters
\
. SALA-HGMF® Series LP Model
214-15-3 (5 units), including:
- magnets;
- canisters;
- matrices;
- power supplies;
- indirect cooling systems;
- instrumentation; and
- support structures.
Backflush Equipment
. screw compressors (2);
. piping;
• valves (5);
. air receivers (5);
. stands and concrete pads; and
. instrumentation.
Filter Piping and Valving
. filter piping and valves;
. chemical rinse piping, tank and valves;
. air line piping and valves;
. structures for piping; and
. test component.
68
-------
TABLE VII-1, Part 2 (continued)
Thickening and De-watering Systems
. thickener and rake;
. sludge pumps (2);
. surge tank;
. vacuum belt filters (2) ;
. conveyor;
. sludge hopper;
. piping for sludge;
. sump pump for thickener overflow (could be designed for gravity flow);
and
. overflow piping.
Disinfection System
. hypochlorite storage tank;
. chlorine diffusers;
. metering pumps;
. residual chlorine analyzer and return pumps;
. chlorine piping;
. contact chambers (3); and
. outfall piping.
Electrical System
. motor control centers;
. wiring and conduit;
. transformer and accessories;
. electrical boxes, etc.
Automatic Process Control and
Monitoring Systems
. micro-processor controller, including:
- flow metering;
- alarms;
- differential recording; and
- interface capability;
. sampler;
. analyzers; and
. monitoring system.
Physical Plant Construction
2
Building of approximately 16,500 ft , including:
. piping;
. electrical system;
69
-------
TABLE VII-1, Part 2 (continued)
lighting;
office facilities;
control room;
laundry;
crane;
etc.
OPERATION AND MAINTENANCE COSTS
Operation and maintenance costs for the 25 MGD facility are estimated in
Table VII-2 in dollars per 1000 gallons of water treated using an assumed
average dry weather flow of 5.7 MGD and a peak wet weather flow of 25 MGD
over the course of a year.
The chemical and electrical costs shown are approximate, current local
prices for the Cambridge, Massachusetts area. Operator labor is based on 24
hours/day monitoring of the facility plus an 8 hour shift for routine main-
tenance. The freight costs for the chlorine are included in the chemical costs;
polyelectrolyte freight costs are considered insignificant.
From the table it will be seen that total estimated operating costs for
seeded water treatment range from $.18 to $.23 per 1000 gallons of treated
water.
No costs have been included for final disposal of sludge and grit, or for
sludge treatment chemical conditioning (should the latter prove necessary).
The chemical demand and net operation costs could change significantly if seed
recycling or alum recovery are incorporated into the flowsheet.
Power consumption for a system running at 5 MGD and at 25 MGD is shown in
Table VII-3. The magnetic filters specified for this system are more efficient
than those units previously specified, and will therefore require only about
12 percent of total plant energy consumption.
DESIGN CONSIDERATIONS
The treatment plant will include three main flow streams for wastewater
purificaiton, although there will be one flow as they enter the facility (pre-
treatment) and as they leave (chlorination and de-watering). The actual water/
suspended solids separation occurs in one of five high gradient magnetic filters
(the fifth is considered a spare). The first filter is available to handle
normal dry weather flow, estimated in this instance at 5 MGD, and the remain-
ing four filters operating at a maximum flow rate of 125 gpm/ft2 are available
to handle the 25 MGD peak storm flow. When all five filters are activated,
the maximum capacity of the system will be 32 MGD, and average annual through
put is assumed to be 5,. 7 MGD. Figure VII-1 is a system design schematic.
70
-------
FIGURE VII-1 PROPOSED DESIGN FOR HIGH GRADIENT
MAGNETIC FILTER BASED 25 MGD/
SEWAGE TREATMENT FACILITY
-------
KEY TO FIGURE VII-1
No. of Units Symbol Description
5 AP Alum Delivery Pump (2 spare)
1 AS Alum Storage Tank
5 AR Air Receiver
1 BS Bar Screen
1 CG Control Gate
1 CP Caustic Rinse Pump
1 CS Chlorination System
1 CT Caustic Storage Tank
6 CV Conveyor
5 FP Feed Pump (2 spare)
3 FT Flocculator Tank System
1 CH Grit Hopper
5 MF SALA-HGMF® Magnetic Filter
3 MP Magnetite Delivery Pump
3 MSF Magnetite Screw Feeder
3 MST Magnetite Slurry Tank
1 OP Over-flow Pump
1 PM Polyelectrolyte Make-up System
5 PP Polyelectrolyte Delivery Pump (2 spare)
6 RS Rotary Strainer
2 SC Screw Compressor
1 SH Sludge Hopper
3 SM Static Mixer
2 SP Sludge Pump (1 spare)
1 ST Surge Tank
1 SW Sump Well
1 TH Thickener
2 VF Vacuum Filter
72
-------
TABLE VII-2 OPERATION AND MAINTENANCE COST ESTIMATES FOR A
HIGH GRADIENT MAGNETIC FILTER BASED 25 MGD
CSO/SEWAGE TREATMENT FACILITY
Costs in $ per 1000 gallons
Item @ 5.7 MGD @ 25 MGD
Chemicals:
. Alum (liquid, 50%) @ 100 mg/£ .042 .042
$100/dry ton
$.110/kg
$.27 /gal
$.071/£
. Magnetite (commercial grade, .030 .030
-325 mesh) @ 200 mg/£
$35/ton
$.018/lb
$.039/kg
. Polyelectrolyte (Hereofloc 831, .011 .011
anionic) @ 1 mg/£
$1.36/lb
$3.00/kg
. Chlorine (15% Sodium Hypochlorite, .004 .004
WilChlor 2) @ 2 mg/l
$.30/gal
$.079/£
$.529/kg available Cl2
(delivered price)
Total Chemical Costs: .087 .087
Chemical Freight Costs:
. Alum for 50 mi or 80 km .007 .007
$.05/gal
$.013/£
. Magnetite for 200 mi or 322 km .008 .008
$10/ton
$.Oil/kg
(estimated)
Total Freight Costs: .015 .015
73
-------
TABLE VII-2 (continued)
Item
Electrical Power @ $.055/kWh
. Magnetic Filter (5) @ 15 kW ea.
. Other Equipment
125 kW @ 5 MGD
404 kW @ 25 MGD
Operator Labor:
32 man-hours/day @ $15/hr
Maintenance Costs:
. Mechanical Equipment and
Physical Plant (3% of equip-
ment cost)
. Electrical Equipment, Instru-
mentation, and Piping (2% of
equipment cost)
Total Labor and Maintenance Costs:
TOTAL OPERATION AND MAINTENANCE
COST PER 1000 GALLONS:
Costs in $ per 1000 gallons
@ 5.7 MGD @ 25 MGD
.032
.024
.039
.042
.011
.092
,226
.039
.008
.002
.049
.175
Item
TABLE VII-3 POWER CONSUMPTION FOR A 25 MGD FACILITY
Energy Consumed in kWh
De-gritting and Pre-screening
Feed Pumping @ 45 ft head
Chemical Makeup and Delivery
Flocculator Chain Mixing
Magnetic Filters
Compressors
Dewatering System
Control Instrumentation and
Building Service
Miscellaneous
TOTAL:
@ 5 MGD
8.2
40.0
2.4
9.0
15.0
12.0
31.0
12.0
10.0
140.0
@ 25 MGD
18.7
200.0
6.5
39.0
60.0
45.0
42.0
23.0
30.0
464.0
74
-------
TABLE VII-4
ASSUMED PARAMETER VALUES FOR A
25 MGD SYSTEM
Item
Magnetic Filters (5 SALA-HGMF®Model
214-15- 3)
. Bore
. Length
. Magnetic Field Strength
. Maximum Flow Rate through Matrix
Waste Water Characteristics
. Maximum Storm Flow
. Nominal Dry Weather Flow
. Average Flow over 1 Year
. Average Influent Suspended Solids
Backflush Parameters
. Backflush FLow Flux Rate through
Matrix
. Backflush Duration
. Backflush Volume
. Maximum Pressure Drop before Backflush
. Caustic Rinse Flow Rate through Matrix
Mixing and Residence Times
. G Factor for Flash Mixers
. G Factor for Flocculator
. Reynolds Number for Static Mixers (21 in)
. Total Mixing Residence Time
Chemical Dosages
. Alum Concentration
. Magnetite Concentration
. Polyelectrolyte Concentration
. Chlorine Concentration
. Chemical Storage
Design Value
84 in (214 cm)
6 in (15 cm)
0- 3 kG
125 gpm/ft2 (18 m3/min)
(4800 gpm)
25 MGD (1.1 m3/sec)
5 MGD ( .22 m3/sec)
5.7 MGD (.25 m3/sec)
100 mg/£
500 gpm/ft2 (20.35 m3/min m2)
(19,300 gpm)
8 sec
2600 gal (10 m3)
10 psi
20 gpm/ft2 (-814 m3/min mz)
300/sec
100/sec
7.5 x 105
5 min
100 mg/£
200 mg/£
1.0 mg/£
2 mg/£
6 days at peak flow; 30
at dry weather flow
days
Other Parameters
. Pumping Head for Feed Pumps
. Pre-screening size
45 ft (33 psi)
20 mesh (.84 mm)
75
-------
Table VII-4 gives the parameter values used in obtaining magnetic filter
and operational cost estimates for a 25 MGD facility. These values were chosen
on the basis of the experimental data, previous experience in the field, flow
requirements, and equipment design limitations.
A combined wet-and dry-weather flow treatment plant has an inherently
large reserve capacity which remains unused for a major portion of the year.
For the 25 MGD peak flow system in question, it was arbitrarily assumed that
in the course of a year there would be 30 storms producing a 25 MGD flow for
an average of 10 hours each, or a total of 300 hours of CSO at a 25 MGD through
put rate, compared to 8460 hours of dry weather flow at 5 MGD. Although these
are assumed values, and may not be representative for a given location, the
fact remains that in this type of treatment facility, a large amount of equip-
ment must be regularly maintained for only sporadic use. As a consequence,
operation and maintenance costs are relatively high in proportion to the num-
ber of gallons treated.
The system shown in Figure VII^l is similar to the seeded water treatment
flowsheets presented in the past (i.e., the Mobile Pilot Plant Trailer design
and the 25 MGD system outlined in Figure X-l of EPA publication EPA-600/2-77-
015), with a few design changes. For example, static mixers have been added
for initial alum dispersal prior to the large alum residence tank. The floe
chain, consisting of an alum residence tank, two flash mixers for polymer and
magnetite addition, and a larger flocculator for final residence are sized for
a total of 5 minutes mixing time. A detail of this system is shown in Figure
VII-2. The pressure head created by the alum mix/flash mix/flocculator tank
unit is now designed to be used in place of the filter pump suction head used
in previous designs to draw the chemically pretreated water through the mag-
netic filters. Thus, there is no longer a need for filter pumps in the system.
Another change is with the magnetic filter design which affects the back-
flush system; each magnetic filter (SALA LP Series) now has its own hydrotank
for backflushing incorporated as an integral part of the magnetic filter canis-
ter (forward flow) plumbing. In this way, much extra plumbing and valving
have been eliminated, as well as the extra control provisions for filling the
hydrotanks with filtered water. This latter operation is now accomplished
automatically, in a passive manner, with each filter cycle.
Other features have been included in the design to make the system both
dependable and foolproof. The design is conservative, and spare pumps assure
adequate back-up; the process control system is versatile and completely auto-
matic; and a by-pass system has been included for emergency situations.
76
-------
TOP
VIEW
POLYELECTHOLYTE
vt"
4.'^'A -'b '•g-Q- O'-oio'
ALUM RESIDENCE
FUOCCULATOR
FIGURE VII-2 TOP AND SIDE VIEWS OF ALUM RESIDENCE/FLASH MIX/
FLOCCULATOR RESIDENCE TANK CHAIN FOR THE 25 MGD
SEEDED WATER TREATMENT SYSTEM
11
-------
REFERENCES
1. Weber, Walter J. Physicochemical Processes for Water Quality Control.
Wiley-Interscience, New York, 1972. 640 pp.
2. Oberteuffer, J. A. High Gradient Magnetic Separation. IEEE Transactions
on Magnetics, Vol. MAG-9(3): 303-306, 1973.
3. Bitton, G. and R. Mitchell. Removal of E. coli Bacteriophage by Magnetic
Filtration. Water Research, 8: 548, 1974.
4. Bitton, G., Mitchell, R., DeLatour, C. and E. Maxwell. Phosphate Removal
by Magnetic Filtration. Water Research, 8: 107, 1974.
5. Mitchell, R., Bitton, G. and C. DeLatour. Magnetic Separation: A New
Approach to Water and Waste Treatment. In: Proceedings of the Seventh
International Conference on Water Pollution Research, Paris, 1974.
6. Oberteuffer, J. A. Magnetic Separation: A Review of Principles, Devices
and Applications. IEEE Transactions on Magnetics, Vol. MAG-10(2):
23-238, 1974.
7. Kolm, H., Oberteuffer, J. A. and D. Kelland. High Gradient Magnetic
Separation. Scientific American, 233(5): 46-54, 1975.
8. Oberteuffer, J. A., Wechsler, I., Marston, P. G. and M. J. McNallan.
High Gradient Magnetic Filtration of Steel Mill Process and Waste Waters.
IEEE Transactions on Magnetics, Vol. MAG-11(5): 1591-1593, 1975.
9. Okuda, T., Sugano, I. and T. Tsuji. Removal of Heavy Metals from
Wastewater by Ferrite Co-Precipitation. Filtration and Separation,
12(5): 472-278, 1975.
10. Allen, D., Arvidson, B., Oberteuffer, J. and R. Sargent. SALA^HGMS™
Filters for the Treatment of Combined Sewer Overflow. In: Proceedings
of the Third National Conference on Complete WATEREUSE, Cincinnati,
Ohio, June 1976. pp. 239-251.
11. Allen, D. Addendum to SALA-HGMS™ Filters for the Treatment of Combined
Sewer Overflow. Presented at the Third National Conference on Complete
WATEREUSE, Cincinnati, Ohio, June 1976, 26 pp.
78
-------
REFERENCES (continued)
12. Mitchell, R. , Bitton, G. and J. Oberteuffer. High Gradient Magnetic
Filtration of Magnetic and Non-Magnetic Contaminants from Water.
Separation and Purification Methods, 4(2): 267-304, 1976.
13. Allen D. M. and J. A. Oberteuffer. Combined Storm Overflow Treatment
with SALA-HGMS®High Gradient Magnetic Filters. In: Proceedings of the
Joint US/USSR Symposium on Physical-Mechanical Treatment of Waste Waters,
Cincinnati, Ohio, April 1977. 32 pp.
79
-------
GLOSSARY
backflush: A rapid reverse flow of water through the separator matrix with
the magnet de-energized following a filtering interval. This serves to
purge the accumulated magnetic floes from the matrix filterbed.
bench test: A test performed manually on an aliquot of sample. More specifi-
cally, a waste water sample that is treated with known concentrations of
flocculants and seed while being agitated, and is then passed through a
gravity feed high gradient magnetic separator, and later analyzed.
canister: The cylindrical copper or stainless steel container holding the
magnetic matrix.
combined sewer overflow, or combined storm overflow (CSO): A flow from a
sewer receiving both surface run off and municipal sewage, in excess
of interceptor capacity, that is discharged into a receiving water.
continuous pilot plant: A small-scale sewage treatment plant set up to cycle
automatically and continuously for feasibility studies.
dry weather flow: The flow of municipal sewage during a dry spell in a sewer
also able to receive run-off.
first flush: The condition often occuring in storm sewer discharges and com-
bined sewer overflows, in which a disproportionately high pollution load
is carried in the first portion of the discharge, or overflow.
flocculation chain (or train): A series of interconnected mixing tanks in
which flocculants and magnetic seed are added to the water to be treated
by magnetic separation.
high gradient magnetic separator or filter: A solenoid electromagnet contain-
ing a filamentary ferromagnetic matrix (proprietary) and encased in a
highly efficient steel magnetic flux return frame.
jar test: A standard sewage treatment test in which several aliquots of
waste water are taken and treated with various concentrations of floc-
culants, etc., to determine optimal dosages necessary for effective
treatment.
magnetic matrix: The ferromagnetic material (steel wool or expanded metal)
which forms the filterbed located within the canister of a high gradient
magnetic separator.
80
-------
matrix loading: The ratio of the weight of solids retained in the magnetic
matrix to the weight of the fibers that comprise the matrix.
pilot plant cycle: A series of automatically timed and executed continuous
cycles including reset, magnet on delay, filter, magnet off delay, flush,
and dump.
seed recycle: The recovery and reuse of the magnetic powder (seed) used in
seeded water treatment from the end-product sludge.
sludge: A semi-solid end product of the magnetic separation process consist-
ing of waste solids, magnetic seed and flocculants.
static mixer: A compact device used for chemical dispersion which works by
repeated division and rejointure of the flow.
surge flow: A rapid increase in incoming volume of waste water to a treat-
ment or detention facility.
wet weather flow: A waste water containing both storm run-off and raw
sewage, resulting from the combining of storm waters with normal dry
weather flow.
81
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-78-209
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
TREATMENT OF COMBINED SEWER OVERFLOWS BY HIGH GRADIENT
MAGNETIC SEPARATION
On-site Testing With Mobile Pilot Plant Trailer
5. REPORT DATE
December 1978 (Issuing Date)
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
David M. Allen
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Sala Magnetics, Inc.
247 Third Street
Cambridge, Massachusetts
10. PROGRAM ELEMENT NO.
1BC611
02142
11. CONTRACT/SKS05CKNO.
68-03-2218
12. SPONSORING AGENCY NAME AND ADDRESS
Municipal Environmental Research Laboratory—Cin.,OH
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TYPE OF REPORT AND PERIOD COVERED
Final 8/76 - 12/77
14. SPONSORING AGENCY CODE
EPA/600/14
15. SUPPLEMENTARY NOTES Project Officers: Hugh Masters, phone (201) 321-6678, FTS 340-6678
and Richard P. Traver, phone (201) 321-6677, FTS 340-6677. See also EPA-600/2-77-015
"Treatment of Combined Sewer Overflows by High Gradient Magnetic Separation"
16. ABSTRACT ~~~~~
Seeded water treatment using a SALA high gradient magnetic separator pilot plant
system was conducted on CSO and raw sewage at SALA Magnetics in Cambridge, MA and at
on-site locations in the Boston area. These tests further built upon the data base
collected under the first phase of this contract (see EPA Publication 600/2-77-015)
with special emphasis on specific design and operational parameters, long term
durability and maintenance problems, and system adaptability to integrated wet and
dry weather flow conditions. The on-site results reported, although not equaling
those obtained on uniform batch samples in house, were nevertheless good and proved
that HGMF magnetic filters are effective on fresh CSO and raw sewage and that the
magnetic filtration treatment system is easily adaptable to the dynamic solids
loading and flow rate conditions typically associated with storm water and integrated
wet and dry treatment systems.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Group
Magnetic separators, Flocculators,
Sewage treatment, Combined sewers,
Water treatment, Filtration
Combined storm overflow,
SALA-HGMS magnetic
separator, Seeded water
treatment, High gradient
magnetic field, Continuou;
pilot plant
13B
13H
13. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
92
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
82
US GOVERNMENT PRINTING OFFICE 1979-657-060/1569
------- |