EPA-R2-72-032
DECEMBER 1972 Environmental Protection Technology Series
Fluidized Bed
Clarification as Applied
to Wastewater Treatment
Office of Research and Monitoring
U.S. Environmental Protection Agency
Washington, D.C. 20460
-------
-------
-------
EPA Review Notice
This report has been reviewed by the Environmental
Protection Agency and approved for publication.
Approval does not signify that the contents nec-
essarily reflect the views and policies of the
Environmental Protection Agency, nor does mention
of trade names or commercial products constitute
endorsement or recommendation for use.
11
-------
ABSTRACT
An experimental study of the application of a fluidized sludge blanket
clarifier to the coagulation and separation of wastewater solids has
been made to determine the effects of controlled process variables on
the treatment achieved.
Experiments using alum and ferric chloride coagulants were carried out
in 12- and 24-inch diameter columns by systematic variation of wastewater
pH, coagulant dose, upflow fluid velocity, and blanket depth. The re-
sults were analyzed using regression analysis techniques, and empirical
relationships were derived relating the variables to the removal of sus-
pended solids, total organic carbon, phosphorus, and coagulant metal ions.
The sludge production rate was also correlated empirically with the op-
erating variables.
A study of the settling rates of discharged sludge and the fluidized
blanket was made by direct observation.
Both alum and ferric chloride were found to be effective coagulants. The
sludge blanket acted as an efficient clarifier up to at least 15 ft/hr
superficial velocity, although best removal efficiencies were achieved
at lower rates.
This report was submitted upon fulfillment of Contract 14-12-912 under
the sponsorship of the Office of Research and Monitoring, Environmental
Protection Agency.
Key words: sludge blanket clarifier, upflow clarifier, sewage treatment,
solids removal, TOG removal, phosphorus removal.
iii
-------
CONTENTS
Section Page
I Conclusions 1
II Recommendations 3
III Introduction 5
IV Equipment 7
V Experimental Procedure 17
VI Experimental Design 21
VII Results and Discussion of Experiments Using Alum 25
VIII Results and Discussion of Experiments Using Ferric
Chloride 49
IX Sludge Settling Studies 61
X Experiment Using Activated Silica as a Sludge
Thickening Aid 77
XI Acknowledgments 79
XII References 81
XIII Publications and Patents 83
XIV Appendices 85
-------
FIGURES
Page
1 12-Inch Diameter Lucite Column 8
2 24-Inch Diameter Fiber Glass Column 9
3 System Flow Diagram 13
4 Photograph of Columns 14
5 Photograph of Mix Tank 15
6A Suspended Solids in Effluent 28
6B Suspended Solids in Effluent 29
7 Total Organic Carbon in Effluent 32
8A Soluble Phosphorus in Effluent 34
8B Soluble Phosphorus in Effluent 35
9A Total Phosphorus in Effluent 36
9B Total Phosphorus in Effluent 37
10A Residual Aluminum in Effluent 41
10B Residual Aluminum in Effluent 42
11 Sludge Production 45
12 Suspended Solids in Effluent 50
13 Total Organic Carbon in Effluent 51
14 Soluble Phosphorus in Effluent 52
15 Total Phosphorus in Effluent 53
16 Residual ~Fe+++ in Effluent 54
17 Settling Curve for Blanket of Test Run Al 62
18 Settling Curve for Sludge of Test Run Al 63
19 Flux Curve 72
20 Sludge Settling Rates vs. Suspended Solids Con-
centration 73
21 Reciprocal Minimum Solids Flux vs. Thickener Solids
Concentration 76
VI
-------
TABLES
Page
1 Analysis of Raw, Degritted Wastewater Influent 19
2 Effect of Alum Dose on Influent pH 30
3 Summary of Regression Analysis of Experimental
Data Using Alum 46
4 Summary of Predicted Treatment Effects Using Alum 47
5 Summary of Regression Analysis of Experimental
Data Using Ferric Chloride 55
6 Summary of Predicted Treatment Effects Using
Ferric Chloride 58
7 Data of SS from the Upflow Clarifier 66
8 Observed Parameters of the Clarifier 68
9 Observed Parameters of Sludge Tests 70
10 Treatment Effects Using Alum with Activated
Silica 78
VI1
-------
SECTION I
CONCLUSIONS
The following major conclusions may be listed:
a) Treatment Effects. The sludge blanket clarifier using either alum
or ferric chloride is an effective means for removing the major contami-
nants from wastewater. Although there is some advantage in a smaller
sludge yield using ferric chloride, the residual F6+++ in the effluent
may make its use less desirable than alum. The major variables affecting
treatment results were found to be pH, coagulant dose, blanket depth, and
upflow velocity, with considerable interactions existing between these
variables. Column diameter was not found to have a significant effect
within the size range studied.
b) Application of the Correlations. The empirical nature of the study
and the resulting correlations do not allow conclusions to be drawn re-
lating to the mechanisms of sludge blanket clarification. Moreover, the
variability of unknown constituents in the wastewater continually shifts
the conditions of optimality.
c) Sludge Production. Minimum sludge production can be achieved by
proper choice of coagulant dose and operating conditions. The use of ac-
tivated silica with alum caused a marked thickening of the sludge, and
its use is indicated as an additional means of reducing sludge volume.
d) Sludge Thickening. Sludges produced during alum treatment exhibited
settling curves in which settling rates were significantly dependent
upon solids concentration. The settling rate of the sludge blanket, how-
ever, was found to be no different than the upflow velocity.
-------
SECTION II
RECOMMENDATIONS
The following recommendations may be made:
a) A study should be made to determine means of reducing the sludge pro-
duction. Both the effects of thickening agents and special clarifier me-
chanical design should be considered.
b) A study should be made to establish methods of instrumentation and
control of a sludge blanket clarifier so that optimal treatment condi-
tions are maintained during periods of changing influent. The dynamics
of influent smoothing should be considered concurrently.
c) A study of further effluent treatment should be carried out to assess
the full potential of the sludge blanket clarifier in an integrated
physical-chemical treatment facility.
d) A large-scale design study by an experienced engineering firm should
be made to establish construction and operating costs.
-------
SECTION III
INTRODUCTION
Suspended and colloidal solids can be effectively removed from raw sew-
age by treatment with a coagulant followed by solids separation. Tradi-
tionally, chemical treatment has been by agitation (rapid mixing) fol-
lowed by slow mixing (flocculation) and sedimentation. In the design of
such systems, the major problem is to achieve the proper time-concentra-
tion-shear relationships in the mixing step so that an easily separated
floe is obtained.
An alternative method of chemical treatment is by use of the fluidized
blanket clarifier. This method seems to have wide use in Europe, but in
this country it is primarily used in water softening plants. The history
of the upflow clarifier appears to date from the late 19th Century. The
first large-scale application was in 1880 in the city of Dortmund, Ger-
many. Numerous patents have since been granted, and several commercial
models are currently available.
The fluidized sludge blanket clarifier (FSBC) has a size advantage over
the conventional system because the floe-supernatant separation occurs in
the same vessel used for the flocculation. The vessel cross-sectional
area required to handle equal flows can be as much as one-third less for
the FSBC than a conventional clarifier.
The sludge production rate from a fluidized sludge blanket clarifier may
be much greater than from a conventional settler unless adequate thicken-
ing area is provided. If an additional thickening vessel is required,
some of the initial size advantage over the conventional process may be
lost, but in terms of total capital investment the cost may be less if
the sludge has better settling characteristics.
Gulp and Gulp (6) state that field experience indicates surface overflow
rates for clarifiers for the removal of floe from chemically treated sew-
age must be less than those recommended for water treatment to prevent
excessive floe carryover. They also state that maintenance of sludge
blankets in upflow clarifiers has proved difficult in sewage treatment
applications. The results of this present study suggest this to be not
always true.
The present study has been made to assess the effectiveness of the FSBC
in the chemical treatment of raw sewage by performing experiments in
which the operating variables were changed in a planned way and changes
in treatment observed. Alum and ferric chloride were used as coagulating
agents, and the effects of pH, coagulant dose, fluid velocity, blanket
depth, and column diameter were measured as they affected the removal of
TOG, phosphates, suspended solids and coagulant metal ions. Additional
studies were made to examine the settling behavior of the blanket and of
the sludge produced.
-------
The approach taken in studying the effects has been largely phenomenal-
istic. That is, the forms of the relationships between causes and effects
have been deduced by statistical methods in an attempt to reduce the un-
accountable variation in observations. Although municipal wastewaters
vary considerably in quality, these data should be applicable to many mu-
nicipal plants which receive a predominantly domestic waste.
-------
SECTION IV
EQUIPMENT
Four clarifier columns were used; two were 12-inch diameter transparent
Lucite, and two were 24-inch diameter reinforced translucent Fiberglas.
All columns were built as an 8-foot straight section with a flanged top
section. The top sections were 1 and 2 feet tall for the 12-and 24-inch
columns, respectively. Threaded taps were located at intervals in the
wall of the 8-foot section for blanket sampling. The columns are de-
picted in Figures 1 and 2.
Column Design
12-inch Diameter Lucite Columns. The 12-inch columns had 60° conical
sheet metal bottoms flanged to the main section with 3-inch standard
threaded unions brazed to the bottom to receive the flow distributor and
sludge removal fittings. Liquid overflow from the top of the column was
removed over an annular weir through the column wall. The weir was fab-
ricated of Fiberglas and flanged between the top and middle column sec-
tions. Use of the weir assured a constant liquid level in the column
during normal operation.
The wastewater was introduced into the column through a single 1-inch
copper pipe which extended vertically to the level of the bottom flange.
A circular copper plate 2-1/2 inches in diameter was mounted at 90° to
the flow 1 inch directly above the pipe opening. This plate acted as a
baffle and forced the incoming liquid to flow radially outward. The
purpose was twofold:
1) to prevent direct upward jetting and destruction of the blanket
2) to provide additional turbulent mixing for the coagulant and the
wastewater feed.
The position of the open end of the distributor at the flange level also
prevented undesirable sludge accumulation which eventually would block
the opening of the distributor. This was found to occur when the pipe
was mounted lower in the conical column bottom.
The copper inflow pipe was mounted inside a 3-inch brass tee by means of
a bushing. The tee was threaded to the union on the conical column bot-
tom. Sludge from the column passed out the column through the remaining
open end of the tee run, as shown in Figure 1.
Lever-handled brass ball valves were used uniformly throughout the system.
These had the distinct advantage of being quick-acting and non-clogging.
The columns were erected on tripods fabricated from steel angle stock.
24-inch Diameter Fiberglas Columns. The 24-inch columns had 45° conical
bottoms of Fiberglas with a 2-inch flanged nozzle for sludge removal.
-------
T
I'O"
-FIBER GLASS WEIR
->- SUPERNATANT OUT
•4-'/2" TAPS
SPACED 2'0"
EXCEPT TOP
8 BOTTOM
WASTE
WATER
IN -
BAFFLE AT FLANGE
LEVEL
60°CONE
3" BRASS TEE
SLUDGE OUT
(SEE DETAIL BELOW )
1
BUSHING
2-'/2" DIAM. CIRCULAR
«O BAFFLE
PIPE
>-FRICTION FIT FOR
EASY REMOVAL
SOLDER
3 BRASS TEE
FIG. I 12 INCH DIAMETER LUCITE COLUMN
-------
ELEVATION
Not to Vertical Scale
SHEET METAL WEIR
I" BULKHEAD FITTING
FOR SUPERNATANT
OUT
4-'/2" TAPS SPACED
2" EXCEPT TOP a
BOTTOM
STEEL SUPPORT RING
I" BULKHEAD FITTING
FOR WATER INLET
(SEE DETAIL "A" "A")
45°CONE
2" NOZZLE
SLUDGE OUT
FLANGE
—LEVEL
BULKHEAD
FITTING
ALL PIPE IS I" PVC
DETAIL "A""A" WATER DISTRIBUTOR MANIFOLD
FIG. 2 24 INCH DIAMETER FIBER GLASS COLUMN
-------
The conical bottom, as well as the 2-foot top section, was flanged to
the 8-foot main section. Liquid overflow was removed over a sheet metal
weir in the top section as in the smaller columns, as shown in Figure 2.
Because of the larger column cross-sectional area, the liquid was intro-
duced through a 4-nozzle manifold built from 1-inch PVC pipe and fittings.
Each vertical nozzle was fitted with a circular metal baffle as in the
12-inch columns, and they were arranged to discharge into equal fractions
of the cross section. The tops of the nozzles were level with the cone-
column flange. Liquid to the manifold entered through a single pipe
through a plastic bulkhead fitting in the column wall about 2 inches
above the flange rather than through the bottom of the cone as in the
smaller columns. These details are shown in Figure 2.
Sludge Removal and Blanket Level Control
Initial attempts to control the blanket level by withdrawing sludge by
means of an adjustable submerged weir were unsuccessful. It was soon
found that the level of the blanket interface could be lowered by simply
removing sludge from the bottom of the cone. This procedure was mecha-
nized into a simple control system by using a photocell and a solenoid
valve. The photocell was arranged to activate when the sludge blanket
interface intercepted a light beam from the source to the cell. This
caused a solenoid valve in the sludge withdrawal line to open, releasing
sludge until the blanket level dropped to expose the cell to the light,
causing the circuit to close the solenoid valve again. An adjustable
timer was built into the circuit to delay activation of the solenoid valve
so that random floes or interface instabilities would not unnecessarily
trigger the control. This system usually provided interface level con-
trol to within 1 inch or closer to the desired level, depending on the
duration of the delay. Delay periods of up to 30 seconds were tried,
but usually a shorter period of 10 to 15 seconds was adequate to prevent
chattering of the valve.
The solenoid valves were Sears washing machine rubber diaphragm valves.
These were eminently satisfactory, although with the slight disadvantage
of being limited to about 2 feet of total fluid head to prevent rupture
of the diaphragm. Accordingly, these valves were mounted at an elevation
to keep the net head less than 2 feet. Actually, this arrangement also
served as a safety device in preventing the complete loss of the column
contents in case of photocell malfunction, since the total liquid level
on the column could not drop below the level of the valve.
For the transparent 12-inch columns the photocell devices were made from
inexpensive burglar alarm kits. These consisted of a light source which
was mounted externally on one side of the column and a detector cell
which was mounted externally on the opposite side. The light beam passing
through the clear supernatnat liquid was intercepted by the upper surface
of the sludge blanket and interrupted the signal. Since the Fiberglas
column walls would not transmit sufficient light, this type of device
could not be used with the 24-inch columns. For these columns commercial
(Keene Corporation Model 8000 SCCS portable sludge level detector)
10
-------
submersible photocell sludge level indicators were used. These were
lowered directly into the blanket. The submersible cells operated best
when a high intensity floodlight was mounted externally to provide suf-
ficient illumination of the interface so that the photocell could detect
the contrast between blanket and supernatant.
Flow Measurement
Owing to the surface-fouling nature of the wastewater, conventional
rotameters were not indicated. Instead the total flow to each column was
measured by timing the overflow in calibrated Lucite cylinders. These
were mounted next to the columns with a ball valve manifold so that one
cylinder could serve two columns. During flow measurement the photocells
were disconnected so that the entire liquid flow passed overhead to the
cylinder. The capacity of the cylinder serving the 12-inch diameter col-
umns was 4.0 liters, while that serving the 24-inch columns was 8.0 liters,
The sludge removal rates were measured by timing the collection of the
total amount discharged in 30-gallon plastic trash cans and weighing on
platform scales.
i
Chemical addition flow rates were set by calibration of the adjustable
chemical feed pumps.
pH Control
pH control was effected by using a Beckman Model 940 pH analyzer and
electrode assembly. The electrode assembly consisted of a general pur-
pose glass electrode, a LAZARAN™ process reference electrode, and a
thermocompensator electrode. The electrodes were mounted in a plastic
flow-through cell supplied by Beckman. No electrolyte is required when
the LAZARAN electrode is used, a particular convenience in pilot plant
operation. This system controlled the pH of the effluent from a 275-
gallon mixing tank by pumping a solution of sodium hydroxide to the tank
as required by the control point limits. A stream of tank effluent was
recirculated continuously through the electrode cell supplied by the manu-
facturer. Initial experience showed that little fouling of the cell or
electrodes occurred when the liquid rate through the cell was maintained
at approximately 1 gallon per minute. Infrequent fouling by fibrous ma-
terial did not appear to affect the controller accuracy. Calibration of
the electrodes with standard buffer solution was made prior to each test
run.
The degree of control achieved was at all times satisfactory, but with
slight overshoot inherent to on-off systems. Deviations of +0.1 pH
unit were normal.
The feed was pumped to each column by flexible impeller pumps run with
DC shunt wound motors operated by electronic speed controllers. The 12-
inch columns were serviced by pumps having a capacity of about 2 gpm at
11
-------
20 feet of head, while the 24-inch columns used larger pumps of about
8 gpm at 20 feet of head. One-third horsepower motors were found to be
adequate.
The smaller pumps were marginally satisfactory. Although most of the
larger grit had been removed from the wastewater by two 55-gallon drums,
impeller abrasion occurred; and each impeller had to be replaced several
times during the course of 6-months operation. No impeller failure oc-
curred with the larger pumps.
It was discovered early that alum could not be added to the pump suction
without causing severe impeller damage. Although this would have been
better from the standpoint of mixing, the pumps had to be protected; and
henceforward, alum was added at the pump discharge.
Raw sewage was pumped to the test site with a submersible centrifugal
pump supplied by the treatment plant.
A system flow diagram is shown in Figure 3.
are shown in Figures 4 and 5.
Chemicals
Photographs of the equipment
Commercial-grade aluminum sulfate, ferric chloride and sodium hydroxide
were used in these tests. The aluminum sulfate was found to have a com-
position consistent with the hydrated molecule A^CSO^) ' IGfUO. This
was determined from the aluminum content of a solution made by dissolv-
ing an accurate weight of the crystals in distilled water. Manufacturer's
(Fisher Scientific) specifications list the ferric chloride as FeClo • 6H20
Sodium hydroxide was obtained in flake form for easy dissolution.
12
-------
SLUDGE BLANKET
SLUDGE
TO
COLLECTION
COLUMN
SUPERNATANT
pH
INDICATOR
CONTROLLER
CAUSTIC
pH CELL
TO
COLUMNS
— SCREENED
SEWAGE
GRIT
REMOVAL
*-SEWER
LEGEND
CHEMICAL
STORAGE
FLOW LINES:
CONTROL LINES:
FP: FLEXIBLE IMPELLER PUMP
MP: METERING PUMP
MT: METERING TUBE
PD : PHOTOCELL DETECTOR
PS : PHOTOCELL SOURSE
S: SAMPLE TAP
FIG. 3 SYSTEM FLOW DIAGRAM
-------
FIGURE 4
Fluidized sludge blanket clarifier apparatus. Column in
left foreground operating with 7-foot blanket. Column in
right foreground in early state of blanket accumulation.
Background columns not in operation.
14
-------
FIGURE 5
Mix tank with pH control and grit removal drums (background)
Sludge collection can (with cover) in left foreground.
15
-------
SECTION V
EXPERIMENTAL PROCEDURE
Wastewater Variation and Sampling
The experimental design plan required that for each experiment the inde-
pendent variables be fixed at prescribed values. Owing to the uncon-
trolled variation in wastewater composition, corresponding variation in
the system response was always present, and over the period of a day's
operation, the response to the controlled variables would tend to be ob-
scured by wastewater variation. It was concluded, therefore, that com-
posite sampling was not indicated and that repeated sampling during a
daily period of minimum wastewater composition variation should be fol-
lowed.
An independent study of the Chapel Hill treatment plant influent by the
UNC Wastewater Research Center (11) showed that there were two periods
during an average day when the quantity and composition of the influent
varied the least. These occurred between 0400 and 0800 hours at the low
daily level and again between 1200 and 1600 hours at the high daily level.
Accordingly, most experiments were scheduled for sampling during these
times. This, of course, had the effect of providing data having the
largest possible variation in wastewater composition, but the effect of
this variation could be examined during the statistical treatment of the
data.
Each experiment was assigned to be conducted at either the steady-state
high feed concentration period or the steady-state low feed concentration
period, the assignment being based upon a randomization of the order of
execution. This was necessary to avoid confounding the desired measure-
ments of performance with changes in feed concentration.
Procedure
Owing to the rapid formation of the sludge blanket under most conditions,
an apparent steady state was usually reached within 4 to 6 hours after
start-up. Occasionally during rainy weather an adequate floe could not
be achieved at a low chemical dose (150 mg/1) or a high controlled pH
(about 9). The usual practice was to begin operation early on Monday and
to take samples on Tuesday, then changing operating variables and sam-
pling again on Thursday. During periods of high experimental activity,
the 24-inch diameter columns were allowed to operate throughout the week-
end, since their larger volume required more time to fill during start-up.
Preceding the expected test period, samples of effluent were measured for
turbidity using a Hach Model 2100 turbidimeter. When the turbidity of 4
to 6 effluent samples taken over an hour appeared to be constant, an ini-
tial set of all stream samples was taken. Effluent turbidity was moni-
tored during the next hour, and if no serious changes occurred, a second
set of samples was taken.
17
-------
A further criterion for steady operation was the continued control of
the blanket level by regular automatic sludge discharge. Occasionally,
from unknown causes, the blanket stopped accumulating sludge to cause the
interface to recede or at least to remain below the desired level for an
extended period. During this time the sludge would thicken to give an
atypical sample. When this occurred, the test run had to be stopped,
since the resulting samples gave inconsistent material balances. Usually
two sets of samples were taken one hour apart, but several runs were made
with a third set after the second hour of operation. The analyses of
these samples were averaged when making calculations.
Sample Analyses
All samples were analyzed for the following:
Total organic carbon (TOC) , mg/1
Total suspended solids (SS) , mg/1
Soluble phosphorus (as P), mg/1
Total phosphorus (as P).mg/1 _^_^
Coagulant metal ion (Al '' or Fe ), mg/1.
Five-day BOD analyses were obtained only on the second sample of efflu-
ent and sewage streams to keep the load on the lab to a manageable level.
Individual TOC analyses were repeated at least twice to minimize error.
Suspended solids were determined for influent and effluent samples by
filtration through glass fiber filters. For the determination of sus-
pended solids in sludge, the fiber mat procedure is not practical.
Sludge samples were centrifuged at 7000 g to separate the solids which
were then dried and weighed. The residual solids in the centrifugate
were determined by filtration on fiber mat filters and the weight of
both solids combined to give the result for the entire sample.
Total phosphorus was determined by the automated stannous chloride
method (15) following digestion of the entire sample by persulfate di-
gestion. Dissolved phosphorus was determined by the same procedure fol-
lowing filtration through a membrane filter. Coagulant metals (Al*""*"4"
and Fe ) were analyzed by atomic absorption spectroscopy. Although
the ferric oxidation state is indicated, the analysis represents the total
iron present in the sample.
All determinations except phosphorus and metals were made within 24
hours of sampling and usually the same day. Phosphorus and metal analy-
ses were run on scheduled days. The phosphorus samples were preserved
by adding mercuric chloride and refrigerating; metals samples were acidi-
fied by adding hydrochloric acid.
Sewage Analyses
Table 1 lists the mean and extreme values for the measured constituents
in raw, degritted wastewater used as influent to the process.
18
-------
TABLE 1
Analysis of Raw, Degritted Wastewater Influent
Suspended Solids
Total
Organic Carbon
Soluble Phosphorus (as P)
Total
i I I
Al^
T, +++
Fe
BOD
Alkali
Phosphorus (as P)
Lnity-, mg/1 CaCOQ
Max. Value
mg/1
328
235
10.7
16.3
10
2
224
166
Min . Value
mg/1
50
31
1.6
2.9
0.05
1.0
42
100
Mean Value
mg/1
176
130
6.4
9.2
0.9
1.4
145
140
19
-------
Sludge Settling Measurements
It was initially proposed to apply a nuclear absorption technique to the
measurement of sludge density variations and to determine sludge set-
tling rates by this method. When sludge samples became available, how-
ever, it was found that the difference in density between the sludge and
solids-free water was too small to give acceptable accuracy with a nu-
clear measurement. Consequently, this method was not attempted, and
sludge settling rates were determined by direct visual measurement of
the sludge interface position with time.
A detailed description of the sludge settling rate experiments is given
in Section IX of this report.
20
-------
SECTION VI
EXPERIMENTAL DESIGN
Plans for experiments were directed toward three goals: to assess the
significance of process variables, to determine the effective ranges of
these variables, and to model the performance of the clarifier. Five
process variables were studied: the column diameter and four opera-
tional variables—upflow velocity, coagulant dose, blanket depth, and
pH. The significance of these variables was judged statistically by
measuring the effects of changes in their levels and comparing the
measured effect with its experimental error. In addition, the signifi-
cance of interactions of the variables was assessed. Effects assessed
included effluent concentrations of suspended solids, total organic
carbon, phosphorus and aluminum, and the volume of sludge produced.
The effective ranges of the operating variables were determined as the
"factor space" within which the clarifier could be operated. Column
diameters of 1 foot and 2 feet were compared, primarily to determine
how readily the smaller diameter system could be scaled up. No attempt
was made, however, to identify or measure all the elements of scale-up.
The performance of the clarifier was modeled by means of modified sec-
ond degree equations, each with an effluent quality parameter (e.g.,
concentration of phosphorus) as a dependent variable, and four operat-
ing factors as independent variables. Modifications of the second
degree equations consisted of extra terms added to account for the ef-
fects of the column diameter, the level of the dependent variable in
the raw sewage and any significant higher order interactions of the
operating variables. The performance models were fitted to the data
by the method of least squares.
The clarifier received wastewater at flow rates which could be con-
trolled. The concentrations of pollutants contained in the raw waste-
water could not be controlled, however. Suspended solids, TOC, phos-
phorus, and aluminum in the wastewater were measured; and the measure-
ments were averaged for each experiment.
In contrast to the uncontrolled variation of the wastewater quality,
the operating variables were set and held at prescribed levels. An ex-
periment consisted of adjusting the pH of the influent to a prescribed
value, adding the prescribed coagulant, and passing it through the
column at a prescribed upflow velocity. The depth of the sludge blanket
was controlled independently.
Experiments with Alum as the Coagulant
A 2^ factorial experimental design was employed to study the four op-
erating variables in each size column. This design inherently met the
data requirements for statistical tests to be applied. It spread the
experiments systematically throughout the factor space so that the
21
-------
limits of good operation could be defined. It could be expanded in a
number of ways to generate data for a second degree model.
The levels of each of the five process variables which were prescribed
for the factorial design were as follows: column diameters of 1 foot
and 2 feet; upflow velocities of 9 and 12 feet per hour; coagulant
doses of 150 and 250 ppm; blanket depths of 3 and 7 feet; and pH values
of 8 and 9.
Upon application of the design, the clarifier failed to function at some
of the prescribed operating conditions at a pH of 9, thus indicating
that this pH was an upper bound for operation of the system. The experi-
mental design was therefore expanded away from this limit by prescribing
experiments at a pH of 7. The previous lower limit for pH was 8.
Additional levels of four of the variables were run to provide data for
fitting the performance model as follows: upflow velocities of 6 and
15 feet per hour; coagulant doses of 200 and 300 mg/1; and a blanket
depth of 5 feet. These additional runs were made at a pH of 8, since
experience showed that this value always resulted in satisfactory op-
eration. Since the data obtained at pH 9 could be used, a fourth
level of pH was not necessary.
Experimental error was measured by repeating 13 of the experiments once
and 3 of them twice. The data from a total of 51 experiments were
analyzed.
Analysis
The data were analyzed statistically using regression analysis (3) tech-
niques to determine the statistical significance of the effects and in-
teractions, and to assess the goodness of fit of selected models.
Several models were fitted to the data. The best model, in terms of ac-
counting for the greatest percent of the total variance in the measures
of performance, contained a term for the effects of the raw wastewater,
all main effects of operating variables and column diameter, all two-
factor interactions, four quadratic terms and the following higher order
interactions: diameter x alum dose x blanket depth; velocity x blanket
depth x pH; diameter x velocity x alum dose x blanket depth x pH.
Experiments with Ferric Chloride as the Coagulant
Only three operating variables were studied: upflow velocity, coagulant
dose and pH. The eight experiments carried out with ferric chloride as
coagulant divide into two sets of four. Each set is a two-level facto-
rial design to study velocity and pH at a constant level of ferric chlo-
ride. The sets employed different levels of ferric chloride dose. Also,
one set employed upflow velocities of 12 and 15; the other, 9 and 12
feet per hour. The data obtained do not provide a self-contained mea-
sure of experimental error which was available for the alum experiments.
22
-------
Therefore, experimental errors were assumed to be approximately the same
as those measured in the analysis of the experiments with alum. The two
sets of four experiments were combined to fit a regression model having
linear terms for velocity, ferric chloride dose, pH and influent concen-
tration of the contaminant and a quadratic velocity term.
23
-------
SECTION VII
RESULTS AND DISCUSSION OF EXPERIMENTS USING ALUM
The original design was followed using pH levels of 8 and 9 until poor
operation at pH 9 forced the design to be altered by substituting pH 7
runs. Actually, nine runs were made at pH 9 before reverting to the
alternate design. Thus, the prescribed two-level design was completed
using pH 7 and 8 with the runs at 9 considered supplementary. Thirteen
of the 53 runs were replicates, and three experiments were repeated
twice to establish a statistic for error. Included in the 53 runs are
six test runs made at an interim point of the design.
Examination of the results disclosed that two experiments were so dif-
ferent from the remainder that they could be legitimately discarded;
poor samples were believed responsible. A total of 51 test runs were
finally used in the correlations. The data from these runs are dis-
played in Appendix A.
During the experimental phases of the study - it was noted that influent
Al1 concentrations determined by analyses of the influent samples
frequently were different from the intended value prescribed by the ex-
perimental design. Actual measured values were used in the regression
analyses.
Data Correlation
The test run data were analyzed using a standard multiple regression
procedure programmed in A programming Language (APL) for the IBM 360/75
computer (9). The original data were stored in program work spaces for
easy access to any alternate regression model desired. Several models
were tried, and the one producing the smallest residual variance was
selected as the best. The same regression model was found best for
all the variables studied except inorganic phosphorus,'which required
a slightly different equation to achieve the best fit.
The first regression model is:
5 5 5 2
Y =b. + c.X.+ E a..X. + E a. ., X.X, + E c. .X.
J 03 oj oj ±=1 ij i ± k=1 ijk iTc i=2 iJ i
(7-1)
This model was used to correlate the effluent concentrations of sus
pended solids, TOG, total phosphorus, residual Al and the volume
percent sludge produced, which are represented by Y..
25
-------
The model fitting the concentration of soluble phosphorus in the ef-
fluent best was found to be:
5 5 5 2
Y. = b . + c .X . + I a. X + £ a XX + Z c X
] oj oj oj i=1 ij i i)k=1 IJK i J 1=2 XJ x
(7-2)
In both of these models the following notation represents the five in-
dependent variables studied:
X = influent concentration of contaminant j
°J
X1 = column diameter, D, inches
X = upflow velocity, U, ft/hr
X = alum dose, Al, mg/1
X = blanket depth, L, ft
X_ = adjusted pH of the wastewater before coagulant
addition.
The effects and interactions of the process variables were estimated in
the form of parameters of the mathematical model. Each parameter so es-
timated was tested statistically to determine the probability of obtain-
ing the experimental value if the parameter had a true value of zero.
The t-test was employed. For the experiments with alum, whenever the
probability of a true value of zero was found to be less than 0.05, the
parameter was judged to have a real effect. The numerical values for
the coefficients are listed in Appendix B. Because these are empirical
formulas, they should not be expected to be accurate outside the range
of the data.
It is not practical to use the correlations for a hand calculation to
predict a result, owing to the fact that a large number of significant
figures must be retained to avoid round-off error. The computer has
prepared a set of graphs from the correlations which should be adequate
for most purposes, however, and provides a visual representation of the
effects of the variables on sludge blanket performance.
The statistical analysis showed that diameter effects were not signifi-
cant except for the case of sludge production; even in this case, the
diameter effect was second order, as it appeared as an interaction factor
with alum dose and blanket depth but not as a main effect.
26
-------
No practical purpose is served by presenting the graphical results show-
ing effects of non-significant variables. Excepting the correlation for
suspended solids in the effluent, the graphs prepared from the correla-
tions present the effects of the significant variables including pre-
dicted maximum and minimum values. These are not confidence limits, but
are the extreme values resulting from the contributions of the non-
significant variables. Thus the charts present a range of results which
might be expected with the particular significant variables chosen for
any value of the remaining non-significant variable within the experi-
mental range.
The above procedure has not been followed for presenting the suspended
solids correlation because it was desirable to draw attention to several
qualitative effects which were of interest despite their lack of statis-
tical significance.
If one wishes to reproduce the curves from equations (7-1) and 7-2) , the
value of Xj = 24 should be used. Other values of X^ will give slightly
different results but not statistically different from those shown on
the curves. Values for any of the variables not within the ranges
studied should be used with caution, as large extrapolations are not
warranted.
Individual data points will generally not agree precisely with the curves
owing to experimental error and operating conditions different from those
chosen in making the plots. In assessing the goodness of fit, the stand-
ard deviation between observed and correlation values should be noted
along with the percent variance removed. These are listed with each cor-
relation in Table 3".
In four of the six correlations, the corresponding value for the quan-
tity in the influent to the sludge blanket had a significant effect on
the same quantity in the effluent. The graphs were prepared using the
average of the influent; thus a correction term is given in the text
and should be applied whenever the influent has a composition different
from the base value.
Suspended Solids in Effluent
Figures 6A and 6B show the graphical representation of the correlation
between the suspended solids concentration in the effluent and the vari-
ables. From the regression analysis the terms in the regression equa-
tion having significant contributions to the correlation in comparison
to their error of estimation are:
X5 = PH
Xg = Al X Al
X| = pH x pH.
Because of the quadratic terms the correlations show that the suspended
27
-------
DIAMETER 24"
UPFLOW VELOCITY 9FT/HR
BLANKET DEPTH 3 FT.
NO
00
pH7
BLANKET DEPTH TFT.
150 200
ALUM DOSE,
250 300
pH8
100 150 200
ALUM DOSE,
250 300
FIG. 6A SUSPENDED SOLIDS IN EFFLUENT
-------
130
120
o» MO
£100
UJ
3 90
u_
u] 80
z
~ 70
Q
H 60
o
w 50
o 40
CL
CO
20
10
0
DIAMETER 24"
UPFLOW VELOCITY I2FT/HR
130
BLANKET DEPTH 3 FT.
pH9
J_
100 150 200 250
ALUM DOSE.mg/^
300
-------
solids in the effluent are the smallest for different values of pH.
The alum dose at the minimum shifts to higher values as the pH increases
from 7 to 9. With the exception of the curves for U = 9 ft/hr and L =
3 ft, pH 8 is seen to produce the smallest effluent suspended solids at
an alum dose between about 200 and 275 mg/1. The curves for both pH 7
and pH 9 give higher effluent suspended solids.
Although the contributions of blanket depth and upflow velocity were
statistically non-significant, it is seen that the correlations predict
qualitatively that the deeper blanket results in a smaller effluent sus-
pended solids, at least for U = 9 ft/hr. The results for the 3 foot
blanket are less reliable because of the difficulty in achieving a
stable interface, particularly at the higher velocities.
For the 7 foot blanket, the qualitative indication is that the lower
velocity produces less suspended solids.
Discussion
A significant effect of both alum dose and pH on the suspended solids
removal corresponds to observations made by Miller and West (16)
although their work was done on river water and not sewage. They noted
minima in the data for effluent turbidity versus alum dose which seemed
to have little relationship to blanket depths in the range of 5 to 9
feet.
Miller and West also concluded that an optimal pH for turbidity removal
occurred between 6.5 and 7.0 as measured in the flocculating mixture.
This agrees generally with the present finding that pH 8 gave a lower
effluent solids than either pH 7 or pH 9. The pH variable of the pres-
ent experiments was measured before the alum addition, and thus the pH
value for the flocculating mixture was lowered by the alum. The average
pH of the influent after: alum addition as measured during the test runs
is shown in Table 2.
TABLE 2
Average pH After Alum Addition
Initial
PH
7
8
9
Alum Dose , mg/1
150
6.1
6.4
6.8
250
5.4
5.6
6.7
30
-------
These final pH values may explain a shift in the minima of the curves in
Figures 6A and 6B to higher alum dose values as the initial pH is in-
creased from 7 to 9, i.e., the minima probably correspond to the optimum
flocculation pH values.
Miller and West also noted the effects of blanket depth and upflow veloc-
ity on effluent turbidity. Generally, deeper blankets produced less tur-
bidity, although the data are scattered. Increasing upflow velocity pro-
duced greater turbidity in the effluent. Both these observations corre-
spond to the qualitative predictions of the present correlations and are
in agreement with the physical point of view that a deeper blanket pro-
vides more opportunity for small particle capture and agglomeration and
lower upflow velocities reduce the tendency for particle elutriation
from the blanket interface.
Although the variables—blanket depth, L; upflow velocity, U; and column
diameter, D—were not found to be statistically significant within the
limited ranges tested, the combined contribution of these variables and
interactions needs to be included to account for the variation remaining
after the major contributions of pH and Al have been removed. The cor-
relation accounted for 78.1 percent of the total variation, and the re-
gression equation fitted the data with a standard deviation of 4.4 mg/1.
A minor effect due to suspended solids in the influent was found. Since
the curves in Figures 6A and 6B were computed for a mean influent sus-
pended solids concentration of 176.3 mg/1, a correction to the predicted
effluent suspended solids should be made if the influent suspended solids
differs from this value. This correction is 0.068(CS - 176.3)mg/l, where
Cs is the influent suspended solids, mg/1, and should be added to the
effluent suspended solids given by the curves in Figures 6A and 6B.
Total Organic Carbon in the Effluent
Figure 7 shows the graphical results of the correlation for TOG. The re-
gression analysis showed the significant term in the regression equation
to be
XX = U x PH.
The remaining variables were not found to contribute significantly to
the regression. The major difference between the predicted results at
U = 9 and U = 12 is seen to be a widening of the limits within which the
effluent TOG may be expected to fall at the given pH. The curves for
U = 9 indicate possibly better TOG removal at values of pH between 7.5
and 8.0, while the curves for U = 12 show generally poorer treatment
with increasing pH. Generally, effluent TOG was unaffected by the vari-
ables except at higher velocity and higher pH.
The correlation accounted for 75.8 percent of the total variation and
31
-------
DIAMETER ! 12 OR 24 INCHES
ALUM DOSE.' 100 TO 300
BLANKET DEPTH : 3 TO 7 FT
70
0
o
20
10
0
7
UPFLOW VELOCITY=9FT/HR
7.5
8.0
pH
8.5
9.0
80^
UPFLOW VELOCITY=I2 FT/HR
MAX
9.0
FIG. 7 TOTAL ORGANIC CARBON IN EFFLUENT
32
-------
gave a standard deviation of 3.5 mg/1 TOG in the effluent. A small cor-
rection of +0.075(CC - 130.1)mg/l TOG in the effluent may be made for
influent TOG concentrations (Cc) different from 130.1 mg/1.
Soluble Phosphorus in the Effluent
Figures 8A and 8B show the graphical representation of the correlation
for soluble phosphorus, in mg/1 as P, in the effluent. The regression
analysis showed the significant terms in the regression equation to be
X3 = A1
X3X5 = Al x PH
X3 = Al x Al
X XJT = Al x PH x PH.
The remaining variables were not found to contribute significantly to
the regression.
The curves represent the maximum and minimum values at the given condi-
tions of alum dose and pH predicted over the experimental range of the
insignificant variables. It can be seen that the effluent soluble phos-
phorus generally decreases with increasing alum dose, except for influ-
ent pH 8 which indicates a minimum between 200 and 250 mg/1 alum.
Soluble phosphorus removal appears to be best at pH 8, worst at pH 9,
with considerable overlapping for alum dose 200 and 300 mg/1.
The correlation predicted negative values for some values of alum dose
and pH because of its inability to distinguish low values of effluent
phosphorus from the regression error. The correlation accounted for
76.9 percent of the variation and gave a standard deviation of 0.94 mg/1,
From the value for the standard deviation, one concludes that no real
difference exists between points predicted to be less than 1 mg/1.
An effect due to the concentration of soluble phosphorus in the influent
was found. If influent soluble phosphorus, Cp, is different from
6.375 mg/1, a correction of +0.369(C - 6.375)mg/l should be made to the
effluent soluble phosphorus.
A discussion of phosphorus removal from wastewaters is presented in a
following section.
Total Phosphorus in the Effluent
Figures 9A and 9B show the graphical results of the correlation for
total phosphorus in mg/1 as P. The significant terms of the regression
equation were found to be
X = pH
33
-------
cr>
E
a:
pH =
± ----- INFLUENT CONCENTRATION
DIAMETER 24 INCHES
UPFLOW VELOCITYiS TO 12 FT/HR
BLANKET DEPTH. 3 TO 7 FT
^ 10
CT>
E Q
0_~ 9
(S)
.<. 8
100
150 200 250
ALUM DOSE.mg/J?
MAX
MIN
300
y 7
CD
O
c/)
pH =
— INFLUENT CONCENTRATION
UJ
cr
0 3
x °
QL
en p
O
x
Q- I
0
100 150 200 250 300
ALUM DOSE.mg//
FIG. 8A SOLUBLE PHOSPHORUS IN EFFLUENT
-------
DIAMETER : 12 OR 24 INCHES
UPFLOW VELOCITY : 9 TO 12 FT/HR
BLANKET DEPTH : 3 TO 7 FT.
pH 9
? „ ------ INFLUENT CONCENTRATION
150 200
ALUM DOSE
250
300
FIG. 8B SOLUBLE PHOSPHORUS IN EFFLUENT
35
-------
LO
CTv
E
Q_"
CO
<
DIAMETER: 24 INCHES
UPFLOW VELOCITY : 9 TO 12 FT/HR
BLANKET DEPTH:3T07 FT.
h-
UJ
:D
_J
u_
u.
UJ
2
CO
DC.
o.
_j
<
; 10
pH 7
INFLUENT CONCENTRATION
8
7
6
5
4
2
I'
0
100
MAX
I
^
C7>
10
9
t" 8
7
6
5
4
3
a.
co
UJ
LU
§
tr
o
CO
pH8
INFLUENT CONCENTRATION
0
100
MAX
MIN
150 200 250 300 I- 100 150 200 250
ALUM DOSE.mgAe ALUM DOSE,
FIG. 9A TOTAL PHOSPHORUS IN EFFLUENT
300
-------
^
E
UJ
CO
r>
o:
o
x
Q_
CO
O
DIAMETER : 12 OR 24 INCHES
UPFLOW VELOCITY : 9 TO 12 FT/HR
BLANKET DEPTH : 3 TO 7 FT.
pH9
--INFLUENT CONCENTRATION
10
9
8
7
6
5
4
3
2
I
0
100 150 200 250 300
ALUM DOSE,mg/£
FIG. 9B TOTAL PHOSPHORUS IN EFFLUENT
MAX
MIN
37
-------
X X5 = Al x pH
X5X5 = pH x pH.
The remaining variables were not found to contribute significantly to
the regression.
It can be seen that increasing alum dose results in lower effluent total
phosphorus except at pH 7 and that the lowest values occur at an influ-
ent pH of 8.
The correlation accounted for 74.5 percent of the variation and gave a
standard deviation of 1.2 mg/1. An effect due to the concentration of
total phosphorus in the influent was found. If influent total phosphorus,
C , is different from 9.246 mg/1 as P, a correction of +0.368(Cp - 9.246)
mg/1 should be made to the effluent total phosphorus concentration.
A discussion of phosphorus removal from wastewater is presented in the
following section.
Discussion of Phosphorus Removal
The correlations for the concentration of soluble and total phosphorus
in the effluent showed that the significant variables were alum dose
and pH, as well as the concentration of phosphorus in the influent. The
effects of these same variables on phosphorus removal have been noted by
other investigators.
In a study using secondary effluent from the same plant, Malhotra et al
(13) studied the removal of phosphorus from samples with and without pH
adjustment. They found that the percent removal increased with alum
dose between 100 and 250 mg/1 but that significantly better removals
were achieved when the initial pH of the effluent was adjusted from 8.0
to 6.0. Further experiments led them to conclude that the optimum pH
zone for removal of phosphorus with alum was 5.50 to 6.0. They found
considerable variation in percent removed at the same alum dose and pH
which was attributed partly to initial phosphorus concentration and
partly to "unknown components" in the sewage. For example, at pH 8.0
and 200 mg/1 alum they observed a range of removals between 81 and 76
percent for initial concentrations of total phosphorus of 7.86 and 10.05
mg/1 as P, respectively. The residual concentrations would have been
about 1.5 and 2.4 mg/1, respectively. Comparing this with the correla-
tions in Figure 9A, at 200 mg/1 and pH 8 the range of residual total
phosphorus is seen to be from 1.4 to 2.0 mg/1, which is in reasonable
agreement with Malhotra's data. At pH 8 and 100 mg/1, the residual phos-
phorus from Malhotra's data is between about 4.6 and 5.5 mg/1, while
Figure 9A shows the expected range of 3.2 to 4.8 mg/1, again in reason-
able agreement. Comparing the correlations at pH 7 and pH 9 with
Malhotra's data, however, does not give as good agreement, since higher
effluent phosphorus concentrations are predicted. The range at pH 7
from Malhotra's data is about 0.6 to 1.9 at 200 mg/1 as compared with
2.2 to 4.8. The range found by Malhotra for pH 9 and 200 mg/1 ±3 from
38
-------
2.2 to 2.8 while the correlations predict a range from 3.3 to 4.9. Al-
though there are differences between the effects of pH noted in this
work and that of Malhotra, it is not unreasonable to expect different
results stemming from widely different wastewater characteristics.
When considering phosphorus removal by A1+++ and ?e+++, it is useful to
distinguish between reaction and removal. The reactions of phosphorus
with Al and Fe''' depend upon the forms of phosphorus present and the
other substances in the water which can also react with these metal ions
(e.g., OH , colloids). The removal of phosphorus depends upon physical
and chemical properties of the solid materials which are formed.
Most of the phosphorus in secondary effluent is orthophosphate (1).
The majority of the investigations of P removal by iron"1"*"1" and
aluminum have dealt with synthetic wastes containing orthophosphate
or with secondary effluent. In contrast, much less (on the order of
40 percent) of the phosphorus in raw domestic sewage is present as
orthophosphate; the balance is comprised of polyphosphates and particu-
late phosphorus (10).
Thermodynamic calculations describing the effects of pH on the precipi-
tation of orthophosphate as FePO^s) and AlPO^(s) have been made by
several authors (19). Experiments are in reasonable agreement with
these predictions (8). AlPO^(s) has a minimum solubility at about pH
6.0, while FePO^ is least soluble at about pH 5.0. The optimum pH in-
creases slightly with increasing metal/phosphorus (Me/P) ratios. Ex-
cess metal ions (Me/P>l) are added because (a) Fe''' and Al' ' f also
react with OH~, and (b) other materials in secondary effluent (e.g.,
some soluble organics, bio- and other colloids) will react with these
metals. At pH levels above these values (pH 5 for FePO^, pH 6 for AlPO^)
more of the iron or aluminum is used to react with OH~ ions, form-
ing Fe(OH)3(s) and Al(OH)3(s). These solids have a positive charge
below their isoelectric points and can adsorb appreciable quantities of
orthophosphate. Alumina columns have been successfully tested as ad-
sorbents for inorganic phosphorus (1). As a result, considerable re-
moval of orthophosphate can be accomplished by Fe''' at pH > 5.0 and
by Al at pH > 6.0 due to this adsorption. Me/P ratios significantly
greater than 1 are used. For example, Me/P ratios of about 1.5 are
often sufficient to achieve good precipitation of orthophosphate at the
pH of minimum solubility, while Me/P ratios of 3 or higher are used
when higher pH levels lead to the formation of the solid Me(OH)3 and
subsequent orthophosphate adsorption. (At 200 mg/1 the aluminum/total
phosphorus ratio was about 2.1 for the present study.)
Precipitation and/or adsorption of polyphosphates are also possible
using Al"1"^" and Fe' ' ' . Relatively few studies have been made. Cohen
e_t_ al (4) noted significant interference by tripolyphosphate in the
coagulation of water using ferric sulfate. Tenney and Stumm (20)
noted that soluble complexes were formed between Al and pyrophosphate
at A1+++/P < 1; when the molar dosage of A1+++ exceeded the molar phos-
phorus concentration, precipitation resulted. Very little information
is available on the coagulation of phosphorus-containing particulates in
39
-------
wastewater by Al''' and Fe'''. It is probable the coagulation is ef-
fective at pH levels in the order of 7.0, somewhat higher than that for
MePCk precipitation. In this case Me/P ratios could be less meaningful;
coagulant dosage would depend more upon the concentration of colloids
to be aggregated.
The precipitates formed by many of the preceding reactions can be col-
loidal. Stated another way, the optimum pH for phosphate precipitation
does not necessarily equal the optimum pH for phosphate removal.
Given this wide variety of materials and reactions, what may happen
when raw sewage is treated with Al"1"1"1" or Fe+++? It seems likely
that in the experiments conducted during this study, some Al"1 ' reacted
with orthophosphate and polyphosphates to form a mixed hydroxophosphato-
aluminum precipitate. It is not possible to evaluate when removal in-
volved phosphate precipitation and when adsorption might have occurred
without data describing the pH of the system after the addition of Al+++.
Some Al+++ may have reacted with soluble organics in the waste, although
this is probably not significant. Additional A1+++ was used to coagu-
late sewage particulates, some of which contained phosphorus. Finally,
some Al"^+ may have been used to coagulate these various solid materials.
Residual A1+++ in the Effluent
Figures 10A and 10B show the graphical results of the correlation for
residual A1+++ in mg/1. The significant terms in the regression equa-
tion were found to be
X3 = Al
X3X5 = Al x PH
X2X2 = u x u-
The remaining variables were not found to contribute significantly to
the regression.
At pH 7 residual Al' ' *~ is seen to increase as the alum dose increases
with considerable overlapping of the ranges indicated at the two upflow
velocities of 9 and 12 ft/hr. For pH 8 and pH 9 there is still some
overlapping for velocity at the lower alum dose, but the trends are
more definite: Increasing the dose and lowering the velocity decrease
the residual effluent A1+++.
The correlation accounted for 70.9 percent of the variation and gave a
standard deviation of 1.4 mg/1 A1+++. No significant effect of A1+++
concentration in the wastewater was found.
Discussion
The effect of pH on residual A.I'^~+ can be explained qualitatively by
40
-------
DIAMETER 24 INCHES
UPFLOW VELOCITY AS SHOWN
10
9
8
pH 7
<
_J
<
UJ
5
4
MIN
/
- /
I2FT/HR/
/
0
100
>:
I
I
150 200 250
ALUM DOSE, mg/ £
300
10
9
8
o>
E- 6
9 3
CO
LJ
cc. 2
I
0
pH8
MAY XXI2FT/HR
MAXv x
V x
^ ^ & j i &. i
MAX
9 FT/HR
MIN
I
100 150 200 250
ALUM DOSE,mg/j£
300
FIG. 10 A RESIDUAL ALUMINUM IN EFFUENT
-------
10
9
8
E 7
6
5
(f)
UJ
(T
2
I
0
DIAMETER: 24 INCHES
UPFLOW VELOCITY AS SHOWN
pH 9
* ^ /MAX
I2FT/HR^- J
.4 — — "*"" \MIN
I
MIN
100
300
150 200 250
ALUM DOSE, mg/jZ
FIG. 10 B RESIDUAL ALUMINUM IN EFFLUENT
42
-------
noting that the range of pH for minimum solubility of aluminum hydroxide
is from about 5 to 7^=418) and that the final influent pH varied with the
dose and the initial pH (cf. Table 2). Thus for the curves in Figure 10A
for pH 7, the increasing dose probably depressed the final pH to values
favoring dissolution of the aluminum and consequent higher residual
values. The curves for pH 8 indicate that increasing alum dose may have
caused the final pH to decrease from near the upper range of relative
insolubility (pH 6.4 or greater) to a value near that for minimum solu-
bility (pH 5.6). The final'set of curves for pH 9 indicates the same ef-
fect, only the final resulting, pH was higher (6.7 to 6.8) causing some
solution of the aluminum.
The effect of upflow velocity on the residual aluminum can be explained
for the pH 8 and pH 9 results by reasoning that once floes have formed
under the proper combination of pH and coagulant dose, the major cause
of residual A1+++ would be from the elutriation of these floes from the
surface of the blanket. Thus a lower upflow velocity will result in
less aluminum loss, which is borne out by the results. If, however, ap-
preciable aluminum remains in solution, as for pH 7, the upflow velocity
will not affect the loss. This is at least partially supported from the
results shown in the plot for pH 7 where the velocity effects overlap
considerably. , •" '.
In their experiments coagulating river water, Miller and West (16) ob-
served that increasing pH of the coagulating mixtures in the range 5.5
to 8.0 increased the aluminum in the effluent slightly. "They observed
that "contrary to expectations" the residual aluminum increased with
dose and with upflow velocity, particularly at the higher up'flow veloci-
ties (18 ft/hr), although the increase with dose was less at upflow ve-
locities of 6 to 10 ft/hr. This agrees with the present1,result shown
for 12 ft/hr. It should be noted that the alum dose in the Miller and
West study did not exceed about 55 to 60 mg/1 and their curves for the
lower upflow velocity decrease in slope with increasing dose. The
slope may well be reversed at higher doses. ;
Sludge Production
The significant terms in the .-regression equation were found to be
X1X3 = D x A1
X;LX4 = D x L
X3X4 = Al x L
X,X0X. = D x Al x L.
134
The remaining variables did not contribute significantly to the regres-
sion. The correlation is relatively poor in comparison with those for
the other variables, accounting for 68.7 percent of the variance and
43
-------
giving a standard deviation of 2.5 volume percent. This result indi-
cates that other factors may have been responsible for the sludge produc-
tion rate, e.g., the differences between 12- and 24-inch diameter column
construction which could have influenced the thickening of the sludge
below the inlet distributor.
The curves shown in Figure 11 are for the 24-inch diameter column only.
Maximum and minimum curves for blanket depths of 3 and 7 feet are shown.
Only results for pH 8 are shown, since the results for both pH 7 and
pH 9 give negative values over parts of the dose range, a result of the
poor correlation.
It is difficult to interpret the sludge curves with a high degree of
confidence. It is seen generally that the deeper blanket produced less
volume of sludge, which probably reflects the better flocculation achieved
by a large number of particles. The sludge volume decreases with increas-
ing dose for the 3 foot depth, but can increase with dose for the 7 foot
depth.
BOD5 Concentration in the Effluent^
A regression between BOD5 and TOG in the effluent was made with the fol-
lowing result:
BOD5 = 2,10(TOC) - 12.1.
This correlation was found to account for 47 percent of the variance
and gave a standard deviation of 3.5 mg/1 ZOD^. While not as good a
correlation as desired, the relationship is of value in predicting the
effect of the operating variables on the BODc via the TOG correlation.
Summary of Results of Experiments Using Alum
Table 3 summarizes the results of the statistical analysis and includes
the statistics representing the goodness of fit of the regression equa-
tions and the statistically significant terms. The final column indi-
cates whether or not the effect of influent on the effluent was signifi-
cant.
Table 4 summarizes the treatment effects predicted at the best combina-
tion of operating variables within the ranges studied. Included in the
table are maximum and minimum predicted values to indicate expected
variation due to the total effect of non-significant variables. In some
cases the predicted value coincides with one of the limits. The values
of the operating variables chosen for the predictions are:
alum dose : 250 mg/1
upflow velocity: 9 ft/hr
blanket depth : 7 ft
column diameter: 24 in.
pH : 8.
-------
DIAMETER: 24"
UPFLOW VELOCITY : 9 TO 12 FT/HR
WASTE WATER pH:8
C/5
h- 20 r-
UJ
ID
O I4l_
UJ
O
o:
UJ
CL
LU Pi-
§
uf
o
Q
100
150 200 250 300
ALUM DOSE, rr\q/jL
FIG. II SLUDGE PRODUCTION
45
-------
D =
TABLE 3
Summary of Regression Analysis of Experimental Data Using Alum
column diameter, in.; U = upflow velocity, ft/hr; Al = alum dose, mg/1;
L = blanket depth, ft; pH = pH
Dependent Variable
Effluent Suspended Solids, mg/1
Effluent TOG, mg/1
Effluent Soluble Phosphorus , mg/1
Effluent Total Phosphorus, mg/1
Effluent Residual Al*4"4", mg/1
Volume Percent Sludge
Percent
Variance Standard
Removed Deviation
78.1 ± 4.4 mg/1
75.8 ± 3.5 mg/1
76.9 ±0.94 mg/1
74.5 ±1.2 mg/1
70.9 ±1.4 mg/1
68.7 ±2.5 percent
Significant Terms
in
Regression Formula
pH
Al x Al
pH x pH
U x pH
Al
Al x pH
Al x Al
Al x pH x pH
pH
Al x PH
pH x pH
Al
Al x pH
U x U
D x Al
D x L
Al x L
D x Al x L
Influent
Effect
yes
yes
yes
yes
no
no
-------
TABLE 4
Summary of Predicted Treatment Effects Using Alum
(Numbers in parentheses are corresponding percent
removals based on average influent concentration)
Contaminant
Suspended Solids
Total Organic Carbon
Soluble Phosphorus (as P)
Total Phosphorus (as P)
BOD
Residual Al*"^1"
Average
Influent
Concentration
mg/1
176.3
130.1
6.4
9.3
145
0.95
Maximum
Predicted
Effluent
Concentration
mg/1
31 (83)
25.5 (80)
0.9 (86)
1.8 (81)
41.4 (71)
3.5
Predicted
Effluent
Concentration
mg/1
2.4 (98)
22.7 (83)
0.8 (88)
0.9 (90)
35.5 (75)
1.3
Minimum
Predicted
Effluent
Concentration
mg/1
2.4 (98)
15.9 (88)
0 (100)
0.8 (92)
21.3 (85)
1.3
Sludge Production, volume % of influent
14.1
14.1
5.1
-------
The predicted effects of different values of the variables within the
ranges studied may be found from the preceding graphs or by using the
regression formulas. Extrapolations are not warranted, however.
48
-------
SECTION VIII
RESULTS AND DISCUSSION OF EXPERIMENTS USING FERRIC CHLORIDE
Limited data were obtained using ferric chloride as a coagulant, the in-
tent of the experiments being to provide a reference basis with which to
compare the results of using alum. Since column diameter was found to
have little effect on contaminant removal effectiveness in the alum ex-
periments, it was not studied using ferric chloride, and all runs were
made in the 12-inch column for convenience. A single level of blanket
depth, 7 feet, was used. Coagulant dose levels were set to approximate
two values, 129 and 212 mg/1, which correspond to a similar molar Fe+++
equivalent as Al44"1" at 150 and 250 mg/1 alum.
Data Correlation
The model chosen to correlate the ferric chloride data included terms
representing those variables shown to be important in describing the alum
data: pH, coagulant dose (Fe), and upflow velocity (U). Neither diame-
ter nor blanket depth was studied. Owing to the limited data only linear
terms were used in the regression model, except for velocity, which is
accounted for by both linear and quadratic terms. The regression model
is:
Y. = b . + c .X . + E a..X. + cn.X?
J °J oj °3 1=1 ij i lj 1
where: Yj = concentration of j measured in the effluent
XQ-S = concentration of j in influent, mg/1
X1 = upflow velocity- U, ft/hr
X? = ferric chloride dose, Fe, mg/1
X = pH of the wastewater before coagulant addition
? 2
Xr^ = (upflow velocity) .
The results of the regression analysis of the data using the above model
are shown graphically in Figures 12, 13, 14, 15, and 16. Values of re-
gression coefficients are given in Appendix B. All effects are shown as
straight lines as required by the model. The quadratic effect of U2 does
not appear owing to the manner of representation.
Table 5 lists the results of the statistical analysis. The standard
deviations are slightly less than those for alum. Although the percent
variance removed is markedly greater, this is due in part to the small
number of degrees of freedom available (six parameters are fitted to
eight data points). The data for the ferric chloride runs are shown in
Appendix A.
49
-------
Ul
o
DIAMETER 12 INCHES
BLANKET DEPTH:7 FEET
r UPFLOW VELOCITY .
9 FT/HR
CO
100 150 200 250
FERRIC CHLORIDE DOSE mg//
en
E
UJ
UJ
CO
Q
O
CO
Q
UJ
Q
Z
UJ
CL
CO
r)
CO
80
70
60
50
40
30
20
10
0
UPFLOW VELOCITY.
12 FT/ H R
I
100 150 200 250
FERRIC CHLORIDE DOSE mg//
FIG 12 SUSPENDED SOLIDS IN EFFLUENT,
-------
DIAMETER: 12 INCHES
BLANKET DEPTH!? FEET
90 r UPFLOW VELOCITY:
80(-^ 9 FT/HR
£60
LU
1550
t40
UJ
z 30
020
o
I- 10
0
90
N80
\
^70
UJ
60
50
40
UJ
o
O
10
0
r UPFLOW VELOCITY:
12 FT/HR
pH9
100 150 200 250
FERRIC CHLORIDE DOSE,mg/£
100 I5O ,200 250
FERRiC CHLORIDE DOSE,mg/j£
FIG.13 TOTAL ORGANIC CARBON IN EFFLUENT
-------
Ln
ro
DIAMETER 12 INCHES
BLANKET DEPTH •? FEET
en „
e 4
LJ
z:
ID
O
Q.
CO
O
X
CL
J
O
UPFLOW VELOCITY: 9 FT/HR
0
1 100 150 200 250
FERRIC CHLORIDE DOSE mg/
0>
E
u.
u_
u
^ 2
CO
:D
cc
o
CO
o
X
Q.
O
CO
0
UPFLOW VELOCITY: 12 FT/HR
100 150 200 250
FERRIC CHLORIDE DOSE mq/S.
FIG. 14 SOLUBLE PHOSPHORUS IN EFFLUENT
-------
UJ
en
cc
o
a.
o
X
Q_
DIAMETER: 12 INCHES
BLANKET DEPTH:7 FEET
Q- 5
0>
E
0
UPFLOW VELOCITY: 9 FT/HR
• 100 150 200 250
FERRIC CHLORIDE DOSE.mgAe
Q- 5
o>
E
UJ
4
uj 3
CO
r>
a:
o
x
a.
0
UPFLOW VELOCITY: 12 FT/HR
~ 100 150 200 250
FERRIC CHLORIDE DOSE, mq/Jt
FIG. 15 TOTAL PHOSPHORUS IN EFFLUENT
-------
DIAMETER: 12 INCHES
BLANKET DEPTH: 7 FEET
O>
LJ
CO
LJ
_ UPFLOW VELOCITY .
9 FT/HR
12
10
8
6
4
0
100 150 200 250
FERRIC CHLORIDE DOSE, mg/£
o>
E
LU
14
12
10
8
Q 2
CO
£ 0
- UPFLOW VELOCITY: 12 FT/HR
pH7
pH8
pH9
100 150 200 250
FERRIC CHLORIDE DOSE, mg/£
FIG. 16 RESIDUAL Fe"^ IN EFFLUENT
-------
TABLE 5
Summary of Regression Analysis of Experimental Data
Using Ferric Chloride
Dependent Variable
Effluent Suspended Solids, mg/1
Effluent TOG, mg/1
Effluent Soluble Phosphorus, mg/1
Effluent Total Phosphorus, mg/1
Effluent Residual Ye+++ , mg/1
Volume Percent Sludge
Percent
Variance Standard
Removed Deviation
91
96
92
99
60
4.2%,
± 3.1
± 2.3
± 0.61
± 0.55
± 1.9
constant within ex
Influent
Effect
yes
yes
no
yes
no
perimenta
error
55
-------
All correlations except the concentration of residual Fe"1"4"4" show that
increasing coagulant dose and decreasing the pH results in a lower con-
centration of contaminant in the effluent. The reverse was found for
the residual Fe'''.
The correlations show that the concentration of suspended solids, total
organic carbon, and total phosphorus all decrease with increasing veloc-
ity, while soluble phosphorus increases. Examination of the regression
parameters a-j_ and C]_ reveals, however, that the effect of velocity is
reversed at 12 ft/hr for suspended solids and TOC, at 12.5 ft/hr for
soluble phosphorus, and at 13 ft/hr for total phosphorus. Extrapolation
of this effect beyond 15 ft/hr is unwarranted. Increasing velocity in-
creases the residual Fe' ' h in the effluent.
Three of the regression analyses showed a significant effect of contami-
nant influent concentration on the concentration in the effluent. The
following corrections should be added to the effluent predicted from
the numerical values shown within the parentheses:
Suspended solids : 4- 0.300(C - 181.6)
s
Total organic carbon: + 0.593(CC - 131.4)
Total phosphorus : + 0.420(Cp - 11.26)
where: Cs = influent concentration of suspended solids, mg/1
C = influent concentration of total organic carbon, mg/1
C = influent concentration of total phosphorus, mg/1, as P.
The correlation for sludge production showed that the sludge production
rate was constant at 4.2 volume percent of the influent within the ex-
perimental error.
Discussion
It is of interest to compare the results using ferric chloride with those
of McLellon et al (14). These investigators studied the coagulation of
secondary effluents using ferric chloride in jar test apparatus with auto-
matic titrimeter pH controls. They concluded that the optimum pH for the
removal of turbidity and phosphorus was between 5 and 6 for any applied
dose of ferric chloride up to about 200 mg F6+++/1. When the pH of the
coagulatingmixture was allowed to vary with dose, the tests indicated
optimum Fe doses depending on the alkalinity and initial turbidity
of the wastewater, showing quite clearly that a critical coagulation con-
centration (ccc) and a critical stabilization concentration (esc) existed
and corresponded to the minimum and maximum zeta potential of the waste-
water colloids. They showed also that the optimum dose resulted in the
optimum pH.
For a secondary effluent having an alkalinity of 130 mg/1, McLellon
56
-------
showed that the optimum ferric chloride dose for turbidity and phosphorus
removal was about 40 to 45 mg Fe+++/l, Or 192 to 216 mg FeCl3 • 6H20/liter,
where phosphorus removal was essentially complete. This dose is close to
the higher level dose of the present study (212 mg/1) at essentially the
same alkalinity (140 mg/1). Smaller doses would therefore be expected to
result in less effective removal as the results of the present study have
shown.- Increasing initial pH values would also decrease the effective-
ness of removal which is also borne out by the present study.
McLellon also showed that the optimum Fe+++ dose for COD removal was es-
sentially the same (47 mg Fe+"H-/l) as for turbidity and phosphorus.
McLellon et al concluded that "pH and the associated buffering effects
of alkalinity were primarily responsible for control of the destabiliza-
tion-restabilization reactions because coagulation of the hydrophilic
colloids was the result of the hydrolysis products of the metallic co-
agulant, which specifically interacted chemically with the ionogenic
groups of the colloidal surfaces to form polymeric bridges."
Since the coagulation mechanism is apparently responsible for major con-
taminant removal as colloids, increasing the effectiveness of coagula-
tion and subsequent flocculation would result in improved removal. This
may explain the lower effluent concentrations of suspended solids, TOC,
and total phosphorus occurring at higher upflow velocity in the present
experiments, since the higher velocity would result in greater shear
within the blanket. Soluble phosphorus removal, however, decreases with
increasing velocity.
The effect of variables on the residual Fe"1""1""1" in the effluent may be the
result of both solubility and adsorption. The residual Fe+++ increases
with increasing ferric chloride dose and velocity and decreasing pH.
Ferric hydroxides become more soluble as pH is decreased.
Summary of Results of Ferric Chloride Experiments
Table 5 summarizes the results of the statistical analysis and includes
the statistics representing the goodness of fit. It should be empha-
sized that the meaning of significance of the individual terms is not
relevant on this correlation owing to the small number of experiments.
Moreover, the regression model was chosen to reflect the influences of
those variables shown to be significant for alum coagulation.
Table 6 summarizes the treatment effects predicted at the best combina-
tion of operating variables within the limited range studied. From the
graphs of the correlations this combination is as follows:
Ferric chloride dose = 212 mg/1
Upflow velocity = 12 ft/hr
Blanket depth = 7 ft
pH = 7.2.
57
-------
TABLE 6
Summary of Predicted Treatment Effects
Using Ferric Chloride
Suspended Solids
Total Organic Carbon
Soluble Phosphorus (as P)
Total Phosphorus (as P)
BOD5
Residual Fe+++
Sludge Production
Predicted Average
Effluent Influent Percent
Concentration Concentration Removed
mg/1 mg/1
13.6
38.5
0.5
1.2
68.8
14.0
182
131
7.0
11.3
150
1.4
92
78
93
90
54
4.2 percent—'
I/
Volume percent of influent
58
-------
The value for pH was taken at 7.2 since this was the lowest value ac-
tually studied. In the graphs which gave predicted results less than
the minimum observed value, the latter value was taken as the predic-
tion and then corrected for influent effect where applicable. The av-
erage influent concentrations are also tabulated together with the per-
cent removal based on these influent concentrations. Sludge production
was assumed to take the mean observed value of 4.2 volume percent of
the influent. The minimum observed value was 3.0 percent.
Comparison of Ferric Chloride Treatment with Alum
Comparison of Tables 4 and 6 reveals that the quality of the effluents
from alum and ferric chloride treatment are generally comparable. The
notable exception is in the residual coagulant ion concentration. The
A1+++ concentration is essentially that of the background wastewater
concentration (0.95 mg/1) while the Fe"1""1""1" concentration is ten times
background (1.4 mg./l) . The amount of this difference is due partially
to the fact that variable combinations that were chosen provided a very
high residual iron. However, the lowest residual iron concentration
found was still 5 mg/1.
It should be noted that the best treatment using ferric chloride occurs
at a higher upflow velocity than for alum, 12 ft/hr versus 9 ft/hr, al-
though it was possible to achieve good treatment with both coagulants
at 15 ft/hr.
59
-------
SECTION IX
SLUDGE SETTLING STUDIES
The sludge from an operating clarifier must be processed to reduce its
bulk and to render it innocuous. It was therefore appropriate to study
the thickening properties of the sludge to see if there were any rela-
tionships between the settling velocity and the mode of sludge blanket
operation. More explicitly, sludge settling tests and suspended solids
analyses were performed
a) to study the character of each sludge
b) to evaluate the effect of changing test run conditions on the sludge
behavior
c) to compare the settling characteristics of the undisturbed sludge in
the upflow clarifier and the sludge obtained as an operational resi-
due
d) to estimate the area-flow-concentration relationships for a gravity
thickener using the solids flux-concentration procedure (7) .
Equipment and Procedure
The settling rate of the undisturbed sludge blanket was measured directly
in one of the 12-inch diameter Lucite columns. Settling tests on sludge
removed from the column were performed in a smaller Lucite cylinder
7.5 inches (18.7 cm) effective diameter and 54 inches (135 cm) in height.
These latter dimensions satisfy the minimal requirements of a settling
column as recommended by Cole (5) to avoid the interference of diameter
and height effects with settling rate.
The residual sludge obtained from the clarifier differed considerably
from the sludge comprising the blanket in the operating column:
The fluidized sludge found in the clarifier consisted of
large floes, had a moderate concentration of suspended
solids and good settling properties. In the following
this sludge is referred to as "blanket."
The collected sludge was of higher concentration than
the blanket owing to some thickening in the bottom of
the clarifier. The original large floes were broken
into very small ones when passing through the pipes
and fittings. This material is referred to as "sludge."
To illustrate the difference between a blanket and a sludge of the same
test run, a height versus time curve is drawn for both slurries in Figures
17 and 18, the slope of the curve being equal to the settling rate of the
liquids-solids interface. The settling curve for the sludge follows the
61
-------
220V-
200 -
40 -
I I I I
0 20 40 60 80 100
TIME (min)
FIG.I7 SETTLING CURVE FOR BLANKET OF TEST RUN Al
62
-------
120
E
i 100
u 80
X
60
40
I I I I I I I I I I I I I I
20 40 60 80 100 120 140
TIME (min)
FIG. 18 SETTLING CURVE FOR SLUDGE OF TEST RUN Al
63
-------
settling pattern described as typical for zone settling. For the blan-
ket the pre-flocculation period is missing, since the particles were
not disturbed and already developed to the maximum size for given op-
erational conditions.
Settling Test for the Blanket
The settling test for the blanket was performed in the clarifier column
itself. Immediately preceding the settling test, samples were with-
drawn from the fluidized bed at four levels, and through the bottom tap
for the lowest, non-fluidized section of the clarifier. After cutting
off the influent stream to the clarifier, the expanded blanket began to
settle. The height of the interface and time intervals were recorded,
and the settling rate of the blanket was determined as the slope of the
straight portion of the curve obtained when height was plotted against
time.
Settling Test for the Sludge
The procedure for the settling test for the sludge was the following:
The collected waste was mixed thoroughly in the collection container and
transferred to four smaller buckets. These were poured rapidly into the
8-inch test cylinder, up to 115 to 130 cm deep, and stirred once more by
hand before the interface readings were taken in regular time intervals,
depending on the settling velocity, between 1 and 10 minutes. The set-
tling rate was then determined by measuring the slope of the straight
portion of the curve of the interface height versus time.
Two sludge samples were taken from the collection bucket while filling
the smaller buckets and sent to the laboratory for suspended solids
analysis.
Since the volume of sludge required for the settling test, about 10 gal-
lons, was obtained over a variable but relatively long period of time,
it had to be of average composition.
Evaluation of the Experimental Data
Over fifty settling tests were performed in the course of this study to
find relationships among the settling rates and the characteristics of
the blanket, sludge, and the different test run conditions. Half the
settling tests were made with dilutions and concentrations of collected
sludges, a step required to estimate solids flux-concentration curves
needed for thickener design.
The following statistical problems were formulated:
a) Correlation between the suspended solids concentration and posi-
tion (height above the distributor) in the fluidized blanket.
64
-------
b) Correlation between the suspended solids concentration in the
clarifier and test run conditions (alum dose and upflow velocity).
c) Correlation of the blanket settling rate and the blanket suspended
solids concentration, alum dose, and upflow velocity.
d) Correlation of the sludge settling rate and the sludge concentra-
tion, alum dose, and upflow velocity.
Problem (a) . The data in Table 7 show the suspended solids values ob-
tained by sampling the blanket at different levels, for different test
run conditions and for replications of the same test run. Samples of
sludge were taken from the sludge discharge line (bottom) and from each
of four taps in the side of the column. Tap 1 was located 1 foot from
the bottom and the remaining three spaced at 2 foot intervals (see
Figure 1) . Sludge settling test runs Bl and C4 have incomplete sample
sets.
It is obvious from the data that the suspended solids concentration is
much higher in the bottom part of the clarif ier, which can be explained
by the thickening effect in the column below the feed distributor. For
the remaining samples, taps 1 to 4, a concentration pattern is not ob-
vious, and for these samples only an analysis of variance was performed,
omitting the data from the incomplete sets Bl and C4, since the analysis
did not permit missing observations.
Based on prior lab experience with suspended solids analysis, a value of
150 mg/1 was adopted for the standard deviation of the suspended solids
values for the blanket. Using this value, the analysis of variance
showed that the difference in sample means for different blanket posi-
tions was not significant at the 5 percent confidence level. This ob-
servation is consistent with Tesarik (21), who observed that the con-
centration of floe throughout the blanket was nearly constant.
Tesarik's data showed the concentration to decrease in the lowest 20
percent of the blanket, but his column was different from the ones used
in this study in that the entire contents were fluidized and no thick-
ening of sludge was achieved in the bottom.
Because of this finding it was then possible to characterize the whole
blanket by a single SS concentration value. Since the sample at any
level was shown to be representative of the entire blanket, it was then
possible to use the incomplete wample sets, Bl and C4, in further evalua-
tion.
Problem (b) . In Table 7, column 8 lists the average SS concentrations
from the different heights in the blanket, excluding the bottom sample
(column 3). Column 9 is the average of concentrations from column 8 for
the given test run replicates. This value is the characteristic blanket
65
-------
TABLE 7
Data of SS from the Upflow Clarifier
Run Run
Condi- Number
tion*
A A1
A2
Bl
B B2
B3
Cl
C2
C C3
C4
Bottom Tap 1
(3) (4)
5,898 1,219
4,624 1,049
614
1,363 898
849 801
752 694
7,884 1,107
Tap 2
(5)
667
1,098
994
697
989
871
714
1,024
1,187
Tap 3
(6)
917
1,231
732
904
846
812
1,137
Tap 4
(7)
486
802
620
782
723
728
1,031
1,144
Average Average
of tap of test run
1 - 4** condition**
(8) (9)
822
1,045
994
666
893
810
737
1,075
1,166
993
803
915
Dl
8,192 1,242 1,367 1,363 1,320 1,323
1,323
* Run Condition
A
B
C
D
Upflow Velocity, U
9 ft/hr
12 ft/hr
12 ft/hr
9 ft/hr
Chemical Dose, Al p_t
250 ppm as alum 8
250 ppm as alum 8
150 ppm as alum 8
150 ppm as alum 8
The SS from bottom samples are not included in the averages,
Remark: All values ± 150 mg/1.
66
-------
concentration for a test run condition.
From column 8 it is clear that replicates of a test run often resulted
in blanket concentrations exceeding the standard deviation of the analy-
sis, and one concludes that the blanket reproducibility is not good.
An analysis of variance made on the average values given in column 9
showed that the differences were not statistically significant. The
differences in upflow velocity between runs may have been too small to
result in a detectable difference due to blanket expansion. For ex-
ample, the formula of Richardson and Zaki (17) which Brown and LaMotta
(2) have shown applicable to alum floes would predict only 5 or 6 percent
change in flow concentration if the ratio in upflow velocities were
only 9/12.
Problem (c) . Table 8 presents the measured settling rate of the blan-
ket together with the SS concentration of the blanket and the test run
conditions (upflow velocity, U, and alum dose, Al). The remaining
test run conditions, pH and blanket depth, were held constant at pH 8
and 7 feet, respectively.
Several forms of statistical relationships were tried between the set-
tling rate (SR) as the dependent variable and SS, U, and Al as the inde-
pendent variables. The most successful was the following determined by
a regression analysis:
log SR = 0.1457 + 0.0003 SS + 0.0286 U + 0.0002 Al . (9-1)
This regression formula accounted for 56 percent of the variations in
the settling rates observed, with the upflow velocity accounting for
47 percent.
The hypothesis that the settling rate was no different from the upflow
velocity was further tested by analysis of the regression equation:
log SR = a + b log U . (9-2)
It was found that (a) was not statistically different from zero and (b)
was not different from 1.0 at the 95 percent confidence level; and
therefore, the settling rate of the blanket was not different from the
upflow velocity. This result is consistent with the findings of Richardson
and Zaki (17) for different types of particles and supports the conclu-
sion of Brown and LaMotta (2), confirming the application of Richardson
and Zaki to alum floes.
It was concluded that the significant variable affecting the blanket
settling rate is the upflow velocity.
67
-------
TABLE 8
Observed Parameters of the Clarifier
Run
Number
Al
A2
Bl
B2
B3
C2
C3
C4
1
Blanket Blanket Up flow
Settling Rate Concentration Velocity
(SR) (SS) (U)
cm/min
3.96
5.0
5.13
5.46
4.85
4.64
5.74
5.12
4.10
ft/hr
7.8
9.85
10.2
10.8
9.6
9.1
11.4
10.1
8.1
mg/1
822
1,045
994
666
893
737
1,075
1,166
1,323
ft/hr
9
9
12
12
12
12
12
12
9
Chemical
Dose
(Al)
mg/1 as alum
250
250
250
250
250
150
150
150
150
68
-------
Problem_(d)_. Table 9 gives the settling rates and SS concentrations
for the sludge produced during the test run conditions shown. As in
the study of the blanket settling rates, pH and blanket depth have been
held constant.
An analysis similar to that of Problem (c) was made with the result that
the following correlation accounts for 96 percent of the variation in
the settling rate of the sludge:
SR = 7.452 - 0.001 SS - 0.266 U - 0.005 Al . (9-3)
Using the logarithm of the settling rate results in a slightly inferior
correlation, accounting for 88 percent of the variation:
log SR = 3.790 - 0.001 SS - 0.207 U - 0.002 Al . (9-4)
It was also determined that the SS term in (9-3) accounts for about 72
percent of the variation, and in (9-4), 77 percent of the variation.
The remainder accounted for by the upflow velocity and the alum dose is
probably the result of a slight change in floe character.
It was concluded that clarifier operating conditions had no practical
effect on the settling rate of the sludge. It must be borne in mind
that the test runs studied were made over several weeks' time and that
differences in the wastewater quality, particularly the concentration
of suspended solids, probably are responsible for much of the variability
in the character of the sludge.
Sludge Thickener Design
Consideration of the means of disposing of sludge from a clarifier leads
naturally to the question of determining the thickener area required to
achieve a thickened sludge of required solids concentration. This sec-
tion presents the results of a study to estimate the area of a thickener
from the settling properties of sludge produced by the clarifier. The
most direct means of doing this is by the solids flux-concentration
method (7) which requires determination of the settling rate as a func-
tion of solids concentration. Having found the settling rate as a func-
tion of concentration, a flux-solids concentration curve is calculated
by plotting the product (solids concentration) x (settling rate) versus
solids concentration. For a given underflow solids concentration, the
minimum flux is then determined as the intercept of the straight line
through the underflow concentration tangent to the flux curve. The area
required is then found from the relationship:
A = ^fo (9_5)
Gmin
69
-------
TABLE 9
Observed Parameters of Sludge Tests
Run
Number
A SR
A SR2
B S
B SI
C SR
C SR2
D S
D SR
D SR3
Sludge
Settling
Rate
(SR)
cm/min ft/hr
0.25
0.05
1.69
1.75
2.38
0.04
2.63
0.54
0.02
0.45
0.10
3.33
3.45
4.70
0.08
5.19
1.06
0.04
Sludge
Concentration
(SS)
mg/1
4,618
4,999
1,559
1,559
752
4,948
2,272
5,411
8,635
Clarif ier
Up flow
Velocity
(u)
ft/hr
9
9
12
12
12
12
9
9
9
Chemical
Dose
(Al)
250
250
250
250
150
150
150
150
150
70
-------
rt
where: A = thickener cross-sectional area, L
Qo = influent rate to the thickener, L3/9
O
C0 = influent solids concentration, M/L
o
G = minimum solids flux, M/L e.
The method assumes an overflow having zero suspended solids. A typical
flux-concentration curve showing the required construction is given in
Figure 19.
The data for the settling rate concentration curves were obtained by re-
peated measurements of the settling rate at different concentrations of
sludges produced from test runs having the operating conditions of experi-
ments A and B (cf. Table 7). These concentrations were varied by remov-
ing part of the supernatant after each settling test and remixing the re-
maining sludge.
Since the settling-remixing operations took considerable time, the ques-
tion of changing settling properties arose. Two sludges were reused two
and three times, respectively; and the settling rates were measured with
the following result:
Settling Test Settling Rate Settling Test Settling Rate
Ba 1.69 cm/min BS1 0.43 cm/min
Ba' 1.75 cm/min BS2 0.46 cm/min
BS3 0.46 cm/min
It was concluded that no substantial difference in settling rates re-
sulted from reuse and aging; and therefore, the remixing procedure could
be used with confidence.
The data for the settling tests are listed in Appendix C and are plotted
in Figure 20. Test runs A and B differed only in upflow velocity, and
since the settling rates for sludge did not appear to depend on velocity
in a significant way, a correlation of settling rate with suspended
solids concentration was obtained by regression analysis. The preferred
correlation was in the form:
log SR = 0.6252 - 0.000300 SS (9-6)
which accounted for 73 percent of the variation.
The form of equation (9-6) is characteristic to many sludges and lends
itself to easy mathematical analysis, allowing the flux-area-relationship
71
-------
1 mm
Cu Underflow Concentration
C | Limiting Sludge Concentration
'mm
Minimum Flux Required for
Clear Overflow
1000
2000 3000 4000 5000 6000
SS (mg/l)
FIG. 19 FLUX CURVE
7000 8000
72
-------
LOG SR= 0.6252-0.000300SS
O
O
0 1000 2000 3000 4000 5000 6000 70QO
SUSPENDED SOLIDS CONCENTRATION, mg/l
FIG.20 SLUDGE SETTLING RATES VS SUSPENDED SOLIDS CONCENTRATION
73
-------
to be expressed by a relatively simple formula. To derive this formula,
note that in Figure 19, the flux curve, the straight operating line must
be tangent at some value of concentration, C]_. The equation for the op-
erating line is:
G = G . (1 - TT-) (9-7)
mm Cu
and its slope is seen to be -Gmin/Cu.
From equation (9-6), putting C = SS, the flux curve equation is simply:
G= a0Ce-b'C, -JpL_ (9-8)
cm min
, (2.303)(.6252) . ,, , .
where aQ = e =4.23 cm/mm
b' = (2.303)(.000300)(106), cm3/gm
C = gm/cm3 = (mg/1 x 10~6)
The condition of tangency at C-. is thus:
(9-9)
-b'C, cl
l = Gmln(l - —) (9-10)
Solving equations (9-9) and (9-10) simultaneously gives the value of
Ci as:
= Cu
b'Cu + \(b'Cu)2 - 4b'Cu
2b'Cu
(9-11)
Solving (9-10) for G . gives:
-b'C
Gmin = x _ c /P,, (9-12)
and from equation (9-5)
74
-------
(1 - C./Cu) ,,_ 2 •
TrV " r1- - a r e ' ' ^^ <9-13>
Q-C G ,- , - a L, , em
^o o mm o 1
Figure 21 shows A/QOCO plotted against Cu for values of Cu in mg/1 and
A/Q0C0 in (ft2)(1000 gal./day)(mg/1). To illustrate the use of Figure
21, consider the problem of estimating the area of thickener required to
produce a thickened sludge of SS concentration of 10,000 mg/1 from a di-
lute sludge of SS concentration of 1000 mg/1 if the sludge volume rate
is 400,000 gal./day. At Cu = 10,000, AO/QOCO = 4.1 x 10~3, and
A = 4.1 10~3 400 1000 = 1640 ft2.
A thickener with circular cross section would need to be 46 feet in di-
ameter.
75
-------
400
300
200 -
"o 100
* 80
*o> 60
l
§ 40
o
o
8
20
o
o
o
o
\
<
II
10
8
6
c
E
8
!0
12
14
16
18
20
Cu UNDERFLOW CONC. SS mg/f xlO
-3
FIG. 21 RECIPROCAL MINIMUM SOLIDS FLUX VS THICKENER
SOLIDS CONCENTRATION
-------
SECTION X
EXPERIMENT USING ACTIVATED SILICA AS A SLUDGE THICKENING AID
A controlled experiment was made in the 12-inch diameter columns using
activated silica with alum to determine the effect of activated silica
in thickening the sludge. Operating conditions were chosen to repre-
sent a situation normally producing a high sludge volume rate so that
a significant effect in reducing sludge volume could be most apparent.
These conditions were
U = 15 ft/hr
L = 7 ft
Al = 200 mg/1
pH = 7.8.
Philadelphia Quartz Company "N"-type sol was activated according to the
manufacturer's instructions using sodium bicarbonate. A dose of 30 mg/1
was used. The activated sol was introduced into the feed line at the
same point as the alum solution in one of the 12-inch columns. The
other column was operated in an identical manner with the same waste-
water but with no silica addition.
Table 10 lists the average influent, effluent and sludge analyses for
the control and the corresponding data using silica. As can be seen,
the effects of using the silica are substantial, the major effect being
the increased concentration of sludge constituents in the silica run.
The sludge volume is 57 percent of the control value.
77
-------
TABLE 10
Treatment Effects Using Alum with Activated Silica-
l/
Parameter
Suspended Solids
Total Organic Carbon
Soluble Phosphorus
Total Phosphorus
Residual Al
BOD5
Volume Percent Sludge
Sewage
165 mg/1
193
8.7
13.3
2.9
126
With Silica
Effluent
12
34
0.8
1.4
1.4
29
Sludge
1,348
404
—
45.5
105
—
19.4
Control
Effluent
18
35
1.0
1.2
1.5
38
Sludge
796
286
—
37.5
57
—
33.6
— All results in mg/1 except sludge volume
78
-------
SECTION XI
ACKNOWLEDGMENTS
The contributions and continued interest of the following persons are
gratefully acknowledged:
B. H. Carpenter, for the experimental design and statistical analyses
of the data;
C. N. Click, for material assistance with the design, construction, and
operation of the apparatus;
I. E. Berninger, for obtaining the sludge settling data and operation
of the apparatus;
Eliza S. Rucker, for secretarial and computational assistance.
The assistance of the staff of the UNC Wastewater Research Center,
Professor J. C. Brown, Director, and W. C. Walker, Chemist, was of par-
ticular importance.
Professor Charles O'Melia, UNC School of Public Health, for helpful
discussions of the mechanisms of phosphorus removal.
The support of the project by the Environmental Protection Agency and
the help provided by Mr. James F. Kreissl, Project Officer, are ac-
knowledged with sincere thanks.
79
-------
SECTION XII
REFERENCES
1. Ames, L. L., Jr., and R. B. Dean, J. Water Pol. Cont. Fed. 42,
R161-R172, May (1970). —
2. Brown, J. C., and E. LaMotta, Proc. Am. Soc. Civil Engrs. 97,
SA2, p. 209-224 (1971). —
3. Cochran, W. G., and G. M. Cox, "Experimental Design," Second Edi-
tion, John Wiley and Sons, New York, 1957, cf. pages 148,342, and
347.
4. Cohen, J. M. , G. A. Rourke, and R. L. Woodward, J. Am. Water Works
Assoc., 51, p. 1255-1267 (1959).
5. Cole, R. F., "Experimental Evaluation of the Kynch Analysis," Ph.D.
Thesis, University of North Carolina School of Public Health,
Chapel Hill, N. C., 1968.
6. Gulp R. L., and G. L. Gulp, "Advanced.Waste Treatment," Van Nos-
trand Reinhold Company, N. Y., 1971, p. 37.
7- Dick, R. I., and B. B. Ewing, Proc. A.S.C.E. (SED) SA4, p. 9-29,
August (1967).
8. Eberhardt, W. A., and J. B. Nesbitt, J. Water Pol. Cont. Fed., 40,
p. 1239-1267 (1968).
9. Falcoff, A. P., and R. E. Iverson, "API Users Manual," IBM, T. J.
Watson Research Center, Yorktown Heights, N. Y., 1968.
10. Grundy, R. D. , Environ. Sci. and Tech., 5_, p. 1184-1190, Dec. (1971).
11. Hanson, R. L., W. C. Walker, and J. C. Brown, "Variations in Charac-
teristics of Wastewater Influent at the Mason Farm Wastewater Treat-
ment Plant, Chapel Hill, North Carolina," Report No. 13, UNC Waste-
water Research Center, Dept. of Environmental Science and Engineer-
ing, University of North Carolina, Chapel Hill, 1970.
12. Lea, W. L. , G. A. Rohlich, and W. J. Katz, Sewage and Industrial
Wastes, 2_6_, No. 3, p. 261 (1954).
13. Malhotra, S. K., G. F. Lee, and G. A. Rohlich, Int. J. Air Water
Pol., j3, p. 487 (1964).
14. McLellon, W. M., T. M. Klinath, and C. Chao, J. Water Pol. Cont.
Fed., 44_, p. 77-91 (1972).
15. "Methods for Chemical Analysis of Water and Wastes," EPA, WQO,
Analytical Quality Control Laboratory, Cincinnati, Ohio (1971).
81
-------
16. Miller, D. G., and J. T. West, Floe Blanket Clarification
Part 1—Effect of physical variables using aluminum sulfate,
Water and Water Engineering, p. 240-242, June (1966)
Part 2—Effect of chemical variables using aluminum sulfate,
Ibid., p. 291-294, July (1966)
Part 3—Effect of physical and chemical variables using ferric
chloride, Ibid., p. 342-346, August (1966).
17. Richardson, J. F., and W. W. Zaki, Trans. Int. Chem. Engrs. , 32,
p. 35 (1965).
18. Sawyer, C. N., "Chemistry for Sanitary Engineers," McGraw Hill,
New York, (1960) .
19. Stumm, W., and J. J. Morgan, "Aquatic Chemistry," Wiley-Interscience,
p. 522 (1970).
20. Tenney, M. W., and W. Stumm, J. Water Pol. Cont. Fed., 37, p. 1370-
1388 (1965).
21. Tesarik, I., Jour. San. Engr. Div. Proc. Am. Soc. Civil Engrs., 39,
SA6, p 105 (1967).
82
-------
SECTION XIII
PUBLICATIONS AND PATENTS
No patents or publications have been produced as a result of this
project.
83
-------
SECTION XIV
APPENDICES
Page No,
A. Test Run Data
Summary of Test Run Data Using Alum
Table A-l: Measured Values of Independent Variables ... 86
Table A-2: Measured Values of Dependent Variables .... 88
Summary of Test Run Data Using Ferric Chloride
Table A-3: Measured Values of Independent Variables ... 91
Table A-4: Measured Values of Dependent Variables .... 92
B. Numerical Values of Coefficients in Regression Model
Table B-l: Alum Coagulant 94
Table B-2: Ferric Chloride Coagulant 97
C. Data for Thickener Design 98
85
-------
TABLE A-l
Summary of Test Run Data Using Alum
Measured Values of Independent Variables
Diameter Upflow Velocity
inches ft/hr
No. D U
1 12 8.6
2 12 9.2
3 12 9.4
4 12 9.0
5 12 8.9
6 12 11.7
7 12 11.6
8 12 12.0
9 12 12.2
10 12 11.8
11 12 12.1
12 12 12.4
13 12 12.1
14 12 12.6
15 12 9.2
16 12 9.0
17 24 8.7
18 24 9.1
19 24 11.6
20 24 11.9
21 24 12.0
22 24 13.7
23 24 11.9
24 24 9.0
25 24 9.0
26 24
27 24 8.8
28 24 9.2
29 24 12.0
30 24 12.0
31 24 12.1
32 24 8.9
33 24 9.8
34 24 11.9
35 12 9.1
Alum Dose
mg/1
Al
258
315
185
259
217
238
294
214
239
232
150
125
216
252
191
273
260
324
300
199
120
136
187
282
120
150
242
235
255
268
158
125
279
222
204
Depth
ft
L
7.0
7.0
7.0
2.75
3.0
7.0
7.0
7.0
3.0
3.3
3.0
3.0
7.3
7.2
3.1
7.0
7.0
3.3
7.3
3.5
3.0
3.5
7.0
3.0
3.5
7.0
7.0
3.0
7.0
2.25
7.0
3.0
7.0
3.0
7.0
PH
8.0
8.7
8.0
7.9
8.0
8.0
7.9
8.0
8.0
8.0
8.0
8.0
8.0
8.0
8.0
8.0
8.0
8.0
7.9
8.0
8.0
8.0
8.0
8.0
8.0
8.0
7.4
7.4
7.1
7.2
7.3
7.2
9.0
9.0
7.4
-------
TABLE A-l (Cont.)
Diameter Upflow Velocity Alum Dose Depth
inches ft/hr mg/1 ft
No. D U Al L PH
36 12 12.0 211 0.75 7.3
37 12 12.6 216 3.0 7.2
38 12 12.2 150 2.5 7.3
39 12 8.8 153 7.0 7.5
40 12 9.1 208 7.1 9.0
41 12 8.9 242 3.0 9.0
42 12 9.1 183 3.0 9.0
43 12 12.0 412 7.25 9.0
44 12 12.0 413 7.0 9.0
45 12 13.7 162 2.6 9.0
46 12 14.9 216 5.0 8.2
47 24 14.5 178 5.0 8.5
48 12 6.0 321 5.0 8.2
49 24 6.0 150 5.0 8.0
50 12 10.0 308 5.0 8.5
51 24 10.5 268 5.0 8.0
87
-------
oo
oo
TABLE A-2
Summary of Test Run Data Using Alum
Measured Values of Dependent Variables
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17 ,
18
19
20
Susp. Solids
i n
effluent
mg/1
18.0
18.0
18.0
34.0
24.0
14.0
13.0
10.0
47.0
37.0
44.0
40.0
20.0
22.0
38.0
14.0
6.5
88.0
42.0
25.0
Susp. Solids TOC
in
sewage
mg/1
208
328
213
274
210
252
198
195
198
196
210
142
196
161
161
180
198
161
196
195
in
effluent
mg/1
41.0
50.0
41.0
62.0
36.0
49.0
19.0
28.0
25.0
24.0
23.0
24.0
19.0
22.0
22.0
39.0
22.0
23.0
27.0
27.0
TOC
in
sewage
mg/1
201
197
192
224
166
176
102
87
186
89
115
151
135
165
102
113
107
214
186
214
Soluble P
in
effluent
mg/1
0.65
0.80
0.35
1.70
0.45
0.60
0.10
0.50
0.45
2.00
0.10
0.10
0.45
0.30
0.55
1.30
0.20
1.05
0.50
1.15
Soluble P
in
sewage
mg/1
10.5
8.60
8.50
10.6
7.70
9.50
4.90
7.40
4.90
8.30
5.90
3.40
5.90
5.60
5.60
8.60
4.90
5.60
5.85
7.40
Total P
in
effluent
mg/1
3.20
3.20
1.45
2.95
2.95
3.60
1.20
0.80
2.20
2.00
2.10
1.30
0.85
0.80
1.45
2.15
1.60
1.60
1.95
1.50
Total P
in
sewage
mg/1
15.2
15.2
14.0
12.6
12.6
16.3
7.25
8.95
7.25
8.25
10.0
5.55
8.25
8.75
8.75
12.40
7.25
8.75
8.25
8.95
Al
in
effluent
mg/1
1.30
2.60
1.30
4.60
6.60
1.50
1.60
1.10
5.00
5.20
3.10
3.10
0.55
0.70
2.55
7.80
1.20
5.50
5.60
3.50
Al
in
sewage
mg/1
1.20
10.0
0.85
1.00
1.55
2.20
2.60
0.50
2.60
0.20
0.50
0.10
0.20
0.50
0.50
0.90
2.55
0.50
0.20
0.50
Volume
Percent
Sludge
9.14
6.25
35.1
9.1
9.96
24.7
13.3
23.1
10.3
20.8
2.50
2.50
24.1
25.6
6.52
11.0
7.00
7.70
7.2
9.00
BOD
in
sewage
mg/1
-_
--
--
--
--
__
150
216
150
150
224
174
150
156
156
198
150
156
150
216
BOD
in
effluent
mg/1
--
--
--
--
--
__
37
45
77
83
24
35
24
28
29
59
37
26
37
36
-------
TABLE A-2 (cont.)
00
No.
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Susp. Solids
in
effluent
mg/1
56.0
17.0
36.0
42.0
40.0
23.0
10.0
53.0
33.5
37.5
29.0
29.0
30.0
92.0
6.50
17.5
32.5
90.0
76.0
10.0
27.0
94.0
11.0
19.0
178
Susp. Solids
in
sewage
mg/1
210
195
112
210
123
84.0
246
246
64.0
63.0
63.0
64.0
263
156
246
50.0
50.0
200
235
173
262
165
263
224
243
TOC
in
effluent
mg/1
29.0
30.0
30.0
28.0
4.50
8.00
28.0
31.5
12.5
8.50
8.00
9.50
39.0
57.0
26.0
6.00
9.00
11.0
57.0
43.0
44.0
64.0
27.0
31.0
101
TOC
in
sewage
mg/1
59.0
186
70.0
52.0
214
102
183
183
41.0
47.0
47.0
41.0
136
128
183
31.0
31.0
138
195
143
235
165
136
160
128
Soluble P
in
effluent
mg/1
0.30
0.80
0.10
0.30
0.30
0.10
0.45
1.05
0.50
0.60
0.30
0.10
1.55
0.90
0.35
0.10
0.10
4.30
3.40
0.45
0.60
4.30
0.40
0.55
7.80
Soluble P
in
sewage
mg/1
5.90
7.40
1.85
5.90
4.05
1.60
6.95
6.95
2.30
4.90
4.05
2.30
9.85
8.90
6.95
2.30
2.30
5.70
8.10
5.50
8.20
9.10
9.85
10.7
8.90
Total P
in
effluent
mg/1
3.10
1.20
0.95
2.10
2.40
0.95
1.20
4.15
1.10
1.40
1.20
1.40
2.75
5.10
0.80
0.45
0.95
6.10
5.80
1.15
1.60
8.40
1.30
1.05
12.2
Total P
in
sewage
mg/1
10.0
8.95
4.50
10.0
6.20
2.90
10.4
10.4
3.90
5.20
5.2
3.90
12.8
12.6
10.4
3.75
3.80
8.60
12.4
10.5
13.2
16.1
12.8
12.8
12.6
Al""
in
effluent
mg/1
3.40
2.00
4.60
3.20
3.40
4.10
1.25
4.75
5.80
4.20
2.75
2.90
3.55
7.70
0.45
0.45
4.20
0.10
5.50
2.80
4.20
10.9
0.60
1.15
11.6
Al
in
sewage
mg/1
0.50
0.50
0.75
0.50
0.20
0.05
0.20
0.20
0.10
0.45
0.45
0.10
0.70
0.45
0.45
0.10
0.10
0.50
0.10
1.10
1.60
1.45
0.70
1.25
0.50
Volume
Percent
Sludge
26.4
17.2
3.00
13.4
6.00
1.70
21.1
9.60
4.90
6.60
2.90
4.90
7.20
12.6
5.10
2.80
3.90
2.90
4.20
5.20
8.40
5.60
11.9
9.60
3.60
BOD
in
sewage
mg/1
224
216
60
224
66
69
204
204
156
--
51
165
174
138
204
42
42
--
96
_-
-_
174
168
138
BOD
in
effluent
mg/1
38
38
14
32
38
17
66
56
35
10
14
38
71
74
42
8
9
--
41
„__
__
._
77
52
114
-------
TABLE A-2 (cont.)
No.
46
47
48
49
50
51
Susp. Solids
in
e f fluent.
rag/1
23.0
11.0
5.50
12.0
19.0
28.0
Susp. Solids
in
sewage
mg/1
163
91.0
163
100
91.0
100
TOG
in
effluent
mg/1
17.0
11.0
14.0
16.7
14.5
13.0
TOC
in
sewage
mg/1
J03
65.0
103
71.0
65.0
71.0
Soluble P
in
effluent
mg/1
0.50
0.10
0.10
0.70
0.10
0. 1.0
Soluble V
in
sewage
mg/1
8.00
4.10
8.40
5.20
4.10
5.20
Total I
in
effluent
rag /I
1.30
1.10
2.00
0.70
0.85
1.10
Total P
in
sewage
mg/1
10.3
4.70
10.3
6.50
4.70
6.50
Al
in
effluent
mg/1
2.60
2.80
1.45
1.20
3.10
4.10
Al
in
sewage
mg/1
0.80
0.80
0.80
0.80
0.80
0.80
Volume
Percent
Sludge
31.0
12.0
4.40
4.20
28.6
6.10
BOD
in
sewage
mg/1
129
81
129
79.5
81
79.5
BOD
in
effluent
mg/1
51
8
68
12
11
17
-------
TABLE A-3
Summary of Test Run Data Using Ferric Chloride
Measured Values of Independent Variables
Upflow Velocity Ferric Chloride Dose
ft/hr mg/1
No. U pH
1 15.1 272 8.1
2 15.0 204 7.2
3 12*3 240 8.3
4 12.0 228 7.3
5 12.1 i 156 8.4
6 12.2 ' » 141 7.3
7 9.0 185 8.0
8 9.04 168 7.2
91
-------
TABLE A-4
Summary of Test Run Data Using Ferric Chloride
Measured Values of Dependent Variables
in
effluent
No. mg/1
1 39
2 16
3 38
4 7
s 45
3
-------
APPENDIX B
TABLES OF REGRESSION MODEL COEFFICIENTS
The following tables list the numerical values of the coefficients in
the regression formulas as shown.
Numerical values are presented in floating point form for convenience
in tabulation. Ten significant figures are reported and should be re-
tained to minimize round-off error during computation. The results re-
ported in the text were computed retaining all ten significant figures.
93
-------
TABLE B-l
Numerical Values of Coefficients in Regression Model-
Alum Coagulant
I/
= b
)
. =
+ d,.X X X + d.XX-X-X.X.
2j 2 4 5 3j 1 2 3 4 5
Effluent
Concentration
Suspended Solids
3 = 1
b = 2.401546952E3
c° = 6.750828988E-2
o
i . = 1.926031772EO
^J = -1.052067640E2
f = 1.745588551EO
2.199752526E2
5.468436417E2
Y. =
XJ -
Xl *
x2 =
X3 =
X4 =
X =
effluent concentration, mg/1
column diameter, D, inches
upflow velocity, U, ft/hr
coagulant dose, Al, mg/1
blanket depth, L, ft
pH before coagulant addition
concentration of j in influent, mg/1
, . -
Total Organic
Carbon
j = 2
1.313209375E3
7.496236799E-2
4.225776396EO
-1.015475574E2
5.645463167E-1
-1.243209470E2
-2.363326236E2
Total
Phosphorus
3 = 3
1.556026732E2
3.681840453E-1
6.074711403E-1
-5.069301940EO
2.417790192E-1
-1.026258881E1
-4.032835919E1
Residual
3 = ^
6.165965239E1
1.542031547E-1
1.679361330EO
-4.268367818EO
3.862958669E-1
-4.718036146EO
-2.656202754E1
Volume
Percent
Sludge
3 = 5
-7.011041523E2
-3.327649793E-3
6.593180037EO
2.163806100E1
8.220314923E-2
8.294901443E1
1.197318401E2
-------
!23j
45j
•
dlj
d2j
d3j
Effluent
Concentration
Suspended Solids
-1.124878379EO
3.248916887E-2
6.546904534E-1
3.463991390E-1
-6.768911337E-2
2.927582096E1
1.999163754E1
1.536683211E-2
-3.172370866E-1
3.044240732E1
-8.748592219E-1
2.429792677E-3
3.369529095E-1
2.743132044E1
-1.233381952E-2
-4.080446929EO
1.098116450E-4
Total Organic
Carbon
i = 2
-2.333677117E-2
7.956793165E-4
-2.742262237E-1
-5.342339549E-1
-1.386573337E-2
1.386364686E1
1.398587744E1
-1.368997021E-2
-3.976917707E-2
1.562904639E1
-2.653454709E-1
-2.399191249E-4
1.527309786EO
8.456590718EO
-8.935182704E-4
-1.866486940EO
1.940932126E-5
TABLE B-l (cont.)
Total
Phosphorus
j = 3
-7.957952689E-2
8.944425800E-4
8.715783150E-2
-1.530079432E-2
-6.135029152E-3
1.220129250EO
1.140034594EO
4.526250503E-3
-2.949983924E-2
1.480293490EO
-5.868436815E-2
8.609914011E-5
-5.499970381E-2
2.193653708EO
-9.485435895E-4
-1.849817831E-1
8.195486081E-6
Residual
Al+4+
j = 4
-6.877459572E-2
-3.347116684E-4
3.242490680E-2
-1.359683740E-1
-4.224053098E-3
1.317492316EO
1.201353847EO
-3.383898144E-3
-4.458780550E-2
1.310640570EO
-1.478010660E-1
9.103543233E-5
-1.796260694E-1
1.798049075EO
-1.086932041E-3
-2.165400309E-1
1.342279013E-5
Volume
Percent
Sludge
j = 5
9.805384525E-2
-2.978614228E-2
-1.496716122EO
-1.684088129E-2
1.427994130E-2
-5.959921912EO
-2.884657222EO
-1.102133105E-1
5.200294512E-2
-8.581309540EO
-9.874144414E-2
-1.0621949 5E-4
8.427113016E-3
-5.824153395EO
9.394378173E-3
9.219416997E-1
-4.492395885E-5
— Numerical values are given in floating point notation, i.e., 2401.547 = 2.401547E3.
-------
TABLE B-l (cont.)
Numerical Values of Coefficients in Regression Model
Alum Coagulant
(soluble phosphorus in effluent only)
Y = b + c X + E a.X. + E a,,X.X, + E c . X^ +
o oo i=i i i i,k=l lk x k 1=2 1 °
2 2
r- + dnX/Xr
5 345
Y = concentration soluble phosphorus in effluent, mg/1
X-L = column diameter, D, inches
y^2 ~ upflow velocity, U, ft/hr
X-j = coagulant dose, Al, mg/1
X^ = blanket depth, L, ft
Xr = pH before coagulant addition
XQ = concentration of soluble phosphorus in influent, mg/1
b = 1.462988190E2
c° = 3.692752427E-1
o
a = 1.069903404E-1
a! = 3.278582248E1
a = -1.768487898EO
a, = -6.782375143EO
a5 = -3.818173316E1
a = -1.042707159E-2
a = 4.676882089E-4
a^, = 2.921856396E-2
al7 = -3.275133404E-2
a^ = 1.083331923E-3
al, = -3.311203967E-2
a = -7.960189863EO
a , = -1.227991728E-3
a^ = 4.323849665E-1
a,l = 1.736353973EO
45
c = -3.841549079E-2
c = 1.289097940E-4
cf = -5.218618145E-2
c^ = 2.503469733EO
d = 4.973385640E-1
d^ = -2.754875542E-2
d = -1.002247620E-1
96
-------
TABLE B-2
Numerical Values of Coefficients in Regression Model
Ferric Chloride Coagulant
Y- =
Cn-Xn-
Oj Oj
Z a-X-
i=1 11
.X
X
V
Oj =
x, =
X2 =
X3
concentration of j in effluent, mg/1
concentration of j in influent, mg/1
upflow velocity, U, ft/hr
ferric chloride dose, Fe, mg/1
pH before coagulant addition
2
X = (upflow velocity)'
ffluent
.centration
nded Solids
j = 1
300.849
0.300
-63.180
-0.421
15.210
Total
Organic
Carbon
j = 2
439.909
0.593
-95.221
-0.183
14.314
Soluble
Phosphorus
j = 3
-14.752
0.058
1.952
-0.010
0.655
Total
Phosphorus
i = 4
9.304
0.420
-1.770
-0.018
0.345
Residual
FeH-f+
j = 5
-79.880
-8.464
20.024
0.016
-1.626
2.651
3.897
-0.078
0.068
-0.876
97
-------
Table C
Data for Thickener Design
Sludge Subsidence Rates and Suspended Solids Concentrations
Settling
Test No.
Settling
Rate
(SR)
cm/min
Susp.
Solids
(SS)
mg/1
A R
A R2
A R4--I
A R4--II
A R4--III
A R4--IV
A R5--I
A R5--II
A R5--III
A R5--IV
B a
B a'
B--I
B--II
B--III
B--IV
B--V
B--VI
B R--I
B R--II
B R--III
B R--IV
B R--V
B R2--VI
B R3--VII
B R3--VIII
B R3--IX
B S 1
B S 2
0.25
0.05
0.03
0.04
0.35
0.40
0.40
0.46
0.49
0.53
1.69
1.75
0.16
0.49
0.57
0.50
0.57
0.82
0.08
0.44
0.52
0.44
0.57
0.44
3.14
2.54
2.06
0.43
0.46
4,618
4,999
5,490
4,652
4,072
3,602
3,781
3,849
2,010
2,621
1,559
1,559
5,222
4,736
4,054
3,844
3,065
2,394
4,136
3,488
3,004
2,852
2,244
2,821
669
847
999
3,070
3,070
98
-------
SELECTED WATER
RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
1. Report No.
4. Title
3. Accession No.
w
FLUIDIZED BED CLARIFICATION AS APPLIED
TO WASTEWATER TREATMENT
7. Author(s)
Orcutt, J. C.
9. Organization
Research Triangle Institute
5. Report Date
6.
8. Performing Organization
Report No.
10. Project No.
17030 EYA
12. Sponsoring Organization
15. Supplementary Notes
11. Contract/Grant No.
14-12-912
13, Type of Report and
Period Covered
Environmental Protection Agency report
number EPA-R2-72-032, December 1972.
16. Abstract
An experimental study of the application of a fluidized sludge blanket
clarifier to the coagulation and separation of wastewater solids has been
made to determine the effects of controlled process variables on the treatment
achieved.
Experiments using alum and ferric chloride coagulants were carried out in
12- and 24-inch diameter columns by systematic variation of wastewater pH,
coagulant dose, upflow fluid velocity, and blanket depth. The results were
analyzed using regression analysis techniques, and empirical relationships
were derived relating the variables to the removal of suspended solids, total
organic carbon, phosphorus, and coagulant metal ions. The sludge production
rate was also correlated empirically with the operating variables.
A study of the settling rates of discharged sludge and the fluidized
blanket was made by direct observation.
17a. Descriptors
* Waste Treatment, * Coagulation, * Solids Contact Process,
*Separation Techniques
17b. Identifiers
*Sludge Blanket, * Mathematical Model, Aluminum Sulfate, Ferric Sulfate,
Activated Silica
17c. COWRR Field & Group 05D
18. Availability
19- Security Class.
(Report)
20. Security Class.
(Page)
21. No. of
Pages
22. Price
Send To :
WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.S. DEPARTMENT OF THE INTERIOR
WASHINGTON, D. C. 20240
Abstractor J. F. Kreissl
I Institution USEPA, NERC-Cinclnnati, Ohio
WRSIC 102 (REV. JUNE 1971)
OU.S. GOVERNMENT PRINTING OFFICE: 1972 514-151/130 1-3
------- |