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

-------