United States
Environmental Protection
Agency
Water Engineering
Research Laboratory
Cincinnati OH 45268
'/•'
'/I
Research and Development
EPA/600/S2-85/140 Feb. 1986
&EPA Project Summary
Automation of Sludge
Processing: Conditioning,
Dewatering, and Incineration
R. C. Polta, D. A. Stulc, and G. A. Mathes
This study developed and tested au-
tomated control strategies for munici-
pal wastewater sludge processing. The
strategies were applied to three unit
processes—chemical conditioning with
lime and ferric chloride, vacuum filtra-
tion, and incineration. The project was
conducted at the St. Paul, Minnesota,
Metropolitan Waste Control Commis-
sion (MWCC) Seneca facility, a 24-mgd
plant with parallel sludge processing
trains.
Several strategies were developed
for controlling lime and ferric chloride
addition and filter cake production rate.
Each strategy was evaluated during ap-
proximately 500 hr of operation.
Automated strategies were also de-
veloped for controlling incinerator
hearth temperatures and air input. The
strategies were designed to maintain
the lowest possible furnace burning
zone and to minimize excess air. The
test data show a moderate cost reduc-
tion for automated control.
On-line sensors were used to collect
data and implement control strategies.
Performance data for these sensors are
summarized, and the digital data acqui-
sition and control system is described.
This Project Summary was devel-
oped by EPA's Water Engineering Re-
search Laboratory, Cincinnati, OH, to
announce key findings of the research
project that is fully described in a sepa-
rate report of the same title (see Project
Report ordering information at back).
Introduction
The costs for solids handling and dis-
posal at a conventional secondary treat-
ment facility are usually 30% to 50% of
the operating and maintenance (O & M)
budget. The solids handling processes
at a St. Paul, Minnesota, Metropolitan
Waste Control Commission (MWCC)
treatment facility were studied to deter-
mine the impact of improved process
control on these costs. The primary goal
was to process the greatest volume of
sludge possible at the least cost. Proc-
ess control can be accomplished by
manual or automated means, but the
complexity and interaction of sludge
dewatering and incineration processes
limit manual control to simplistic
schemes. In addition, the variable na-
ture of the raw sludge requires consid-
erable control flexibility. Thus auto-
matic digital control techniques were
used,
Study Objectives
The overall study objective was to
model and eventually control the three
unit processes for sludge treatment—
conditioning, dewatering, and incinera-
tion—to achieve near-optimum per-
formance for a particular operating
goal. This goal will change throughout
the year and may vary from maximizing
the volume of liquid sludge processed
to maintaining specified limits for par-
ticulates in the stack discharge. The
original reasoning was that if adequate
models could be generated, it would be
relatively easy to prepare an algorithm
to minimize the total cost for any given
operating scheme. Though the overall
project objective was not met, the proj-
ect demonstrated that considerable cost
savings could be generated by using au-
tomated process control schemes even
when they are less than optimum.
-------
Facility Operations
All of the work described was con-
ducted at the Seneca Wastewater Treat-
ment Plant on the Minnesota River in
Eagan, Minnesota. This facility has a de-
sign capacity of 90,840 m3/day (24
mgd).
Approximately 10% of the influent
flow is from industrial sources. The
treatment scheme consists of screen-
ing, grit removal, primary sedimenta-
tion, complete-mix activated sludge,
final sedimentation, and chlorination.
The solids processing operations in-
clude floatation thickening of waste-
activated sludge, vacuum filtration, and
incineration. Ferric chloride and lime
are used to condition the sludge before
dewatering. All of the primary and
waste-activated sludges generated at
Seneca are processed along with
sludges hauled from other MWCC facili-
ties. During 1980, the plant processed
approximately 215,745 m3 (57 million
gallons) of liquid sludge, of which ap-
proximately 93,490 m3 (24.7 million gal-
lons) were hauled to the plant.
Because of the variable nature of the
raw sludge, the conditioning and filtra-
tion processes are extremely difficult to
operate at a steady output rate with a
cake of uniform burning characteristics.
The measured daily variation was 14%
to 25.5% for cake solids and 53% to 68%
for volatile solids. Filter yield varied by a
factor of three.
Because of the lack of cake storage
capacity at Seneca, the incinerator must
burn the cake as produced, maintain
hearth temperatures below 1100°C
(2000°F) to protect the furnace structure,
and meet air pollution standards. These
goals are difficult to achieve given the
variable nature of the filter output.
The total operating cost for sludge de-
watering and incineration at Seneca, in-
cluding chemicals, energy, and labor,
was $166/metric ton ($151/ton) or ap-
proximately 47% of the total plant oper-
ating budget for 1981.
Chemical Conditioning Control
To control chemical conditioning, it is
necessary to know the impact of chemi-
cal addition on the dewatering charac-
teristics of sludge. The two most com-
mon methods of estimating sludge
dewaterability are capillary suction time
(CST) and specific resistance (SR).
Ideally, an on-line measurement
would produce a signal that could be
used to vary the chemical dose and
maintain the desired dewatering char-
acteristics. Unfortunately, neither CST
nor SR is directly measurable on-line.
However, a correlation was sought be-
tween CST or SR and other parameters
(Table 1) that could be measured on-
line.
Rheological data were analyzed and
yielded a relationship between sludge
SR (after conditioning with ferric chlo-
ride and lime) and several other sludge
parameters. This analysis indicated the
feasibility of using on-line data to mea-
sure SR and CST.
Because the conditioning process at
Seneca uses both ferric chloride and
lime, two control systems and hence
two models were required. Since others
have demonstrated that SR adequately
describes sludge dewatering character-
istics for vacuum filtration, this parame-
ter was selected for study. Initial at-
tempts at determining the SR of
ferric-conditioned sludge yielded ex-
tremely long filtration times, and in
most cases, no cake was formed. Since
the laboratory data demonstrated a
strong relationship between the SR of
lime-conditioned sludge (SRL) and the
CST of lime-conditioned sludge (CSTL),
the CST of the ferric-conditioned sludge
(CSTF) was used to characterize ferric
conditioning.
The CST model was constructed with
a total of 124 data sets collected over a
7-month period. A number of relation-
ships were developed to gauge the po-
tential for success and the need for ad-
ditional data. Multiple linear regression
analysis techniques were used to ana-
lyze the data.
The modeling efforts resulted in the
following equations for predicting the
CST of ferric-conditioned sludge
(R = 0.82).
CST2 = 66-0.104(ORPD) (1)
- 1.8(PF) + 6.0(SSR)
+ 0.05 (ORPF) - 3.5 (PHD)
The addition of the last two variables
did little to improve the goodness of fit.
Thus for model CST2, 67% of the vari-
ability in the actual CST is explained by
the independent variables. The unex-
plained variability may be related to the
measurement precision of the field
monitors or to one or more of the im-
portant (but unknown) parameters that
were not monitored.
Three control schemes were consid-
ered for FeCI3 addition: pH (PHD) con-
trol, CST control, and sludge mass ratio
control. Tests demonstrated that both
the pH and the ORP sensors responded
to a change in FeCI3 dose, but that no
universal setpoint would maintain con-
slant dewatering characteristics. Thus
the feedback approach required addi-
tional parameters to make the strategy
responsive to sludge conditioning
needs.
Previous experience had demon-
strated that the FeCI3 dose required for
good vacuum filtration increased as the
raw sludge solids decreased. To reflect
this physical characteristic, the control
scheme was modified to calculate and
maintain PHD based on the raw sludge
solids concentration.
PHD setpoint = K2 + 1C, (SSR) (2)
and
PHF setpoint = PHR - PHD setpoint (3)
Test results demonstated that this
strategy could control PHF. However,
the ability to maintain PHF at the set-
point did not insure proper condition-
ing, as over- and under-dosing resulted
when raw sludge characteristics other
than SSR changed. Proper conditioning
required manual entry of new values for
KT and K2. The control system was thus
reduced to a semi-automated scheme
requiring periodic judgments by the
operators.
CST uses a cascade control to adjust
the ferric pump. The inner loop controls
the pH of ferric-conditioned sludge by
adjusting pump speed. The pH setpoint
is generated by a CST controller.
The model receives inputs from proc-
ess sensors and predicts a CST value. If
the prediction differs from the desired
value, a new pH setpoint value is calcu-
lated. To avoid coupling problems, the
CST control model is recalculated infre-
quently, thereby appearing constant to
the pH controller.
Evaluation of the on-line CST control
proved a constant CST setpoint to be
unworkable, as the FeCI3 pump ramped
to maximum or minimum speed if the
desired CST value was not attained.
Thus a variable CST setpoint was ini-
tiated based on raw sludge solids.
CST setpoint = K3 (SSR) - K4 (4)
The variable CST setpoint provided a
degree of flexibility in the control strat-
egy, but the constants K3 and K4 re-
quired adjustment during control runs
because of changes in sludge proper-
ties not totally reflected in solids con-
-------
Table 1. Sludge Modeling Parameters
Symbol
Definition
SSR Suspended solids concentration in raw sludge, % solids
PHR pH of raw sludge
PHF pH of ferric-chloride-conditioned sludge
PHL pH of lime-conditioned sludge
PHD Differential pH
ORP Oxidation—reduction potential
ORPR ORP of raw sludge, mv
ORPF ORP of ferric-chloride-conditioned sludge, mv
ORPL ORP of lime-conditioned sludge, mv
12 Shear stress of lime-conditioned sludge, 12 rpm, dyne/cm2.
30 Shear stress of lime-conditioned sludge, 30 rpm, dyne/cm2.
60 Shear stress of lime-conditioned sludge, 60 rpm, dyne/cm2.
CST Capillary suction time
CSTR CST of raw sludge, sec
CSTF CST of ferric-chloride-conditioned sludge, sec
CSTL CST of lime-conditioned sludge, sec
SRL Specific resistance of lime-conditioned sludge x 70~" m/kg
SSL Suspended solids concentration in lime-conditioned sludge, % solids
TL Temperature of lime-conditioned sludge, °C
PF Ferric chloride feed, % by weight of dry sludge solids
PL Lime feed, % by weight of dry sludge solids
VAC Vacuum in form zone, in. Hg
VAT Submergence of filter drum, in.
DRUM Rotational speed of filter drum, rph
TPH Wet cake production rate, ton/hr
SSF Total solids concentration in cake, % solids
FLOW Raw sludge flow, gpm
CLOTH Operating hours since media acid cleaned, hr
SR3 Specific resistance of conditioned sludge calculated from the yield equation,
m/kg
tent. Though the predicted CST closely
followed the CST setpoint value, signifi-
cant deviations between the setpoint pH
and the measured pH remained.
These problems were related to the
fouling of the pH electrodes and
changes in the sludge characteristics.
Although short-term control was
achieved using the model, a suitable
setpoint for all sludges was not devel-
oped. Eventually, dose limits of 7% to
13% FeCI3 were implemented to elimi-
nate severe excursions.
The FeCI3 dose ratio control strategy
is defined as follows:
Ferric chloride mass flow = (5)
K5 (sludge mass flow) + K6 (SSR-4)
This control strategy uses a selected fer-
ric chloride dose (K5) and reflects
changing sludge characteristics
through K6. The advantages of the dose
ratio strategy are that sensor mainte-
nance is eliminated and a given FeCI3
dose is maintained. Disadvantages are
that there is no feedback adjustment,
that the accuracy of the sludge solids
analyzer is critical, and that a steady-day
tank ferric chloride concentration is
required.
A number of control schemes for lime
addition were considered during the
course of the project, but the filter yield
dropped off dramatically when the pH
of the lime-conditioned sludge de-
creased much below 12.0. Thus the only
strategy used for the test runs was con-
trol of pH to 12.0 or above. Although
this strategy did not address the initial
intent of controlling the SR of the condi-
tioned sludge, it had two major advan-
tages: It was simple and it worked.
Vacuum Filter Control
Control Strategies
Initial filter control strategies were de-
signed to reduce the variation in cake
yield and maximize the production rate
by stabilizing and controlling filter oper-
ating parameters.
Analyses of vat level adjustments on
filter operation indicated that low vat
levels and higher drum speeds im-
proved cake quality because of the
lower moisture content. However, to re-
duce the risk of vacuum loss and pro-
mote efficient cake release, most opera-
tors maintained the vat at high levels.
Under manual control, the vat level
varies in response to any change in the
filtration rate. For any reasonable com-
bination of parameters, the vat level
reaches an equilibrium value between
the lower limit and the vat overflow.
Any change in dewatering characteris-
tics destroys this equilibrium, and the
vat level changes to a new point.
A cascade algorithm is used to control
the vat level. The internal loop consists
of a flow controller that adjusts the
sludge pump motor speed. The external
loop provides the flow setpoint.
Only two variables that influence the
cake production rate are adjustable: vat
level and filter speed. Form zone vac-
uum is usually maintained at the
highest value possible.
When automated control of the vat
level and drum speed was initiated, the
vat level setpoint was a function of the
cake thickness.
Vat level setpoint = K7 - K8
x (thickness in inches/100)
(6)
Because attempts at thickness measure-
ment were not successful, the cake
thickness was calculated from the pro-
duction rate, drum speed, and cloth
area.
By judicious choice of the constants
K7 and K8, reasonable setpoint values
for a wide range of sludge characteris-
tics can be established. The control ac-
tion also reduces the chance of filter up-
sets by increasing the level setpoint
when lower cake production rates (thin
cakes) occur because of low sludge
solids or filter media blinding.
Drum speed control used a cake pro-
duction setpoint adjusted as a function
of raw sludge solids concentration:
Cake production setpoint = Kg (7)
+ K10 (SSR)
A filter recovery mode was added,
which decreased the drum speed to a
fixed value whenever cake production
fell below a predetermined value, and
an upper limit was placed on the drum
speed to prevent filter failure as a result
of cake discharge problems. The upper
limit also facilitated the use of a con-
stant cake production setpoint without
frequent upsets.
Evaluation of Automated Vac-
uum Filter Control
Approximately 2800 hourly average
data sets were generated for automated
and manual filter control. These data
sets fell into three main categories: Data
3
-------
from filter No. 1 under computer con-
trol, data from filter No. 1 under manual
control, and data from filter No. 2 under
manual control.
Performance data for the three modes
show that filter No. 1 under computer
control processed more sludge with
lower chemical doses. If it is assumed
that the average raw sludge solids did
not vary for the three categories, then
the computer control mode also pro-
duced a drier cake.
Although the wet cake yield was es-
sentially constant, the variability of
chemical feed was considerably greater
in the manual modes, especially for
lime. These data show that control of
the lime dose was the major advantage
of the computer control mode.
In incineration operation, the ratio of
water to volatiles is important to fuel
consumption and steady hearth tem-
peratures. Although all attempts to esti-
mate cake moisture by direct and indi-
rect methods were unsuccessful, this
ratio varied more with manual opera-
tions and was directly related to the
chemical feed rates. Control of the
chemical feed systems would provide a
reasonably conditioned sludge without
large excursions in cake moisture and
volatile content. Table 2 summarizes
the average filter performance by
mode.
Incinerator Control
Control Strategies
Since the input to the incinerator can
vary, even with automated control of
the dewatering train, some control ac-
tion is necessary to maintain steady in-
cinerator operation. The primary objec-
tive for instituting new incinerator
controls was to minimize the use of aux-
iliary fuel while burning all cake pro-
duced and meeting stack discharge
standards.
A series of tests was conducted to de-
termine whether the Seneca incinera-
tors could be operated at rates near ca-
pacity and still meet the stack discharge
standard. The final control strategy was
based on these tests.
This strategy used the calculated in-
cinerator load as a feedforward signal.
This signal was delayed for each burn-
ing hearth to account for process dead-
time. The temperatures on the con-
trolled hearths (Nos. 3, 5, and 6) were
used as feedback signals to tempera-
ture controllers. Fuel flow was in turn
regulated by the air flow rate. The tem-
perature controllers received their set-
points from a temperature adjustment
supervisory program. The output from
the temperature controller was com-
bined with the delayed loading signal at
adjustable ratios to throttle air valve po-
sitions. A burner sequence program
could start or stop the burners individu-
ally if necessary. The temperature con-
trol program also prevented imbalances
between hearths.
Furnace pressure was controlled
using three signals—furnace load, fur-
nace draft, and atmospheric air damper
position—to calculate the position of
the induced-draft fan damper.
The air for drying, burning, and cool-
ing was provided through an atmo-
spheric air damper on hearth No. 7 and
the return of the center shaft cooling air.
The atmospheric air damper was ad-
justed to maintain a fixed oxygen level
in the stack gas, and the incinerator load
was used as a feedforward trim. Hearth
temperatures are used as a second trim
to detect conditions such as too much
cooling air entering the incinerator.
In addition to the on-line sensors, vi-
sual observations are also required. For
example, dark smoke indicates insuffi-
cient combustion air. Cake volatile
solids data are manually entered.
Table 2. Average Filter Performance by Mode—Complete Sets
Parameter
Sludge feed. Us (gpm)
Cake yield, mt/hr (ton/hr)
Cake solids, %
Volatile solids, %
FeCly, % of sludge solids
CaO, % of sludge solids
pH of ferric cond. sludge
pH of lime cond. sludge
Ib H2O/lb volatile
Vat level, cm (in.)
Drum speed, rph
Filter No. 1
Computer
5.6 (89)
4.8 (5.3)
20.5
56
11.2
18.3
4.9
12.0
6.9
43 (17.1)
15.3
Manual
5.7 (90)
5.2 (5.7)
20.8
48
11.5
24.3
4.8
12.1
8.0
53 (21)
15.3
Filter No. 2
Manual
4.5 (71)
4.4 (4.8)
20.3
50
12.9
26.0
-
-
8.1
37 (14.8)
13.3
Tests were conducted to characterize
the dynamic response of the process to
variations in temperature and gas flow. A
The tests demonstrated that three dis-
tinct zones should be maintained to
maximize fuel efficiency: cake drying on
hearth Nos. 1, 2, and 3, complete com-
bustion on hearth Nos. 4 and 5, and ash
cooling on hearth Nos. 6 and 7. The
tests further determined that the above
zones could be maintained by operating
as follows:
Furnace draft 0.4 cm (0.15) H20
Shaft cooling
air 100% return to
hearth No. 7
Ambient air 0-20% open as re-
quired
Hearth 6 burner .. Off
Hearth 5 burner .. 100% capacity
Hearth 3 burner .. Control to 650°C
(1200°F)
By eliminating burning on the upper
hearths, the thermal efficiency was in-
creased, and overall thermal efficien-
cies of approximately 4420 kJ/kg (1900
BTU/lb) water evaporated could be ob-
tained. Typical values fell in the range of
4200 to 5800 kJ/kg (1800 to 2500 BTU/
Ib).
The control strategy was organized
into two areas—temperature and gas
flow. The recommended temperature
profile appears in Table 3. As previously
demonstrated, these temperatures
could be maintained without firing
hearth No. 6, only the burners on hearth
Nos. 3 and 5 were used regularly.
Gas flow through the furnace was di-
vided into draft and oxygen control. The
draft setpoint was manually entered,
and the controller adjusted the position
of the induced-draft fan damper.
Gas flow through the furnace was
controlled to maintain a manually en-
tered value for flue gas oxygen concen-
tration. Tests indicated that all shaft
cooling air should be returned to the
furnace before opening the atmo-
spheric air damper. This goal was ac-
complished by using two independent
controllers. Subsequent tests demon-
strated that sufficient oxygen was usu-
ally obtained without opening the atmo-
spheric air damper.
Evaluation of Automated
Solids Processing
Tests were conducted to operate
process train No. 1 by computer and
process train No. 2 by hand. In one such
test, the initial setpoints for process
train No. 1 were as follows:
-------
Parameter
Wet cake yield
FeCI3 dose
pH of lime-conditioned sludge
Incinerator draft
Flue gas 02
Hearth 2 temperature
Hearth 4 temperature
Setpoint
5.4 mt/hr (6TPH)
11% of dry sludge solids
12.0
5 mm (0.2 in.) H20
7.4%
565°C (1050T)
980°C (1800°F)
The performance data for the two de-
watering trains during the 11-hr test
No. 2 are summarized in Table 4. As in
other tests, the vacuum system limited
the performance of both trains. During
the ninth hour of the test, both filters
lost vacuum, were shut down, and were
restarted manually.
Filter No. 1 produced a cake with a
higher volatile solids content, primarily
because of the reduced lime dose. The
FeCI3 dose in train No. 2 was lower than
normal because the actual discharge ca-
pacity of the chemical feed pump was
only 45% of its nominal capacity.
The fuel efficiencies of the two proc-
ess trains are summarized in Table 5.
The total energy used per mass of water
fed was significantly lower than normal
for both trains, but because of the
higher volatile content, furnace No. 1
was more efficient.
The improvement in the efficiency of
furnace No. 1 can be attributed to set-
ting an upper temperature limit for
hearth No. 4. Initially this limit was
980°C (1800°F), and it was changed to
950°C (1750°F) at 1610 hr.
The results show that automated con-
trol of solids processing can yield eco-
nomic savings. For Seneca, the annual
savings are estimated to be $50,000 for
chemicals and $48,000 to $127,000 for
energy, depending on whether natural
gas or oil is used.
Recommendations
Control Strategy for Sludge
Conditioning
Changes in the raw sludge character-
istics have a dramatic impact on the per-
formance of a vacuum filter and, subse-
quently, the incinerator. Thus operators
tend to include a significant factor of
safety when selecting dose rates for
chemical conditioning. This project
demonstrated the need for an on-line
method of identifying the dewatering
characteristics of chemically condi-
tioned sludge. However, since such a
method is not presently available, the
recommended control strategy is out-
lined as follows:
1. FeCI3—Maintain a constant dose
of 6% to 12% of dry solids. The
specific dose is a function of the
raw sludge variability and the op-
erator time available to monitor fil-
ter operation.
2. Lime—Maintain a constant pH of
12 for Seneca after lime addition.
The above strategies require only
three field sensors—raw sludge
flow, raw sludge solids, and lime-
conditioned sludge pH.
Control Strategy for Sludge
Dewatering
The operating strategy is based on
the assumption that the dry yield of
sludge processing must be maximized.
Because an on-line cake moisture ana-
lyzer was not found, the recommended
strategy is based on maximizing the wet
cake yield without exceeding the incin-
erator capacity. After selecting the wet
cake yield setpoint the recommended
strategy is as follows:
1. Yield—Adjust the drum speed to
obtain the yield; however, to main-
tain a reasonable cake thickness,
apply an upper limit of 18 rph.
When the yield is below setpoint,
the drum speed will increase to in-
crease yield. As the drum speed in-
creases, the cake thickness will de-
crease. To eliminate problems
with a thin cake, the following
equation is used to calculate a new
vat level setpoint:
Vat level setpoint = K7 - K8
x (thickness)
Increased vat level will increase
cake thickness.
2. Vat Level—The vat level is main-
tained by varying the raw sludge
flow rate, which in turn causes the
feed rate of FeCI3 to vary.
To implement this strategy, the opera-
tor must enter the setpoint value for
yield along with the constants K9 and
KIQ.
Once the conditioning strategy is
adopted, the dewatering strategy re-
quires only three additional sensors—a
belt scale to determine cake yield, a
level detector to measure vat level, and
a tachometer to determine drum speed.
Incinerator Control Strategy
The recommended incinerator con-
trol strategy is divided into the follow-
ing two areas:
1. Temperature—Only the burners
on hearth Nos. 3 and 5 are used in
this strategy. Hearth No. 3 burners
are operated to maintain a temper-
ature of approximately 540°C
(1000°F) on hearth No. 2, and the
burners on hearth No. 5 are con-
trolled to maintain a temperature
of approximately 950°C (1750°F) on
hearth No. 4. For the typical load-
ing rates observed at Seneca,
these setpoints make hearth No. 4
the main burning hearth.
2. Air Flow—The air flow strategy
consists of maintaining the fur-
nace pressure in the range of
-0.38 to -0.51 cm (-0.15 to -.20
in.) H2O. For cake loadings below
approximately 4.1 wet mt/hr (4.5
wet ton/hr), the lower (absolute
value) setpoint is sufficient. A
lower limit of approximately 20%
open should be established for the
induced-draft damper. This limit
will prevent problems associated
with low draft values. The atmos-
pheric air damper should remain
closed, and all air not entering at the
burners or through uncontrollable
leaks should be provided as recycled
shaft cooling air. The flow of re-
cycled cooling air should be con-
trolled to maintain the desired O2
concentration on hearth No. 1. The
Oz setpoint should be determined
from visual observation on hearth
Nos. 1 through 4. Sufficient O2
should be provided to minimize
smoke generation.
During the test, it appeared that air
may have entered through the sludge
feed drop gate at the top of the incinera-
tor. For this reason, the O2 concentra-
tion should be determined on hearth
No. 1 as opposed to the downstream
location used during the tests. Also, a
continuous measurement of opacity
should be used to provide feedback for
adjusting the oxygen setpoint or the fil-
ter yield or both.
The incinerator control strategy uses
the standard temperature and pressure
sensors found on most furnaces. In ad-
dition to these instruments, an oxygen
analyzer is required on hearth No. 1.
5
-------
Summary and Conclusions
The study demonstrated that automa-
tion of the sludge treatment facilities at
the Seneca plant would yield annual
savings on the order of $100,000 based
on 1981 costs for conditioning chemi-
cals and fuel. This estimate is based on
implementation of the following control
strategies:
1. Control of the chemical condition-
ing process consists of maintain-
ing an FeCI3 dose in the range of
6% to 12% (based on dry sludge
solids) and adding sufficient lime
to maintain a pH of 12.0.
2. The vacuum filter yield (wet
weight/time) is controlled by ad-
justing the drum speed. Vat level is
adjusted to maintain a cake thick
enough to discharge properly.
Level is maintained by varying raw
sludge flow rate, which in turn
varies the chemical feed rates.
3. The hearth temperatures are ap-
proximately 540°C (1000°F) on
No. 2 and 950°C (1750°F) on No. 4.
Temperature is controlled by ma-
nipulating the burners on hearth
Nos. 3 and 5, respectively. Furnace
pressure is maintained at -0.38 to
-0.51 cm (-0.15 to -0.20 in.) by
manipulating the induced-draft fan
damper. The flow of recycled cool-
ing air is controlled to maintain a
setpoint oxygen concentration in
hearth No. 1. Additional air is gen-
erally not required.
The following is a summary of the
performance of the sensors used during
this study:
Table 3. Recommended Temperature Profile for Controlled Incineration
Temperature, °C (°F)
Hearth No.
1
2
3
4
5
6
7
Minimum
400 (750)
450 (850)
650 (1200)
760 (1400)
590(1100)
200 (400)
80 (180)
Average
430 (800)
480 (900)
700 (1300)
870 (1600)
650 (1200)
260 (500)
90 (200)
Maximum
450 (850)
540 (1000)
760 (1400)
980 (1800)
700 (1300)
310 (600)
90 (200)
Table 4. Dewatering Performance During Test No. 2
Parameter
Hours
Sludge flow, Us (gpm)
Raw sludge solids, %
FeC/a, %
CaO, %
Yield, wet mt/hr (TPH)
Cake solids, %
Cake volatile solids, %
Process Train
No. 1 (Computer)
11
4.4 (70)
4.1
11
18
4.7
20.5
53.8
Process Train
No. 2 (Manual)
11
4.2 (66)
4.1
8.5
25
4.4 (4.9)
21.2
44.1
Parameter
Hours
Gas use, sm3/hr (SCFH)
Gas use, sm3/wet mt (SCF/wet ton)
Gas use, sm3/dry mt (SCF/dry ton)
Total energy, kJ/kg H^ (BTU/lb H2O)
Gas, kJ/kg H2O (BTU/lb H£»
Process Train
No. 1 (Computer)
11
184 (6,500)
39 (1,250)
184 (5,900)
4,885(2,100)
1,860 (800)
Process Train
No. 2 (Manual)
11
218 (7,700)
50 (1,600)
240 (7,700)
4,885(2,100)
2,325 (1,000)
Acceptable
Vat level
Temperature
Rotational speed
Rotational position
Sludge flow
pH
Solids
Vacuum
Cake weight
Questionable
ORP
Oxygen
Turbidity
Unacceptable
Viscosity
Cake thickness
Cake moisture
Tank level
sorship of the U.S. Environmental Pro-
tection Agency.
Several studies were conducted at
Seneca to determine whether commer-
cially available sensors could provide a
continuous measure of cake moisture.
All those tested were unacceptable, as
they failed to provide reliable estimates
of cake moisture.
Even acceptable sensors require sig-
nificant maintenance. The extreme is
the conditioned sludge pH sensor,
which required cleaning at least once
per shift. Although the maintenance re-
quired for the other sensors was lower.
both the raw sludge solids sensor and
the belt scale require frequent calibra-
tion to assure success of the control
system.
Control of sludge-handling processes
should be designed to allow operator-
entered setpoints to maintain operation
without going to complete manual
operation.
The full report was submitted in fulfill-
ment of Grant No. S803602 by the
Metropolitan Waste Control Commis-
sion of St. Paul, Minnesota, under spon-
•&U. S. GOVERNMENT PRINTING OFFICE:!986/646-116/20775
-------
R, C. Polta and D. A. Stulc are with the Metropolitan Waste Control Commission,
St. Paul. MN 55101; and G. A. Mathes is with EMA, Inc., St. Paul. MM 55101.
Waiter W. Schuk is the EPA Project Officer (see below).
The complete report, entitled "Automation of Sludge Processing: Conditioning,
Dewatering, and Incineration," (Order No. PB 86-138 963/AS; Cost: $22.95.
subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA2216J
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Water Engineering Research Laboratory
U.S. Environmental Protection Agency
Cincinnati. OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use $300
EPA/600/S2-85/140
LCD W TILLEY
REGION V EPA
LIBRARIAN
230 S DEAR30RN
CHICAGO
ST
IL 60604
------- |