RESEARCH NEEDS
FOR
AUTOMATION
OF
WASTEWATER TREATMENT
SYSTEMS
PROCEEDINGS of a WORKSHOP
sponsored by the U.S. ENVIRONMENTAL PROTECTION AGENCY
in cooperation with CLEMSON UNIVERSITY
-------
14S54
RESEARCH NEEDS
FOR
AUTOMATION
OF
WASTEWATER TREATMENT
SYSTEMS
PROCEEDINGS of a WORKSHOP
held at Clemson, SC, September 23-25, 1974
Sponsored by the U. S. ENVIRONMENTAL PROTECTION AGENCY
in cooperation with CLEMSON UNIVERSITY
H. O. Buhr, J. F. Andrews and T. M. Keinath, editors
Clemson University
Clemson, South Carolina
1975
-------
ORGANIZING COMMITTEE
Cochairmen:
A. W. Breidenbach, Environmental Protection Agency
J. F. Andrews, Clemson University
Arrangements:
G. D. Barnes, City of Atlanta
T. M. Keinath, Clemson University
Members:
W. W. Eckenfelder, Jr., Vanderbilt University
C. F. Guarino, City of Philadelphia
J. F. Roesler, Environmental Protection Agency
W. A. Rosenkranz, Environmental Protection Agency
J. R. Trax, Environmental Protection Agency
D. R. Wright, Environmental Protection Agency
-------
CONTENTS
FOREWORD
Page
. 5
WORKSHOP OBJECTIVES 7
Andrew W. Breidenbach, Environmental Protec-
tion Agency
WORKSHOP SUMMARY 9
William A. Rosenkranz, Environmental Protection
Agency
WASTEWATER COLLECTION SYSTEMS
Automated Collection Systems
Instrumentation Needs 12
Curtis P. Leiser, Municipality of Metropolitan
Seattle
Present Practice and Research Needs in Wastewater
Collection System Design and Operation 18
James J. Anderson, Watermation, Inc.
Discussion 19
Report of Working on Wastewater Collection Systems . 21
John A. Lager, Metcalf & Eddy, Inc.
Harold Torno, Environmental Protection Agency
BIOLOGICAL TREATMENT PROCESSES
Dynamics and Control of Biological
Treatment Processes 26
John F. Andrews, Clemson University
Automation of the Activated Sludge Process 38
Michael J. Flanagan, Brown & Caldwell Consulting
Engineers
Discussion 46
Report of Working Party on Biological
Treatment Processes 52
Paul H. Woodruff, Roy F. Weston, Inc.
A. W. West, Environmental Protection Agency
PHYSICOCHEMICAL PROCESSES
Field Experiences with a Pilot-Physical-Chemical
Treatment Plant 56
Walter W. Schuk, Environmental Protection
Agency
Automation of Physical and Chemical Processes 58
Thomas M. Keinath, Clemson University
Discussion 64
Report of Working Party on
Physicochemical Processes
Wesley W. Eckenfelder, Jr., Vanderbilt University
John Stamberg, Environmental Protection Agency
Page
Report of Working Party on Physicochemical Processes 66
Wesley W. Eckenfelder, Jr., Vanderbilt University
John Stamberg, Environmental Protection Agency
SLUDGE PROCESSING, TRANSPORT AND DISPOSAL
Automation of Sludge Processing,
Transport and Disposal 70
Bart T. Lynam, Raymond R. Rimkus and Stephen
P. Graef, The Metropolitan Sanitary District of
Greater Chicago
Control of Sludge Handling:
Some Successes and Problems 72
Joseph B. Farrell. Environmental Protection
Agency
Discussion 76
Report of Working Party on Sludge Processing,
Transport and Disposal 79
Richard I. Dick, University of Delaware
John R. Trax, Environmental Protection Agency
COMPUTER APPLICATIONS IN AUTOMATION
Current Practice in Instrumentation and Computer
Application at the County Sanitation Districts of
Los Angeles County 84
Walter E. Garrison, Kip Payne and Tim Haug,
County Sanitation Districts of Los Angeles County
Computer Applications in Automation ....
William E. Dobbins. Teetor-Dobbins, P.C.
99
66
Discussion 106
Report of Working Party on Computer Applications . . 109
Carmen F. Guarino, Philadelphia Water Depart-
ment
John M. Smith, Environmental Protection Agency
EVALUATION OF THE EFFECTIVENESS OF
AUTOMATION
Evaluation of the Effectiveness of Automation 114
Joseph F. Roesler, Environmental Protection Agency
Field Evaluation of the Effectiveness of Automation ..118
Allen E. Molvar, Raytheon Company
Discussion 128
Report of Working Party
on Evaluation of Effectiveness 130
Walter G. Gilbert, Environmental Protection
Agency
James A. Mueller, Manhattan College
LIST OF PARTICIPANTS 133
-------
FOREWORD
Improvement of the operation of municipal wastewater
treatment systems is one of the nation's major environmental
problems. Possible solutions to this problem are increases in
the quantity and quality of personnel involved in operations
and/or increasing use of automated systems. The need for an
increase in the quantity and quality of personnel is well known
and has been the subject of numerous conferences; the
prospect of improving operations through the use of automa-
tion, however, is not as well recognized. There are many
reasons for this, but one of the most important is that research
efforts expended on the automation of wastewater treatment
systems are relatively minor when compared with research on
other aspects of water pollution control.
There is currently a great interest in the automation of
wastewater treatment systems as evidenced by the attendance
of more than 200 persons at an International Workshop on
Instrumentation, Control and Automation of Wastewater
Treatment Systems which was held in London during the fall
of 1973. The automation systems reported on at that
workshop for cities such as Atlanta, Chicago, London, Los
Angeles, Paris, Philadelphia, etc., will greatly affect the
performance of wastewater treatment systems valued at many
millions of dollars. However, few of the papers presented at
the workshop were oriented toward research, which again is an
indication that relatively few researchers are currently engaged
in this area. Most automatic control installations for waste-
water treatment systems are, of necessity, designed on an
empirical basis because of a lack of more fundamental
knowledge concerning such factors as dynamic behavior,
control strategies, component reliability and cost/benefit
analysis.
The above statements illustrate the great need for research
on the automation of wastewater treatment systems. However,
in order to accomplish such research in the most effective
manner, it is necessary to first clearly define and establish
priorities for the research needed. Recognizing this, the U. S.
Environmental Protection Agency requested Clemson Univer-
sity to organize and conduct a Workshop on Research Needs
for Automation of Wastewater Treatment Systems.
In order to insure the incorporation of all viewpoints,
participants were invited from government regulatory and
research agencies, universities, operating engineers and man-
agers of large treatment systems, consulting engineering firms
and equipment manufacturers. Extensive and lively discussion,
which forms the heart of a workshop, was encouraged by
dividing the workshop into three portions. The first portion
consisted of formal presentations of present practice and
current research on each of six topics, by authorities on these
specific topics. These formal presentations, and the discussions
associated with them, served to set the stage for individual
meetings of working parties on each of the six topics, where
special attention was devoted to stating the problems and
specifying research needed to solve these problems. The
cochairmen of the working parties then prepared brief
documents summarizing the discussion in their session and
orally presented these for discussion at a reassembly of all of
the Workshop participants. Finalization of these documents,
representing the deliberations of each working party and the
viewpoints expressed at the reassembly, was accomplished at a
meeting of the cochairmen of the individual working parties in
Washington on November 19, 1974. Special thanks are due to
the cochairmen of the six working parties for their per-
formance of a difficult and time-consuming task.
These proceedings represent the integrated best judgement
of the experts gathered at the workshop as to research needs
for the automation of wastewater treatment systems. It should
provide a firm foundation for the development of a national
research program in this important area.
John F. Andrews, Cochairman
Workshop on Research Needs
for Automation of Wastewater
Treatment Systems
April 1975
-------
WORKSHOP OBJECTIVES
Andrew W. Breidenbach
Director, National Environmental Research Center, Environmental Protection
Agency, Cincinnati, OH 45268
This workshop was developed to provide an opportunity to
discuss problem areas and research needs for the automation
of wastewater treatment plants. The discussion generated
should have significant impact on future research and should
ultimately affect the design and operation of wastewater
treatment plants.
Clemson was selected as the site for this Workshop because
of its leadership in the area of control engineering as applied to
environmental systems. Recognizing automation as an area of
important research, the United States Environmental Protec-
tion Agency at the National Environmental Research Center-
Cincinnati (NERC) initiated a research program in automatic
control of wastewater treatment plants a little over two years
ago. Because of mutual interests, a jointly sponsored workshop
to define research needs seemed essential if we were to achieve
our goal of having fully automated wastewater treatment
plants on stream in the 1980's.
Looking back only ten years, it is apparent that automation
of wastewater treatment systems is a recently developed
technology that has yet to be fully exploited.
Ten years ago, the dissolved oxygen (DO) probe was
emerging from the laboratory for application to relatively
clean water. The thought of placing a DO probe in the hostile
environment of a wastewater treatment plant appeared to be
an unworkable concept. Today, it is a reality. A recent survey
sponsored by the U. S. Environmental Protection Agency
found that 12% of the 50 plants surveyed have automatic
on-line DO control.
Total organic carbon (TOC) was virtually an academic
curiosity ten years ago. At that time, the Robert A. Taft
Laboratory had just received the prototype of a commercial
TOC analyzer. However, less than ten years later, continuous
on-line TOC analyzers have become available on the open
market. The same can be said about process control com-
puters; they were available ten years ago, but they were
expensive and difficult to program. Yet, some process com-
puters can be programmed easily in languages as simple as or
simpler than FORTRAN. The development of the 1C chip in
the sixties gave birth to micro-computers. Properly applied,
these computers are very effective for data acquisition and
process control in situations where previously they were not
cost-effective. Today, more than two dozen plants are on
stream or will be on stream with process computer installa-
tions. Some cities such as Seattle and Minneapolis have been
using a computer for several years to control the stormwater
flow in their combined sewer systems. They are now expand-
ing their computer installations to control other aspects of
wastewater processing as well.
Although progress is being made, it is being made at a
relatively slow pace. One of the questions we must face at this
workshop is: What is delaying the implementation of automa-
tion in wastewater treatment plants? Is it the lack of suitable
sensors for automatic on-line control? For example, consider
the recently developed ammonia probe: It does not appear to
be applicable for automatic on-line control for nitrogen
removal because the probe requires considerable maintenance
by a skilled technician. The same can be said for the
continuous on-line TOC analyzer. It is known that many
plants in the country simply cannot afford to hire such skilled
technicians. The lack of sensors and trained technicians may
be the major deterrent to the automation of treatment plants.
If so, is there a solution to these problems?
To control a property of a process stream, an engineer must
be knowledgeable in several areas. He must be familiar with
the characteristics and the limitations of measuring devices,
with the treatment plant's chemical and biological reaction
kinetics and with the process equipment in which these occur.
A knowledge of the plant's design limitations and its opera-
tional stresses in terms of the loadings and environmental
changes to which this equipment will be subjected is essential.
Control theory, computer technology, and systems analysis are
important tools in this field. Thus, special training is often
necessary for an effective environmental systems and control
engineer.
Until recently, there were very few doctoral degrees
awarded in the area of systems analysis. Now, a whole new
field has opened. Clemson University was one of the first
universities in the United States to apply systems analysis
techniques to the automatic control of wastewater treatment
plants. The first doctoral degree in the area of environmental
systems control engineering was awarded here in the middle
sixties. This is a new field that has grown very rapidly with
continuing growth potential for further development.
If automated treatment is to be applied, the existing
technology must first be examined. An open attitude towards
automation is a requirement. Evaluation of this technology
and adaptation of it to our needs will then follow. Finally, we
must improve the technology where necessary. Thus, the
development of automation will require resources, hard work,
and research.
The first objective of this workshop is to instill enthusiasm
in each of us. We must convince ourselves (and then others)
-------
that resources, hard work, and research will be well-spent on
automation of wastewater treatment plants. Today, I am
confident that we have accomplished the first objective merely
by having carried out this workshop. Next, we must exchange
information and experiences, communicating our successes
and failures.
The second objective of this workshop, and one that vitally
concerns EPA's National Environmental Research Center in
Cincinnati, is to assess the research necessary to bring about
successful and cost-effective automation of wastewater treat-
ment plants throughout the United States. We must establish a
channel of communication between the users of our research
(especially the designers, operators, engineers, and managers of
our wastewater treatment plants) and the persons who
influence the drafting and enforcement of state and federal
regulations. One of the ultimate research goals of the Center is
to develop fully automated wastewater treatment systems.
This includes flow routing and storage in the collection
systems as well as processing technology such as biological
treatment, physical-chemical treatment and sludge handling
and disposal. We are interested in defining the most rewarding
areas for further research to extend the present state-of-the-art
and develop practical control technology. We need to evaluate
control techniques such as F/M control, with on-line respir-
ometers or feedforward TOC measurements. We wish to
discuss treatment plant reliability ideas such as control of
toxic wastes entering the plant. We wish to explore the
possibility of how best to utilize a process computer for
automating a wastewater treatment plant. We also wish to
discuss the effectiveness of various automatic control strategies
and how these control strategies would improve the per-
formance or lower the cost of operating a wastewater
treatment plant. And finally, we would like to discuss methods
of improving the plant operators' attitudes towards automa-
tion. Ultimately, it is the operator who must live with the
automated plant. His attitude and his approach towards
operating an automatic plant will certainly help determine the
success or failure of automation. What additional research do
you want to see done in this area? What are the problems that
your particular group are experiencing on a day-to-day basis?
And finally, do you have any ideas that would prove
potentially profitable for automation of wastewater treatment
plants in this countryideas that you would like to see
developed, even though you have neither the time or the funds
to do so?
If we have identified and prioritized, through some logical
process, a set of research needs, we will have taken a stride
toward more cost-effective wastewater treatment.
-------
WORKSHOP SUMMARY
William A. Rosenkranz
Director, Municipal Pollution Control Division,
Environmental Protection Agency, Washington, DC 20460
The automation of wastewater treatment systems offers a
number of potential benefits including improved performance,
reduction in size and construction cost of new systems,
improved reliability, more efficient use of operating personnel,
and minimized operating costs. These benefits are clearly
"potential" since application of instrumentation and automa-
tion in the wastewater field is still minimal. Compared to most
industrial processing, automation of wastewater treatment
systems is in its infancy. The purpose of this Workshop was to
define how to move this specialized technology progressively
through adolescence and into adulthood.
The philosophy for addressing wastewater management
systems requires change. A treatment system should no longer
be considered as a marginal water pollution control facility,
but rather as a production facility for wastewater refining or
renovation. The medical and chemical industries developed the
philosophy of pushing the newest technology into their
respective fields many years ago. It appears now that basic
water pollution control technology is sophisticated enough to
adopt such a philosophy and thus attain this new goal.
In order to accomplish the automation of wastewater
control and treatment systems in the most effective manner, it
is necessary to first clearly define the research and develop-
ment needed and then establish priorities for implementation.
More than one hundred participants attended the "Workshop
on Research Needs for Automation of Wastewater Treatment
Systems". In order to insure a broad coverage of all view-
points, the participants represented government regulatory and
research agencies, universities, operating engineers and man-
agers of large treatment systems, consulting engineering firms,
and equipment manufacturers. The first day of the Workshop
was devoted to the presentation and discussion of current
practice and research activities directed towards automation of
wastewater treatment systems. Experts from both government
and private industry gave these presentations in order to
identify needed research and development. From these presen-
tations, it can be concluded that most instrumentation for
control of wastewater systems is, of necessity, designed on an
empirical basis because of a lack of more fundamental
knowledge concerning such factors as dynamic behavior,
control strategies, component reliability and cost-benefit
analysis. The effect of automation on the design and operation
of wastewater recycle systems and manpower requirements for
treatment plant operation was also considered.
Working parties were organized to address the following
subject areas: (1) automation of wastewater collection sys-
tems, (2) automation of biological treatment processes, (3)
automation of physical-chemical processes, (4) automation of
sludge processing, transport and disposal, (5) computer appli-
cations, and (6) evaluation of the effectiveness of automation.
A number of research needs were identified which were
common to most or all of the six subject areas. These common
research needs were:
(l)The Workshop recommended that an information clear-
ing-house dealing with instrumentation and automation
be established within EPA. For example, a great deal of
instrument testing on a specific case basis is being done,
yet results are not available to the technical community
on a broad basis. A central location for gathering and
dispensing such information would be of considerable
assistance to those planning and using instruments in
municipal wastewater systems.
'(2)Development of efficient and dependable sensors is
needed. This is a prerequisite to the implementation and
verification of virtually all control strategies. Some of
the sensors needing improvement or development in-
clude: sludge blanket level indicator, settling velocity
indicator, respiration rate sensor, suspended solids sen-
sor, on-line replacement for the BOD test and on-line
analysers for ammonia, nitrate and phosphorous.
(3)Performance specifications should be developed for
sensors and instrumentation as a guide to the user
community. This could be in the form of a testing
protocol and procedures for evaluation of sensor and
instrumentation packages.
(4)With respect to treatment processes and treatment
systems, it was determined that a logical progression of
research and development should be as follows:
a. Development of dynamic mathematical models for
individual processes. In several cases, this has been
partially accomplished although many of the models
should be further refined using the latest data
available. An important long-term goal is incorpora-
tion of the individual process models into an overall
mathematical model for treatment plants.
b. Tentative control strategies based on computer simula-
tion using mathematical models should be developed.
A small amount of work is now on-going in this area.
c. Control strategies should be evaluated at pilot scale to
select the most promising for future demonstration.
Such evaluation has frequently been limited in the
past by lack of adequate sensor capability.
-------
d. The most promising control strategies should be
demonstrated at full scale, including cost/benefit
analysis.
(5)A protocol should be established to evaluate instrumen-
tation and automation system design and selection. This
protocol should take into account all relevant aspects of
the problem such as direct cost savings, system per-
formance and reliability, minimization of the consump-
tion of energy and other resources, and the man-machine
interface.
(6)Instrumentation and automation control strategies
should be expanded to include interrelationships be-
tween liquid and solids processing, stormwater and
dry-weather flow control and treatment, and eventually
area-wide wastewater management.
Each individual working party identified research and
development needs specific to the assigned subject area.
Highlights from the six groups are as follows:
The importance of addressing the area of instrumentation
and automation in terms of a total system was identified by
the working party dealing with automation of wastewater
collection systems. It is very important that the demonstration
of total control systems with the necessary operating strategies
incorporating the newest technologies be accomplished. This
could even include sophisticated weather forecasting and
tracking. The role and benefits of flow equalization need to be
better documented and more widely applied. Coupled with
automation of treatment processes, this technique may have
significant plant performance advantages.
The automation of biological treatment processes group
identified a need for great improvement in information
exchange, particularly from the operator to the manufacturer
and from the manufacturer back to the operator. Development
of performance specifications and acceptance standards for
instruments was also emphasized. In addition, the develop-
ment of adequate mathematical process models, practical
operating control strategies, real-time monitoring, and im-
provements in the man-instrumentation interface were areas of
research that were identified.
The matter of dynamic models for process control and
monitoring functions was also highlighted in the area of
physical-chemical systems. Control strategies and instrumenta-
tion development to facilitate implementation of control
strategies were two key needs identified. Specific needed
research was identified for several treatment processes includ-
ing chemical clarification, the deep-bed filtration process and
granular carbon adsorption.
Working party deliberations indicated that sludge process-
ing, transport and disposal should be placed at the highest
priority level. The need for development of a variety of sludge
quality sensors was identified, particularly sensors to measure
settleability, dewaterability and other similar parameters.
The computer applications working party indicated that
improved capability to measure flow is needed throughout the
wastewater control and treatment system. This is a difficult
technical problem and, although advances such as the use of
sonic devices have been made in recent years, no major
technological breakthrough appears to be on the horizon.
Improved flow measuring capability is a key to total system
management. Development of control strategy models, includ-
ing reliability testing, was highlighted as a research need.
Determination of specifications for computer selection and
utilization of a centralized computer for data acquisition and
report preparation were areas deemed worthy of research. The
need for increased educational activities to produce personnel
who understand both the computer and treatment plant
operations was also stressed.
The group dealing with evaluation of the effectiveness of
automation, indicated that a cost-effectiveness evaluation
protocol is needed so that meaningful comparisons of instru-
ment applications and control systems can be made. Such
comparisons are vital to selection of control systems for
specific cases and for evaluating cost-effectiveness of alternate
systems of control and treatment. The concept of using a
man-in-the-loop to control several satellite plants through the
use of a centralized computer and terminals was proposed as a
potentially cost-effective technique.
The Workshop also discussed the matter of eligibility of
instrumentation and automation within EPA's Construction
Grants program. Recognizing that many treatment plants
involving large sums of money are funded under the EPA
grants program, a need for detailed guidance concerning
eligibility conditions for instrumentation and automation was
identified. Computer applications are of specific concern.
Although not identified as a research need, the Workshop
noted that the United Kingdom has recently established an
instrumentation and automation Working Party for the pur-
pose of establishing and advancing the state-of-the-art. Based
upon this information, the Workshop recommended that EPA
explore the possibility of setting up a similar panel in the
United States in order that we can have a direct interchange of
information, ideas and technology with the United Kingdom
working group and similar groups throughout the world.
In conclusion, the Workshop developed many specific
research needs related to the instrumentation and automation
of wastewater treatment systems. It indicated a need for an
information clearinghouse, international exchange of data, and
projected a new philosophy of wastewater renovation as
opposed to processing wastewater to the minimum quality
requirements. The cost-effective application of instrumenta-
tion and automation to wastewater management systems will
be a key to implementing this philosophy.
10
-------
AUTOMATION
OF
WASTEWATER COLLECTION SYSTEMS
Workshop on Research Needs
Automation of Wastewater Treatment Systems
11
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
AUTOMATED COLLECTION SYSTEM
INSTRUMENTATION NEEDS
Curtis P. Leiser
Manager of Computer Services, Municipality of Metropolitan Seattle,
410 West Harrison Street, Seattle, WA 98119
INTRODUCTION
One need only review recent publications and conference
agenda of technical organizations such as ASCE, WPCF and
APWA to appreciate the growing interest nationwide and, to
some degree, internationally in wastewater collection system
control and treatment research. Centralized collection system
automation is in its infancy. The number of operating
computerized systems can be counted on one hand. However,
many more are on the drawing boards for consideration or
construction in the near future. The computerized systems
have generally struggled through technical and funding prob-
lems typical of research and development work. To this date,
the systems which have been developed are not only atypical
in design but also contain significantly different degrees of
monitoring, control and the ultimate form of automation
which involves optimized simulation or modeling.
This paper will concentrate on the hardware instrumenta-
tion considerations of these systems. Primary reference will be
to the results and recommendations developed during a
six-year demonstration grant study between the Environ-
mental Protection Agency and Seattle Metro which culminated
in a report titled "Computer Management of a Combined
Sewer System" (1). It was found during this study that there
was a considerable amount of research that could be accom-
plished to overcome many of the instrumentation problems
and shortcomings which were revealed during the development
of Seattle's computer-controlled system. It is appropriate, at
this time, to present a short summary of the Seattle system, its
objectives and its equipment. More detailed information can
be obtained from the list of references at the end of this paper
(2-6).
SEATTIi METRO SYSTEM
In January of 1971, the Municipality of Metropolitan
Seattle (Metro) first placed into operation a monitoring and
control facility termed CATAD or "Computer Augmented
Treatment and Disposal System." This centralized control
system encompasses approximately a ten-square-mile com-
bined sewer system in and around the City of Seattle. The
main objectives of the CATAD system are:
1. To utilize the maximum storage capability of trunk and
interceptor lines within a combined sewer system built
to ultimate capacity so that overflows caused by storm
inflow are reduced or eliminated.
2. To regulate daily flows to treatment plants, thereby
aiding in the stabilization of the treatment processes and
effectively increasing the dry-weather capacity of exist-
ing plants.
3. To select the overflow points which will cause the least
harm to receiving waters, beaches and marine life during
intense storms when overflows cannot be avoided.
4. To eliminate the need for or reduce the cost of total
separation of combined sewers which would be espe-
cially costly and disruptive to commercial and industrial
areas of the city.
5. To monitor and control mechanical equipment within
remote stations while accomplishing the above objectives
and,
6. To retain dry- and wet-weather flow data of component
collection systems for subsequent identification of infil-
tration problems and for potential charges in accordance
with those flows.
At the time the construction of the computerized monitor-
ing and control system began, a consideration portion of the
remote equipment was already installed and operating. It
seemed logical to incorporate as much of that system as
possible and interface the computer-control equipment to the
existing regulator controls. Thus, two objectives would be
achieved, (a) a minimum disruption of existing control
equipment which was proven to be reliable and (b) the local
control system within each station could serve as a backup
12
-------
WASTEWATER COLLECTION SYSTEMS
control capability in the event of computer or communica-
tions equipment failure.
The existing control equipment was primarily pneumatic,
while newer station designs included electronic control equip-
ment to facilitate computer interface installation. Modifica-
tions to older stations included new precision sensors, com-
munication and control equipment, to adapt existing regulator
and pumping stations for remote control capability.
Key elements of the system are pictured in Figure 1. At the
heart of the CATAD system is the real-time process control
computer and background devices which allow for the
programming of that computer. Specially designed interfacing
or connection equipment was installed to allow the computer
to communicate through cables to many mechanical devices.
At the central console, information is collected from remote
stations and is converted to visual displays. Control push-
buttons, alarm and status indication lights, radio and tele-
phone communication equipment and logging teletypes com-
plete the components of the central console. In addition, a
wall map showing the geographic locations of the collection
system together with general status-indication lights for each
station combine to offer the human console operator all
information and control features which could be expected of a
computerized wastewater collection system.
A second interface "black box" connects a series of water
quality monitors to the computer. These monitoring stations
are located on the Green-Duwamish River south of Seattle and
monitor such conditions as temperature, dissolved oxygen,
conductivity, pH and solar radiation. These data are fed to the
computer and are logged on a permanent paper form as well as
stored on magnetic tape for later statistical processing. The
information thus becomes available to enable either the human
operator or, through programming, the computer, to select
overflow points which would provide the least impact upon
receiving waters.
A third interface connects the computer to two satellite
consoles which are compact versions of the central console.
These satellite consoles are able to display the same data and
to command remote station actions independently of the
central console. The Renton satellite console is the newest
system addition and has been given more flexibility by
utilizing a mini-computer rather than a fixed logic terminal, to
perform many functions. The mini-computer was expected to
be adapted for use in process control loops during the
expansion of Metro's Renton secondary treatment plant, but
the $100,000 estimated cost for the control interface aborted
that plan until interfacing prices become more realistic.
Thirty-seven remote stations communicate to the computer
through the last interface. A telemetry control unit (TCU) at
each remote station acts as an interpreter to collect, convert
and assemble station information into a form which can be
transmitted over telephone lines to the computer. Figure 1
shows the types of information collected from three different
representative remote stations passing through a series of
electronic units for transmission to the computer. Commands
return in the opposite direction to control regulation or
pumping stations in the combined sewer portion of the
collection system.
INSTRUMENTATION AND SENSORS IN
CONTROL SYSTEMS
The objective of the data collection system is (1) to make a
continuous, accurate measurement of all information neces-
sary to monitor the safety and operation of a station and
calculate required flow and storage data to facilitate logical
control decisions and (2) to transmit that information with
some degree of security to the central computer system.
Sensors, or measuring devices, must be present to continuously
monitor hydraulic, meteorologic, atmospheric and water qual-
ity parameters for optimum control of a combined sewer
system. A minimum system must be capable of monitoring the
hydraulic parameters described below.
Flow Calculations
Obviously the most important parameter to measure in a
sewage collection system is flow. Generally the more flow data
accumulated from the various elements of a collection system,
the better will be the control and predictive modeling
capabilities of that system. A measurement of water depth
together with known sewer configuration data can provide
open-channel conduit flow values by incorporating one of the
variations of Manning's equation. More accurate system flows
are calculated by measuring depth and using metering sections
such as the Palmer-Bowles flume where space is available or
Parshall flumes where head losses are not critical.
A measured depth may also allow the calculation of flow
through sluice gates which may be manually or mechanically
driven. Added to the measurement of depth must be the
position of the control gate in the stream flow. By incorporat-
ing these values with standard orifice formulae, the flow may
be calculated with some degree of reliability. Flow calculations
however, can be complicated by upstream or downstream
backwater and submergence effects.
Different measurements are necessary to calculate pumped
flow in a collection system. Four measuring techniques are
currently being utilized in Metro's CATAD system. Wetwell
water level and pump rotational speed allows calculation of
flow through standard pump curves provided by the equip-
ment manufacturer. Where the pumping station force main is
long enough to cause sufficient head loss, force main pressure
may be measured and flow calculated with calibration curves.
More expensive but direct-reading meters such as a magnetic
flow meter or a calibrated velocity meter are additional means
of obtaining flow information from pumping stations or
submerged conduits.
A continuous water depth measurement at one or more
locations will also provide system storage data for use in
controlling and minimizing overflows. A single depth measure-
13
-------
JACKGROl
PERIPHERAL
.DEVICES
INTERFACE
M
LEGEND
O COMPUTER SUB SYSTEM
CD CENTRAL CONSOLE SUB SYSTEM
C3 RIVER MONITOR SUB SYSTEM
£\ SATELLITE CONSOLES
[""I REMOTE TERMINALS
- TELEPHONE LINES
INTERFACE "BLACK BOX"
CABLING
I
>
5
o
V
INTERFACE
m
33
3D
m
TELEPHONI
LINES
1
WATER QUALITY
MONITORS
jn fcOMPUTEFf F TELEPHONE
1 V
INTERFACED
CWALL MAP^
ALARMS a STATUS
LIMITED COMMANDS
PLANT DATA ENTRY
1
r
7 X
/WEST POINT >
/ SATELLITE
/ CONSOLE
i
-"
* mmmmmm
/^CENTi
"V CONSC
. . f
WES
REG
STA
}
i r ~
i
i
i
t i
.T POINT
ULATOR
TIONS
rcu
| A T ^\ (REMOTE- LOCAL)
IAL \ v
>LE } \
L
CRT DISPLAYS \ /
LOGGERS
Al ARMS
STATUS
1 CONTROLS
r~
1
1
L-
< /RENTON
\ / SATELLI
\ / CONSOL
GATE
POSITIONS
' ' WATER-TDE
LEVELS
-L ' EXPLOSION
| HAZARD
\ viisc.
A ALARMS
L\
ACU
1
J
__.
»
DETACHABLE
CABLES
" ~l~
1
1
1
1
r
WEST POINT
PUMPING
STATIONS
TCU
(REMOTE-LOCAL)
\A DETACH ABLE
/ CABLES
ecu
KATAD- LOCAL)
PUMP
SPEEDS
1
WATER 1
LEVELS | "
FORCE MAM
PRESSURES
EXPLOSION
HAZARD
MISC.
ALARMS
1
1
1
1
1
f
MENT SYSTEMS
r ~ - -
i
RENTON
PUMPING
STATIONS
TCU
(REMOTE -LOCAL)
\/
ecu
(CATAD-LOCAL)
PUMP
SPEEDS
WATER
LEVELS
FORCE MAM
PRESSURES
EXPLOSION
HAZARD
MISC.
ALARMS
-------
WASTEWATER COLLECTION SYSTEMS
ment within a storage tank is a simple method of calculating
storage. In circular conduits the calculation is more complex
but can be simplified by having more than one depth
measurement available.
Rainfall is the minimum meteorologic data to be collected
and supplied to an automated control system. Rainfall data
from each sub-basin will allow the analysis of infiltration and
will provide some predictive information for control functions
in a combined sewer system. The rain data also provides basic
design information for future construction.
Non-Essential Measurement Devices
A sewage collection control system must be primarily
concerned with hydraulic measurements described previously,
but depending upon the level of sophistication of that system
the measurement equipment described below may or may not
be present.
Continuous monitoring of water quality within the collec-
tion system and in its receiving waters would be of value, in a
control decision to select a point of overflow, for example.
Quality parameters such as dissolved oxygen level, suspended
solids, temperature and biochemical oxygen demand are a
sampling of the parameters that would be especially valuable
in determining control and design decisions. Automatic
sampling equipment may be part of this system and might be
triggered either by local conditions or remotely by the
computer control center. Sampling devices may be an alter-
native to continuous monitoring and if so, their operation and
performance should be monitored as if the sampling equip-
ment were an integral part of the collection system.
The collection system may contain some type of treatment
process entirely separate from the major treatment plant
facilities at the end of the collection system. Small local
treatment plants or similar facilities to treat overflows or
bypasses from the system should be monitored for chlorine
residual, for example, or other appropriate performance
criteria.
Two or more gaseous parameters may be measured in the
collection system. Explosion hazard monitoring is a feature of
the Metro Seattle System. Its purpose is two-fold: (1) to
provide a safety warning to protect personnel and facilities and
(2) to supply information for tracing sources of illegal dumps
of hazardous volatile material into the collection system. No
method of treating or disposing of hazardous materials within
the system is currently being applied. A second valuable
measurement which currently is very difficult to obtain
without the use of an elaborate series of chemical tests similar
to an auto-analyzer is hydrogen sulfide. The purpose of this
measurement is to monitor both odor production and corro-
sion within the sewage collection system.
CATAD INSTRUMENTATION PROBLEMS
As one might expect, there are a great many difficulties
encountered during a research and development effort on the
scale of the Seattle Metro CATAD control system. The many
instrumentation and control problems have been condensed to
three general areas:
1. Looking back on the development effort, a timing
problem was evident. The majority of design and
construction work was done between 1968 and 1970
while some major changes were taking place in the
electronics and computer industry. The CATAD system
was too far along to reap many of the benefits of these
advancements.
2. The tendency of engineers or designers to rely on
existing technology, in Metro's system, meant specifying
equipment which was generally being successfully ap-
plied in industrial situations at the time. This resulted in
a larger burden being placed on the interfacing and
conversion tasks. The goal of having a local independent
control system at each remote station was excellent.
However, the dissimilarity between equipment created a
great number of difficulties which will be described later
in this paper.
3. A final source of problems was the selection of a
different set of designers and suppliers for remote
station industrial equipment and centralized computer
control equipment. An aero-space firm, ready with all
' the latest electronics technology, was on one side of the
interface while an industrial, wastewater-oriented design
firm was ready with older, but safer and more reliable
equipment, at the other side of the interface.
At this point, the types of equipment problems will be
discussed in more specific terms, leading into a list of
comments on the type of research which should be contem-
plated to alleviate these problems.
One broad category of problems developed due to the
application of sensitive instruments and sensors in a hostile
environment for which they were not designed. Since most of
the industrial controls would be adversely affected by humid-
ity and explosive or corrosive atmospheres, a controlled
environment contained within an expensive structure was
generally provided. These remote stations, in the 1960's, cost
an average of $200,000 each. The new station price has now
been inflated to an average of $500,000. These high costs limit
the number of-sites where monitoring and control can take
place. In Seattle's system, there are between two and three
hundred overflow points (some admittedly very small); how-
ever, only approximately ten percent of those sites are
monitored and controlled by the CATAD system. Some
sensors, such as gate-position indicating devices, would be an
excellent selection for a normal environment but in the humid,
occasionally explosive atmosphere of the sewage collection
system, tests indicated that the equipment could not cope with
these conditions without a prohibitive amount of money and
space being allocated to provide a special enclosure. Normal
utilization voltages, radio-frequency and other electromagnetic
disturbances that the heavy industrial-type equipment was
15
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
immune to, played havoc with sensitive computer com-
ponents. Costly damage done to five-volt diode-transistor logic
(DTL) by misapplication of line voltage during routine
servicing pointed out a need for additional technician training,
warning messages and protective devices wherever possible.
When applying sophisticated computer control to a sewage
collection system, one naturally thinks of improved precision
capability. To obtain five-percent precision, some sensors and
control equipment, relatively new on the market, were
incorporated into the system. These devices tended to be
candidates for extensive and repetitive calibration and de-
manded a great deal more maintenance than was routinely
scheduled for station equipment.
Interfacing, or tying together the computer and remote
industrial-grade devices, was another major source of diffi-
culties. The fact that an interface existed was immediate cause
for problems because the separation resulted in frequent
disputes as to whose responsibility a particular problem was.
Since the computer equipment generally requires a specific
range for analog measurements, it turned out that a great
number of transducers was required because of a distinct
absence of commonality between instruments offered by
different manufacturers and sometimes even within the sepa-
rate lines of instruments available from year to year from a
single manufacturer. In an early attempt to prevent excessive
voltage from harming the solid state electronic equipment at
each remote station, a general-purpose industrial interposing
relay was installed at every contact-sensing point. Unfor-
tunately, the relay coil itself (approximately 50 per station)
generated an extremely fast induced electrical surge during
contact activity. This electrical "noise" caused a disturbance
within the solid-state logic, which occasionally was of a serious
nature and was both difficult to pinpoint and to correct. After
a series of studies of the electrical noise problems, remedial
shielding measures were successfully implemented (7, 8).
RESEARCH NEEDS
A battery of well designed and tested instrumentation and
sensors must be developed. These devices should be modular,
heavy-duty, exhibit low long-term drift, and should be easy to
calibrate, maintain and replace. They must be designed with
enclosures and other protective features suitable for a hostile
sewer environment; meaning they should be explosion-proof,
resistant to electrical noise, power system voltage, large
magnetic fields and power failures. The instrumentation
should be digitally oriented and standardized, with no special
one-of-a-kind components, so that interfacing and replacement
problems are minimized. Ideally, this instrumentation should
have small space and power requirements so that $500,000
structures are not necessary to contain it. The new equipment
should also be thoroughly tested in non-critical field locations.
A parallel effort in this area would be to establish uniform
standards or guidelines for instrumentation and sensors to be
used in collection system monitoring and control applications.
The Instrument Society of America and American Public
Works Association have made some fine efforts towards
surveying the instrument field recently (9) but there has been
little or no attempt to establish standards in computer control
applications in wastewater systems. The Environmental Protec-
tion Agency sponsored an excellent summary of automatic
sampling equipment (10). However, the report would have
made an even greater contribution if it had set some standards
and enforcement guidelines. Sampler manufacturers, as well as
instrument suppliers, will continue to manufacture an infinite
variety of equipment until uniform industry-wide standards
and guidelines are set by the Environmental Protection Agency
or a similar funding agency on a national scale.
Another research effort tied closely to the previous
suggestion is an analysis of the potential for applying
pre-programmed mini-computers in monitoring and control
applications in the collection system. The mini-computers
would replace complex arrays of relays, which are expensive to
install and maintain, and prone to failure. Industrial control
manufacturers have been making greater use of mini-
computers in various applications. However, this equipment
has not commonly been applied to the wastewater collection
and treatment field.
Flow measurement in the collection system needs a great
deal more research. Metro has relied heavily on pneumatic
bubblers for water elevation measurement but these devices
cost an average of $3-5 per month to operate. Sonar water
level sensors which operate for about ten cents per month have
been employed by Metro with a mixed degree of success and
failure. A thorough investigation of the proper application of
sonar and other innovative methods of non-contact measure-
ment would provide a beneficial service to the wastewater
collection automation goal. Depth measurement, however, has
a drawback in that flow calculation is then required through
the use of largely empirical Manning's, orifice or weir
formulae. A "characterization" technique should be re-
searched which would allow the direct output of flow from a
sensor. By incorporating LSI (Large Scale Integrated) elec-
tronic modules into the primary sensor or tapered capacitance
probe techniques with direct digital output, a heavy calcula-
tion burden on the control computer is thus avoided. A
researching of velocity measuring sensors, which might be
applied to open-channel flow within sewer systems, would be
beneficial in achieving more reliable flow calculation. Magnetic
and ultrasonic flow meters have been found very useful in
generating reliable flow data where a full pipe is available, such
as in a pump force main. But precalibrated and tested
ultrasonic flow meters are not currently available and some
additional research in the application of these devices would
prove valuable.
Machine vibration and heat measurement in pumping unit
control systems has been applied successfully at Metro as well
as in other agencies. However, a limited amount of application
information is available on vibration monitoring of low-speed
16
-------
WASTEWATER COLLECTION SYSTEMS
equipment (less than 1,000 RPM).
Considerable research has been completed and workable
equipment is now available to handle petroleum products
dumped into open waters. But oil spills are a more frequent
problem in combined sewage collection systems. Reliable
explosion-hazard monitors can now detect the presence of
volatile products. Research is necessary to detect other illegal
petroleum fractions, and, even more important, to recommend
methods of handling illegal discharges before they can enter
and foul treatment processes.
Precise gate position measurement is another area which
requires some additional research. Mechanically or hydrauli-
cally driven sluice gates are easier to monitor than inflatable
gates. However, some problems still exist which could be
improved through additional research. Metro has employed
multi-turn potentiometers connected to gate drive gears to
produce an analog signal proportional to gate travel. The
aerospace industry has used slide wire technology with a great
deal of success. However, collection systems would benefit
from additional study and the incorporation of digital encod-
ing to minimize calibration and drift problems.
A great deal of research is required to determine what type
of monitoring and control equipment is necessary for treat-
ment facilities within the collection system. Such facilities
would, for example, provide sedimentation, chlorination or
other form of treatment to combined sewage overflows or
pump station by-passes. Water quality monitoring and
sampling within the collection system to determine treatment
levels and influence storage and control decisions is currently
not being done in a continuous on-line mode. These research
efforts are quite similar to treatment plant control and
monitoring problems and will be left for later speakers in this
workshop to discuss.
CONCLUSIONS
There have been a significant number of promising develop-
ments in the computer and electronics fields which should
make wastewater collection system automation much easier in
the future. Direct digital output sensors, noise-resistant solid-
state electronics, photo-optic isolators and other devices now
need to be assembled into a completed system to serve as a
guide to other agencies seriously considering automation. The
EPA's Taft Center should serve as a storehouse for the latest
automation technology and be the headquarters for the
issuance of new standards and guidelines for control and
monitoring instruments and sensors. Seattle Metro's pioneering
efforts in wastewater collection system control have shown
that once the research and development hurdles are overcome,
the automation concept holds great promise toward solving
the problem of handling urban drainage within combined
sewers. Metro will continue to cooperate with the Environ-
mental Protection Agency and other municipal agencies in
furthering wastewater collection system automation efforts.
References
1. Leiser, C. P., "Computer Management of a Combined Sewer 6.
System", USEPA Report No. EPA-670/2-74-022, National En-
vironmental Research Center, Cincinnati, OH (1974). 7.
2. Gibbs, C. V. and Alexander, S. M., "CATAD System Controls for
Regulation of Combined Sewage Flows", Water & Wastes Eng., 6,
No. 8,46-69(1969). 8.
3. Leiser, C. P., "Maximizing Storage in Combined Sewer Systems",
Water Poll. Control Res. Ser., Project No. 11022-ELK, Washing-
ton, DC (1971). 9.
4. Gibbs, C. V., Alexander, S. M. and Leiser, C. P., "System for
Regulation of Combined Sewage Flows", Jour. San. Eng. Div.,
Proc. Amer. Soc. Civil Engr., 98, SA6, 951-972 (1972). 10.
5. Mallory, T. W. and Leiser, C. P., "Control of Combined Sewer
Overflow Events", Paper prepared for the 1973 APWA Public
Works Congress and Equipment Show, Denver, CO, September,
1973.
Mallory, T. W., "Seattle's CATAD", APWA Reporter, 41, No. 18,
26-27 (1974).
Metropolitan Engineers, "Preliminary Final Report-CATAD
Noise Analysis", Report to the Municipality of Metropolitan
Seattle, June, 1973.
Boeing Company, "Electromagnetic Compatibility Study of the
CATAD System", Report to the Municipality of Metropolitan
Seattle, June, 1973.
American Public Works Association, "Feasibility of Computer
Control of Wastewater Treatment", EPA Contract 14-12-580,
December, 1970.
Sheiley, P. E. and Kirkpatrick, G. A., "An Assessment of Auto-
matic Sewer Overflow Samplers", USEPA Report No. EPA-R2-
73-261, Washington, DC (1973).
17
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
PRESENT PRACTICE AND RESEARCH NEEDS IN
WASTEWATER COLLECTION SYSTEM DESIGN AND OPERATION
James J. Anderson
President, Watermation, Inc., 1404 E. 9th Street,
Cleveland, OH 44114
INTRODUCTION
The task of defining research needs concerning collection
systems seems somewhat difficult, especially in the light of
current practice. Most current sewer design is based on
steady-state, mostly linear relationships, and sewer systems
operate with only gravity in control.
Because of my experience, I will confine my remarks to
existing sewer systems.
The major problems in existing sewers today, in order of
importance to the homeowner-taxpayer, are 1) stoppage, 2)
inadequate capacity, and 3) discharge of untreated polluted
water. In most cases, the untreated discharge is because of
inadequate capacity, resulting in the placement of relief
outlets.
Malfunctioning of sewer systems has primarily been a result
of design methods, the long life of underground structures,
and unpredictable urban growth and development. Current
design methods have already been mentioned. Many of the
sewers in use today were sized with now antiquated methods,
fifty to one hundred years ago. It is rather surprising how well
such systems are performing today. In many cities, the
combined sewers have served effectively because they were
sized on the basis of estimated runoff from the tributary area,
and not on the basis of estimated wastewater flow.
With the advent of the first major interceptor sewer and
treatment plant construction early in this century, and even in
recent large systems constructed in the 1950's, engineering
judgement, the public, and economics allowed the use of
interceptor sewers which captured only 90 to 95 per cent of
the dry-weather flow. Treatment plants that were constructed
had very limited capacity to treat excess flows during runoff.
Our work has been to provide controlled dispatching
techniques to better use existing sewer systems and to
determine what minimum improvements can be made to
nearly eliminate untreated overflow. In order to accomplish
this, we have developed dynamic analysis methods for design
and operation. We have also reviewed new and old sewer
system components and selected those which can be effec-
tively used in dynamic operation of sewers. As one client put
it, we are assisting Newton in his previously unassisted control
of sewers!
COLLECTION SYSTEM COMPONENTS
Components which can be used in a controlled combined
sewer system include automated regulators, relief sewers, and
off-line storage. Wet-weather treatment can be accomplished
by utilizing excess capacity in the dry-weather treatment plant
as well as additional wet-weather treatment facilities.
CONTROL SYSTEM COMPONENTS
The control system components include a rain gauging
network, sewer level measurement, a system simulation model,
a system optimization model, data acquisition and control
equipment, and a process control computer.
DESIGN METHODS
Because we propose to take advantage of the dynamic
characteristics of rainfall, runoff, and of the sewer system, we
must use dynamic design methods. Very briefly, we use design
dynamic rain events in the simulated model to generate the
hydrographs and materials balance in an existing sewer system.
The output from the simulation is used as input to the
optimization model. The results of the optimization are
combined with engineering judgement, which cannot be
computerized, to select a final design. The expected per-
formance of the final design is then analyzed by simulating a
number of design rain events, ranging from light to heavy,
which represent the spectrum of local rain events over a period
of years.
RESEARCH NEEDS
Assuming that research will precede practice by a number
of years, and that dynamic design and control methodology is
in its infancy, a number of research opportunities can be
foreseen in the area of our experience.
Among the most significant are:
1. Determine where dynamic design methods apply and where
steady-state or handbook methods will suffice.
2. Develop dynamic design methods not requiring the designer
of a small system to use a computer.
3. Assess actual hydraulic conditions found in sewers and
generate new equations describing these.
4. Define mixing, dilution and transport of dissolved and
suspended materials in sewers. Perhaps develop new
methods for obtaining dynamic materials balances and for
tracing pollution loads through the sewer system.
5. Develop techniques to predict likely amount and distribu-
tion for a short future period (hours) during rainfall, based
on actual measurements, for real-time control.
18
-------
WASTEWATER COLLECTION SYSTEMS
6. Determine the tolerance of treatment systems to shock
hydraulic and pollution loading, to assist in determining
economic trade-off between additional sewer control and
degraded plant performance.
CONCLUSION
Of course, there are many other general subject areas
needing research. In addition there is need for various
hardware and devices. For example, on-line on-stream sensors
have been a matter of interest for some time.
However, it seems to me that until there is more widespread
application of dynamic design and control of sewer systems,
the hardware, devices, and sensors will remain on the shelf or
inoperative in the field. The latter will not generate the need
for their existence.
The conversion of knowledge-in-being to knowledge-in-use
seems to be a stumbling block in the application of dynamic
methods to sewers. Perhaps this matter might itself benefit
from research.
DISCUSSION
L. A. Schafer:
The CATAD system is an outstanding example of a
monitoring and control system. In the area of future needs,
however, why is digital compatability stressed? Although this
may be desirable, development of digital-type instrumentation
has proven to be expensive and generally unsuccessful in the
instrument industry and can be split off as a separate problem.
C. P. Blakeley:
There is a distinct need for an early-warning system that
advises the waste treatment plant operator of unusual pollu-
tants in the collection system. This may not be essential for
the very large municipalities where dilution of a chemical
dump or spill in the collection system negates detrimental
effects on biological treatment, but in smaller systems-
probably up to 50 MGD and even largerit may be of value.
No really good sensors are available. Conductivity measure-
ment will provide a warning if the pollutant is ionized. A
sampler, activated by the alarm could provide a means to
identify the nature of the pollutant while the record of
conductivity provides a reasonable handle on the magnitude
and volume of the spill.
H2S analysis of the atmosphere above the lift station
wetwell could indicate unusual contaminants if, under normal
operation, aerobic conditions maintain at the station. Again, a
sampler can be activated by the analytical device.
These devices are available and can operate unattended for
extended periods of time. Their use now may help plant
operators and may also influence design in the future.
Stephen P. Graef:
One 01 the problems in a collection system is that of
obtaining flow-proportional samples on each of several inter-
ceptors leading into a wetwell. Needed is a method for
estimating flow in each interceptor so that a flow-proportional
composite sample can be obtained. The device must be
upstream from points of recycle.
Kelly M. PeU:
An Army report, due for publication in October 1974,
concerns side-by-side evaluation of various types of samplers,
tested on six different classes of water. Objectives of the
report were to provide methodology to compare samplers
under standard conditions, and provide a basis for a customer
to select a sampler best suited for his requirements. The report
may also provide information for manufacturers to build a
better sampler.
Russell H. Babcock:
The motivation for major manufacturers to design and
produce instruments for the special conditions encountered in
waste treatment facilities is not present. The market is very
small. According to a report by A. D. Little in November
1966, treatment plant and equipment represents about 20% of
the total expenditure for collection and treatment facilities.
An article entitled "The Worst Public Works Problem" by
Edward T. Thompson appearing in Fortune for December
1958 used figures from Black and Veatch to estimate that
treatment plant and equipment represents 13.5% of the total
expenditure. The state of the art survey concluded that 3.3%
of plant cost was for instruments. This included installation
which is not normally a profit center for a manufacturer.
Product shipment records from the Department of Commerce
show about 2% of plant cost is for instruments. Most
manufacturers will agree that approximately 50% of an
instrument system sale is spent on panels and their mounting,
piping and wiring. This is a very low profit item normally
furnished only because it is essential to a sale. These figures
multiply out as follows:
20% X 2% X 50% = 0.2% of total expenditures using A. D.
Little's data.
13.5% X 2%X 50% = 0.135% of total expenditures using
Fortune's data.-
These figures are for process instruments and do not include
laboratory instruments or supervisory telemetry which are not
part of the market that can support development of special
instruments. Conclusion: the market is not sufficiently large to
support an independent instrument program.
19
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
Harry Torno:
Would Mr. Leiser please describe the algorithm for fully
automated control under CATADfrom rainfall and flow
input through control actions taken?
Joseph F. Roesler:
Can you describe the storm water treatment centers? What
is needed to eliminate the storm water discharge into Puget
Sound? How can automation play a role here?
Mohamed Elsahragty:
In order to achieve the highest benefits from the automa-
tion of wastewater collection systems, it is very helpful to use
optimization techniques, like dynamic and linear program-
ming, to screen the alternative system operating strategies.
These techniques can be used to select the real-time operating
rules from the infinite number of alternatives that are
technically feasible. Existing optimization models are either
very expensive to use (require several hours on the largest
computers commercially available) or too simple to capture
the main features of the system. It is my opinion that the lack
of adequate optimization models is a serious limitation to
achieving the maximum benefits from the existing automated
wastewater collection systems at minimum cost.
John F. Andrews:
Would you comment on the question of improving plant
performance by regulating plant inflow as against accepting
variable inputs and improving performance by better control
of the plant?
Both authors have raised the importance of adequately
considering the interface between the collection system and
the treatment plant. In this connection, there are those who
might leap to the conclusion that it would be highly desirable,
through a combination of sewer control and balancing tanks,
to obtain a constant wastewater flow rate and composition
into the treatment plant. On the other hand, there are those
who would concentrate entirely upon control systems for the
treatment plant for handling of the "wild" or uncontrolled
inputs from the collection system.
The discusser would like to point out that there may be a
"best" waveform and input frequency for the inputs to
wastewater treatment plants and this may be neither the
"wild" nor constant input. For example, there are instances in
other fields of science and engineering where cyclic inputs at a
particular frequency have been found to give improved
performance over constant inputs. Examples are the use of
alternating instead of direct current in electrical systems,
certain chemical reactors which perform best with cyclic
inputs and the surface renewal theory for oxygen transfer. The
possibility therefore exists that there may be a "best"
waveform and input frequency for the inputs to wastewater
treatment plants.
Joseph F. Roesler:
What difficulties do you face in explaining to the "City
Fathers" the type of computer that must be used to
accomplish the job? Would it be feasible to describe the
planned system in terms of functional use, but yet in the
layman's language so that the City Fathers can understand?
Jack Matson:
Has the question of cost effectiveness been studied? What
are the quantifiable benefits of an automated wastewater
collection system, and do they exceed the costs?
CLOSURE
Curtis P. Leiser:
Reply to L. A. Schafer: Data transmission is almost
universally done in digital form. By providing all original data
signals also in digital form, we bypass the need to provide,
install and maintain expensive interfacing and conversion units.
Reply to R. H. Babcock: I cannot argue with the analysis.
Perhaps EPA involvement in research for standardized instru-
mentation would offset development costs.
Mr. Torno is referred to EPA publication EPA-
670/2-74-022, "Computer management of a combined sewer
system," pp. 140-152.
Reply to J. F. Roesler: There currently are no overflow
treatment centers in Seattle. A rotary screening facility at the
King Street outfall was once proposed but rejected as too
costly to operate. The EPA is currently sponsoring research on
numerous overflow treatment techniques. Overflows could be
eliminated by providing additional offline storage, construct-
ing parallel interceptors at critical locations and by limiting
storm inflow through surface stormwater retention practices.
The computer would be very useful in measuring flow,
overflow and storage rates, and controlling or recommending
control actions in offline storage facilities.
Reply to J. Matson: The reference given above devotes an
entire chapter to costs of automated control. Quantifiable
benefits of improved water quality are obviously difficult to
measure. It is easier to base such decisions on a comparison of
alternatives for achieving a certain degree of overflow reduc-
tion. Comparisons from the referenced report show that the
first 60-80% reduction is far less expensive using automated
control.
James J. Anderson:
Reply to S. P. Graef: We have used stage-discharge relation-
ships, taking advantage of the computer's capability to solve
complex equations, to estimate flow. The equations of
hydraulic flow have been well defined in the literature and can
be used with good engineering judgment for practical applica-
tions.
Reply to M. Elsahragty: We have optimized a number of
combined and separate sewered areas to obtain cost-effective
combinations of relief sewers, storage, control, and treatment
20
-------
for overflow pollution reduction. Our early modelling was very
complex until we learned which parameters and considerations
were significant.
Reply to J. F. Andrews: Perhaps a new "figure-of-merit" or
"annual utilization rate" might be used to answer this
question. The plant removal in pounds per hour could be
accumulated annually and the total divided by the design
removal capacity in pounds per hour times the number of
hours of operation. High removals as a per cent of influent
may be achieved at light loads, but the excess treatment
capacity has been lost forever. It is my belief that a
combination of load control and process control are required.
Reply to J. F. Roesler: At one time the word "computer"
meant "automation = loss of jobs." It still has a negative
meaning to a few people. By and large, however, "City
Fathers" are progressive people who see the computer as
WASTEWATER COLLECTION SYSTEMS
another of the complex tools used in municipal administra-
tion. In my opinion, computers are more readily accepted by
elected officials than by the engineering profession which is
only slowly learning how to use them. There are a number of
primers on computer science which could be modified for use
in the water pollution control field.
Reply to J. Matson: The capital cost of an optimized
automated collection system in the Cleveland Southerly
District was approximately $50 million. These construction
costs are the least possible to eliminate 90 per cent of
combined sewer overflow over a ten-year period. The value of
benefits for this reduction have not, and, in my opinion,
cannot be quantified. For example, how can a value be placed
on the aesthetics of a creek without raw sewage flowing
through the city? The selection of the automated system was
based on a tenfold reduction of overflow at a cost of about
one-tenth of other standard engineering approaches.
Report of Working Party
on
RESEARCH NEEDS
FOR
AUTOMATION OF WASTEWATER COLLECTION SYSTEMS
John A. Lager
Vice President, Metcalf & Eddy, Inc., 1029 Corporation Way, Palo Alto, CA 94303
Harold Tonio
Office of Research and Monitoring, Environmental Protection Agency
Washington, DC 20460
The following outline presents the problems and associated
research needs on automation of wastewater collection sys-
tems as determined by the working party. A section on
objectives, applicable to both combined and separate sewer
systems, has been included to place the problems in perspec-
tive.
OBJECTIVES
1. Reduce or eliminate untreated overflows or bypasses.
2. Provide measurements and data logging of flow and
pollution characteristics for:
a. NPDES compliance
b. Legal protection
c. Normal state and federal reporting purposes
3. Minimize flooding.
4. Equalize flow and pollutant loadings arriving at the
treatment facilities.
5. Improve system operationthat is, provide for the opti-
mum utilization of the existing sewer system and its
inherent hydraulic capacity in such a way that the
treatment facilities are most effectively used and that
overflows or bypasses are minimized.
6. Detect and monitor infiltration and/or inflow.
7. Detect system abnormalities, including stoppages and mal-
functions in regulating devices or other system components,
and hazardous conditions in the sewer system, such as the
presence of H2S or other gases.
8. Provide a .data base for the cost-effective planning of
improvements to, or extensions of, existing systems.
9. Save on costs and improve system utilization by pumping at
highest efficiency, early detection of faults or malfunctions
in the system, and source detection of contaminants
entering the system at random times or places.
PROBLEMS
The following is a list of problems considered to be
important by the working party if the stated objectives are to
be met. These problems are not necessarily in priority order.
1. Equipment limitations, particularly the sensors required to
meet specific needs, such as flow or load equalization: Of
21
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
primary importance are flow measurement capabilities and
the necessary sensors to operate regulating devices within
the sewer system. Considerable attention was directed at
the conference toward the limitations of sensors. The
working party believes, however, that the sensors available
on the market are adequate, except for flow measurement
devices, and that the primary problem is an engineering one
of devising the proper equipment to convert the sensed
signal at a data point to data in a central computer system.
2. Lack of understanding the impact of automation on system
design and operation: Most sewer systems are designed on
the basis of peak flow assumptions when, in fact, the urban
runoff phenomena are highly time-variant. The uneven
nature of the rainfall runoff process makes it extremely
attractive to use automated systems in order to maximize
flow through that system and/or to minimize the impact of
pollutant discharges from overflows.
3. Inadequate rainfall forecasting: The primary problem is one
of predicting on a very short-term (1- or 2-hour) basis what
the impact of rainfall will be on a catchment: How intense
will the rain be? From which direction will it come? In
which direction will it travel as it crosses the catchment?
How quickly will it abate? And what is the impact of this
rainfall travel and intensity on the runoff from that
catchment?
4. Lack of real-time data on the cost-effectiveness of auto-
mated control of collection systems, and hence inadequate
justification of the use of an automated system: The
working party believes, however, that there are opportun-
ities for the joint automation of treatment facilities and the
collection system so that these systems are jointly cost-
effective. These opportunities should be given careful
consideration.
5. The need to improve existing regulators and their design
and, in some cases, to provide better regulators, such as a
replacement for fabridams: While some work is under way
in this direction (the swirl concentrator is an example),
much still remains to be done on improvement of regulators
and the control and sensing devices associated with them.
Moreover, the effect of the imposition of controls, such as
rapidly closing gates, on the physical integrity of the system
has been inadequately defined. For instance, such gates can
introduce hydraulic transients and water hammer condi-
tions, or pressure conditions within the piping that could
cause serious damage to the system. Better definition is also
needed of the limitations on control devices that must be
applied when they are installed in collection systems.
6. The general reliability of automation systems and the cost
of systems related to their overall reliability has been
inadequately defined. It is possible, of course, to design
control and telemetry equipment that will provide per-
formance similar to that used in submarines or missiles.
However, the cost may be prohibitive and unjustified.
7. There is a general lack of standardization of regulators,
hardware, computer hardware, and the programs or even
the programming languages used when developing models,
for operating in an automatic model wastewater collection
system: The performance requirements for system com-
ponents are not adequately defined, nor are system
specifications available so that purchasers of computing
equipment are adequately protected in case of either
software or hardware failure. In this regard, it is important
that there be an interface with other engineering disciplines
which have already solved many of the problems in process
control of systems. For example, the petroleum and textile
industries run highly automated process control operations.
Their engineers have largely solved many of the same
problems that we will face, such as telemetry and analog-
to-digital conversion devices (and their interface with small
computing systems).
8. Maintenance and personnel requirements that are unique to
automated wastewater collection systems are largely unde-
fined, not only in numbers but also in the skills required
when such automated systems are installed. Finally, while
not a research problem, the working party believes that
state and federal organizations should better define the
standards on combined sewer overflows and stormwater
discharges, so that those who must design collection
systems to deal with pollution abatement from these
sources know what their objectives are and have something
on which to base valid cost comparisons.
PRIORITY NEEDS
The following needs have been ranked in priority sequence
as determined within the working party through open discus-
sion and critique.
1. A determination should be made of the relative balance
between the benefits to be accrued from the automation
of the collection system and from treatment facilities
improvements (automation of treatment facility or some
other upgrading). To do this, some definition of overflow
standards will have to be considered in terms of their
ultimate impact on receiving water quality.
2. A set of uniform standards and guidelines for instrumen-
tation and sensors must be established. In this regard, the
standards developed, particularly for sensors, should be
directed toward operation and performance, and not
toward technical details on how a piece of equipment is to
be manufactured. An excellent example is the publication
"Specifications for an Integrated Water Quality Data
Acquisition System," (Eighth Edition, January 1968)
developed by the Instrumentation Development Branch,
Methods Development and Quality Assurance Research
Laboratory, USEPA, National Environmental Research
Center, Cincinnati, Ohio. Computer standards should
include a language specification.
3. The demonstration, on some existing wastewater systems,
of the integration between collection system automation
22
-------
WASTEWATER COLLECTION SYSTEMS
and plant performance in a plant that is at least partially
controlled by the same computer system as the collection
system, is needed. Present opportunities for doing this
exist in Seattle, and future opportunities will be available
in Cleveland, San Francisco, Detroit, and perhaps Phila-
delphia.
4. Operating strategies must be defined for a variety of
automated systems, considering both the system as a
whole and the responses of the existing treatment
facilities to the varying loads that could be expected in
the treatment of combined sewer overflows. These strate-
gies should reflect the difficulty of operating a large-scale
model on a small process computer. One such strategy is
now being developed by Colorado State University for the
City of San Francisco and should be considered.
5. The range of system sizes or characteristics that lend
themselves to automation of the collection system needs
to be identified. Obviously, some systems are too small to
warrant automation and others have a physical makeup
that is unsuitable for automation. Also, the software, in
the form of simplified models to drive the control
systems, needs to be developed. Such models as the EPA
stormwater management model are much too complicated
and require too much computing power to be useful as
control models.
6. There should be a determination whether the equalization
of flow or pollutant loading is a valid objective in separate
sanitary sewer systems that may become overloaded by
virtue of infiltration or inflow, or by inadequate capa-
cities.
7. The effects on various wastewaters of in-system storage
for a range of time periods must be determined. For
instance, hydrogen sulfide production, solids deposition
and resuspension, odors, and the possibility of slug
loadings and their influence on the treatment facilities,
should be studied for both combined and separate
sanitary wastewaters.
8. Existing combined sewer regulators should be further
evaluated with a view toward automation. Research in
improving these existing regulators and also in the
development of new regulators, such as a replacement for
the fabridam, needs to be accomplished. The fabridam, in
spite of its simplicity, places in the system some unde-
sirable hydraulic characteristics and is not totally amen-
able to automatic control (it is very difficult to determine
the actual level of the fabridam crest).
9. Improved rainfall prediction or projection techniques
(anticipatory modeling) must be developed. The use of
radar as a predictive device for short-term pattern and
direction-of-rainfall relationships should be investigated.
10. A vigorous technology transfer program in the combined-
sewer area and opportunities for programmed collection
system control should be stressed. As a result of the past
seven years of the combined-sewer research program,
much technology has been developed which needs dis-
semination in the same way that treatment process
technology is broadcast to the profession by the Tech-
nology Transfer group.
23
-------
AUTOMATION
OF
BIOLOGICAL TREATMENT PROCESSES
Workshop on Research Needs
Automation of Wastewater Treatment Systems
25
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
DYNAMICS AND CONTROL
OF BIOLOGICAL TREATMENT PROCESSES
John F. Andrews
Environmental Systems Engineering, Clemson University, Clemson, SC 29631
INTRODUCTION
The need for consideration of dynamic behavior in both the
design and operation of biological processes used for waste-
water treatment is frequently greater than that for industrial
processes because of the large temporal variations which occur
in wastewater composition, concentration, and flow rate.
However, our understanding of this dynamic behavior and how
it may be modified through the application of modern control
systems is in its infancy. Gross process failures are all too
frequent and even when these are avoided, it is not unusual to
find significant variations in process efficiency, not only from
one plant to another but also from day-to-day and hour-to-
hour in the same plant.
Dynamic mathematical models are usually necessary for the
description of time-variant phenomena, as commonly encoun-
tered in biological processes, and increasing efforts are being
devoted to their development. The models usually consist of
sets of non-linear differential equations for which analytical
solutions are not available and this is one of the major reasons
why the dynamic behavior of biological processes has not been
adequately considered in past years. Being practical people,
most environmental engineers have said "Why develop a
dynamic mathematical model when it is not possible to obtain
a solution?" However, computer simulation has largely
eliminated this bottleneck and the current problem is not so
much one of being able to obtain a solution as it is to insure
that the model adequately describes the dynamic behavior of
the process being simulated.
When the dynamic behavior of a plant has been defined, the
environmental engineer then should become interested in
modifying this behavior so that it will conform to some
desired behavior. As illustrated in Figure 1, this can be
accomplished through both process design and the incorpora-
tion of control systems. When the characteristics of the plant
to be controlled are fixed, as for existing plants, the
improvements which can be obtained may be limited, fre-
quently because of a lack of flexibility in the plant design.
However, in the design of a new plant it is possible to strike a
better balance between the effort and expense devoted to the
plant and that devoted to the control system. The major point
of significance is that both the control system and the
controlled plant affect the plant outputs and the two should
therefore be designed as an integrated system.
A literature review of present practice and current research
on the dynamics and control of all biological processes is not
possible within the space constraints of this paper and
attention will therefore be devoted to only the two more
common processes used in the U. S., these being the activated
sludge process and anaerobic digestion.
ACTIVATED SLUDGE PROCESS
The activated sludge process is the most commonly used
process for the treatment of wastewaters and consists of two
units, an aeration basin and a sedimentation basin. The three
major inputs to the aeration basin are the wastewater,
concentrated activated sludge from the sedimentation basin,
and air. The microorganisms in the activated sludge react with
the organic pollutants in the wastewater and oxygen in the air
to produce more activated sludge, carbon dioxide, and water.
The effluent from the aeration basin flows to the sedimenta-
tion basin where the activated sludge is separated from the
liquid phase. The process effluent consists of the clarified
overflow from the sedimentation basin. This basin also serves
to concentrate the solids which settle to the bottom of the
tank for recycle to the aeration tank.
The recycle of concentrated sludge from the sedimentation
basin to the aeration basin is an essential feature of the
process. Recycle serves the purpose of both increasing the
concentration of microorganisms in the aerator and maintain-
ing the organisms in a physiological condition such that they
will readily flocculate. However, recycle has also resulted in
difficulties in understanding and modeling the process since it
26
-------
TIME VARYING
INPUTS
(a)
TIME VARYING
OUTPUTS
INPUTS
UNCONTROLLED
PROCESS
OUTPUTS
(b)
10
n 3
Hz
'
-------
AUTOMATION OF WASTE WATER TREATMENT SYSTEMS
o
UJ
-------
BIOLOGICAL TREATMENT PROCESSES
300
BATCH REACTOR SIMULATIONS
DEMONSTRATING THE EFFECT OF
INITIAL STORED MASS CONC.
initial concentration
750 mg/1
no rapid transfer of
substrate to stored mass
8
12
TIME, HRS
16
20
24
Figure 3. Batch Reactor Simulations Demonstating the Effect of the
Initial Stored Mass Concentration
change in the settling characteristics of the sludge since these
are dependent upon, among other factors, the specific growth
rate of the sludge which is a function of the sludge
concentration in the aeration basin. The relative proportions
of stored, active, and inert mass may also change if there are
substantial changes in the specific growth rate.
Bryant (2) was the first to develop a dynamic model of the
secondary settler to express the above interactions on a
quantitative basis. He also incorporated in his model an
expression presented by Pflanz (3) for prediction of the
suspended solids in the settler effluent as a function of the
solids flux for the settler. A modificatioof Bryant's model has
been incorporated in the process model developed by Busby
and Andrews (1).
Control Strategies
In the conventional activated sludge process, the operator
has a relatively limited choice of control actions, these being:
(1) air flow rate, (2) sludge recycle flow rate, and (3) sludge
wasting rate. Of these, air flow rate is primarily of importance
in controlling process economics and does not appear to have
much effect on process efficiency as long as the dissolved
oxygen in the reactor remains above some minimum level.
Exceptions to this may be encountered in the control of
filamentous organism growth and the use of high purity
oxygen in lieu of air. Common strategies for the control of air
flow rate include the regulation of the air flow rate in
proportion to the wastewater flow rate or the regulation of air
flow to maintain a constant dissolved oxygen concentration in
the aeration basin.
Variation of the sludge recycle flow rate in proportion to
the wastewater flow rate is also a common control strategy.
However, this is not always successful, especially when poorly
settling sludges are encountered, since it does not take into
account possible changes in the concentration of sludge in the
recycle flow. An increase in the return sludge flow rate can,
within limits, increase the mass of sludge in the aeration basin;
however, it also increases the solids loading to the secondary
settler as well as creating additional turbulence in the settler
and can sometimes result in an increased carryover of solids in
the process effluent.
There is a net growth of sludge in the activated sludge
29
-------
INFLUENT
WASTE WATER
H
5
o
n
S
m
3)
S
m
CLARIFIED
EFFLUENT '
SEPARATOR
GATE VALVE
-------
BIO LOGICAL TREATMENT PROCESSES
process and sludge must therefore be intermittently or
continuously wasted from the system. One control strategy is
to waste that amount of sludge each day which will maintain a
constant mass of sludge in the aeration basin; another is to
waste sludge whenever the sludge blanket level in the settler
exceeds a certain depth. Sludge wasting may also be used when
poorly settling or bulking sludge is encountered in order to
prevent the sludge blanket level from rising in the separator
until sludge is discharged in the effluent. However, on a
long-term basis this may be the wrong control action, since, if
the bulking is due to an overload of organic materials, as is
frequently the case, sludge wasting will decrease the amount of
sludge in the reactor thus resulting in still further overloading
with possible process failure.
Two or more of the above mentioned control actions may
be combined and an example of this has been given by Brouzes
(4) in which he controlled sludge wasting to maintain a
constant specific growth rate of the sludge. Using a special-
purpose analog computer, he regulated and measured the air
flow rate required to maintain a constant dissolved oxygen
concentration in the aeration basin. Assuming a constant
oxygen transfer efficiency, the computer then calculated the
oxygen uptake rate and related this, using empirical coeffi-
cients, to the sludge production rate. The computer then
calculated and controlled the rate of sludge wasting to
maintain a constant specific growth rate of the sludge. A
safety override control was provided to actuate sludge wasting
whenever the sludge blanket level exceeded a certain depth in
the settler.
The step feed activated sludge process (Fig. 4) permits a
fourth control action to be taken, this being the ability to
regulate the points at which wastewater is added along the
length of the aeration basin. The basin is usually constructed
in the "folded" fashion shown for economy of construction.
This control action is especially effective for poorly settling or
bulking sludge which can lead to process failure by loss of the
activated sludge in the overflow from the sedimentation basin.
Andrews and Lee (5) have illustrated the value of step feed as
a control action by computer simulation using a simplified
dynamic model. Figure 5 shows the transient effect of
suddenly shifting from an operational mode where all of the
wastewater is admitted to stage one (see Fig. 4) to an
operational mode where all of the wastewater is equally
divided between stages two and three. The sludge in the settler
is rapidly transferred to the aerator and there is also a rapid
decrease in the solids flux to the settler. Both of these
responses would have the short-term effect (hours) of decreas-
ing the mass of solids carried over in the effluent from the
settler. These predictions are qualitatively verified through the
field studies reported by Torpey (6) in his work on the step
feed process at the Bowery Bay plant in New York City.
Torpey also demonstrated that there was a long-term (days)
improvement in the settling characteristics by application of
this control action.
The increasing use of high-purity oxygen processes in the
U. S., coupled with the increased installation of computers in
wastewater treatment plants, offers the possibility of control
from a variable more fundamental than any now in common
use. This is the specific oxygen utilization rate which is
expressed as mass of oxygen utilized per unit of time per mass
of sludge in the aeration basin. This is a direct indicator of the
"activity" of the microorganisms in the aeration basin. It
could be calculated on a continuous basis from oxygen and
solids balances on the aerator. The oxygen balance would be
relatively easy to perform for the closed type of high purity
oxygen processes since these are, in effect, on-line respir-
ometers.
ANAEROBIC DIGESTION
The anaerobic digestion process is widely used for the
treatment of organic sludges from municipal wastewater
treatment plants. The process has several significant advantages
over other methods of waste treatment and among these are
the formation of useful by-products such as methane gas and a
humus-like slurry well suited for land reclamation. Unfortun-
ately, even with these advantages the process has in general not
enjoyed a good reputation because of its poor record with
respect to process stability as indicated through the years by
the many reports of "sour" or failing digesters. The major
problems with the process appear to lie in the area of process
operation as evidenced by its more successful performance in
large cities where there is less variation in the influent sludge
and skilled operation is more prevalent. At the present time,
operating practice consists only of sets of empirical rules and
there is a significant need for a rational control strategy to put
process operation on a quantitative basis.
Dynamic Model
The present dynamic model of the process has evolved over
the past ten years and is discussed in more detail in the
publications of Andrews (7, 8), Andrews and Graef (9), and
Graef and Andrews (10, 11). The model, summarized in Figure
6, was developed from material balances on the biological,
liquid, and gas phases of a continuous-flow, complete-mixing
reactor. The components on which material balances are made
are given below:
Biological Phase
1. Organisms
Liquid Phase
1. Volatile acids.
2. Conservative toxic agent.
3. Cations.
4. Bicarbonate.
5. Dissolved carbon dioxide.
6. Methane.
Gas Phase
1. Carbon dioxide.
2. Methane.
31
-------
Is Ha
,. 1300
00
a:
o
1200
> 200
£
150
to 1100
1000
900
to
100
50
0
t « 0; ALL FEED TO REACTOR I '
t 0 ; FEED DIVIDED EQUALLY
JO REACTORS 2 AND 3
ORGANISM MASS
SEPARATOR,
"Sec
-i
200 HG/L
200 MG/L
KD - 0.005 HRS'
fl - 0.2 HRS"1
0.5 GH X/GM S
150 CU M/HR
131* SQ M
V,,V2,V3,Vi, - 150 CU M
KE 10
R - 0.25
S0
KS
Y -
F -
A -
SOLIDS FLUX TO
SEPARATOR, SF
ORGANISM MASS IN
REACTORS,
2.0
o
o
O m
CO »
0.5 1.0 1.5 2.0 2.5 3.0 3.5
ELAPSED TIME AFJIR CHANGE IN FEED PATTERN, HRS'
1.5
C?
f»
m
1.0
0.5
CO
o
I
TO
-------
BIOLOGICAL TREATMENT PROCESSES
V, V6
PT.D
(C02)DO
Txo *
F-v .
Ka, K, K| , KJ_Q
x0 r
So
Y
>A.KS,K,
F.V.D
GAS PHASE
dPC02 y PC02
dt T VG o v8 °
Q = OCOt+ OCH4+ QH20, QC02 « - DVT0
»
M T
o s ^" IX ~ X ) T MX ~ K 1
dt v Yx/s
^ ~ Ka HS
R»=YCOt/X>* X. ' OCH4= D V YCH4/X P X,
Q
PC02
z,
(CO^,
H*
x, r
S' t
Figure 6. Summary of Mathematical Model and Information Flow
33
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
There are strong interactions between the phases as well as
internal to each phase. These interactions must be considered
if the model is to predict the dynamic response of the five
variables most commonly used for predicting process condi-
tion which are: (1) Volatile acids concentration; (2) alkalinity;
(3) pH; (4) gas flow rate; and (5) gas composition. The
following relationships were used to express these interactions
on a quantitative basis:
1. Yield coefficients.
a. Moles organisms produced/mole volatile acid utilized.
b. Moles carbon dioxide produced/mole organisms pro-
duced.
c. Moles methane produced/mole organisms produced.
2. Kinetics of organism death due to a conservative toxic
agent.
3. Inhibition function for relationship between organism
growth rate and un-ionized acid concentration.
4. Equilibrium relationship between ionized acid, un-
ionized acid and pH.
5. Equilibrium relationship between dissolved carbon diox-
ide, bicarbonate and pH.
6. Charge balance on ionic species in solution.
7. Henry's law.
8. Mass transfer equation for transfer of carbon dioxide
across the gas-liquid interface.
The model has been kept as simple as possible by
considering the conversion of volatile acids to methane and
carbon dioxide as the rate limiting step. It is also assumed that
there is no lag phase, endogenous respiration, or inhibition by-
products. The model is also restricted to a pH range of 6 to 8
and does not consider the precipitation or dissolution of solid
chemical phases such as calcium carbonate.
Two key features of the model are the use of an inhibition
function in lieu of the Monod function to relate volatile acids
concentration and specific growth rate for the methane
bacteria and consideration of the un-ionized fraction of the
volatile acids as both the growth limiting substrate and
inhibiting agent. The use of an inhibition function is an
important modification since it enables the model to predict
process failure by high concentrations of volatile acids at
residence times exceeding the washout residence time. Con-
sideration of the un-ionized fraction of the volatile acids as the
inhibiting agent resolves the conflict which has existed in the
literature as to whether inhibition is caused by high volatile
acids concentration or low pH. Since the concentration of
un-ionized acids is a function of both total volatile acids
concentration and pH, both are therefore of importance.
Digital computer simulation studies provide qualitative
evidence for the validity of the model by predicting results
which have been commonly observed in the field. Among the
results predicted by the model are: (1) At steady state, an
increase in the alkalinity concentration in the digester results
in an increase in the operational levels of pH and volatile acids;
(2) failure of the process can occur through hydraulic, organic,
and toxic material overloading; (3) the course of failure, as
evidenced by the behavior of the operational variables, pH,
alkalinity (HCOs), volatile acids concentration (S), and gas
composition is qualitatively the same as that observed in the
field; (4) stopping or reducing the flow to the reactor, the
addition of base, or recycle of sludge from a second-stage
reactor are effective techniques for curing failing digesters.
Process Stability
Hybrid computer simulations were used to analyze process
stability by simulating digester overloading and observing what
changes in design and operational characteristics provided the
best buffer against process failure. The analysis procedure
involved making a change in a digester parameter, such as
increasing the residence time (0), followed by simulating larger
and larger step increases in digester loading until failure
occurred. By plotting the locus of points of critical substrate
loading rates vs. reactor residence time or other parameter, it
was possible to obtain a semiquantitative measure of digester
stability. An example of a stability analysis for the effects of
residence time and alkalinity is given in Figure 7.
In addition to the increases in stability which are obtained
by increases in residence time or alkalinity, stability also
increases sharply with an increase in the concentration of
methane bacteria in the digester. This increased concentration
can be attained by increasing the influent substrate concentra-
tion (sludge thickening) or by recycle of concentrated digested
sludge from a second stage. It is significant that three of the
measures for improving stability, increased residence time,
alkalinity, and influent substrate concentration, can be at-
tained by sludge thickening. Other simulations indicated that
process stability could be enhanced by the incorporation of
suitable control systems.
Control Strategies
Operating practice for the anaerobic digester consists
primarily of manual procedures, with notable exceptions being
closed-loop temperature control and the use of density control
on the feed sludge. However, an outstanding step toward
automation of these manual procedures is the near-real-time
computer control system developed by Philadelphia (12). This
system uses a remote terminal as an interface between the
plant operator and the computer. The operator inputs his data
into the terminal and the computer then reacts to the input
data using the internal program of empirical and theoretical
algorithms and the previous data inputs to determine the new
operating conditions. Sludge charging rates, mixing require-
ments, etc., are printed out at the terminal and the operator
then implements these instructions. This near-real-time control
represents a valuable first step toward automation.
Using the dynamic model previously described with simula-
tion by the hybrid computer, Graef and Andrews (10, 11)
have explored a variety of control signals, controller modes,
and control actions for the anaerobic digester. The simulations
34
-------
s.?
(P C
O o
&§
i
i.
/«
8
3-
<= >
£E
-n 2
> O
C >
xD -H
60.0
OJ
en
g ~O
o- co m
CD "^3
CN3
3 <
3 O
o r"
m
m
50.0
fe
SO.PA
9 * 15 Days
01
0 mM
10 ^m/L as Acetic
597 gm/L as C
1
0 1000 2000 3000 4000 5000 6000
ITIAL STEADY STATE BICARBONATE ALKALINITY,HC03~, mg/L as CaC03
n
6
r
33
m
-I
D
3)
O
O
m
m
en
-------
,,2500-
faw-
8
tisoo-
5100C-
500-
j 15
s
I 10
?Ts
NO CONTROL
S02-40gm/L
10 gm/ L as ACETIC
ON-OFF GAS SCRUBBING
STAGED RECYCLE
FIRST STAGE QR 50 L/DAY
SECOND STAGE QR . 175 L/DAY
60 -
30 iS
S 50
UJ
u
«
^ 30
LU CVJ
s °
IOC;
en
=>
20
b.
S02 * tOgrn/Las ACETIC
JL
J_
10
7.0
6.8
6.6
6.4
-400?
e
E
3»oP
200|
Q.
<
100 I
S02 40 gm/Las ACETIC
I I 1 L
J L
J L
S02 * 40 gin/Las ACETIC _
S02 "27gm/L
UIO CONTROL
FEEDBACK CONTROL SIGNAL IS pH
FIRST STAGE ON-OFF BAND
OFF pH - 7.00
ON pH « 6.75
SECOND STAGE ON-OFF BAND
OFF pH - 7.00
ON pH « 6.65
1 I I I I
TIME, Days
-TIIVIE, Days
-------
BIOLOGICAL TREATMENT PROCESSES
indicated that the most effective control strategy was directly
dependent on the type of overloading. The recycle of gas from
which carbon dioxide has been scrubbed, a new control action
proposed as a result of this research, and base addition, both
using pH as the feedback signal, were best suited for the
correction of organic overloading. The simulated response of a
digester to this control action is presented in Figure 8. A
simple on-off controller mode was used and the dashed lines
indicate the controlled response to a step forcing in influent
substrate concentration insufficient to cause process failure,
while the solid lines indicate the response to a step forcing
which would cause failure.
Failure by an overload of toxic materials was best pre-
vented by the recycle of concentrated sludge from a second
stage using the rate of methane production as a feedback
signal. Although the control action proposed is not new,
having first been proposed by Buswell (13) in the 1930's, the
proposed control signal, rate of methaiie production, is new
and should be one of the best indicators of digester condition
with respect to overloading with toxic materials. The rate of
methane production can be easily calculated from the com-
mon measurements of flow rate and composition of the gas
phase and would be an excellent indicator of the activity of
the methane bacteria which are the most sensitive and critical
organisms in the digester. An analogue can be drawn between
the use of the rate of methane production as an activity
indicator in the anaerobic digestion process and the use of
oxygen utilization rate as a measure of microbial activity in
the activated sludge process.
RESEARCH NEEDS
The dynamic models, process stability characteristics, and
control strategies summarized herein are by no means com-
plete and still require further development. Some additional
factors which should be incorporated into the models are as
follows:
Activated Sludge Process
1. An improved procedure for predicting the concentration
of suspended solids in the effluent from the secondary
settler. An important fraction of the effluent BOD is due
to these suspended solids.
2. Establishment of a quantitative relationship between the
settling characteristics of activated sludge and the
process parameters. At present only an empirical relation-
ship between interface settling velocity and sludge age is
available and there is some doubt as to the validity of
this relationship.
Anaerobic Digestion
1. Incorporation of the effects of changes in temperature
into the dynamic model. The process has a reputation of
being unstable with respect to sudden changes in
temperatures; however, this has not been expressed on a
quantitative basis. Process stability with respect to
temperature changes would be of special importance for
the thermophilic version of anaerobic digestion.
2. Establishment of a quantitative relationship between the
solids-liquid separation characteristics of the digested
sludge and the process parameters. This is of importance
in connection with vacuum filtration or centrifugation.
Also of importance in this respect would be the
establishment of a quantitative relationship between the
quality of the supernatant, filtrate, or centrate and the
process parameters.
In addition to the above specific research needs, it should
also be mentioned that the dynamic models, process stability
characteristics, and control strategies summarized herein have
only been validated by simple laboratory experiments, litera-
ture searches, and discussion with knowledgeable operations
engineers. This is, at best, only semi-quantitative validation
(responses are in the right direction and right order of
magnitude) and both pilot and full-scale field experimentation
will be needed for quantitative validation. In this respect, it
should be emphasized that models are evolutionary in nature
and will change as more knowledge is gained about the
process. A model which is quite adequate as a first approxi-
mation may be replaced at a later date by a more exact model
with better estimates of the coefficients, fewer empirical
relationships, and inclusion of more variables. This evolutionary
nature of models is not always recognized and can lead to
reluctance on the part of an investigator to either modify or
discard a model in the same fashion that investigators in past
years have sometimes been reluctant to modify or discard
verbal hypotheses.
Still another very significant research need is the combina-
tion of dynamic models for the individual processes into an
overall dynamic model for a wastewater treatment plant with
subsequent use of the model to explore computer-compatible
control strategies for the plant. A preliminary effort in this
direction is the work of Bryant (2) who coupled dynamic
models of the primary settler, aeration basin, secondary
settler, and chlorine contact basin. Such a model, after
appropriate validation, could then be used to explore inter-
actions between the individual processes with the ultimate
objective of establishing an optimal control strategy for the
plant. The author has been working toward this objective for
the past six years.
ABBREVIATED GLOSSARY
A concentration of anions other than HCO5, CO 3, S" and OH
C concentration of cations other than the hydrogen ion
F liquid flow rate to reactor
HS un-ionized substrate concentration
Q total dry gas flow rate from reactor
S total substrate concentration
S" ionized substrate concentration
Ty concentration of conservative toxic chemical agent
V reactor liquid volume
VQ volume of gas space in reactor
X organism concentration
Z net cation concentration
9 hydraulic residence time
37
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
REFERENCES
1. Busby, J. B. and Andrews, J. F., "Dynamic Modeling and Control
Strategies for the Activated Sludge Process," Jour. Water Poll.
Control Fed. (In Press).
2. Bryant, J. O., Jr., "Continuous Time Simulation of the Conven-
tional Activated Sludge Wastewater Renovation System," Ph.D.
Dissertation, Clemson University, Clemson, SC (1972).
3. Pflanz, P., "Performance of (Activated Sludge) Secondary Sedi-
mentation Basins," Advances in Water Pollution Research (Jen-
kins, S. H., ed.), Pergamon Press, New York (1969).
4. Brouzes, P., "Automated Activated Sludge Plants with Respiratory
Metabolism Control," Advances in Water Pollution Research
(Jenkins, S. H., ed.), Pergamon Press, New York (1969).
5. Andrews, J. F. and Lee, C. R., "Dynamics and Control of a
Multi-Stage Biological Process," Proc. Ivth Intern. Fermentation
Symp. (Terui, G., ed.), Society of Fermentation Technology,
Japan, Yamada-Kami, Suita-shi, Osaka, Japan (1972).
6. Torpey, W. N-, "Practical Results of Step Aeration," Sew. Works
Jour., 20, 781 (1948).
7. Andrews, J. F., "A Mathematical Model for the Continuous
Culture of Microorganisms Utilizing Inhibitory Substrates," Bio-
technol. Bioeng., 10, 707 (1968).
8. Andrews, J. F., "Dynamic Model of the Anaerobic Digestion
Process," Jour. San. Eng. Div., Proc. Amer. Soc. Civil Engr., 95,
SA1, 95 (1969).
9. Andrews, J. F. and Graef, S. P., "Dynamic Modeling and Simula-
tion of the Anaerobic Digestion Process," Anaerobic Biological
Treatment Processes, Advan. Chem. Ser., No. 105, American
Chemical Society, Washington, DC (1971).
10. Graef, S. P. and Andrews, J. F., "Mathematical Modeling and
Control of Anaerobic Digestion," Water-1973 (Bennett, G. F.,
ed.),AIChESymp. Ser. 70, No. 136, 101 (1974).
11. Graef, S. P. and Andrews, J. G., "Stability and Control of
Anaerobic Digestion," Jour. Water Poll. Control Fed., 46, 666
(1974).
12. Guarino, C. F., Oilman, H. D., Nelson, M. D. and Koch, C. M.,
"Computer Control of Wastewater Treatment," Jour. Water Poll.
Control Fed., 44, 1718(1972).
13. Buswell, A. M., "Septic tank to controlled digestion," Biological
Treatment of Sewage and Industrial Wastes, Vol. II. (McCabe, J.
and Eckenfelder, W. W., eds.), Reinhold Publ. Co., New York
(1958).
AUTOMATION OF THE ACTIVATED SLUDGE PROCESS
Michael J. Flanagan
Brown and Caldwell, Consulting Engineers, 66 Mint Street, San Francisco, CA 94103
INTRODUCTION
The conventional activated sludge process consists of two
main units, an aerobic biological reactor (oxidation tank)
followed by a solids-liquid separator (solids concentrator/sedi-
mentation tank). As shown in Figure 1, process inputs include
(1) wastewater, usually effluent from a primary sedimentation
tank; (2) return activated sludge (RAS), which is a concen-
trated suspension of organisms and substrate recycled from the
sedimentation tank; and (3) an air or pure oxygen supply to
provide oxygen and mixing in the oxidation tank. Process
outputs include (1) process effluent; (2) waste activated sludge
(WAS); and (3) carbon dioxide.
The optimum performance goal of the activated sludge
process can be stated as the production of a specific and
consistent quality of process effluent with a minimum
Air or
02
1 1
CO2
Primary
Effluent
i
OXIDATION
TANK
t
I
RAS
STORAGE
SEDIMENTATION
TANK
^
i
* Activat
Under ft
(RAS +
*
Process
Effluent
ed Sludge
ow
WAS)
IS ^ To
Disposal
I I
Figure 1. Activated Sludge Process
38
-------
BIOLOGICAL TREATMENT PROCESSES
consumption of manpower, materials and energy. In practice,
this goal has not yet been attained. However, a number of
activated sludge treatment plants are currently being designed
or constructed which incorporate modern automatic control
systems for monitoring and controlling the activated sludge
process. Simplified versions of these control systems are
described in this paper, together with suggested improvements
for attaining increased process performance.
The most critical step in the activated sludge process is
usually the removal of biological solids in the secondary
sedimentation tank. Unfortunately, the importance of gravity
sedimentation in the activated sludge process is often over-
looked. Provided that the oxidation tanks have sufficient
capacity, virtually complete utilization and conversion of the
degradable organic matter to biological solids will take place.
The process effluent contains both dissolved and suspended
oxidizable organic matter. In many plants, the major portion
of the oxidizable organic matter in the process effluent is due
to suspended solids which have not been removed by the
sedimentation process.
Strong interactions exist between the oxidation and
sedimentation tanks. For example, the flocculation and
settling characteristics of the oxidation tank mixed liquor
influence the quality of the process effluent and the solids
concentration of the RAS. The solids concentration, in turn,
influences the solids concentration that can be maintained in
the oxidation tank, the sludge age and the solids flux to the
sedimentation tank. An automatic control system for the
activated sludge process should therefore monitor and control
both process units and their mutual interactions.
PROCESS CONTROL
The following basic control actions can be exerted in the
activated sludge process:
1. Control of the point(s) of addition of primary effluent
to the oxidation tank (i.e. step feed process).
2. Control of an inventory of activated sludge in a storage
tank to offset the effects of diurnal variations in
wastewater flow and strength.
3. Control of the air or oxygen flow rate to satisfy the
mixed-liquor oxygen demand.
4. Control of the RAS flow rate.
5. Control of the sludge wasting rate.
The extent to which control actions 1 and 2 can be
implemented is determined by the configuration of process
units and equipment in a given plant. However, control actions
3, 4 and 5 can be readily incorporated in existing plants
without making any major modifications to the process
equipment. Several different control systems that include
some or all of control actions 2, 3, 4 and 5 are described in
this paper. All of the process sensors are presently available
and either digital or analog logic can be employed to solve the
control equations, although, in most instances, digital logic is
the preferred technique.
The control systems shown in Figures 4 through 9 all
employ a system of RAS pump control which is designed to
maintain the sludge blanket in the associated sedimentation
tank at a preset level or within preset limits. The blanket levels
are set low to minimize the occurrence and duration of anoxic
(zero DO) conditions.
Under conditions of limiting solids flux, the operator must
override the automatic controls and reduce.the solids flux to
the sedimentation tanks by taking the appropriate remedial
action.
A description of each of the control systems is given below.
Instrumentation symbols and identification are in accordance
with Standard S5.1 (1973) of the Instrument Society of
America.
Control of Oxygen and Air Dissolution in Mixed Liquor
(Figures 2 and 3)
Control of air and oxygen dissolution in the mixed liquor is
an important parameter in the activated sludge process. The
normal strategy is to add sufficient air or oxygen to meet the
time-varying oxygen demand of the mixed liquor. Because
electrical energy is one of the major operating costs of the
activated sludge process, there is an economic incentive to
minimize unnecessary aeration.
Most activated sludge plants use air as an oxygen source.
The preferred method for air flow control is to maintain the
desired dissolved oxygen (DO) level in each oxidation tank.
The design of a DO control system depends on the oxidation
tank configuration and the aeration method. Numerous
examples are provided in the literature. A DO control system
that has been successfully applied by Brown and Caldwell to a
number of diffused-aeration type activated sludge plants is
given in Figure 2.
In the high-purity oxygen process, the oxidation tanks are
covered and thus function as on-line respirometers. Four stages
are normally provided within each oxidation tank. A cryogenic
oxygen plant supplies oxygen to the oxidation tanks in
proportion to the oxygen uptake rate as reflected by changes
in oxidation tank gas pressure. The oxygen is introduced to
each of the oxidation tanks in the first stage only, via the
first-stage recycle compressors, and is exhausted as vent gas
from the fourth stage. The system is designed to utilize 90
percent of the oxygen supplied. A submerged turbine aerator
is installed in each oxidation tank stage to maintain DO and
keep solids in suspension in the mixed liquor.
The oxygen supplied to the intake of the first-stage
compressors is controlled to maintain a constant gas pressure
in the oxidation tanks by modulation of a control valve on the
oxygen supply header. The oxygen purity control system is
designed to maintain constant oxygen purity in the gas venting
from each oxidation tank by modulation of the vent gas valve.
39
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
V
TM @ &-
Centrifugal
Blower (typ)
From
Other
Blowers
Oxidation Tank
Pass (typ)
To OfAer
> Oxidation Tank
Air Headers
-Header
Figure 2. Diffused Aeration DO Control System
Effluenl
Stage I
Stage 2 Stay* 3
OXIDATION TAHK (Tfp)
Figure 3. Oxygen Dissolution Control System
40
-------
Constant Mixed Liquor Solids Concentration Control (Figure
4)
A common method of solids control in the activated sludge
process is to maintain a constant mixed-liquor suspended
solids (MLSS) level and thus a constant mass of sludge in the
oxidation tank. This method provides good results if the
strength of the incoming wastewater remains reasonably
constant. Under these conditions, the amount of sludge that is
wasted equals the net sludge growth in the system. This
method is not recommended for activated sludge systems that
treat wastewaters with widely fluctuating characteristics be-
cause the food-to-microorganism (F/M) ratio is not held
constant.
Two infrared sludge blanket sensors are provided in each
sedimentation tank to provide on-off control of the WAS
pump and thus maintain the sludge blanket between preset
high-low levels.
Constant Food-to-Microorganism Ratio Control (Figures 5 and
6)
Two constant F/M ratio control systems are described. In
the control system shown in Figure 5, the F/M ratio is
calculated by measuring the respiration rates of the return
activated sludge and the mixed liquor (Genthe, Arthur and
Srinivasaraghavan (1). In the control system shown in Figure
6, the F/M ratio is calculated by measuring the total TOC in
the primary effluent, the soluble TOC in the process effluent
and the average volatile suspended solids in the mixed liquor.
BIOLOGICAL TREATMENT PROCESSES
Both F/M ratio control systems incorporate a sludge
blanket level control system whereby the RAS pumps are
controlled to maintain a constant blanket level in the
associated sedimentation tank. The sludge blanket measure-
ment system comprises a hoist-driven infrared blanket sensor
which is track-mounted on the sedimentation tank fixed
bridge. Under program control, the sensor measures the
blanket level at 6 (say) locations during a radial traverse of the
tank, thus enabling the sludge blanket profile and volume to
be automatically computed. The sludge blanket level/volume
signal so derived serves as the process variable for the RAS
pump control system.
A biochemical oxygen demand meter, such as the Badger
model OD-2000, can be employed to measure respiration rates
of the RAS and the mixed liquor. These measurements,
together with measurements of plant flow and RAS flow, can
be employed to calculate the F/M ratio.
The microorganism concentration (M) can be obtained
from the oxygen demand rate of the RAS. The oxygen
demand rate is a measure of the viable microorganism
concentration because respiration is essentially endogenous.
When the RAS is returned and mixed with the primary
effluent, the RAS biomass is provided with the organic
substrate present in the primary effluent. Measurement of the
oxygen demand of the mixed liquor downstream of the point
of RAS-primary effluent mixing reflects the additional
oxygen required by the biomass to utilize the new substrate in
support of new cell synthesis. Thus the difference between the
Plant Flaw )
ritnary _
ffluent
u-"-J
'
1
1
1
J^ao
©
OXIDATION TANK (Typ)
Process
Effluent
7°
Disposal
Figure 4. Constant MLSS Control System
41
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
Plant Flo*(Flnf) »
I
Primary
Effluent
To
Disposal
Figure 5. F/M Ratio Control System Using Respirometry
Primary
Efflutut
Disposal
Figure 6. F/M Ratio Control System Using TOC and MLVSS Measurements
42
-------
oxygen demand of the mixed liquor and the RAS provides a
measure of F, the metabolizable substrate in the primary
effluent. As shown in Equation 1, the F/M ratio can be
calculated from two flow measurements and two respiration
rate or oxygen demand measurements.
(1)
F/M
ODMLFlNF
OD
where:
F/M
K
RAS RAS
the food to microorganism ratio (preset
by operator)
constant determined by operating experi-
ence for a given plant
oxygen demand of mixed liquor
BIOLOGICAL TREATMENT PROCESSES
microorganisms, and average MLVSS as a measure of the
microorganism population. According to Weddle and Jenkins
(2), volatile suspended solids is an excellent index of the viable
microorganism content of activated sludge for the practical
operating range of activated sludge plants treating domestic
sewage. Because the hydraulic regime of most oxidation tanks
is between complete mixing and plug flow, several measure-
ments of MLVSS may be required to obtain an average value
of the MLVSS concentration in the oxidation tank.
Because the total TOC measurement includes oxygen
demand of substrate which cannot be assimilated by the
microorganisms, a measurement is made of the soluble TOC in
the process effluent, and this value is subtracted from the
primary effluent TOC value to give a measure of the substrate
available to the microorganisms. The F/M ratio is given by
Equation 2 as follows:
F/M =
^I)RAS = oxygen demand of RAS
E
Total TOC[NF - Soluble TOC£FF
MLVSS
,(2)
influent flow
RAS flow which is indirectly controlled
by varying the controlled variable
"INF
RAS
A second method of F/M ratio control is based on the
measurement of primary effluent TOC less the process effluent
soluble TOC as a measure of substrate available to the
where:
Total
= constant determined by operating experi-
ence in a given plant
= total TOC in the primary effluent
Solu- TOCEFF = soluble TOC in the process effluent
MLVSS
mixed liquor volatile suspended solids
To
Disposal
Figure 7. Hydraulic SRT Control System
43
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
Solids Retention Time Control (Figures 7 and 8)
Two solids retention time (SRT) control systems are
described. Solids retention time is defined as the average
retention time of biological solids in the system and includes
solids in the secondary sedimentation tank and, if necessary,
solids in the mixed liquor and RAS transportation facilities. It
is calculated as the sum of the volatile suspended solids (VSS)
in the oxidation and secondary sedimentation tanks divided by
the sum of pounds of VSS intentionally wasted and those
unintentionally lost over the effluent weirs. A distinction is
drawn between SRT and sludge age; calculation of sludge age
is based only on the solids in the oxidation tank.
The simplest SRT control method is known as the hydraulic
control method (Figure 7) and was first proposed by Garrett
(3) in 1958. More recent applications of this method have
been described by Walker (4) and Burchett (5). If the solids in
the sedimentation tank overflow are ignored and mixed-liquor
wasting is employed, then the quantity of mixed liquor that
must be continuously wasted is given by Equation 3 as follows:
'ST
SRT
sedimentation tank volume in cubic feet
solids retention time in days.
rWML
7.48 OT
ST
(3)
where:
WML
'OT
SRT
waste mixed liquor flow in gpd
oxidation tank volume in cubic feet
Despite the fact that this process suffers from inflexibility
and requires the wasting of a greater volume (approximately 4
times) of waste compared to RAS wasting, the control system
is very simple and requires only measurement and throttling of
the waste mixed liquor stream.
A more exact and flexible method of SRT control is shown
in Figure 8 and is based on computing the total solids in the
activated sludge system and the total solids wasted from the
system, both intentionally and unintentionally. For example,
if the SRT is selected as 5 days, then 20 percent of the total
solids in the system are wasted continuously. As the influent
BOD varies, the WAS wasting rate varies but the percent
removal rate from the system remains constant. Furthermore,
because a fixed percentage of the total solids are wasted daily,
the percentage of viable organisms in the system is not
important because if 20 percent of the total solids are being
wasted, then 20 percent of the viable microorganisms are also
being wasted. The relationship between SRT and F/M ratio is
shown in Equation 4:
SRT4 = y(F/M)-b (4)
where:
y =
cell growth constant
and b = endogenous respiration rate
Process Effluent
Flow (Feff)
Plan, Flo* (Finf) -
To
Disposal
Figure 8. SRT Control System
44
-------
Thus it can be seen that if the SRT is held constant, the
F/M ratio remains constant. Using this method of SRT
control, sludge is wasted from the RAS channel. The wasting
rate, F^^g, is controlled in accordance with Equation 5:
(5).
hWAS
where:
M
SRT
F
8.33 . F
EFF
ssEFF
EFF
SS
EFF
SS
RAS
controlled WAS flow, mgd
total solids in system, Ibs
5-20 days (system set point)
effluent flow, mgd
process effluent suspended solids, ppm
RAS suspended solids, ppm
Combined SRT and F/M Ratio Control System
The combined SRT and F/M ratio control system shown in
Figure 9 incorporates features provided in Figures 5 and 8,
together with RAS storage facilities. Although this process
configuration suffers from the disadvantage that the RAS is
pumped twice, it should be capable of responding to diurnal
BIOLOGICAL TREATMENT PROCESSES
variations in BOD loading in addition to maintaining the
desired SRT. Basically, the SRT control is used to control
solids wasting and the F/M ratio control is used to control the
rate of return of RAS to the oxidation tank to match the
time-variant BOD load in the primary effluent. If the level in
the RAS storage tank reaches a preset high level, an override
level control system increases the sludge wasting rate to
prevent the sludge level from rising.
CONCLUSIONS AND RECOMMENDATIONS
Several control systems for the activated sludge process
involving conventional measurement and control techniques
have been described. The performance and productivity of
most existing activated sludge plants can be increased by the
installation of one or more of the control systems that have
been presented in this paper. The degree to which control
systems can be retrofitted to existing plants is usually limited
by a lack of flexibility in the plant design. For new plants,
however, an opportunity exists to incorporate useful control
systems. It is recommended that process and instrumentation
diagrams (P and ID's) be prepared during the functional design
stage to help ensure that a proper balance is established
between the process and the control system.
, The principal justification for automation is to ensure that
treated effluents comply with discharge regulations for various
pollutants. Secondary advantages of automation include in-
creased performance and productivity, minimized costs and
the like. Presently, the principal obstacle associated with
treatment plant automation is the people problem. It is natural
for people to think and act in terms of their own expertise and
to resist acceptance of the unfamiliar. Because the need for
Plant FH,.(F:nl) -
WAS Pump lT,f)
Figure 9. Combined SRT and F/M Ratio Control System
45
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
treatment plant automation has only occurred in recent years,
it is not surprising that most people associated with the design
and operation of wastewater treatment facilities are skeptical
of automation. Poorly designed and maintained automatic
control systems will tend to reinforce this skepticism.
It appears reasonable to assume that, as effluent standards
become more stringent and wastewater reuse for non-potable
purposes increases, automation will play an increasing role in
the operation of the nation's municipal wastewater treatment
plants. Furthermore, as the market for wastewater instru-
mentation expands, we can expect new and improved on-line
sensors, designed specifically for the measurement of waste-
water parameters, to become increasingly available in the
future. The measurement problem is compounded by the
heterogeneous nature of wastewater and the tendency for
probes and sample lines to foul up. Measurements such as
pressure, temperature, flow and level that are common in the
process industries are relatively easy since an extensive
collection of sensors has been developed by the process
industries and may be used in wastewater treatment plants.
The central problem is in the area of analytical measurements.
The use of digital data transmission and control systems is
becoming increasingly cost-effective for wastewater treatment
plant control and data management. Currently, it is estimated
that digital systems are more cost-effective than hardwired
systems for new plants or plant expansions larger than 10 mgd
(6). Another advantage of digital control systems is that
advanced control strategies can be readily implemented. Most
unit processes encountered in wastewater treatment are
characterized by long dead-times and thus feedforward control
can be used to great advantage in minimizing process upsets
and reducing process excursions.
Dynamic mathematical models can be employed in digital
computer control systems for real-time optimizing control.
Brown and Caldwell (7) is currently designing a distributed
digital control system for a 125 mgd pure oxygen plant; the
design calls for running process models in a supervisory level
computer that in turn adjusts the control strategies in the unit
process level computers for performance optimization. As has
been pointed out by Andrews (8), dynamic mathematical
models are usually necessary to describe time-variant waste-
water treatment processes and usually consist of sets of
non-linear differential equations. However, construction of a
dynamic mathematical model requires a number of approxima-
tions about process variable interrelationships. If the approxi-
mations are incorrect or if unanticipated changes occur in the
process, the model is no longer useful.
One method which can overcome this problem is to employ
a self-learning adaptive control system. With such a system, it
is necessary to specify the important process variables but it is
not necessary to specify the interrelationships between the
variables. Providing a good process data base is maintained and
updated, the model develops and continuously updates process
equations which produce the least error in predictions.
Examples of such predictions would be process effluent
chlorine demand and suspended solids concentration.
The firm of Adaptronics Inc. of Virginia, has developed an
adaptive nonlinear modeling system for the predictive control
of complex multivariable processes such as the basic oxygen
furnace and hot strip steel mill runout table cooling sprays.
Background information on nonlinear and adaptive control
techniques is given in references 9 and 10.
REFERENCES
1. Genthe, W. K., Arthur, R. M., and Srinivasaraghavan, R., "On-line
Oxygen Demand Rates of Primary Influent and Effluent and Their
Relation to 5-Day BOD", Presented at 45th Annual Conf., Water
Poll. Control Fed. (1972).
2. Weddte, C. L., and Jenkins, D., "The Viability and Activity of
Activated Sludge", Water Res., 5, 621 (1971).
3. Garrett, M. T., Jr., "Hydraulic Control of Activated Sludge
Growth Rate". Sew. & Ind. Wastes, 30, 253 (1958).
4. Walker, L. F., "Hydraulically Controlling Solids Retention Time in
the Activated Sludge Process", your. Water Poll. Control Fed., 43,
30 (1971).
5. Burchett, M. E., "Physical Facilities for Controlling the Activated
Sludge Process by Mean Cell Residence Time", Presented at
Northern Regional Conf., California Water PolL Control Assn.
(1973).
6. Systems Control Inc., "Phase IV Report, Palo Alto Wastewater
Treatment Plant Automation Project" (1974).
7. Brown and Caldwell, "Process Control Report; Sacramento Re-
gional Treatment Plant" (1974).
8. Andrews, J. F., "Review Paper; Dynamic Models and Control
Strategies for Wastewater Treatment Processes", Water Res., 5,
261 (1974).
9. Control Engineering, "Nonlinear and Adaptive Control Techniques",
Proceedings of the First Annual Advanced Control Conference
Sponsored by Control Engineering and the Purdue Laboratory for
Applied Industrial Control (1974).
10. Anon, "Controls that Learn to Make their Own Decisions", Bus.
Week, April 6, 1974.
DISCUSSION
J. O. Bryant:
The need for good data has been stressed. Would the authors
comment on the status of instrumentation to obtain these
data?
Michael du Cros:
What specifically is required in the way of new and
improved on-line sensors? Is it new parameters, new analytic
techniques, better reliability, or a combination of these?
L. A. Schafer:
Can you tell me of any wastewater processes now function-
ing that successfully utilize composition analysis of the liquid
stream (other than residual chlorine)?
Stephen P. Graef:
Would Mr. Flanagan please indicate the application status
of each of the control schemes illustrated in his paper? Where
have they been tried, or what must be done to implement
46
-------
BIOLOGICAL TREATMENT PROCESSES
them on a full scale field unit?
John K. Nelson:
I have three questions for Mr. Flanagan
1. In measuring oxygen uptake rates for RAS, is consideration
given to solids concentration, so that variations in sludge
stability can be taken into account?
2. When measuring clarifier sludge blankets, how is allowance
made for hydraulic expansion of the blanket, so that a false
clarifier solids inventory is not obtained?
3. An objective of process control is to monitor process
variables, and to use these variables to control the process.
However, how does one "monitor" a factor K (Equations 1
& 2) which is based on operating experience?
Richard I. Dick:
The critical nature of solids separation in control of the
activated sludge process is noted in the paper. However, the
control systems presented for maintaining MLSS, F/M, and
SRT would not seem to consider this importance of solids
separation. Rather, an unlimited capacity for solids separation
was tacitly assumed. With real solids separators of limited
capacity, this control strategy might result in overload of the
final tank. The control systems do have high level blanket
regulations, but these controls might contradict the require-
ments for maintaining the required MLSS, F/M, or SRT.
Poul E. Sorensen:
Six years ago Westberg from Sweden introduced the
concept of the totally controlled process, giving a constant
effluent BOD. This process required storage of sludge. One of
the possibilities for sludge storage is the step-feed process, but
has this process in practice been operated as a totally
controlled process?
N. J. Biscan:
I agree that sludge storage for additional recycle in times of
high loading is an important concept. It is one that needs more
field application and testing.
In a recent study at Dow, "Optimizing a Petrochemical
Waste Bio-oxidation System Through Automation" (EPA
Grant No. S800 766), we developed an on-line analog F/M
control system based on repetitive total carbon measurements
of the feed and of a diluted and homogenized aeration basin
mixed liquor sample. The automated control system gave an
updated value of F/M every 12 minutes. The F/M signal
proportionally controlled the fraction of the recycle sludge to
be wasted.
The system was tested on a glycol feed in a 250 gallon pilot
plant. The response time for F/M to return to within 20
percent of its steady-state value after a 50 percent step
increase in feed concentration was 17 hours for the controlled
system (with no sludge storage capabilities) versus 46 hours in
the uncontrolled system. Although there would be faster
response times in municipal systems due to the higher yield of
bacteria, response times for F/M to return to its steady-state
optimum value would still be on the order of hours if sludge
storage facilities are not provided.
I think that F/M is one of the most important parameters
to be controlled in the activated sludge process. Whether the
control action is taken on the basis of total carbon measure-
ments, total oxygen demand measurements, or respirometry
(i.e., oxygen uptake) measurements, our research needs should
include field testing of incorporating sludge storage capabilities
in the F/M control system.
M. Dolan McKnight:
Separating sedimentation from the reactor by interposing a
sludge storage tank seems worthwhile. How would the design
engineer determine the capacity of such a tank, could the
aeration tank size be minimized, and how would this affect
clarifier size? Should sludge wasting be constant or variable in
such a system?
C. S. Zickefoose:
With respect to the storage of sludge in the activated sludge
process, the plant under construction for the City of Portland,
Oregon will have the capability to store up to 800,000 gallons
in four separate tanks of return sludge and/or high-strength
liquors generated in-plant.
The contents of these tanks are aerated and can be
introduced to the aerators by pumps; by monitoring TOC and
solids at various points in the process, in-plant loads and return
flows can be introduced at optimum times during the day.
The plant has a design capacity of 100 MGD average flow
and is due to start up in the fall of 1974.
Heinrich O. Buhr:
The flow diagrams presented by Mr. Flanagan illustrate
some of the strategies currently proposed for control of
activated sludge plants. It seems likely, however, that research
on detailed process models which fully recognize the time-
varying nature of treatment plant operations might lead to a
re-examination and reformulation of many of these concepts.
As an example, consider SRT and F/M ratio control: It is
generally agreed that it would be desirable to maintain a
constant biological growth rate, or F/M ratio, in the aeration
system. Further, it is often assumed that this may be achieved
through maintaining a constant SRT, based on relationships
such as equation 4 in Mr. Flanagan's paper:
SRT-1 = y(F/M)-b
(4)
However, this equation is based on steady-state considerations,
and in a plant which is subject to diurnal flow variations the
relationship is true only on a "daily average" basis. In practice
F/M will vary throughout the course of the day, typically from
50% to 170% of the average value, even when the sludge
wastage rate is held constant.
47
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
If closer control of F/M is required, a more direct approach
than SRT control must be adopted. For this purpose, "food",
F, may be measured by the techniques suggested by Mr.
Flanagan, or estimated from the air supply rate as done by
Brouzes (Andrews, Ref. 4); then mass of micro-organisms, M,
may be adjusted in accordance with the variation in F, to
maintain F/M constant. In the absence of sludge storage, the
most effective control of M may be achieved by manipulating
the return sludge (RAS) rate. Some schemes, however, propose
to keep total RAS flow rate from the sedimentation tanks
constant, and to attempt F/M control by varying the waste
(WAS) flow rate instead (cp. Brouzes, and Figures 5 and 6,
Flanagan). A typical system material balance will show that
the expected WAS flow rate for long-term steady-state
operation is only about 3% of the RAS rate, so that a control
strategy which utilizes WAS as the manipulated variable, can at
best exert a marginal influence on the mass of solids in the
aerator. This is not to say that variation of sludge wasting rate
will not cause changes in biological growth conditions, but the
effect is exerted through slow changes of sludge inventory on a
time scale measured in days. For F/M control on an
hour-by-hour time scale, a more forceful control action is
required, and this usually means direct manipulation of RAS
flow rate.
The main purpose of sludge wasting control should be to
maintain sludge inventory at a desired value on a day-to-day
basis and this may be achieved by simple level control as
illustrated in Figure 4 (Flanagan). Even in this case, however,
dynamic studies will show that the time when the sludge
blanket is at its highest (and the waste sludge pump will tend
to switch on) may not necessarily be the best time to
withdraw sludge. For example, with diurnal inflow fluctua-
tions and, say, ratio RAS control, high blanket levels will
occur when inflow is highest; however, since the controDed
RAS flow rate will also be relatively high at this time, the
recycle sludge concentration will be at its lowest for the day.
When using straightforward level control, waste pumps will
thus switch on during periods of low sludge concentration,
which will tend to maximize, rather than minimize the
volumetric quantity of waste sludge to be withdrawn daily. A
preferable strategy would be one which programs sludge
withdrawal to take advantage of periods of high sludge
concentration.
These points illustrate some of the information which can
be gained from a study of dynamic process models. Research
in this direction is clearly a prerequisite for the design of
adequate control strategies.
GustafObson:
In order to control a process a reasonable quantitative
performance index is needed. Of course we know that the goal
is the cleanest possible water at the smallest possible cost. The
problem is what parameters are really affecting the effluent
quality. This also relates directly to the choice of proper
models. More research is needed to establish what setpoints on
DO, MLSS, F/M, settling characteristics, etc., should be used.
Thus a good model is needed for better understanding of the
quantitative performance index, and both such an index and a
model is needed to obtain good control.
J. O. Bryant, Jr.:
Few treatment plants operate as they were designedSRT,
MLSS, etc., usually cannot be adequately, and accurately,
predicted during the design phase. Hence, operation of most
facilities becomes a function of the operator's ability to "learn"
what the best parameter values for his plant are. Adaptive,
non-linear control has in reality been applied routinely by the
good operator.
Wayne C. Smith:
As more and more batch-type discharges from industries are
going to an activated sludge system, how do we handle these in
the models and how do we predict the effects of these
discharges on the treatment system, both as regards effluent
quality and upsetting the system? Also, how do we include
non-domestic materials that degrade very slowly in this model?
James A. Mueller:
With respect to the pure oxygen system, three problems
exist with use of data to verify our mathematical models,
namely:
1) no gas flow measurements between stages
2) no data on the amount of gas lost from the various
stages due to leakage, and
3) when dealing with the pure oxygen system, our models
must not only include the biological growth functions
but also the system chemistry due to CO2 dilution of
the gas phase.
This makes our models more complex and more difficult to
verify. However, system understanding is increased by making
the attempt.
In general if mathematical models are to be effectively used
in design and control of biological treatment systems, the
model should include measureable parameters and have re-
ceived experimental or field verification. In most cases a
simplified model which approximately describes the system
will be more effectively used than a complex model that more
closely describes the actual system. The next viable phase of
research in the modelling area appears to be verification and
practical utilization instead of complex model development.
P. M. Berthouex:
The problem of pilot plant and field scale testing has been
mentioned. I would like to propose that we need more
partnerships between model-builders and others. Dr. Andrews
has been in such partnerships and I'm sure would agree that
more of this involving others will be profitable. It seems that
48
-------
BIOLOGICAL TREATMENT PROCESSES
more people using the iterative cycle theorydataanalysis-
theory are needed.
Some data we at Wisconsin have collected shows that
certain mechanistic models predict too much variation in
effluent quality. We have seen systems that show no significant
dynamics in the output in spite of fairly strong input dynamics
(10 to 1 diurnal variation in organic load). Of course, not all
plants show such stability. Can we predict which plants
(processes) will exhibit strong dynamics?
I have heard you say before that a good model will predict
system failure. Do you think we need models to predict
system recovery after failure? My thought is that a control
strategy designed to prevent failure may not be sufficient to
bring a "sour" system back into service. Do you think we can
predict, or need to predict, the dynamics of recovery?
Second, we need to ask, "How much control is needed?"
and "How will load balancing help us?" The answer, I expect,
will be different for small plants than for large plants. For
small plants it might be nice to know how to design so that no
elaborate control strategies are needed.
Another area where attention should profit us is stochastic
analysis of actual operating data and to work toward stochas-
tic dynamic models. To some this approach seems to conflict
with mechanistic modeling. Really they should complement
each other; each has strengths and disadvantages. We need
more experience with both, but of the two the stochastic
methodology is less well understood, probably because it is
less intuitively satisfying. In particular, these models are
developed from "normal" operating dynamics. I doubt they
will predict failure and I know they are not useful once a
system has gone far out of control.
Gustaf Olsson:
There is a clear need for a whole spectrum of models
describing, for example, an activated sludge plant. Which
description to use would depend on the purpose for which the
model is required.
System identification is a very powerful technique to
establish what is the proper degree of complexity required.
Using this technique, theoretical models can be compared and
adjusted to real plant performance. Not only input-output
relations (ie. the black box approach) but also parameters for
physical a priori models can be calculated. Moreover, more
reliable models of the inherent system noise can be estab-
lished.
Lars Pallesen:
The topic of modeling has been brought up several times
this morning. In this connectfon I would like to make the
following point: It is possible to drive a car without
understanding why it works, that is, without knowing how
each segment of an automobile functions in detail! It may be
interesting to understand how, say, the engine is working, but
if the objective is to drive the car, such knowledge is of limited
usefulness. What you do need to know is some "macro"
aspects of its total behaviordon't bother about how the
steering gear is put together, but concentrate on learning how
the car responds to turning the steering wheel. In short: worry
selectively.
Similarly, the kind of mechanistic modeling well exempli-
fied in the paper presented by Dr. Andrews, is clearly
scientifically very interesting. At the present stage of develop-
ment, however, these "micro" models appear to be unable to
completely explain the behavior of sewage treatment plants,
and, in the mechanistic way of thinking, further refinements
are therefore needed-making the models even more compli-
cated.
Fortunately mechanistic models are not needed, if our goal
is to control. Much less complicated empirical models,
describing macro features will be adequate, particularly if
models reflecting the stochastic nature of wastewater treat-
ment systems are employed.
This approach to the control problem requires, of course,
that real plant data be available to do the modeling. The very
fact that the empirical modeling procedure forces the re-
searcher to compare his models with real data, must be
considered an inherent strength of the method.
Russell H. Babcock:
With respect to automation of biological wastewater treat-
ment processes, it should be noted that present practice in
other fermentation industries such as brewing and the produc-
tion of antibiotics is still very crude. According to one major
instrument manufacturer, the manufacturing of antibiotics is a
simple batch process. Temperature is controlled and pH is
measured. The principal control is by laboratory analysis of
successive samples. Laboratory animals are still used as means
of testing the final product.
Pilot plant work is underway which includes DO control
and pH control. This work, however, is in its infancy and has
not yet been applied to routine production. The process will
remain batch with the instrumentation being used to minimize
laboratory control. Conclusion: most fermentation processes
are still controlled by the basic techniques of laboratory
analysis which have been in use for many years.
Carmen F. Guarino:
The automation of anaerobic digestion should not be made
complicated. The C02 content of the gas, plus the quantity of
CO2 generated, goes a long way towards automating the
digestion process.
K. J. Jacobson:
In response to Mr. Babcock's comment, the unwillingness
of the fermentation industry to implement warranted and
available sophistication in control strategies should not be used
to excuse the same deficiency for wastewater treatment.
Sophisticated control instrumentation is in fact being applied
49
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
to new ventures in the fermentation industry (protein manu-
facture), and gradually to the classic ventures, drug and
alcohol production. At Penn we are developing a mini-
computer interface for a pilot-size fermentor. At present data
acquisition and data handling capabilities are being established
for model development purposes. The eventual goal is auto-
matic control of the important (and previously unaccessible)
fermentation parameters. Continuous measurement of cell
mass using fluorometry and of glucose substrate by enzymic
oxidation are being developed. These quantities, along with
the more standard measurements, will enable automatic
calculation of growth rate and yield coefficients for use in
process control.
The dynamic character of biological processes has been
treated this morning, but one important addition has not been
discussed. The Monod growth kinetics, or some derivative
which accounts for inhibition, is empirically applied and in
general is a steady-state model since it ignores time lags. It
does not fit dynamic data nearly as well as a first-order
dynamic kinetic model, as has been shown by Young,
Bruleynd Bungay (1). As has been amply pointed out today,
this model refinement is hardly necessary for implementing
practical control strategies. Indeed steady-state models can do
a remarkable job of predicting actual plant performance for
activated sludge wastewater treatment (2). On the other hand,
accurate predictive models will provide important insights to
aid understanding and designing systems, and hence assist in
designing the most useful control strategies. Therefore, al-
though sophisticated dynamic models may never be used for
process control, their development is an immediate and
essential research need in order to establish optimal designs for
processes and control strategies.
REFERENCES:
1. Young, T. B., Bruley, D. F. and Bungay, H. R. Ill, "Dynamic
Mathematical Model of the Chemostat," Biotechnol. Bioeng., 12,
747-769 (1970).
2. Landis, J. G., Casey, J. P., Hartzog, D. G-, "Oxygen Activated
Sludge: Design by Waste Analysis and Modeling," presented at 47th
Annual Conference, WPCF, Denver, CO (1974).
CLOSURE
John F. Andrews:
Reply to J. O. Bryant, Jr.: The author is not engaged in
research on the development of sensors and therefore cannot
comment in detail on the status of instrumentation for making
measurements in wastewater treatment plants. Throughout
this workshop there have been many comments regarding the
nonavailability of sensors for making these measurements.
However, there are some basic questions which should be
asked before we rush into an accelerated program on sensor
development. Among these are:
1. What measurements are really needed in order to exert
effective control over the plant?
2. How frequently should these measurements be made in
order to exert effective control?
3. How accurate do the measurements need to be in order
to exert effective control?
4. How much time delay, between the time the sample is
taken and the time when the results are known, can we
tolerate and still exert effective control?
The answers to these questions can best be obtained by a
dynamic analysis of the plant and its associated control
system. As an example, there is a tendency to search for
continuous, on-line sensors whereas in many instances the
dynamic response of a process is sufficiently slow so that
discrete measurements at relatively long time intervals would
be adequate. The allowable period between samples could be
defined using sampled data control theory if the dynamic
behavior of the process could be quantitatively defined.
Reply to P. E. Sorensen: The author is familiar with
Westberg's work which was a pioneering effort in dynamic
modeling and control of the activated sludge process. As given
in the paper, the author believes the step feed process offers
great potential for process control. However, the author knows
of no plant currently using the potential of step feed in an
automatic control loop. This is one of those areas where pilot
or field studies are needed.
Reply to Wayne C. Smith: The specific oxygen utilization
rate (mass of oxygen utilized per unit mass of sludge per unit
of time) should be a valuable means of detecting batch
discharges of concentrated organic wastes or toxic wastes. The
dissertation by Busby (1) should be referred to for detail on
the use of this parameter as a signal for control purposes.
Calculation of the specific oxygen utilization rate would be
relatively easy for the high purity oxygen process where the
reactors are covered and the flow of oxygen into the reactors
are metered. Coupling of the oxygen balance with a solids
balance would then permit calculation of the specific oxygen
utilization rate which should be a direct measure of sludge
activity.
Reply to James Mueller: The author is in strong agreement
with Dr. Mueller concerning the need for model verification.
The ease and speed with which computer simulations can
frequently be made may lead to a neglect of this very
important aspect of model development and, in the extreme,
can result in one becoming so enamoured with the techniques
that the purpose for using them is almost forgotten. This can
lead to the generation of large quantities of worthless results if
the model is not a reasonable representation of the real
system.
Mathematical modeling, computer simulation and physical
experimentation are not exclusive but rather complement one
another and should therefore be used in an iterative manner. It
is obvious that results of physical experimentation can provide
better numerical values for parameters in computer simula-
tions using mathematical models; however, knowledge gained
in simulation is also useful for modifying the mathematical
model, guiding physical experimentation, and establishing the
type and frequency of field observations needed. Expressing
50
-------
BIOLOGICAL TREATMENT PROCESSES
relationships in mathematical terms, computer simulation and
physical experimentation are all part of the same problem,
model development, and to a large extent can proceed
simultaneously.
The author is also in agreement with the need for model
simplification for field use. A model which is too complex is
subject to either "misuse or disuse." Sensitivity analysis can be
used for this purpose by indicating those variables which have
little effect on the outputs and can therefore be considered as
constants.
A more detailed discussion of some of the above mentioned
points is given in (2).
Reply to P. M. Berthouex: The author is in agreement with
the points raised by Dr. Berthouex. For example, the inputs to
wastewater treatment plants are comprised of both determin-
istic and stochastic components and our most recent work is
considering both.
The development of models, to a certain extent, is
dependent upon the tools with which the investigator is most
familiar. Just as Dr. Berthouex has suggested that we need
more partnerships between model builders and others, I would
also suggest that we need more partnerships between those
concerned with deterministic models and those working with
stochastic models.
REFERENCES:
1. Busby, J. B., "Dynamic Modeling and Control Strategies for the
Activated Sludge Process," Ph.D. Dissertation, Clemson University,
Clemson, SC (1973).
2. Andrews, J. F., "Application of Some Systems Engineering Con-
cepts and Tools to Water Pollution Control Systems," Proceedings
of the Symposium on the Use of Mathematical Models in Water
Pollution Control (A. James, ed.), University of Newcastle upon
Tyne, England (1974).
Michael J. Flanagan:
Reply to Michael du Cros: A combination of these, for
example, sample conditioning equipment for automatic wet
chemistry analyzers requires improvement. Real-time or near-
real-time measurements that are required include viruses,
enzyme activity, nitrogen and bioassays.
Reply to L. A. Schafer: Yesa number of Brown and
Caldwell-designed plants have successfully employed on-line
sensors for the measurement of composition variables such as
pH, ORP, conductivity, DO, etc., for up to 10 years.
Reply to S. P. Graef: All of the control systems shown,
except those that include F/M ratio control, have been
successfully applied. We are currently designing an activated
sludge plant that will incorporate the control concepts shown
in Figure 9. Sludge storage will be accomplished in the
oxidation tanks by using the step-feed mode of operation.
Power-operated slide gates will be employed to modulate the
quantity and point of addition of primary effluent to the
oxidation tanks.
Reply to J. K. Nelson:
1. As we have not tried F/M ratio control using respirometry,
I cannot give a definite answer to this question.
2. In a properly designed clarifier, hydraulic expansion of the
blanket should not occur within normal operating limits.
Temporary expansion of the blanket caused by rotation of
the sludge collector does not affect the blanket level
measurements because the level programmer only operates
when the sludge collector is at right angles to the horizontal
travel axis of the level sensor.
3. In Equation 1, for example, K is an empirically determined
constant which relates F/M, the computed variable, to the
measured variables on the right-hand side of the equation.
The computed value of F/M serves as the process variable
for controller UIC (Figure 5), the output from which is
proportional to the difference between the desired and
computed F/M ratios. I would also point out that other
measured variables may be required in actual practice;
however, the basic measurements required are oxygen
demand and flow.
Reply to R. I. Dick: I agree-the activated sludge process
control systems given in the paper are not valid under limiting
solids flux conditions. The way we would approach the
problem is to generate a velocity-solids level relationship
through a series of mixed liquor settling tests conducted daily
at different solids concentrations (v = a x c'n). A computer
program is used to produce the limiting solids flux and its
associated limiting underflow and mixed-liquor concentrations
as a function of sedimentation tank overflow rate and return
sludge ratio. In actual operation, if limiting solids flux
conditions are approached, the operator will exercise remedial
action such as reducing the total sludge inventory in the
system, placing idle tanks in service, changing operating mode
to reduce solids flux, etc. When normal operating conditions
are restored, the automatic control system is placed back in
operation.
Reply to P. E. Sorensen: Yes-at the Renton Treatment
Plant in Seattle, Washington.
Reply to M. D. McKnight: For a given plant and associated
F/M ratio, storage for M can be estimated from the average
time that F exceeds F average.
51
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
Report of Working Party
on
RESEARCH NEEDS FOR AUTOMATION
OF BIOLOGICAL TREATMENT PROCESSES
Paul H. Woodruff
President, Roy F. Weston Inc., Weston Way, West Chester, PA 19380
A. W. West
Office of Enforcement and General Counsel, National Field Investigations
Center, Environmental Protection Agency, Cincinnati, OH 45268
INTRODUCTION
Biological processes are expected to continue to be the
principal means of removing biodegradable dissolved organic
matter and reducing the volatile solids content of putrescible
solids produced by wastewater treatment facilities. Biological
processes are capable of high dissolved organic removal
efficiencies and generally have a substantial cost-effectiveness
advantage over alternative methods. Engineered systems utiliz-
ing biological processes for wastewater treatment have been
under development for the past 100 years. Much progress has
been made during this period of time in adapting the process
to overcome the various treatment problems which have been
encountered. A considerable amount of research has gone on,
the preponderance of it in the past thirty years. Yet, the full
potential of biological processes to produce the high-quality
effluent demanded by today's standards has not been
achieved. One of the major problems in achieving reliable peak
performance is insufficient understanding of process dynamics
and a lack of adequate instrumentation to improve treatment
plant performance through adoption of automation for pro-
cess controls.
PROBLEMS
The following are major problems relative to automation of
biological treatment processes, as brought out at the work-
shop.
1. Lack of dynamic models of biological processes (activated
sludge, aerated lagoon, trickling filter, stabilization ponds,
rotating filter media, anaerobic sludge process, aerobic
sludge digestion, and anaerobic sludge digestion) which
will adequately identify the essential process variables
and, therefore, the needed process measurements. One
area which typifies this problem is the poor understanding
of the impact of variables in the biological reaction
process on solid-liquid separation.
2. Lack of accepted and field-tested process control strate-
gies.
3. Poor understanding of biological treatment processes.
4. Lack of knowledge regarding available instrumentation
and its application, adaptation and performance in this
area.
5. Lack of incentive for manufacturers to develop new
instrumentation.
6. Insufficient appreciation of the benefits of instrumen-
tation and automation and the consequent inability to
provide a strong incentive based on a strict cost/benefit
analysis.
7. High-quality treatment results are obtained at the expense
of relatively high consumption of energy and natural
resources.
8. Unavailability of necessary sensors in sufficiently reliable
and appropriate form to meet process monitoring and
control needs.
9. Lack of industry-wide standard instrumentation specifica-
tions.
10. Generally low operating and maintenance skills, particu-
larly with respect to understanding and proper utilization
of additional instrumentation and control equipment.
11. Inadequate communication among operators, designers,
equipment suppliers, researchers, and regulatory agency
personnel.
12. Lack of definition of necessary and discretionary monitor-
ing instrumentation and automation which can be sup-
ported by the available human resources at a small
treatment plant.
RESEARCH NEEDS
The following research needs for the automation of
biological treatment processes have been listed in order of
priority:
1. Develop improved means of information exchange:
There is a dire need for making better use of past and
present experience. This can be done only by setting up
substantially improved means of sharing ideas and experi-
ence between those groups (operators, designers, equip-
ment suppliers, researchers and government regulatory
personnel) who are involved in this area of technology.
There is a particular need to develop new mechanisms to
encourage the rapid exchange of experience with existing
instrumentation and automation.
There is a need for development of a standardized plant
52
-------
BIOLOGICAL TREATMENT PROCESSES
performance data-logging report. Standard formats should
be developed for each type of biological process. In order
to have such standardized reports gain uniform acceptance
and use, they would have to be adopted by EPA and their
use required as part of the NPDES permit and monitoring
process.
Feasibility of a private "Underwriter Laboratories" type
of evaluating and testing organization should be con-
sidered for independent evaluation. Such an organization
could be supported by users of the instrumentation that is
tested.
2. Develop a thorough assessment of available instrumenta-
tion:
A thorough assessment should be made of presently
available instrumentation with respect to its applicability
for the purpose intended, reliability, serviceability, and
consistency.
EPA has recently sponsored a study that was intended to
provide such an assessment. Many believe, however, that
the scope of the study was far too limited to provide
conclusive information.
3. Develop adequate mathematical process models:
a. Functioning of biological processes must be under-
stood sufficiently well to relate process variables in a
mathematical model which will accurately simulate
the real-world process. Such a model would contain
many sub-models and would have to consider such
wastewater input characteristics as flow, temperature,
dissolved organic strength, nature of dissolved organ-
ics, suspended solids, dissolved solids, pH, nutrients,
toxicants, etc. Other parameters, such as the amount
of active biomass, detention times, degree of mixing,
gas diffusion rates, organic removal kinetics, solid-
liquid separation rates, suspended solids carryover
from final clarifiers, clarifier underflow, solids con-
centration, sludge recycle rate, etc., would also have
to be considered. The preceding parameters apply to
the activated sludge process; additional parameters
would have to be considered for other types of
biological processes.
b. The advantages of flow control and the means of
accomplishing it also need to be researched.
c. For the present state-of-the-art a particularly vexing
research need is to improve the ability to predict and,
therefore, hopefully control the secondary clarifica-
tion processes. Basic research is required to better
understand the solid-liquid separation phenomenon.
Sludge dewatering operations must be considered an
integral part of the biological process when part of
the liquid portion (e.g., filtrate or centrate) is
returned to the head of the treatment system; thus
sub-models satisfactorily describing the several sludge
dewatering methods must be developed.
d. Mathematical models for nitrification-denitrification
biological processes should also be developed.
e. Research is obviously required to better understand
and manage the interactions of all elements of the
total treatment system, i.e., sources of wastewater,
collection system, pretreatment, primary treatment,
secondary treatment, tertiary treatment, reintroduc-
tion of treated water to the environment and operat-
ing, maintenance and management manpower.
4. Develop practical operating control strategies:
A very important research need is to devise and test, by
mathematical models and plant control tests, the strat-
egies for biological processes. A full-scale demonstration
facility is requked.
5. Develop means to provide economic motivation for
process equipment and instrumentation improvements
and use (invention, development, application):
a. The lack of fundamental knowledge concerning bene-
fits vs. costs of automated biological treatment
processes is a major obstacle to the general applica-
tion of these automated systems. An important
research need is to define the potential improvements
in biological process performance in a quantitative
manner and the potential costs or savings in biological
treatment process construction and operation. This
cost-benefit analysis is of utmost importance in
making the decision to automate.
b. Automation has considerable potential for minimiz-
ing consumption of energy and chemicals while, at
the same time, optimizing treatment performance
when compared to manual methods. This potential
should be addressed as part of the preceding research
areas; however, due to the current national and
international concern for minimizing energy and raw
material consumption (even when used for environ-
mental betterment), this area deserves special atten-
tion. Although not directly related to automation per
se, basic research is needed on methods to increase
the performance yield as a function of energy and
materials consumed. This research needs to consider
the total system, le., human labor, energy, raw
materials, and pollutants created in the manufacture
of energy and materials required for the treatment
facility, as well as use of same for operating purposes.
c. A thorough market research effort is needed to define
the size and time element for various instrumen-
tation/automation components to speed up develop-
ment where market forces seem to justify it.
6. Develop sensors for real-time monitoring and control of
biological processes:
A detailed instrumentation-needs survey should be made,
with input and suggestions solicited from operating
personnel on a national basis. At least the following types
of sensors are needed:
a. Sludge blanket indicator that would give in-situ
53
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
readings and would be readable over at least a six-foot
span.
b. Automatic settleability indicator for process control
use.
c. A respiration-rate sensor integrated with a suspended
solids sensor to give unit oxygen uptake rates.
d. Reliable suspended solids sensors for a range of 0 to
5,000 and 4,000 to 20,000 milligrams per liter.
e. On-line TOC analyzer.
f. An on-line replacement for the BOD test.
g. On-line analyzers for ammonia, nitrate and phos-
phorus.
h. Sensors for achieving methanol dosage control on the
denitrification stage of nitrification-denitrification
treatment.
7. Develop standard specifications, acceptance standards and
application techniques:
Research is needed to develop standards that include
specifications for application, acceptance, and mainte-
nance that are related to users' needs.
8. Develop improvements in the man-instrumentation inter-
face:
Research is needed to clarify the operator's needs for
instrument support and the best means of extending his
abilities through instrumentation. There is a strong need
for improved educational materials for operations and
maintenance personnel relative to the application and care
of instrumentation for process control.
9. Develop improved means of effecting solid-liquid separa-
tion:
This item is actually not one that involves instrumentation
and automation so much as that it deals with an
important component of biological processes. Significant
improvement is needed for effecting improved solid-liquid
separation for both the. forward-flow process and for
solids-processing flow streams. Return streams from solids
processing can drastically affect biological systems. (It
should also be noted that present solids dewatering
systems are inordinately costly.) Many biological pro-
cesses depend on the effective separation of the biomass
from the wastewater. Insufficient attention has been paid
to this portion of biological processes. More research is
needed to better understand the relationship between
waste characteristics, biological reactor performance, and
sludge settling characteristics, which in turn would im-
prove prediction of effluent suspended solids from final
clarifiers.
10. Develop instrumentation and automation particularly
suited to small treatment plants:
Small plants present a particularly difficult problem with
respect to instrumentation. This is due to several things.
For one, instrumentation and automation equipment in a
small plant usually accounts for a larger percentage of the
total construction costs, and secondly, operating and
maintenance personnel are usually substantially less quali-
fied than personnel available at larger plants. However,
this may also suggest that automation under the proper
circumstances could provide a means of improving treat-
ment plant performance and help to compensate for the
shortage of trained operators.
Research is needed to better adapt present instrumenta-
tion and automation technology to small plants. The role
of the operator and degree of manual control may be
more significant in small plants than large. Thus, adaption
of present technology that still requires significant opera-
tor involvement may represent a long-term optimum.
SUMMARY
Priorities have been listed based on the following overall
philosophy. Improved communication is fundamental to op-
timizing present knowledge and intelligently directing all
future activity. Likewise, better knowledge of present automa-
tion capabilities/equipment would speed the return from
existing investment. Future development of technological
improvements would seem to be best directed if basic system
models were available. They should tell us what we need to
measure and suggest operating control strategies. Once we
know what we need and how it is to be used, we need to
determine the economic incentives to insure that all are aware
of economic feasibility and opportunities. Development of
sensors and needed man-instrumentation research would seem
the next logical priority, followed by specific consideration for
the many small treatment plants.
54
-------
AUTOMATION
OF
PHYSICOCHEMICAL PROCESSES
Workshop on Research Needs
Automation of Wastewater Treatment Systems
55
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
FIELD EXPERIENCES
WITH A PILOT PHYSICAL-CHEMICAL TREATMENT PLANT
Walter W. Schuk
Environmental Protection Agency, 5000 Overlook Avenue S.W., Washington, DC 20032
INTRODUCTION
The automation of process monitoring and control has
played a major part in research performed at the EPA-DC Pilot
Plant in Washington, DC. Pilot plant instrumentation includes
conventional analog controls and sensors, on-line wet chemis-
try analyzers, and a sensor-based digital computer (IBM S/7).
A recent pilot plant study compared the effects of manual,
analog, and digital process control strategies on the operating
cost and product quality of a sewage treatment process. The
pilot plant physical-chemical system, with programmed diurnal
flow variation, was selected for this study.
PROCESS DESCRIPTION
The pilot plant physical-chemical process consisted of
screening, two-stage (pH 11.5) lime precipitation with inter-
mediate recarbonation, dual-media filtration, breakpoint
chlorination, and downflow granular carbon adsorption. The
process was designed for a nominal capacity of 50,000 gpd
with an optional diurnal variation of 3:1 maximum to
minimum flow.
Test Procedure
The process was operated for three sequential two-week
periods. The process was controlled manually by pilot plant
operators for the first operating period,automatically controlled
by analog controllers for the second operating period, and
controlled by a sensor-based digital computer for the third
operating period. Grab samples of the influent and effluent of
each stage of the process were collected and composited by
pilot plant operators. Daily laboratory analysis of organic
carbon, phosphorus, and nitrogen concentrations produced
process loading and removal efficiency data for the three
operating periods. Chemical usage was calculated by manually
measuring the volumes of solutions removed from the chemi-
cal feed tanks each day.
Control Description
The control strategies applied can be divided into two basic
groups: those that are flow-proportional with manual updating
such as flocculant aid addition or sludge wasting, and those
that are flow-proportional with on-line sensor-based feedback
updating such as pH control or free chlorine control. In the
presence of diurnal flow variation, manual or sensor-based
feedback control of the process could not be accomplished
without automatic flow proportioning.
With the exception of breakpoint chlorination, the feed-
back control signals were developed by analog proportional-
integral controllers or the digital equivalent of proportional-
integral control produced by the sensor-based digital com-
puter. For control of breakpoint chlorination, feedforward
NH3 proportioning was added to the analog control algorithm,
and a steady-state control equation was used for digital
operation.
RESULTS
The product quality produced by the three control
methods indicated only a marginal difference in favor of
automatic controls for the operating periods studied.
Due to an extreme change in the influent water character-
istics caused by extended heavy rainfall during one portion of
the study, direct comparison of chemical usage for the three
control modes was impossible. As an alternate to measuring
the actual volume of chemicals dispensed, the deviation of the
controlled variable from the set point was analyzed to estimate
the probable cost effect. With the exception of breakpoint
chlorination, both automatic control strategies used approxi-
mately 10% less chemicals than the manual strategy to
maintain the system.
At the time this test was performed, the digital control
equation for breakpoint chlorination was designed for steady-
state process flow. Linear flow-proportioning was added to the
56
-------
PHYSICOCHEMICAL PROCESSES
equation to compensate for diurnal flow during the test. Later
testing of the breakpoint process showed a non-linear response
to flow changes (probably caused by changes in mixing
energy). As a result, the digital control mode provided the
poorest operation of the three control modes tested. While
flow changes also caused upsets during manual and analog
control, recovery was much faster than with digital control.
The manpower required to operate and maintain the
physical-chemical process, excluding solids handling, was
significantly greater for manual control. For either analog or
digital control, 24-hour operation of the pilot plant process
required approximately 12 hours per day of operator time,
and 8 hours per day of maintenance time. For manual
operation, operator time increased to 36 hours per day, and
maintenance time decreased to 4 hours per day. The net
difference of 20 hours per day was a 100% increase in
manpower to produce an equal product quality.
CONCLUSIONS
The development of automation in physical-chemical pro-
cess control has both improved process operation and in-
creased process maintenance problems. The best control
algorithm cannot improve a process when supplied with
inaccurate or unreliable data, nor can optimum results be
expected when final control elements malfunction or change
characteristics. Present knowledge of breakpoint chlorination
is inadequate to develop an optimum control equation for this
process; however, for other processes, proven, viable control
strategies were not implemented because of inconsistent
on-line sensor data. For example, the lime feed control
strategy could not be used until the original pH monitoring
system and the lime feed system were replaced with equip-
ment of different design. Prior to these changes, the excessive
maintenance time required, and process upsets caused by
equipment failures, precluded^ automatic pH control of the
lime clarification process.
Equipment failure not only impacted process operation but
also complicated process research efforts. To date, most
process evaluation has depended on manual data analyses and
long-term operation to overcome the lack of reliable on-line
information. It seems futile to expend research effort on the
cost-effectiveness of automation, or the impact of automation
on product quality and reliability, when the data produced by
on-line sensors cannot be accepted as reliable.
It is also frustrating to put forth the maintenance effort
necessary to produce reliable automatic process control and
then evaluate control effectiveness by manual analysis of grab
or composite samples, made hours or even days after the
samples were collected.
To simply ask manufacturers for reliable instrumentation is
equally futile. At present, consistent specifications for sewage
treatment instrumentation are non-existent. Typically, they
include some blend of the electrical code, the plumbing code,
and the whim of the originating engineer.
RECOMMENDATIONS
The process just described required accurate, reliable
measurement of four parameters for optimum process control:
flow, pH, free chlorine and ammonia, and the ability to
accurately dispense four chemicals: lime, carbon dioxide,
chlorine and caustic soda. To expand this process to full-scale
operation, improvements should be made in most of the
equipment just named, and the breakpoint chlorination
control strategy must be improved.
Although recent improvements in flow metering may have
improved the reliability of flow measurement, as yet there is
no documented method of checking large flow meters other
than by shipping them to a test facility at considerable
expense and lost time. A self-cleaning pH assembly is needed
to reduce both maintenance time and process upsets caused by
fouled electrodes. Both the free chlorine and ammonia
analyzers are time-consuming, high-maintenance items. The
throttling of chemical feeds by control valves and metering
pumps must be linear and repeatable over a broad range
(approximately 10:1) if flow-proportioning control is to be
used. Few valves or metering pumps meet these standards.
With the increasing incidence of power shortages, reduced line
voltages are occurring more often. This results in equipment
being operated under borderline conditions more often and for
longer periods of time. To maintain process control accuracy
and reliability, instrument power supplies must be improved.
As some areas must meet effluent quality standards for
carbonaceous, phosphorus, and nitrogen contaminants, reliable
on-line sensors must be developed that can measure these
parameters in raw wastewaters and plant effluents.
One major problem exists that has not yet been mentioned.
In order for automation to become an effective tool for
wastewater treatment process control, it is mandatory that a
program be implemented to recruit and train personnel to
operate and maintain automated waste treatment systems.
Without well-trained personnel, reliability in data evaluation
and achievement of rigid process control will be impossible.
57
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
AUTOMATION OF PHYSICAL AND CHEMICAL PROCESSES
Thomas M. Keinath
Environmental Systems Engineering, Ctemson University, Clemson, SC 29631
INTRODUCTION
Physical and chemical wastewater treatment processes have
found application both in biological and physicochemical
wastewater treatment systems. Although several of the P/C
processes are primarily employed only in physicochemical
process trains (chemical clarification and adsorption on acti-
vated carbon), most are commonly found in both types of
systems; e.g., gravitational sedimentation and chlorination.
Accordingly, this treatise has not been limited solely to those
processes commonly found in physicochemical treatment
trains. All major physical and chemical wastewater treatment
processes will be considered, with the exception of chlorina-
tion, which has been considered in detail in a companion paper
by W. W. Schuk.
Minor physical treatment operations including flow routing
and equalization, comminution, screening and degritting have
also been omitted from this discussion. Automation or control
of several purely chemical operations such as pH adjustment
and recarbonation have purposefully not been considered
herein because they have been routinely automated through-
out the water and wastewater treatment industry.
It is to be noted, furthermore, that this treatise is not
meant to be an exhaustive review of the pertinent literature.
Rather, literature citations have been made only to illustrate
the general scope of research that is currently being con-
ducted.
CLARIFICATION AND THICKENING OF BIOLOGICAL
SLURRIES
Present Practice and Current Research
Feed-forward control of the activated sludge process can be
achieved only when a descriptive dynamic mathematical model
for the process is available. For this system it is mandatory
that the dynamic model for the aeration basin be coupled with
a complementary dynamic model for the secondary settler.
Development of time-dependent models for the aeration
chamber has occurred primarily during the past six years.
During this period these models have evolved to a relatively
structured status. Nonetheless, they lack verification at both
the pilot- and prototype-scale levels.
Dynamic models for the secondary clarifier/thickener have
not yet attained this level of development. Further, none of
the existing models have been used as part of the control
strategies that currently are employed in full-scale systems.
Control algorithms that recently have been used in practice
generally focus on one of four control objectives: (1) ratio
proportioning of the sludge recycle flow to the influent flow
rate; (2) MLSS control; (3) F/M or PLI control; or (4) SRT
control. The control action normally taken is to adjust the rate
of recycle flow in response to some measured variable such as
MLSS, TOC, or sludge blanket level in the secondary clarifier.
Without reference to a dynamic model of the secondary
clarifier, one cannot ensure that a specific control action will
result in the desired response. This can be accomplished only
when a dynamic model of the secondary clarifier/thickener is
used to: (1) predict concentration of the biological solids in
the underflow; (2) predict solids blanket height; and (3)
predict the solids concentration profile in the settler in
response to dynamic changes in the influent and underflow
rates of flow and the influent solids concentration.
Any dynamic model of the secondary clarifier also must be
able to predict the concentration of suspended solids in the
overflow in addition to the three factors enumerated above.
The first attempt to develop a dynamic model for continuous
thickening was that of Bryant (1). Through application of the
Kynch analysis of zone settling and considering that solids are
transported to the bottom of a settler by bulk flow and
gravitational sedimentation, the following expression was
derived:
at
ac
where,
z =
C =
u =
Gs =
t =
vertical distance in settler
concentration of solids
bulk (downward) flow velocity
gravitational solids flux
time
Solution of the partial differential equation was effected
through spatial lumping into ten ordinary differential equa-
tions. Bryant also incorporated an empirical relationship (2) to
simulate the response of the clarification function of the
secondary settler.
Alkema (3) provided a slightly different approach to
dynamic modeling of the continuous thickening process. The
model which he developed provided for the movement of
concentration discontinuities between the predominant layers
in the sludge blanket.
Tracy (4) and Tracy and Keinath (5) were the first to
implement the lumped-parameter model for the entire clarifier
both above and below the feed point. Furthermore, they were
58
-------
PHYSICOCHEMICAL PROCESSES
the first to provide for laboratory verification of the dynamic
model. They found that the model accurately predicted the
response of the thickener to transient inputs for the under-
loaded case. Moreover, their model appeared to give an
adequate representation of the dynamic response of the
clarifier when it was overloaded, or when it was forced from
an overloaded to an underloaded condition or vice versa.
Research Needs
(1) The existing dynamic mathematical model for the
secondary settler must be modified to account for the
movement of displaced water upward through the
settler.
(2) A structured clarification model must be coupled
with the continuous thickener model to enable
prediction of effluent suspended solids levels. This
must, of course, account for scour of biological solids
from the sludge blanket into the effluent.
(3) The existing models have been developed in terms of
one spatial variable (depth). These must be modified
for geometrical effects; e.g. radial profiles in circular
settlers and longitudinal profiles in rectangular set-
tlers.
(4) Laboratory, pilot- and full-scale verification must be
provided for the models.
SETTLING CHARACTERISTICS OF BIOLOGICAL
SLUDGES
Background and Current Research
Solution of any dynamic model of the secondary clari-
fier/thickener can be achieved only when an appropriate
expression for the relation dGg/dC is available. Bryant (1),
Alkema (3), and Tracy (4) obtained expressions for this term
by experimentally determining a relation for the initial inter-
facial settling velocity as a function of the concentration of
suspended solids. From this relation an expression for Gs vs. C
was derived as detailed by Dick (6).
One recognizes, however, that in an activated sludge system
the settling characteristics and, therefore, the term 3Gs/9C
change continually. Consequently, if a dynamic model of the
activated sludge process is to be fully useful from a control
viewpoint, it is imperative that the expression for 9GS/9C be
updated continuously either by (1) off-line measurement of
the settling properties of the biological slurry or (2) prediction
of the settling properties by reference to various biological
processes parameters. The latter approach is of course prefer-
able as one can then formulate overall feed-forward control
strategies that include the dynamics of sludge settling charac-
teristics. Conversely, the former approach provides a means for
feedback control.
A digital solids-liquid interface settling monitor has been
developed by George (7) for the off-line measurement of the
settling properties of biological slurries. The device, however,
has not been subjected to field trials. Other approaches to
on-line measurement of settling characteristics have yet to
proceed beyond the conceptual stage.
Considerable research has heretofore been directed toward
determining the biological factors that affect sludge settling
relationships, both clarification and thicking. These have
focused on delineating the effects of organic loading and
oxygen tension on sludge settleability and on effluent clarity.
Lesperance (8), Logan and Budd (9), Stewart (10), Bisogni
and Lawrence (11) and Chao (12) showed that for conven-
tional air activated sludge systems several optimum ranges of
organic loading intensities (F/M or PLI) existed with respect to
sludge settleability. At very high and intermediate PLI's sludge
settleability was observed to deteriorate due to filamentous
and zoogleal bulking, respectively. Similar studies on high-
purity oxygen systems conducted by Jewel et al. (13), Okum
(14), Chao (12), and Albertsson et al. (15) showed that the
sludge settling characteristics were materially better than for
conventional air systems. Chao's studies showed, furthermore,
that the sludge settleability for the high-purity oxygen system
was relatively constant over the entire PLI spectrum and was
not subject to either zoolgleal or filamentous bulking.
The effects of biological factors on clarification efficiency
were also investigated by Chao (12). His studies showed that
the suspended solids concentration in the overflow of the
secondary clarifier increased with increasing organic loading
intensities. In contrast, his studies on high-purity oxygen
activated sludge systems showed that the clarification effi-
ciencies were materially poorer than for conventional air
systems.
It must clearly be recognized that virtually all studies
conducted regarding the effects of biological factors on sludge
settleability and effluent clarity were conducted under
pseudo-steady state conditions. Consequently, these studies
can only be employed to indicate trends. Essentially no studies
have been conducted relative to the dynamics of settleability
and the clarification of biological suspensions vis-a-vis various
process parameters. One notable exception is the work of
Chudoba, et al (16-18) who studied the effect of microbial
population dynamics on sludge settleability in systems that
were operated at steady state with respect to input conditions,
but were operated under different mixing conditions. While
their results were inconclusive, they were able to demonstrate
certain interesting trends in sludge settleability as a function of
population dynamics. Moreover, they qualitatively determined
that the propagation of filaments proceeded much more
rapidly than their suppression.
Research Needs
(1) Various off-line and on-line approaches for determin-
ing by experimental measurement the expression for
9GS/9C must be developed and evaluated.
(2) A quantitative dynamic relationship must be estab-
lished between the clarification and settling character-
istics of activated sludge and the process parameters.
59
-------
AUTOMATION OF WASTE WATER TREATMENT SYSTEMS
PRIMARY CLARIFICATION
Current Research on Clarifier Dynamics
Bryant et al. (19), in their transient simulation of a
wastewater treatment plant, employed a "hybrid" clarifier
model. The model employed a correlation of the fraction of
suspended solids removed in a full-scale clarifier versus
overflow rate to yield the solids separation (essentially a
steady-state model) while assuming a number of completely-
mixed tanks in series to provide the time response. Provision
was made for sludge hold-up in the clarifier, although the
solids concentration in the sludge and the withdrawal rate
were not taken as functions of the hold-up. More recent
transient simulations of wastewater treatment plants by Naito
(20) and Johnson and Yang (21) employed steady-state
models for clarifiers, neglecting, without justification, their
contribution to the transient behavior. One might conclude,
therefore, that primary clarifier models that are employed as
part of overall dynamic treatment plant models are primitive
at best. The reason that physical phenomena associated with
clarifier dynamics are not considered is probably because they
are not understood.
Steady-state operation of clarifiers has been studied exten-
sively. For the sizing of clarifiers a plug-flow model proposed
by Camp (22) is often employed to calculate the vessel surface
area. Because the plug-flow model is not realistic, investiga-
tions in recent years have ranged from measurements of
mixing [e.g. Wills and Davis (23), Murphy (24)], to research
into the causes of mixing [e.g. Fitch and Lutz (25), Takamatsu
and Naito (26)], to the development of performance models
allowing for mixing.
One of the earliest models proposed for primary clarifiers is
that of Hazen (27). Much of the recent research that has been
directed toward dynamic model development has been con-
ducted by Silveston and co-workers (28-32). Their approach
has been to apply stochastic methods and power spectrum
analysis to primary clarifier dynamics. Smith (33) also has
provided a stochastic model which is based on flow rate and
the clarifier surface area.
Current Research on Chemical Clarification
Convery et al. (34) have described extensive studies on
automation of the lime clarification process at EPA's Blue
Plains Pilot Plant. The studies focused on four potential
control schemes: (1) conductivity-ratio, (2) flow-proportional,
(3) pH feedback trim plus flow-proportional, and (4) alkalinity
feedback trim plus flow-proportional. Although the latter
control scheme provided the best results, it was judged
unsuitable for general application because of difficulties
encountered in maintaining the alkalinity sensing system. The
pH feedback trim plus flow-proportional system was recom-
mended for general use. In similar pilot-plant studies con-
ducted at the Cleveland-Westerly Wastewater Treatment Facil-
ity (35), simple pH feedback control was employed.
Each of the control schemes detailed above provide for
control only of the chemical dosing function. No control is
provided specifically for an effluent quality parameter, e.g.,
suspended solids or soluble orthophosphate concentration. To
accomplish this, one would first need to mathematically
characterize the chemical precipitation and chemical coagula-
tion/flocculation processes and then would need to couple
these with a suitable dynamic model for the primary clarifier.
The same approach would be required in the case of
chemical clarification using either aluminum sulfate or ferric
chloride. For these cases, however, it would be much more
difficult to properly describe the chemistry and chemical
interactions of the system. Perhaps the most feasible control
algorithm would be one of the following simplified strategies:
(1) flow-proportional with pH feedback trimmingwhere
the chemical feed and pH set points would be routinely
updated using off-line jar coagulation studies.
(2) mass proportional to suspended solids with pH
feedback trimming.
(3) mass proportional to soluble orthophosphate with pH
feedback trimming.
Research Needs
(1) Development of a comprehensive dynamic mathe-
matical model of primary clarification.
(2) Development of a simplified but realistic dynamic
model of the chemical precipitation and chemical
coagulation/flocculation processes (pH-dosage do-
main).
FILTRATION
Present Practice
Automation of the deep-bed filtration process has not
advanced materially beyond what has been common practice
in the water treatment industry for several decades. Control
schemes that are currently employed in practice generally
relate to two functions: (1) backwash initiation; and (2)
influent flow splitting.
Four different backwash initiation control modes were
described by Convery, et al., (34). These modes, which were
evaluated at the Blue Plains Pilot Facility, included headless,
high-level (influent), programmed time interval, and manual.
The headless control mode consisted of a headless sensor
which initiated the backwash cycle when the headless across
the filter bed exceeded a preset value (e.g., 9 feet of water).
Control using the high-level scheme was accomplished through
the use of a high-level controller which served to open the
effluent control valve so as to maintain a constant level of
water above the filter. When the effluent control valve had
been opened entirely, the backwash cycle was initiated. For
these two modes, time-delay circuits prevented the premature
triggering of the backwash cycle due to accidental or momen-
tary events.
60
-------
PHYSICOCHEMICAL PROCESSES
The programmed time-interval controller simply initiated
the backwash sequence after the expiration of a preset number
of operating hours. An operator manually changed the
operating time set-point depending either on the flow rate
through the plant or the rate of headloss build-up. Convery, et
al. (34) concluded that the programmed time-interval control
mode with a back-up headloss indicator alarm (which pre-
vented flooding when system upsets caused increased solids
loading) and a high-level alarm (which indicated equipment
failure) provided for peak operating efficiency at the lowest
possible operating cost.
At the Cleveland Westerly Advanced Wastewater Treatment
Facility (35) individual filter backwash sequencing will be
accomplished by continuously sensing and scanning filter
headloss and filter effluent turbidity for each of the filters in
operation. When either of these values exceeds a set-point, the
backwash sequence for the critical filters) will be sequentially
initiated according to the incoming error signals. Convery, et
al. (34) emphasized, however, that turbidity could not
properly be used as a control criterion because changes in
clarifier efficiency resulted in marked changes in filter effluent
turbidity.
Apportionment of influent flow between the operating
filters was accomplished at the Blue Plains facility (34) by a
mechanical splitter box which equally distributed the flow to
each of the filters. At the Cleveland Westerly facility the
influent will be apportioned among the operating filters so as
to minimize the overall headloss.
Current Research
Much of the current research relative to the automation of
deep-bed filtration has been directed toward the development
of descriptive dynamic models for the process. The majority of
the developments in this research area have been contributed
by Ives and co-workers (36^4) although Deb (45) has also
been active in the development of a dynamic model of the
process. Qualitatively, these models provide for description of
headloss and breakthrough turbidity as a function of time.
based on measurements of a variety of parameters that
characterize the filter media and influent turbidity.
None of the dynamic models have yet been fully verified
for suspensions of particulates found in wastewaters, although
Payatakes, et al (48) and Mehter, etal. (49) have conducted
laboratory experiments toward this end. When this is accom-
plished, however, the dynamic models could be employed in a
feedforward manner to determine optimal control strategies
which will maximize filter run times, thereby minimizing
backwash water requirements and recycle flows. The filtration
model, of course, should be coupled with the primary
(chemical) clarification model because of the strong inter-
actions between them. A preliminary effort in this regard was
conducted by Kriegsman (46) who coupled the models of
Bryant, etal. (19) for the primary clarifier and Diaper and Ives
(44) for the deep-bed filter to simulate the performance of,
and the interactions between, these processes.
Research Needs
(1) The existing dynamic mathematical model for the
deep-bed filtration process must be adapted for the
filtration of suspensions of wastewater particulates.
(2) The deep-bed filtration model must be coupled with
the primary clarification model.
(3) Laboratory-, pilot-, and full-scale verification must be
provided for the coupled dynamic models.
(4) Control strategies using the coupled clarification-
filtration models must be developed.
(5) Backwash supervisory control strategies must be
developed to prevent cascading and minimize effluent
deterioration effects during periods of high flow and
to minimize backwash water storage requirements for
periods of low flow.
ADSORPTION
Present Practice and Current Research
In practice (34, 35) the adsorption process has been
controlled in essentially the same fashion as has filtration.
That is, control has been exerted only with respect to
distribution of flow to the adsorbers and initiation of the
backwash sequence. At the Cleveland Westerly plant (35),
furthermore, an activated carbon column that is exhausted, as
determined by TOC measurements on the influent to and
effluent from the column, will be automatically removed from
operation for spent carbon discharge and refill with regener-
ated and virgin make-up activated carbon. Removal and refill
operations will be automatically sequenced, monitored, and
controlled.
Research conducted on automation of the adsorption
process has been directed primarily toward dynamic mathema-
tical modeling. Keinath and Weber (47) were the first to
provide a simplified dynamic model of the process when used
for the removal of organic contaminants from wastewaters.
Under contract to the Environmental Protection Agency, Tien
and associates (48-51) developed a structured dynamic model
which attempted to account for the adsorption of soluble
organics, the filtration of particulates, and the biodegradation
of both soluble and particulate organics in a packed-bed
activated carbon adsorber. Due to the complex interactions
between the adsorption, filtration, and biodegradation func-
tions, this dynamic model only very qualitatively represents
columnar adsorption dynamics. For municipal wastewaters, it
is questionable whether a fully quantitative dynamic model
can realistically be developed or whether such a model, if
developed, would be particularly useful from a control
viewpoint. Because of the large damping capacity in columnar
adsorbers, simple feedback control strategies may well be
entirely satisfactory.
61
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
In the case of industrial wastewaters where chromato- municipal wastewater treatment systems should be
graphic or reversible displacement effects are observed, formulated. For example, the treatment objective
however, a dynamic model is mandatory such that feed- might be to minimize carbon loading rates while
forward control can be effected to prevent solute displace- meeting effluent standards. This could be achieved by
ment. A preliminary dynamic model developed by Carnahan by-passing a portion of the flow around the carbon
(52) semi-quantitatively simulates the chromatographic effects contactors and blending it with the effluent from the
that have been observed in the field. Carnahan assumed that contactors to produce the desired quality. These
the filtration and biodegradation effects were negligible. For studies should be conducted on a full-scale level to
many industrial wastewaters this assumption is realistic. establish the operational cost benefits that might be
experienced through application of this control stra-
Research Needs tegy.
(1) The dynamic model which incorporates the adsorp-
tion, filtration, and biodegradation functions should OVERALL CONTROL STRATEGIES
be verified at either the pilot- or full-scale level. Another significant research need is the coupling of
(2) The dynamic model which accounts for chromatogra- dynamic models for individual processes into an overall
phic effects should be verified at either the pilot- or dynamic model for an entire physicochemical wastewater
full-scale levels and appropriately refined. treatment facility. Such a model could subsequently be used
(3) Control strategies using both of the dynamic mathe- to explore control strategies for the entire plant. As indicated
matical models should be formulated. above, Kriegsman (46) coupled the primary clarification and
(4) Simple feedback control strategies for adsorbers in filtration processes as a preliminary effort in this direction.
62
-------
PHYSICOCHEMICAL PROCESSES
REFERENCES
1. Bryant, J. O., Jr., "Continuous Time Simulation of the Conven-
tional Activated Sludge Wastewater Renovation System," Ph.D.
Dissertation, Clemson University, Clemson, SC (1972).
2. Pflanz, P., "Performance of (Activated Sludge) Secondary Sedi-
mentation Basins," Advances in Water Pollution Research
(Jenkins, S. H., ed.), Pergamon Press, New York (1969).
3. Alkema, K. L., "The Effect of Settler Dynamics on the Activated
Sludge Process," M. S. Thesis, University of Colorado, Boulder,
CO (1971).
4. Tracy, K. D., "Mathematical Modeling of Unsteady-State Thick-
ening of Compressible Slurries," Ph.D. Dissertation, Clemson
University, Clemson, SC (1973).
5. Tracy, K. D. and Keinath, T. M., "Dynamic Model for Clarifica-
tion and Thickening of Activated Sludge," Amer. Inst. of Chem.
Engrs. Symposium Series, Water 1973 (1973).
6. Dick, R. I., "Role of Activated Sludge Final Settling Tanks," Jour.
San. Engr. Div., Proc. Amer. Soc. Civil Engrs., 96, 423 (1970).
7. George, D. B., "Ditital S/L Interface Settling Monitor," Special
Development Report, Clemson University (1973).
8. Lesperance, T. W., "A Generalized Approach to Activated Sludge:
Part III. Process Variants, Why They Occur," Water Works and
Wastes Eng., 2, No. 7, 34 (1965).
9. Logan, R. P. and Budd, W. E., "Effect of BOD Loadings on Acti-
vated Sludge Plant Operation," Biological Treatment of Sewage
and Industrial Wastes. Vol I Aerobic Oxidation, (McCabe, J. and
Eckenfelder, W. W., eds.), Reinhold Publ. Co., New York (1956).
10. Stewart, M. J., "Activated Sludge Process Variations, The Com-
plete Spectrum: Part III. Effluent Quality-Process Loading
Relationships," Water & Sew. Works, HI, R-260 (1964).
11. Bisogni, J. J., Jr. and Lawrence, A. W., "Relationship Between
SRT and Settling Characteristics of Activated Sludge," Water Res.,
5, 753(1971).
12. Chao, A. C., "Destabilization of Activated Sludge Suspensions
with Ferric Chloride: Clarification and Thickening," Ph.D. Dis-
sertation, Clemson University, Clemson, SC (1973).
13. Jewell, W. J., Eckenfelder, W. W. and CavaHer, M. E., "The Use of
Pure Oxygen for the Treatment of Brewery Wastes," Proc. 26th
Ind. Wastes Cow/., Purdue University (1971).
14. Okun, D. A., "System of Bio-Precipitation of Organic Matter from
Sewage," Sew. Works Jour., 21, 63 (1949).
15. Albertsson, J. G., McWhirter, J. R., Robinson, E. K. and Vahl-
dieck, N. P., "Investigation of the Use of High Purity Oxygen
Aeration in the Conventional Activated Sludge Process," FWQA,
Water Poll. Control Res. Ser. 17050 DNW (1970).
16. Chudoba, J., Ottova', V. and Madera, V., "Control of Activated
Sludge Filamentous Bulking-I. Effect of Hydraulic Regime or
Degree of Mixing in an Aeration Tank," Water Res., 7, 1163
(1973).
17. Chudoba, J., Grau, P. and Ottova, V., "Control of Activated
Sludge Filamentous Bulking-II. Selection of Microorganisms by
Means of a Selector," Water Res., 7, 1389(1973).
18. Chudoba, J., Blaha, J. and Madera, V., "Control of Activated
Sludge Filamentous Bulking-III. Effect of Sludge Loading,"
Water Res., 8, 231 (1974).
19. Bryant, J. O., Jr., Wilcox, L. C. and Andrews, J. F., "Continuous
Time Simulation of Waste Water Treatment Plants," Presented at
the AIChE National Meeting, Cincinnati (1971).
20. Naito, M., "Dynamic Model for Wastewater Treatment Plant,"
Presented at Fourth Joint AIChE/CSChE Meeting, Vancouver
(1973).
21. Johnson, A. I. and Yang, J. T., "Computer Simulation and Control
Studies for a Sewage Plant," Presented at Fourth Joint
AIChE/CSChE Meeting, Vancouver (1973).
22. Camp, T. R., "Sedimentation and the Design of Settling Tanks,"
Trans. Amer. Soc. Civil Engrs., Ill, 895 (1946).
23. Wills, R. F. and Davis, C., "Flow Patterns in a Rectangular Sewage
Sedimentation Tank," Advances in Water Pollution Research, 2,
335 (1962).
24. Murphy, K. E., "Tracer Studies in Circular Sedimentation Basins,
Proc. 18th Ind. Wastes Con/.. Purdue University (1963).
25. Fitch, E. B. and Lutz, W. P., "Feedwells for Density Stabiliza-
tion," Jour. Water Poll. Control Fed., 32, 147(1960).
26. Takamatsu, T. and Naito, M., "Effects of Flow Condition on the
Efficiency of a Sedimentation Vessel," Water Res., 1, 433 (1967).
27. Hazen, A., "On Sedimentation," Trans. Amer. Soc. Civil Engrs.,
53, 63 (1904).
28. Silveston, P. L., "Computer Simulation of Waste Treatment
Plants," Presented at the ACS/CIC Symposium on Pollution
Problems of Our Environment, Toronto (1970).
29. Silveston, P. L., "Simulation at the Mean Performance of Munici-
pal Waste Treatment Plants," Water Res., 6, 1101 (1972).
30. Singh, D. P., Bryson, A. W., Jr. and Silveston, P. L., "A Stochastic
Model for Primary Settlers," Proc. 5th Annual Symp. on Water
Poll. Res., Waterloo, Ontario, Canada (1970).
31. Sakata, N. and Silveston, P. L., "Exponential Approximation for
Settling Rate," Water Res. (In Press).
32. Sakata, N., Cordoba-Molina, J. F. and Silveston, P. L., "Studies on
the Application of Power Spectrum Analysis to Primary Clarifier
Dynamics," Presented at the 9th Canadian Symposium on Water
Pollution Research (1974).
33. Smith, R., "Preliminary Design and Simulation of Conventional
Wastewater Renovation Systems Using the Digital Computer,"
Water Poll. Control Res. Ser. Publ. WP-20-9 FWPCA, Cincinnati,
OH (1968).
34. Convery, J. J., Roesler, J. R. and Wise, R. H., "Automation and
Control of Physical-Chemical Treatment for Municipal Waste-
water," in Applications of New Concepts of Physical-Chemical
Wastewater Treatment, (Eckenfelder, W. W. and Cecil, L. K., eds.),
Pergamon Press, New York (1972).
35. Zurn Environmental Engineers, "Westerly Advanced Wastewater
Treatment Facility: Process Development and Preliminary En-
gineering Design," Zurn Environmental Engineers and Battelle
Memorial Institute (1972).
36. Ison, C. R. and Ives, K. J., "Removal Mechanisms in Deep-Bed
Filtration," Chem. Engr. Sci., 24, 717 (1969).
37. Ives, K. J., "Rational Design of Filters," Proc. Inst. Civil Engrs.,
16, 189 (1960).
38. Ives, K. J., "Simulation of Filtration on an Electronic Digital
Computer," Jour. Amer. Water Works Assn., 52, 933 (1960).
39. Ives, K. J., "Simplified Rational Analysis of Filter Behavior,"
Proc. Inst. Civil Engrs., 25, 345 (1963).
40. Ives, K. J. and Sholji, I., "Research on Variables Affecting
Filtration," Jour. San. Engr. Div., Proc. Amer. Soc. Civil Engrs.,
91, 1 (1965).
41. Ives, K. J., "The Use of Models in Filter Design," Effl. & Water
Trt. Jour., 6, 522 (1966).
42. Ives, K. J., "Advances in Deep Bed Filtration," Symposium on
Advances in Filtration, Inst. of Chem. Engrs., Manchester (1968).
43. Ives, K. J., "Theory of Filtration," Special Subject No. 7,
International Water Supply Congress, Vienna (1969).
44. Diaper, W. J. and Ives, K. J., "Filtration Through Size-Graded
Media," Jour. San. Engr. Div., Proc. Amer. Soc. Civil Engrs., 91,
89 (1965).
45. Deb, A. K., "Theory of Sand Filtration," Jour. San. Engr. Div.,
Proc. Amer. Soc. Civil Engrs., 95, 399 (1969).
46. Kriegsman, G. R., "Simulation of the Dynamics of Clarification
and Filtration," M. S. Special Problem Report, Clemson Univer-
sity, Clemson, SC (1973).
47. Keinath, T. M. and Weber, W. J., Jr., "A Predictive Model for the
Design of Fluid-Bed Adsorbers," Jour. Water Poll. Control Fed.,
40, 741 (1968).
48. Payatakes, A. C., Turian, R. M. and Tien, C., "Integration of
Filtration Equations and Parameter Optimization Techniques,"
Research Report 70-4, Syracuse University, Syracuse, NY (1970).
49. Mehter, A. A., Turian, R. M. and Tien, C., "Filtration in Deep
Beds of Granular Activated Carbon," Research Report 70-3,
Syracuse University, Syracuse, NY (1970).
50. Hsieh, J. S., Turian, R. M. and Tien, C., "Experimental Investiga-
tion of the Adsorption of Organic Contaminants in Waste Water
on Granular Activated Carbon," Research Report 69-1, Syracuse
University, Syracuse, NY (1969).
51. Hsieh, J. S., Turian, R. M. and Tien, C., "Batch Adsorption
Kinetics with Chemical Reaction," Research Report 69-2, Syra-
cuse University, Syracuse, NY (1969).
52. Carnahan, R. P., "Mathematical Modeling of Heterogeneous Sorp-
tion in Continuous Contactors for Wastewater Decontamination,"
Ph.D. Dissertation, Clemson University, Clemson, SC (1973).
63
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
DISCUSSION
Robert A. Ryder:
Would Mr. Schuk please comment on the type of ammonia
measurement and sensing system utilized, and its reliability?
In connection with breakpoint chlorination, was there any
attempt to control overdosage of chlorine, to minimize
wastage as well as the effect of excess residual on the sorption
capacity of the granular activated carbon filters?
Robert H. Wise:
We in the municipal wastewater treatment field tend to
believe that most of the many novel sensors and instruments
needed to implement WWTP automation will almost auto-
matically be forthcoming from instrumentation suppliers once
a need for such devices has been conclusively shown. I have
discussed this "philosophy" with representatives of many
instrumentation manufacturers, all of whom assured me that
this prevailing concept is false. Mr. Russell Babcock has
already commented on this topic.
Once the need for a particular sensor or instrument has
been conclusively shown, development of a reliable "device"
to fill this need in the hostile environment of a sewage
treatment plant requires years of intensive research and
development. Therefore, if we postpone development of such
devices (using the rationale that such equipment might not be
really needed at some future date), we thereby throw away
lead time corresponding to the developmental time required to
reduce sensor concept (or instrument concept) to practice.
Typically, such lead time is 4 to 8 years*. Can we afford this?
'Example Lead Times for Development of Existing Devices
1) On-line TOC analyzer: about 8 years
2) On-line polarographic DO probe: about 10 years
3) On-line glass electrode: about 10-15 years
4) On-line NH3 probe: after 3 years, still not acceptable for on-line
use.
L. A. Schafer:
In the automation of continuous processes, the development
of new instruments usually occurs in three interacting phases,
performed by three separate and distinct groups. The first
phase (innovation) carries the basic idea through bench-scale
testing. The second phase (development) transforms the
concept into marketable hardware. The third phase (applica-
tion) adapts the instrument to its proper function by
modification and the addition of sub-systems. A prime
example is the development of the gas chromatograph, where
the three phases were worked out by chemists and other
researchers, instrument manufacturers, and industrial users,
respectively. No one group was capable of producing a
working instrument; all three performed their function and
interacted to develop an effective machine.
In the municipal wastewater field, the first two groups are
generally available, but there is no one with the ability,
motivation, and funding to perform the third function. As has
been pointed out, instrument manufacturers have little incen-
tive to put their funds into application. In the treatment
facilities themselves, there are no funds for such work, the
operating staff has no motivation, and the one possible
innovation group, the analysts or technicians, are trained
toward invention and laboratory work and are therefore
conditioned toward the innovative phase, rather than the
necessary and different application effort.
Application, like innovation and development, is almost
never achieved by any one single group, no matter what the
capability or motivation. Technology transfer is effective only
when it involves the user, or the person who has to live with
the results. Instrument application by salesmen, design en-
gineers and consultants has, all too often, led to dismal and
expensive failures, with the taxpayer taking the loss.
It would seem, therefore, that to develop instruments such
as nutrient meters (TOC, BOD, respirometers, and the like) or
suspended solids meters, a group or groups must somehow be
established to promote attempts at application, evaluate the
results, disseminate and promote the findings, and encourage
the selection and adoption of proven instrument systems.
(The word instrument refers to many classes of devices. In
this case I am talking about process-type instruments, suitable
for continuing service for automatic control. I am particularly
excluding laboratory instruments.)
John F. Andrews:
It would appear to me that many of our problems with
sensors could be reduced by using simple, on-line recalibra-
tions and performing tteese more frequently. Has anyone had
any experience with techniques suitable for on-line calibration
of large flow meters?
Robert H. Wise:
On-line (in-place) calibration of flow meters, regardless of
size, can be accomplished relatively easily with fluorimetric
tracer techniques. Equipment and/or procedures for making
such measurements can be obtained from Turner Designs, Palo
Alto, Calif. Mr. Ron Doty (one of the participants in this
workshop) has had first-hand experience with this calibration
technique and with some of Turner Designs' flow-calibrating
equipment; his comments on both were favorable.
P. M. Berthouex:
Mr. Schuk spoke of many problems with on-line sensors. I
agree that to demonstrate the feasibility of on-line control you
need (a) working hardware, (b) a usable control algorithm, and
(c) performance data. To learn about process dynamics,
however, we need performance data, preferably data taken
64
-------
PHYSICOCHEMICAL PROCESSES
during controlled experiments. We need more of this kind of
data and we need to properly analyze such data. Data may be
collected manually or by any other convenient means, and
methods are available to fit such data in an attempt to derive
the dynamic model, to identify the control problem, and to
design the control algorithm. This approach, to which George
E. P. Box at the University of Wisconsin has contributed
substantially, has the advantage of minimizing the need for
assumptions about mechanism, although known mechanisms
can and should be incorporated. Basically it lets the process
tell its own story.
Aside from the data analysis for which alternatives exist, I
should like to encourage more effort toward manual collection
of data since heavily instrumented plants are not yet common.
We can learn some things about automation and control
without having automated plants, if we have the energy to do
the data collection necessary.
Harry Fertik:
By developing and using scientific models one can learn
how various physical and chemical phenomena produce
process behavior, but scientific models are of limited useful-
ness in developing and testing real-time control strategies. For
example, in developing computer control algorithms for
cement kilns more than a decade ago, the initial effort was
placed in writing a scientific description of the process
dynamics. This work was dropped when it became clear that
the model, although complex and of high order, still would
not exhibit the behavior familiar to plant operators (material
buildup into "rings" in the kiln producing a typical tempera-
ture behavior, etc.). The control strategy was developed from
input-output analysis using process measurements obtained
from dynamic testing of the process. Low-order dynamic
models that incorporate process stochastic behavior and
nonlinear characteristics were the basis for this development.
The control strategy was refined and initially tested on these
models, with final testing on the process.
Optimization techniques have been applied to various
classes of problems with varying degrees of success. They can
be applied to design problems, the determination of steady
state operating conditions of a plant and to the calculation of
the dynamic response following a process disturbance or a
setpoint change. Design optimization is an accepted, successful
technique, and steady-state optimizers are working routinely
in various industries to calculate setpoints, but control design
by dynamic optimization in process industries has not in
general resulted in control system performance with any
significant improvement over conventionally designed control
systems.
Richard I. Dick:
Much of the discussion at this conference relates to the
conflict between complex theoretical models and practical
control procedures. A middle ground in some cases would be
to modify the design and/or operation of processes to make
them conform to the theoretical model. This sounds like
idealism, but in some cases it may be realistic. An example is
in the design of sedimentation basins. The theoretical analysis
of steady-state thickening in basins such as the final settling
tank in the activated sludge process is well established, and the
performance of closely controlled laboratory units conforms
well to the theoretical predictions. Parallel measurements with
full-scale tanks indicate that they often perform far less
satisfactorily than is theoretically possible. Professor Keinath
has correctly suggested a need for modifying theoretical
predictions in comparison with the actual performance of real
tanks. However, an alternative and more satisfying approach
would be to change the design of sedimentation basins to
make their performance approach that which is theoretically
achievable. It would seem that sedimentation equipment
manufacturers and design engineers might give some attention
to the differences between the predicted and actual per-
formance of their installations.
Harold D. Oilman:
The discussions so far appear to me to have focused too
strongly on the computer application in process mathematical
modeling. We should not neglect the basic computer capabil-
ities of logging, reporting, data management, alarming, and its
response to the information needs of operators, engineers,
management, and maintenance.
Research should be undertaken on how to take advantage
of the new dimensions of information processing and display,
notably with the cathode ray tube, graphics and colorjust
start with improved and tailored information for operators.
Poul Sorensen:
In considering the needs for research, I think a point should
be made about optimizing the chemical post-precipitation
process.
In Scandinavia, alum is normally used in this process. Some
provisional experiments carried out in Denmark have shown
that a pH adjustment with sulfuric acid has improved the
process in lowering the P content in the effluent from about 1
ppm to 0.2 ppm. The reason for this could be the high
alkalinity in the wastewater, but the process requires further
investigation.
John F. Andrews:
Physicochemical processes offer the advantage, in compari-
son with biological processes, of removing a greater percentage
of pollutants from wastewaters. They should also be more
amenable to variable-efficiency operation. However, their
operating costs are usually higher than those for biological
processes. These characteristics point to a potential research
need, this being selection of the proper combination of
biological and physicochemical processes for operating at a
variable efficiency and interfacing with the time-dependent
65
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
requirements of the receiving stream.
For example, treatment by a biological process may be
adequate for high flows in the receiving stream but inadequate
during low flow periods. A proper time-dependent basis for
plant operation might therefore be a biological process
followed by a physicochemical process, with the proportion of
the total flow treated by the physicochemical process being
controlled in accordance with the flow rate in the receiving
stream. An extension of this would be to regulate the
proportion of flow going to the physicochemical process in
accordance with the pollution load in the stream above the
plant discharge. The ultimate application of this concept
would be to use feedforward control of the flow proportion-
ing, based on real-time computer simulation using a dynamic
model of the stream below the point of discharge.
Report of Working Party
on
RESEARCH NEEDS FOR AUTOMATION
OF PHYSICOCHEMICAL PROCESSES
Wesley W. Eckenfelder, Jr.
Professor, Environmental and Resources Engineering, Vanderbilt University,
Nashville, TN 37235
John Stamberg
Office of Research and Monitoring, Environmental Protection Agency.
Washington, DC 20460
INTRODUCTION
An extensive Advanced Waste Treatment research program
sponsored by the Federal Water Pollution Control Administra-
tion during the 1960's delineated a variety of physical and
chemical wastewater treatment processes that are capable of
producing very high quality product waters. Subsequent
demonstrations at pilot plants and small full-scale plants
showed the utility of these processes, defined their respective
operational parameters and provided economic data.
Automation of these physicochemical processes has for the
most part been limited to studies conducted on a pilot scale
under contract to the Environmental Protection Agency.
Although these studies have defined several control strategies
that apparently are applicable to full-scale systems, they have
been evaluated only on a limited scale at small wastewater
treatment plants. Because the process of automating any
treatment system is an iterative procedure whereby strategies
or algorithms are proposed and then repetitively evaluated and
refined, it is clear that automation of physicochemical
processes is yet in its infancy.
It is noteworthy, furthermore, that most physicochemical
processes are not yet fully understood with respect to process
variables and operational parameters. This is due to a lack of
long-term operating experience at large wastewater treatment
plants. Until such information becomes available, strategies
directed toward optimal control cannot be formulated and the
full potential of physicochemical processes for reliably produc-
ing high-quality product waters cannot be attained. According-
ly, it is apparent that research needs in the area of automation,
instrumentation, and process control often cannot be divorced
from process research needs.
Physicochemical processes considered by this working party
include chemical clarification, filtration, adsorption, and nitro-
gen removal. Another extremely important application of a
physical process, clarification and thickening of biological
slurries, was considered as a primary subject by several other
working parties and, therefore, was not included in the
discussions of this group. Let it suffice to point out that great
potential payout exists for automating activated sludge bio-
logical systems so as to optimize the performance of secondary
clarifiers.
PROBLEM AREAS AND IDENTIFIED RESEARCH NEEDS
Relative to the automation of physicochemical treatment
processes the following problem areas were delineated:
1. Although existing data on lime clarification would in-
dicate that pH is the only parameter required to prop-
erly control the process, it is questionable
whether such a control strategy is universally transfer-
able to all geographic locations. For example, in areas
where the carriage water is relatively soft, it may be
necessary to add a polymeric flocculant to enhance
solids-liquid separation. In such a case a simple pH
control algorithm would not suffice and a more
complex control strategy would be required.
2. Chemical clarification using either Al(III) or Fe(III)
requires, at a minimum, control of both pH and
66
-------
PHYSICOCHEMICAL PROCESSES
dosage of the chemical in response to the concentra-
tion of soluble orthophosphate, turbidity or sus-
pended solids, and alkalinity. Even though the litera-
ture is replete with appropriate parametric relation-
ships, the task of formulating, evaluating and refining
a suitable control algorithm remains.
3. With respect to chemical clarification, furthermore.
each of the control schemes evaluated heretofore
have provided only for control of the chemical dosing
function in response to a readily measured parameter
such as pH. No scheme has been provided to control a
specific effluent-quality parameter, e.g., suspended
solids or soluble orthophosphate concentration. To
accomplish this, the chemical precipitation and chem-
ical coagulation/flocculation processes would first
have to be mathematically modeled. The resultant
models would then have to be coupled with a suitable
model for the primary clarifier. Until such an overall
model is available, optimal control of the chemical
clarification process cannot be established.
4. Automation of the deep-bed filtration process has not
advanced materially beyond what has been common
practice in the water treatment industry for several
decades. Control schemes that are currently em-
ployed in practice generally relate to two factions:
(1) now splitting; and (2) backwash initiation. While
these are very useful, they are not amenable to
establishing optimal control algorithms. If descriptive
dynamic models of the filtration process were avail-
able, they could be employed in a feedforward
manner to determine optimal control strategies so as
to maximize filter run times, thereby minimizing
backwash water requirements and recycle flows. Such
a dynamic model would have to consider filter
loading, bed penetration, head loss, and effluent
turbidity. Moreover, any filtration model would have
to provide for prediction of the effects of adding a
filter aid and be capable of controlling the dosing
function.
5. Operation of a series of deep-bed filters that are part
of wastewater treatment systems frequently is
plagued either by cascading or effluent deterioration
effects during periods of high flow. Conversely.
during periods of low flow, backwash water storage
requirements increase due to the lack of water which
can be obtained from the main flow. Suitable
backwash supervisory (executive control) strategies
can materially enhance operation of deep-bed filtra-
tion systems by either preventing or minimizing the
problems mentioned.
6. Although a dynamic mathematical model which
describes the adsorption of organics in municipal
wastewaters onto granular activated carbon was devel-
oped under an EPA contract, it has not been
employed to evaluate the feasibility of alternative
control strategies. For example, a treatment objective
might be to minimize carbon loading subject to the
constraint of meeting effluent standards. Using the
existing model, this as well as other control alterna-
tives could be evaluated.
7. Treatment of industrial wastewaters using granular
activated carbon is commonly beset with problems of
chromatographic displacement of adsorbed organics.
Solute displacement can be prevented through the
application of feedforward control using a descriptive
dynamic model. A need exists, therefore, for the
development of a dynamic model and for formulation
of an appropriate feedforward control strategy.
8. Because adsorption contactors are generally designed
and operated in a fashion analogous to deep-bed
filters, considerations regarding cascading and efflu-
ent deterioration effects also apply. Accordingly,
executive control strategies must be developed for
adsorption systems as well,
9. Because of the lack of a complete understanding of
process kinetics, removal of nitrogen from waste-
waters either by air stripping, selective ion exchange or
breakpoint chlorination cannot yet be reliably accom-
plished over the long term. This shortcoming can be
resolved by a complete parametric description of
these processes, by dynamically modeling the pro-
cesses and then by developing and evaluating control
strategies.
10. Process monitoring and control can only be accom-
plished if reliable sensors are available. Although
many sensors required to automate physicochemical
processes are available, the need exists to improve
their reliability and to reduce the maintenance
requirements. This is particularly important for the
measurement of flow, pH, phosphorus, ammonia, and
organic carbon (TOC, TOD, COD, etc.)
PRIORITIZED RESEARCH NEEDS
1. Develop and/or refine structured dynamic mathematical
process models for:
a. Lime clarification with or without addition of polymeric
flocculants.
b. Chemical clarification using either Al(III) or Fe(III).
c. Deep-bed filtration of wastewaters (single or dual
media).
d. Adsorption of organics from municipal wastewaters.
e. Adsorption of organics from complex industrial waste-
waters.
f. Removal of nitrogen from waste streams by either air
stripping, selective ion exchange, or breakpoint chlorina-
tion.
2. Develop practical control strategies for process operation
for each of the processes listed in 1. above.
67
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
3. Develop executive or supervisory control strategies for: organic carbon sensors and reduce their respective mainte-
a. Backwashing of a series of deep-bed filters. nance requirements.
b. Backwashing of a series of adsorption contactors. 6. Couple dynamic models for individual processes into an
4. Evaluate process models and candidate control strategies on overall dynamic model for an entire independent physico-
a pilot scale (Blue Plains) and subsequently at a full-scale chemical wastewater treatment facility and subsequently
wastewater treatment plant. explore and evaluate cost-optimal control strategies for the
5. Improve reliability of flow, pH, phosphorus, ammonia and entire plant.
68
-------
AUTOMATION
OF
SLUDGE PROCESSING, TRANSPORT & DISPOSAL
Workshop on Research Needs
Automation of Wastewater Treatment Systems
69
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
AUTOMATION OF SLUDGE PROCESSING,
TRANSPORT AND DISPOSAL
Bart T. Lynam, Raymond R. Rimkus and Stephen P. Graef
The Metropolitan Sanitary District of Greater Chicago,
100 East Erie, Chicago, IL 60611
INTRODUCTION
The Workshop mission of defining research needs and
establishing priorities in the area of wastewater treatment
systems automation is a positive step toward improving facility
operation. The benefits are especially germane to large
municipal utilities such as the Metropolitan Sanitary District
of Greater Chicago. In recent years it has been the District's
experience that process automation, where feasible, yields the
following benefits:
1. Performs routine tasks
2. Provides process information
3. Enacts control action decisions based on process status
4. Provides information for a planned maintenance pro-
gram
These benefits in effect
Enable the operator to cover larger service areas
Improve process quality and stability
Reduce capital and M & O costs
The District has either implemented or experimented with
automation in all three areas of sludge processing, transport
and disposal. The District's experience suggests that the area of
sludge processing has more potential application for automa-
tion than does sludge transport and disposal. Transport and
disposal operations are usually carried out by multiple,
single-unit pieces of equipment mobilized over a broad
territory and are thus more difficult to regulate automatically.
The areas of sludge transport and disposal do have a serious
need for basic development work of a general nature: however,
this is beyond the scope of this workshop.
SLUDGE PROCESSING
The objective of sludge processing is to convert raw sludges,
which are produced in wastewater treatment, into a form
which can be disposed of in an environmentally compatible
fashion. Raw sewage sludge production fluctuates somewhat
proportionally to the changes in raw sewage flow rate. The
amplitude is damped and out of phase compared to that of
raw sewage flow rate variations, because of the dynamics of
the settling and biological oxidation processes. Unfortunately,
the discontinuous schedule for transferring sludges, which
many plants follow, causes abrupt changes in volumetric and
mass loading rates to sludge processing units. Equipment is
available, such as variable speed pumps and surge tanks, which
can smooth the volumetric loading rates to sludge processing
units.
Fluctuations in mass loading, on the other hand, are more
difficult to regulate. First of all the dynamics of few, if any. of
the sludge handling processes have been experimentally tested
on a full scale basis in the field. This is in spite of the fact that
numerous dynamic studies on sludge processing have been
performed on a pilot basis or by computer simulation
techniques. These field studies are very important for imple-
menting automatic control and should be undertaken. Sec-
ondly, solids concentration and/or sludge density have not
been reliably measured on-line under long-term field condi-
tions. Entrained gas, stringy materials, inadequate velocity
gradients and grease coatings are several problems which
invalidate the accuracy of measured solids concentration. Field
testing and development is needed to make the sensors work.
For example sensor redundancy, sensor orientation, and
preconditioning of the sludge before meeting; e.g., degasifica-
tion and grinding should all be explored as means for
improving the solids concentration measurements.
Research needs for implementing automation in the various
types of solids handling processes are presented below. The
needs outlined are primarily oriented toward automation of
field units which process municipal sludges on an continuous
basis.
Concentration
Sludge concentration is usually accomplished by two types
70
-------
SLUDGE PROCESSING, TRANSPORT AND DISPOSAL
of thickeners, (1) gravity and (2) flotation. The research needs
for these units are as follows:
Gravity Concentration
1. Perfect devices for automatic monitoring and control
of underflow sludge concentration.
2. Develop a field-verified dynamic model for the
gravity thickening process.
Flotation Thickening
1. Develop feedforward control of flocculant addition
based, for example, on measured influent sludge
concentration.
2. Perfect feedback trimming of polymer addition rate
and pressurized recycle flow rate based on a process
variable; e.g., subnatant turbidity.
3. Develop a field-verified dynamic model for the
flotation thickening process.
De watering
Municipal sludges may be dewatered mechanically via
centrifuges, vacuum filters and filter presses, or by heat
treatment. Control actions for mechanical dewatering involve
either changes in mechanical settings and parameters or
variation in the rate of sludge conditioner chemical addition.
In most instances, chemical addition control can be adjusted
more readily than mechanical control settings. Potential
control actions for heat treatment include reactor residence
time and reactor temperature. The research needs for mechani-
cal dewatering and dewatering by heat treatment include:
Mechanical Dewatering
1. Develop feedforward and/or feedback control schemes
for addition of chemical conditioners. The District.
for example, controls ferric chloride addition to
waste activated sludge based on the conditioned
sludge pH. Other potential feedback control signals
and sensors may be on-line viscometers monitoring
the conditioned sludge stream and on-line turbidi-
meters monitoring the filtrate or centrate streams
from these units.
2. Develop a reliable field measurement system for esti-
mating solids concentration of sludge cakes.
Heat Conditioning
1. Refine a feedforward/feedback control system to
regulate reactor temperature and residence time based
upon influent total and volatile solids concentration.
Several manufacturers have steady-state control sys-
tems and these could be augmented to provide
dynamic control. Since the settling rate of heat
treated solids is rapid, a settlometer could potentially
be modified to obtain an effective feedback signal.
Combustion
Manufacturers of most wet and dry combustion processes
have incorporated steady state control into their combustion
systems. These systems may include feedforward control of
thermal energy input, air input rates and reactor pressure.
based on total and volatile solids loading rate measurements.
Stack gas composition and temperature, ash content of the
combustion residue and organic content of the recycle liquor
measurements have been used for feedback trimming control.
In general, equipment and technology for automatic regulation
of combustion processes are more developed than for other
sludge processing methods.
Heat Drying
Like other mechanical equipment for processing sludge,
automatic regulation has also been incorporated into proprie-
tary heat drying systems. Research needs for improving upon
present techniques for the heat drying process are as follows:
1. Develop means for determining an on-line water balance
around the reactor.
2. Develop an automatic control scheme for blending dry
solids with feed sludge.
3. Formulate a dynamic control strategy for regulating
energy utilization, stack gas quality and volatile content
of the dried sludge.
Digestion
Plant scale automatic control of biological processes is
practically non existent. There have been, however, several
dozen pilot plant and computer simulation studies in which
automatic control strategies have been formulated for these
processes. These studies have indicated that both aerobic and
anaerobic digestion appear suited for automatic control
actions and the research needed to achieve implementation
include the following:
Aerobic Digestion
1. Perfect sensor installations which yield reliable mea-
surements for obtaining organic carbon or similar
balances on a full scale reactor in the field.
2. Develop aeration or oxygen supply rate control
techniques for plant scale digesters. Potential feedback
control signals include off-gas composition from and
dissolved oxygen concentration in the reactor. An
aerobic digester could potentially be operated as a
continuous flow respirometer with respiration rate
- governing the rate of aeration.
3. Develop a field-verified dynamic control model for
the aerobic digestion process.
Anaerobic Digestion
1. Develop equipment and instrumentation systems
which would automate accurate digester sludge feed-
ing and withdrawal on a continuous basis. The system
should be reliable when handling the heterogeneous
materials in municipal sludges.
2. Automate digester stability evaluation based on mea-
surements such as gas composition and flow, in-line
71
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
pH and short chain fatty acid concentration in the
liquid phase.
3. Develop a feedforward off-line fermenter for screen-
ing the raw sludge for inhibitory characteristics.
4. Develop a field-verified dynamic control model for
the anaerobic digester.
Composting and Air Drying
Sludge handling processes such as composting and air
drying are not especially suited for automation. Both processes
incorporate natural forces, have extended reaction times, and
utilize mobilized solids handling machinery which are operated
by one or more persons.
SLUDGE TRANSPORT
At the present state of the art, automation in the area of
sludge transport may be applicable in only a few situations.
The best suited areas for control implementation seem to be
pipeline transport of sludge slurries and conveyor transport of
sludge cake solids and dry sludge solids. Other transport
methods such as barge, train and truck are not well suited for
automation since control of such equipment requires direct
human intervention. However, in moderate to large sludge
transport operations which utilize barges, trains and/or trucks,
a small computer or time-shared computer terminal can be
very beneficial for
(a) Monitoring sludge shipment and storage
(b) Developing operations research models for simulating
and evaluating sludge routing strategies
(c) Establishing a sludge management information
system.
Pipeline Transport
The District has experimented with automatic control of
pipeline transport both inside and outside of the plant.
Automatic regulation of fluid flow can be very effective. The
research needs to implement automatic control of sludge
transport include the following:
1. Develop reliable apparatus and techniques for starting
and stopping pumps, opening, modulating and closing
valves and routing sludge through a network of piping
based upon volume transferred and instantaneous flow
rate.
2. Develop automatic control of sludge viscosity reduction
techniques such as (a) high energy mixers for increasing
the rate of shear and (b) addition of chemicals which
reduce viscosity. On-line viscometers could serve as a
feedback control sensor for regulating both mixer speeds
and rate of chemical addition.
Conveyor Transport
Conveyors have been used extensively in the automatic
mode for feeding dry chemicals to assorted unit processes.
Feedback control of belt speed based upon belt scale
measurements insures a desired mass feed rate. Additional
work is needed to perfect sensing devices which estimate the
moisture content of both sludge cake and dried sludge solids.
Such a sensor could provide a feedback control signal for
regulating the rate of sludge delivery on a dried solids basis.
SLUDGE DISPOSAL
Sludge disposal by land spreading, landfill or ocean disposal
is not especially suited for implementing automatic control.
Individual pieces of machinery and equipment used in sludge
disposal may be equipped with some automatically controlled
mechanisms but total automatic control has not been imple-
mented. Although considerable research is needed in the broad
area of sludge disposal few avenues for applying automatic
control techniques seem available.
CONTROL OF SLUDGE HANDLING:
SOME SUCCESSES AND PROBLEMS
J. B. Parrel!
Chief, Ultimate Disposal Section, Advanced Waste Treatment Research
Laboratory, National Environmental Research Center, Cincinnati, OH 45268
INTRODUCTION
Steady advances, dating back to the late 1930's, have been
made in process control-advances in instruments for measur-
ing process variables as well as in control systems.
Instrument development has progressed beyond the mea-
surement of primary variables such as pressure, temperature,
and density, and instruments can rapidly measure complex
properties such as flow rate of heterogeneous mixtures or
composition of gas or liquid mixtures. Instrument develop-
ment sometimes seems slow or inadequate because there are
almost as many needs as there are processes. Complex
properties must be measured: the quality of a bread dough, or
the plastic and thixotropic behavior of a paint, or the
filterability of a sludge. As the needs become more specialized,
the economic motivation to instrument developers becomes
72
-------
less because the numbers of potential installations are fewer.
Nevertheless, the need exists, and improved control of
complex processes depends on such developments.
The chemical and petroleum industries offer excellent
examples of successful application of process instrumentation
and control. Several factors have played a part in this
development. Very often the materials that are processed
are simple fluids. Almost always the feedstock is quite uniform in
composition, the flow rate is held constant or is controlled at
will, and the chemical reactions are straightforward. As under-
standing of the process steps diminishes or the feed or
intermediate streams become complex fluids, control dimin-
ishes as well.
CONTROL IN WASTEWATER PROCESSING
Wastewater processing combines a number of factors that
work against satisfactory control. Flows vary widely and un-
predictably. Feedstock composition is poorly defined and fluc-
tuates greatly. Process steps are poorly understood. Residue
streams (sludges) that result are complex in physical and chem-
ical characteristics. It is no surprise that wastewater processing.
particularly sludge processing, has not seen extensive applica-
tion of instrumentation and control systems. Some reasons are
related to the nature of the "products", the scale of operation.
and the interest of top management. The "products" of waste-
water treatment do not return a profit, so there is little incent-
ive to produce a superior product. Most plants are too small to
justify instrumentation for optimum performance. Top man-
SLUDGE PROCESSING, TRANSPORT AND DISPOSAL
agement (the municipal authorities) is understandably more
likely to be concerned about creative innovation in areas other
than wastewater treatment.
SOME SUCCESSES
The most extensive and successful use of sophisticated con-
trol techniques in sludge processing is in sludge incineration.
There have been very good reasons for this development. In-
cinerators work poorly and use excessive quantities of fuel
when they are controlled manually, and manufacturers would
have seen their business disappear under the combined on-
slaughts of a populace interested in clean air and strict air pol-
lution codes. A second reason is that significant process varia-
bles can be measured accurately, and relationships between
variables are well understood.
The manner in which excess air is controlled in a multiple-
hearth furnace (MHF)* is illustrated in Figure 1. Control is ex-
ercised by pressure at the outlet of the furnace or by oxygen
content of the cooled exhaust gas. In one mode, pressure is
sensed at the furnace outlet and transmitted to a two-mode
controller, which positions a damper in the exhaust line. When
conditions have stabilized, control is manually transferred to
the second mode of the controller. Oxygen content of the ex-
haust gas is measured and a proportional signal is transmitted
"Information on control systems has been obtained from the Enviio-
tech Corporation. Mention of products or use of such information
does not indicate EPA endorsement.
FEED
TO STACK
SCRUBBER
Fgure 1. Draft Control for MHF.
73
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
to the controller, which controls the damper (and the exhaust
gas pressure) to give the desired air flow and exhaust gas oxy-
gen content. This is an example of cascade control.
Temperature on hearths is controlled by a conventional
feedback circuit (Figure 2). Temperature is sensed, and if it is
off the setpoint, the air flow rate is changed in the proper di-
rection. A ratio controller changes fuel flow in proportion to
air flow to give a constant percent excess air.
Variations in flow rate of solids can cause large fluctuations
in the fuel requirements on the different hearths. For example,
an increase in flow rate can cause the burning zone to travel
down the furnace and start at a lower hearth; this would sub-
stantially change the fuel demand on the hearth. A feed-
forward system has been devised (Figure 3) that increases the
rabble arm speed as feed rate increases. The increased agitation
of the rabble teeth exposes more surface, which tends to make
the sludge start burning sooner. The two tendencies counter-
balance each other, and the burning zone remains fixed.
Successful applications of instrumentation are not limited
to incineration. Magnetic flow meters and sludge density
gauges are used in open-loop control of sludge streams. Con-
stant-displacement pumps and conventional pressure, tempera-
ture, and flow controls are used to achieve closed-loop control
of the heat treatment process.
A DIFFICULT CONTROL PROBLEM
Some sludge handling processes and operations could be
profitably controlled, but so far, satisfactory feed-forward or
feedback control techniques have not been devised. One ex-
ample is vacuum filtration. The operation is costly, so there is
incentive to reduce costs. The major expense is the cost of
chemical conditioner, which conceivably could be reduced by
proper control procedures. The scale of operation is import-
ant. Some filters are too small to automate; they do not con-
sume enough conditioning agent to warrant the cost of a con-
trol system. However, a 500-square-foot filter, which is a com-
monly used size, can process over 4,000 dry tons of sludge per
year and can consume $40,000 worth of conditioning poly-
mer. If a control system could reduce this cost by 10 percent,
it would be a sound investment.
A vacuum filtration installation is illustrated in Figure 4.
Best operation is achieved when each filter has its own sludge
FEED
EXHAUST
(TRd
TRQ
AIR
\
ASH
Figure 2. Temperature Control on MHF Hearths.
74
-------
conditioning system. In most plants, control is manual. Sludge
flow rate is controlled by varying the speed of a constant dis-
placement pump. Conditioning agent is fed in proportion to
the sludge feed rate by changing the adjustments on a chemical
proportioning pump. The proportion between polymer and
sludge is established by bench-top filtration tests.
If the filter capacity is adequate to handle the plant capa-
city in the allotted operating period, a suitable control object-
ive would be to minimize polymer consumption. A simple
automatic control procedure that is used in some plants, parti-
cularly those in the chemical industry, is to control sludge feed
by the level in the sludge feed pan and match the polymer
dose to sludge flow by means of a ratio controller. This proce-
SLUDGE PROCESSING. TRANSPORT AND DISPOSAL
dure does not account for changes that occur in the solids con-
tent of the liquid sludge and in the filterability of the sludge.
An alternative procedure, which is under test by a filter manu-
facturer, is to add to this control system a continuous mea-
surement of solids content of the sludge by means of a nuclear
density gauge. The polymer dose is matched by the ratio con-
troller to the product of volumetric flow rate and sludge solids
content; i.e., the mass flow rate of dry solids. The control pro-
cedure now accounts for changes in sludge solids content but
still does not account for changes in the filterability of the
sludge.
Swanwick (1) describes the use of a rapid filterability test
to control dose of chemical conditioning agent to a sludge that
BELT SCALE
BELT CONVEYOR
FEED
I WEIGHT TRANSMITTER
*
RATIO CONTROLLER
FURNACE
SPEED CONTROLLER/" A
TRANSMITTER T
ASH
SHAFT DRIVE
Figure 3. Control of MHF Rabble Arm Speed.
75
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
CONDITIONING
TANK
FILTER
POLYMER
CAKE
TO VACUUM
SLUDGE
Figure 4. Vacuum Filtration of Sludge.
was to be pressure filtered. The test, which combined hand-
operated mixing operations with an automated version of the
capillary suction test (CST), took 4 minutes to complete and
produced a recorded output. The sludge was dosed with suffi-
cient lime to give a CST value below an empirically determined
critical value. This method eliminates the need for a measure-
ment of the solids content of the sludge. If sludges are blended
before filtration so that changes in solids content and filter-
ability are slow, the time required for the number of CST tests
needed to establish optimum conditions should not seriously
limit the ability to use these measurements to automatically
control conditioning chemical dosage.
FUTURE DEVELOPMENTS
It seems certain that the pace at which instrumentation and
control systems are introduced into sludge handling processes
will accelerate, particularly in plants that handle large waste-
water flows (for example, in excess of 30 million gallons per
day). The primary problems to be overcome are inadequate
understanding of complex processes, such as digestion, and the
lack of suitable instrumentation for measuring complex
phenomena.
REFERENCES
1. Swanwick. J. D.. Water Poll. Control. 72. 78-86 (1973).
DISCUSSION
Richard I. Dick:
Is there any potential for improving and automating the
performance of land systems for sludge disposal by use of in
situ sensors for constituents such as soil moisture, ORP, oxy-
gen concentration, ammonium, and nitrates? It would seem
that sludge application schedules could be optimized, ground
water contamination could be minimized, and possibilities for
reduction of the nitrogen load by controlled nitrification-
denitrification could be realized if such a system were
implemented.
Ted Lejeune:
At the R. M. Clayton plant in Atlanta it is planned to pro-
vide polymer dosing control based on centrate density. So far
we have no operating experience with this method. As an alter-
nate, a ratio proportioning meter is available.
We have experienced continual problems in trying to use
sludge density and flow meters, due to blockages and materials
such as sticks in the sludge. Because of these problems, the At-
lanta computer will be used mainly for logging, rather than
control purposes.
76
-------
SLUDGE PROCESSING, TRANSPORT AND DISPOSAL
The multiple-hearth furnace at Atlanta is operated on an in-
termittent basis, and we feel that control of the final burn-out
is best done manually. This also points out the need for better
training of operators for automated plants.
Carmen F. Guarino:
In contrast to the experience in Atlanta, I want to point
out that there has also been good experience in operating
sludge density gauges and flow meters in Philadelphia and Los
Angeles.
Richard I. Dick:
Traditionally, problems of sludge treatment and disposal
have been addressed independent of the problems related to
the wastewater treatment process which generates the sludges.
This is regrettable, for overall optimization would require
simultaneous consideration of all aspects of the entire treat-
ment process.
It would be unfortunate to extend this traditional approach
to automation. Shouldn't attempts to automate sludge handl-
ing extend beyond the sludge treatment and disposal area and
go back to the wastewater treatment process itself? For ex-
ample, are there opportunities for controlling the quantities
and physical, chemical, and biological characteristics of sludges
by feedback control of the processes which generate the
sludges?
John F. Andrews:
I agree that one should guard against looking at individual
processes in isolation. Work is in progress on the development
of an overall plant model, which will enable us to explore in-
teractions between units, and start to think about optimiza-
tion of the total operation.
P. .M. Berthouex:
There are two ideas I invite Steve Graef to comment on.
First, we have talked mostly about automation to monitor or
control traditional processes. Can't we use control problems to
direct our thinking toward design changes and the develop-
ment of new processes. One use for theoretical models is to
explore process and operational changes. Promising ideas are
identified for pilot plant or full-scale trials. This is true even if
the model is never implemented for control, or, perhaps, even
if the model is later found to have inadequacies.
Secondly, the operator himself may be a good sensor. Some
operators can adjust to color and sludge appearance. This is
not optimal control, but it is control. As an example I cite the
use of closed-circuit television at the Blackbirds Sewage Treat-
ment Works, England. Sludge flows over a circular wire and
the operator can estimate sludge density by appearance. It is
simple and apparently helpful. It is also automation in a sense.
Clever design ideas such as this may be an alternate-perhaps
an intermediate solution-to more sophisticated automation
systems.
John F. Andrews:
A sensor which could be of significant value in anaerobic di-
gestion is an on-line calorimeter for continuous monitoring of
the caloric value of digester gas. The discusser has observed the
use of such an instrument at the San Jose, California, waste-
water treatment plant in which the caloric value of the digester
gas was measured and this signal used to control the amount of
natural gas blended with the digester gas. The result was a gas
with a constant caloric value for improving operation of the
gas engines used in this plant.
The discusser would like to suggest that the rate of energy
production, as calculated from the calorimeter and digester gas
flow rate measurements, would be a valuable indicator of the
condition of the anaerobic digestion process. This is analogous
to the rate of methane production which has been proposed as
an indicator of digester condition by Graef and Andrews (1).
Moreover, this measurement, when used in combination with
COD measurements on the feed to the digester, could be used
to compute the efficiency of the digester with respect to
energy removal from the feed sludge, since one pound of COD
is approximately equivalent to 6,300 BTU of energy (2).
REFERENCES:
1. Graef, S. P. and Andrews. J. F., "Stability and Control of Anaero-
bic Digestion," Jour. Water Pott. Control Fed., 46, 666, (1974).
2. Andrews. J. F. and Kambhu, K., "Thermophilic Aerobic Digestion
of Organic Solid Wastes," Final Report to Office of Solid Wastes,
U. S. Public Health Service, Environmental Systems Engineering
Department, Clemson University, Clemson, SC (1971).
CLOSURE
Stephen P. Graef:
There is a potential for improving land application systems
via soil moisture, ORP, oxygen concentration, ammonia, ni-
trate, etc., surveillance. Moreover, the benefits recognized by
Dr. Dick can and have been realized. Automation, however,
would not enhance a land application operation. Most land ap-
plication variables change quite slowly; Le., on the order of
weeks and months rather than hours and days. Although com-
plete and accurate surveillance is essential for gauging the con-
dition and performance of a land application program, the
monitoring does not have to be continuous or on-line. Periodic
sampling and analysis both in situ and at the laboratory can
provide the information needed to effectively manage a land
application program.
Automation and continuous on-line surveillance can be a
distinct advantage in controlled small-scale pilot studies. In a
research environment as opposed to a large-scale production
effort, agricultural instrumentation and control apparatus can
be serviced efficiently and can provide the large volume of
data needed for investigative studies.
In his second question, Dr. Dick states an important con-
cept. Efforts to automate sludge handling as well as other
wastewater treatment processes should not be confined to the
process itself. The control strategy should take into considera-
77
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
tion the interactions between the process under consideration
and others throughout the plant.
Such a system control strategy is developed on two levels.
First a strategy must be formulated for each component of the
system. The component control scheme must enable the pro-
cess to meet its objectives, given the range of inputs which can
be expected from the system. A more encompassing system
strategy should then be developed which balances the inter-
actions and tradeoffs among processes to meet the objectives
of the system.
As an illustration consider a system, two components of
which are the activated sludge and anaerobic digestion pro-
cesses. Assume that under severe conditions the digester may be
required to process 0.45 Ibs. of volatile solids/ft.3 day for a
short period. A strategy of pH, temperature, and gas recircula-
tion regulation must be formulated if the activated sludge pro-
cess does indeed provide such large masses of volatile sludge to
be processed from time to time. Under normal conditions a
system strategy could control the sludge produced by the acti-
vated sludge process within 0.25 to 0.3 Ib VSS/ft.3 day. By
regulating SRT and sludge blanket depth in the final clarifier,
the organism loading to the digester will also be regulated. A
system strategy should be formulated which will control and
balance the interactions so that the total output of the system,
e.g. digested sludge and clarified secondary effluent, meet or
exceed the quality criteria.
In response to Dr. Berthouex's first question, math models
and computer simulations are powerful tools for evaluating de-
sign changes and development of improved process trains, if
not improved processes as well. It is probably the only way a
designer can answer dozens of questions such as, "What would
happen if...", without an enormous hand calculation effort.
Moreover, some questions of this sort could not be answered
at all without computer simulations. As indicated, several par-
ticipants in this workshop have successfully utilized theoretical
models to formulate productive process and operational
changes.
In response to the second question, there is no reason for
intentionally making automation and control techniques com-
plicated. Process output criteria or objectives should be speci-
fied for the designer. He should in turn provide the control
scheme to meet the requirements. The scheme should be reli-
able, easy to use by operations personnel and easy to maintain.
J. B. Fairell:
In response to Mr. Lejeune's comment, the use of centrate
density (i.e., solids content) is an interesting way to control
polymer dose for centrifugation. Unfortunately, gamma ray
density gauges are not very sensitive indicators of sludge solids
content in the range of 0 to 1% solids. Mr. Lejeune might want
to consider using a turbidity meter to indicate solids content.
If the centrate is too opaque, the centrate stream could be di-
luted in a fixed proportion with clear water to get an on-
stream reading.
Mr. Lejeune observes that the multiple-hearth furnace at
Atlanta will be operated on an intermittent basis, and that
burnout (in preparation to shutdown) is best done manually.
He is probably right that during burnout, when sludge feed is
being shut off, control should be turned over to manual opera-
tion. I suggest that Mr. Lejeune should do all that he can to re-
duce the number of shutdown-startup cycles for the incin-
erator. Operation of a multiple-hearth incinerator for one shift
a day with shutdown for two shifts consumes excessive quanti-
ties of supplementary fuel and disproportionately increases
maintenance cost. We should really be thinking of continuous
operation of multiple-hearth incinerators with shutdown at
most once a week rather than once a day. Perhaps we should
design to run the dewatering room during the first part of the
week, provide intermediate sludge cake storage, and run the in-
cinerator continuously with the staff of the dewatering room
during the second half of the week. Alternatively, the dewater-
ing and incinerating equipment could be operated continu-
ously only during the first few days of the week, but this pro-
cedure presents awesome personnel scheduling difficulties.
In response to Professor Dick's question, I believe that the
design of the water purification processes of a sewage treat-
ment plant and the sludge treatment and disposal processes
should be considered simultaneously, with the intention of
minimizing cost and optimizing performance. Processes
selected and conditions at which the processes are operated
have a strong effect on sludge quantity and properties. As an
illustration, it is well known that the quantity of biological
sludge produced in the activated sludge process depends on the
operating conditions of the process. However, these conditions
are fixed within relatively narrow limits by the initial design. It
is difficult to picture a control scheme, for example, which
would feed back into the water purification processes when
the sludge concentration at the sludge thickener outlet starts
to decrease or when the sludge production rate exceeds a de-
sired figure. The water purification processes should primarily
be controlled to produce high quality water. Control schemes
for sludge handling and disposal should probably not feedback
into the water purification processes.
78
-------
SLUDGE PROCESSING, TRANSPORT AND DISPOSAL
Report of Working Party
on
RESEARCH NEEDS FOR AUTOMATION
OF SLUDGE PROCESSING, TRANSPORT AND DISPOSAL
Richard I. Dick
Professor, Department of Civil Engineering, University of Delaware, Newark, DE 19711
John R. Trax
Office of Water Programs, Environmental Protection Agency, Washington, DC 20460
In spite of the fact that sludge treatment and disposal has
been a problem since the first wastewater treatment plant was
built, there is probably a greater backlog of unsolved problems
related to automation in the area of sludge treatment and dis-
posal than in other areas. Development of technology in sludge
treatment and disposal has largely been ignored in the past and
even comparatively new processes (such as the chemical-physi-
cal processes developed in recent years) are probably closer to
being capable of precise automation than are processes for
sludge treatment and disposal.
Because of the larger number of unsolved problems relating
to sludge treatment and disposal, and because of the accom-
panying difficulties in accomplishing effective automation of
sludge treatment and disposal processes, the area deserves high
priority in competition for research funds. There is appreciable
potential for significant returns on research money spent in
this area.
STATEMENT OF PROBLEM
Attempts to effectively and extensively automate sludge
treatment and disposal schemes might be frustrated by a num-
ber of problems. Some of these are indicated below.
1. Automation of sludge treatment and disposal facilities
inevitably involves sensing of sludge properties and con-
trol of sludge flows. To accomplish this, it is necessary
that sludges be free of gross solids, rags, sticks, stones.
etc. This problem is not expanded in the later list of re-
search needs because it is not felt that research is wai-
ranted on the subject. However, in the considerations
which follow, it is assumed that complications caused by
such materials in sludges have been eliminated. Ques-
tions remain as to whether the solids can best be elimi-
nated from the raw waste flow or from the sludge itself.
2. To automate sludge treatment and disposal schemes it
will be necessary to sense appropriate properties of the
sludge. In the case of sludges, some of the important
properties are less well defined than with wastewater.
For example, automated procedures for sensing physical
parameters such as settleability and dewaterability may
be difficult to develop.
3. Attempts to effectively automate sludge treatment and
disposal processes may be expected to be hampered by a
heritage of ineffective integration of the various steps in-
volved in sludge treatment and disposal with each other
and with the wastewater treatment processes which gen-
erate the sludges. The relationships between various pro-
cesses of sludge treatment and disposal are less well un-
derstood than the relationships between various stages in
wastewater treatment.
4. Temporal variations in sludge characteristics may
hamper the performance of many sludge treatment and
disposal schemes. An example is the anaerobic digestion
process which, particularly in cities of small and moder-
ate size, often does not operate under stable conditions.
5. Although the performance of all processes of sludge
treatment and disposal is related to the suspended solids
content of the sludge, it is difficult to automatically
measure those concentrations.
6. Because processes for sludge treatment and disposal in
general have not been developed to as high a degree as
those for wastewater treatment, appreciable work is
needed to develop predictive models describing the per-
formance of these processes.
RESEARCH NEEDS
The following specific research needs are noted.
1. Investigation of the interaction between various processes
used in sludge treatment and disposal must be undertaken
so that the overall process of sludge management can be op-
timally integrated. In the area of automation of sludge
treatment and disposal, this is considered to be a major re-
search need. Additionally, benefits are to be derived from
better integration of sludge treatment and disposal facilities
with the processes which generate the sludges.
2. Methods for sensing suspended solids concentration over a
wide range of sludge consistency are needed. Three ranges
of suspended solids concentrations might arbitrarily be
identified:
a. Dilute concentrations (filtrates, centrates, etc.): turbidi-
meters offer possibilities for sensing in this region.
b. Intermediate range (concentrations typical of gravity
thickener feed and underflow): some need for improve-
ment of available instrumentation in this area is seen.
c. High concentrations (filter cakes, centrifuge cakes, etc.).
79
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
In spite of the importance of these measurements, means
for sensing them in real time are not available.
3. There is a need for sensors to continuously monitor the
physical properties of sludge which are important to various
sludge treatment and transportation methods. Specifically:
a. On-line monitoring of the rheology of sludges.
b. On-line measurement of the settleability of sludges.
c. Continuous measurement of the dewaterability of
sludges.
d. Continuous measurement of the calorific value of
sludges.
4. Existing steady-state models and developing non-steady
state models of the performance of gravity thickeners
would seem ready for implementation. The final step of
field testing of these models to demonstrate the control
strategy could seem to be important for future progress in
sludge thickening and automation of sludge thickening
processes.
5. The following research needs exist in the area of biological
stabilization of sludges.
a. Improve the stability of the anaerobic digestion process
for stabilization of sludges-Representatives of larger cities
did not feel that this need deserved high priority because
their sludges tend to be more uniform in quantity and
quality. The need was considered to exist for moderate
and small-sized cities, however.
b. Field scale evaluation of available models of anaerobic
digestion performance should be undertaken. Sensors
such as those for gas composition, gas production rate,
volatile acid concentration, alkalinity and suspended
solids concentration would be useful in undertaking such
field scale evaluation and refinement of models of anaer-
obic digester performance to develop a demonstrated
control strategy.
c. Similarly, existing models of the performance of aerobic
digesters also should be evaluated in field scale studies.
d. The effects of aerobic and anaerobic digestion on solids
separation and dewaterability need to be evaluated and
developed into performance models.
6. Sludge conditioning processes are not fully understood and
are very difficult at present to automate. Two phases of re-
search are required in this area:
a. Stage one would involve empirical, short-term evalua-
tions of sludge conditioning processes. Control pro-
cedures would then be developed on an empirical basis
to regulate machine variables such as submergence or
drum speed and control coagulant doses.
b. Long-term research should be conducted to improve fun-
damental understanding of conditioning and dewatering
processes. This research should lead to development of
models based on fundamental mechanisms involved in
conditioning and dewatering.
7. In the area of incineration, the prospect for avoiding short-
term transients by.storage and blending of dewatered sludge
prior to introduction into incinerators to assure a constant
supply of sludge of uniform quality should be undertaken.
Existing incinerator monitoring techniques for automation
procedures are considered to be relatively well developed
except that additional work is needed on suspended solids
monitoring devices, and devices for sensing the calorific
value of sludges need to be utilized.
8. Currently, little research for automation of ultimate dis-
posal schemes such as land and ocean disposal is considered
to be of high priority. Much additional work, however, is
needed in developing the ultimate disposal processes them-
selves. In the area of automation, there is need for a means
of sensing the numbers of pathogenic organisms and viruses
in sludge.
Priority of Research Needs
As argued in the Introduction, all research needs in the area
of sludge treatment and disposal have a comparatively high pri-
ority because technology in this area has lagged behind that of
other wastewater treatment processes. Following is a ranking
of priorities within this high priority area:
1. The interrelations between the various processes used for
sludge treatment and disposal need to be explored to al-
low integration of the processes with aid of automation
procedures in an optimal way.
2. Develop reliable suspended solids concentration sensors
for the entire range of solids encountered in sludge treat-
ment and disposal. Sensors for the low range of suspended
solids concentrations (for flows such as filtrates and cen-
trates), for the intermediate range of concentrations (typi-
cal of liquid sludges), and for the high range of concentra-
tion (corresponding to dewatered sludges) will operate on
different principles.
3. Develop a "settleability meter." Such a device might be
based on actual measurement of settling properties or by
sensing a related physical property.
4. Develop a "dewaterability meter." Such a device is essen-
tial for automation of sludge dewatering processes and
monitoring of sludge conditioning processes.
5. Carry out field-scale testing of gravity thickening models.
Such models currently are available but require extensive
full-scale testing and possible modification.
6. Develop control strategies for improving the stability of
anaerobic digesters. This is not a high priority item for
large installations, but it is important for smaller installa-
tions receiving occasional toxic discharges.
7. Develop an empirical model for control of conditioning
and dewatering facilities. This basis for automatic control
of conditioning and dewatering processes is needed until
more fundamental models can be developed.
8. Develop and apply sensors for sludge rheology. Such sen-
sors should be useful for controlling pumping and piping
operations and, potentially, other operations whose per-
formance is related to the physical properties of sludges.
80
-------
SLUDGE PROCESSING, TRANSPORT AND DISPOSAL
9. Carry out field-scale evaluation of anaerobic digestion
models. Such models currently are available, but lack the
extensive field-scale verification and modification required
to achieve a demonstrated control strategy. Effects of an-
aerobic digesters on solids separation and dewaterability
should be included.
10. Develop rational models for control of conditioning and
dewatering facilities. This is a more long-term solution to
the same problem addressed in high priority item number
7 above.
11. Develop a sensor for pathogenic organisms and viruses in
sludge.
12. Conduct field-scale demonstrations of controlled "steady-
state" incineration accomplished by storage and blending
of dewatered sludge prior to incineration.
13. Evaluate the transferability of industrial solids-handling
technology to municipal and industrial sludge treatment
and disposal.
14. Conduct field-scale evaluations of models for aerobic di-
gester performance. Effects of aerobic digestion on solids
separation and dewaterability should be included.
15. Develop and test on-line sensors for the calorific value of
sludges.
81
-------
COMPUTER APPLICATIONS
IN AUTOMATION
Workshop on Research Needs
Automation of Wastewater Treatment Systems
83
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
CURRENT PRACTICE IN INSTRUMENTATION AND
COMPUTER APPLICATION AT THE COUNTY SANITATION DISTRICTS
OF LOS ANGELES COUNTY
Walter E. Garrison, Kip Payne and Tim Haug
County Sanitation Districts of Los Angeles County, 1955 Workman Mill Road
Whittier,CA 90601.
INTRODUCTION
The Sanitation Districts of Los Angeles County plan, de-
sign, construct, and operate a sewerage system serving almost
4,000,000 people residing in all or part of 71 incorporated
cities and adjacent County territory. The basic operations are
supported by a substantial research and development effort di-
rected toward improving existing design and operational tech-
niques and to pilot testing of advanced treatment processes for
future water quality upgrading. District management is dedi-
cated to a well-financed program to improve plant reliabilily
and performance through development and application of any
advanced technology which will aid in achieving this goal.
The purpose of this paper is to present a manager's over-
view of the present "State-of-ihe-Art" technology of the appli-
cation of instrumentation and computers in the Districts1
wasiewater treatment plants and collection system. The text
has been prepared to briefly summarize salient features of the
existing system, to explain the purpose and function of auto-
mation in routine operation al the plants, to describe the cur-
rent practice in computer application and, finally, to explore
potential uses for future application of instrumentation and
computers.
DESCRIPTION OF EXISTING SYSTEM
All service areas of the Districts are served by separate sew-
erage systems with minimum storm water infiltration. Rainfall
generally occurs for about three months during the year. Thus
inland plants discharge to dry water courses most of the year
or to reclamation projects whenever available. Public contact di-
rectly with plant discharges is common, requiring the mainte-
nance of high-quality effluent, adequately disinfected at all
limes.
In the principal Los Angeles drainage basin, approximately
420 million gallons per day of sewage is processed by six
wastewater treatment plants. Five of the plants utilize the acti-
vated sludge process and are strategically located inland to
withdraw raw sewage from existing trunk sewers which are
tributary to a large primary plant near the ocean. The inland
plants are unique in that all solids removed in the treatment
process are relumed to the exJsting trunk sewers for central-
ized processing al the primary treatment plant. Table I sum-
marizes the flow parameters of the treatment plants on this
system.
Six other District treatment plants ranging in size from 0.2
MOD to 5 MOD serve other inland drainage basins. Solids pro-
cessing at these planls is handled conventionally by use of two-
stage anaerobic digestion followed by onsite digested sludge
drying beds or truck transport to off site drying and/or
disposal.
Al the large primary plant (JWPCP) 500 tons per day of
raw sludge is stabilized in a highly automated single-stage, high
loading anaerobic digestion process followed by a (wo-siage di-
gested sludge dcwatering station. The dewalering station con-
sists of a firsl stage, continuous-flow solid-bowl centrifuge fol-
lowed by a second slage, baich-type basket centrifuge, the lat-
ter currently under construction. Centrate from the dewater-
ing station, at about 1000 mg/l, will be recycled to the raw
sewage channel, with the cake'discharged lo an open-bed com-
posting operation. Mosl of the dried solids will be sold as a fer-
tilizer, with the balance hauled to landfill.
Expansionof the above described system, based upon curreni
federal legislation, will consist of construction of secondary
treatment at the existing primary plant. Pilot-plan! testing is
currently in progress for (his 300 million dollar expansion. An
interim short-term improvement in effluent quality will be
achieved by chemical treatment using polymers, lo achieve
80% removal of suspended solids, followed by fine screening
with a 20 mesh travelling water screen to remove remaining
floatables prior to ocean discharge.
At the inland plants, planning for upgrading is concentraied
on advanced treatment techniques to reduce coliform bacteria
84
-------
COMPUTER APPLICATIONS IN AUTOMATION
TABLE 1
LACSD TREATMENT PLANTS IN LOS ANGELES BASIN
TREATMENT PLANT
Primary Plant
Joint Disposal
Plant (JWPCP)
Activated Sludge Plants
Whittier Narrows WRP
San Jose Creek WRP
Pomona WRP
*Los Coyotes WRP
Long Beach WRP
DESIGN FLOW
MGD
450
12.5
38
14
38
13
"Includes 25 MGD currently under construction.
AVERAGE DAILY FLOW
MGD
(AS OF JULY 1973)
350
12
31
8.5
8.5
9.1
RATIO OF PEAK
TO MINIMUM FLOW
2.6
1.5
3.1
4.3
1.9
4.3
to less than a median value of 2.2 per 100 ml and to the re-
moval of virus below levels of detection. Such sophisticated
treatment processes will not only assure greater public health
protection where human contact with undiluted effluent oc-
curs, but also will upgrade water quality to a level where reuse
projects will gain much wider public support. Currently, the
Districts' staff is designing a carbon-adsorption advanced treat-
ment module for the Pomona WRP which will provide 10
MGD of water meeting the above criteria. Award of contract
for construction is scheduled for December 1974.
AUTOMATION IN ROUTINE OPERATION
The following discussion is intended to provide a general
summary of those instrumentation systems which are in rou-
tine use to provide operational control of District plants. A
considerably more detailed report is currently being prepared
under an EPA research grant titled. "State-of-the-Art Tech-
nology for Semi-Automatic Control of Activated Sludge Treat-
ment Plants". This detailed report is scheduled for completion
by March 1,1975.
The management policy of the Districts encourages maxi-
mum utilization of advanced control systems where it can be
demonstrated that improved reliability and performance will
result. A continuing effort is being made to field-test new and,
hopefully, improved control equipment to clearly establish
capital, operation and maintenance costs. Backup equipment
and/or manual override of all automatic controls is obviously
necessary to assure continuous compliance with requirements
of regulatory agencies. Equipment described in this paper is
satisfactorily controlling plant operations, but improvements
are being sought which will reduce operational manpower re-
quirements and improve effluent quality at a reasonable capi-
tal cost and without excessive maintenance costs.
Inlet Pumping Controls
All of the influent pumping plants in the LACSD system
are designed with variable speed pump and liquid level control.
In general, water surface elevations in the incoming sewers are
maintained near normal depth by varying pump speed with
depth of flow. Although this proportional relationship be-
tween depth and pump speed does not exactly produce normal
depth for all flows, it approximates it closely enough to avoid
adverse effects caused by excessively high or low velocities in
the sewer. Utilization of the potential storage capacity in in-
coming trunk sewers to reduce peak flow has not been aug-
mented since the warm climate and correspondingly high tem-
perature of the sewage would cause odor problems and higher
hydrogen sulfide levels than already exist. Variable-speed
pumping to match normal water evaluations eliminates
"breathing" of wet wells with attendant discharge of odorous
air to the neighborhood.
In a typical recent design, smaller pump units consist of a
vertical 10-inch centrifugal pump, flexible shafting to a close
coupled magnetic drive, and electric motor. Larger pump units
85
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
are similar except for the addition of a right-angle gearbox
speed-reduction unit. Major components of the control system
consist of a liquid level sensor and transmitter, liquid level con-
trol unit, and the motor/variable speed drive units used to
power the pumps.
The level control system controls the influent pumps so that
the rate of discharge from the pumping station is approxi-
mately equal to the varying rate of sewage inflow. Level con-
trol in the wet well is initiated by means of proper pump se-
quencing and by varying the pump speed. The system is capa-
ble of sequencing pump operation and varying the speed of all
pumps as necessary to pump a variable station flow without
storage in the wet well.
Primary intelligence for the level control system is obtained
from an electronic differential pressure transmitter or trans-
ducer. This unit is used to sense rising and falling liquid level in
a stilling well which is directly connected to the influent wet
well. Pressure sensed in the stilling well is converted to a 4-20
milliampere direct-current control signal, proportional to the
stilling well water level.
The liquid level control unit receives the control signal gen-
erated by the differential pressure transmitter. This signal is
used as a pilot signal for operation of the control equipment.
These operations include indication of the liquid level in the
wet well, initiation of start and stop functions for all pumps.
modulation of pump speed, and actuation of alarms for high
and low water levels in the wet well.
In a special situation where available flow in a trunk sewer
greatly exceeded treatment plant capacity and the plant was
being operated for reclamation purposes, pump controls were
set to vary pump speed to maintain a constant water surface
elevation in a channel feeding the treatment plant. Thus, a
constant flow could be maintained through the plant. General-
ly, however, existing inlet pump controls function well and no
special need exists for more sophisticated control. Later in this
paper the potential for flow equalization in existing CSD
plants will be discussed and, in that instance, more compli-
cated controls would be warranted.
Process Air Controls for Activated Sludge Plants
Basically the quantity of oxygen (air) required to satisfac-
torily operate an activated sludge plant will vary depending
upon the organic load in the primary effluent, the cell resi-
dence time at which the plant is operated, and whether nitrifi-
cation does or does not occur. Oxygen requirements to stabi-
lize a wastewater can be determined within reasonably narrow
limits depending primarily upon influent COD, NH3-N concen-
trations and mean cell residence time. However, air quantities
required to satisfy oxygen demand are more variable depend-
ing upon the type of oxygen transfer equipment used and the
efficiencies at which they operate. In all but completely mixed
systems, rates of oxygen consumption will vary along the
length of the aeration tank. Therefore, process air control is
critical to the stable operation of the activated sludge system.
Not only must total air quantities be controlled, but the rate
at which it is supplied along the aeration tank must be varied
to provide for stable and efficient treatment.
Process air control systems used at CSD activated sludge
plants utilize either a control signal to throttle a centrifugal
compressor or preset timers to control the number of on-line
positive displacement compressors. In the former case, dis-
solved oxygen probes, cam programmers, and plant flow rate
are currently used to provide the command signal. The control
method used at various CSD activated sludge plants is shown
in Table 2.
None of the above control systems have operated without
some disadvantages, discussion of which is too lengthy to be
included in this short paper. From a theoretical standpoint
control by use of D.O. should be the most desirable since such
a system can respond to any diurnal fluctuation in oxygen de-
mand as well as shock loads, whether they are caused by flow
rate or waste strength. Until recently, a reliable D.O. probe has
not been available which was capable of holding calibration to
the extent that air compressor programming could be depend-
ably set by the sensor. Recent developments in this field now
make D.O. sensing a viable control scheme.
Experience with D.O. probe control systems has revealed
several operational characteristics which must be considered in
any application. D.O. levels below 1 or 2 mg/1 will affect nitri-
fication kinetics by imposing a rate limitation. Based upon ex-
perience in Southern California, if the mean cell residence time
is above 5 to 6 days (dependent upon temperature) and the
D.O. set point is below about 0.5 mg/1, nitrification may not
occur because of the D.O. rate limitation. Conversely, if the
D.O. set point is increased much above 0.5 mg/1 the rate limit-
ation is removed and nitrification will proceed. This will pro-
duce a sharp increase in oxygen demand and process air flow
will increase dramatically to try and match the demand. If ni-
trification is not desired, the best operational procedure is to
keep the MCRT below about 4 days. However, this may not
produce the best quality effluent in terms of suspended solids.
In the D.O. probe system currently used at the CSD plants,
only one D.O. probe can be selected at any time to control
process air flow. The D.O. probe system does not regulate the
distribution of air flow along the aeration system, but only
regulates total air flow to the aeration tank in order to main-
tain a dissolved oxygen concentration at the site of the con-
trolling probe.
Another operational consideration with regard to the D.O.
probe control system is the added maintenance required, com-
pared with other control techniques in use at the CSD plants.
Probes in current use require weekly routine maintenance, ad-
justments and calibration by instrument repair personnel and
daily comparative checking by the plant operator. If reliable
response is obtained, this level of maintenance is not con-
sidered to be excessive.
If the full potential of the D.O. probe system is to be rea-
lized, total air flow and distribution of air throughout the aera-
86
-------
COMPUTER APPLICATIONS IN AUTOMATION
TABLE 2
ACTIVATED SLUDGE PROCESS AIR CONTROL
Treatment Plant
Whittier Narrows WRP
San Jose Creek WRP
Pomona WRP
Los Coyotes WRP
Long Beach WRP
Dist. 26 WRP
Dist. 32 WRP
Average
Flow MOD
12
31
8.5
8.5
9.1
3.3
1.3
Air to
Flow Rate
X
X
D.O.
Probe
X
X
Cam
Programmer
Timers
X
X
tion system must be controlled while still preventing closure of
the inlet guide vanes beyond the point that would cause com-
pressor surge. This may well represent a potential application
for a mini-computer.
Waste Activated Sludge Control
Control of mean cell residence time requires wasting of a
predetermined mass of activated sludge each day. Two alter-
nate control schemes haye been proposed to accomplish solids
wasting, either wasting directly from the mixed liquor or wast-
ing from the return sludge line coming from the secondary sedi-
mentation tanks. The latter method has been in dependable
use at CSD plants. Propeller meters are used for waste acti-
vated sludge measurement. The propeller meter has proven to
be cheaper than other types of flow measurement devices and
is relatively maintenance-free, provided that effective primary
sedimentation tanks precede the secondary system and that no
solids recycle from aerobic or anaerobic digestion is discharged
into the secondary system. Dependable 24-hour composite
sampling is also mandatory, but such sampling should be pro-
vided regardless of the wasting system.
In the CSD design, a pulse signal generated by the propeller
meter is sent to a pulse-to-current converter. Converter output
is a 4 to 20 ma signal, proportional to flow rate. This signal is
compared with a set point and used to control the position of
a motorized throttling valve. The signal also actuates a re-
corder, flow totalizer and a no-flow alarm. Thus, if something
should accidentally stop the propeller, an alarm is immediately
sounded, alerting the operator.
The technique of wasting a constant flow rate from the re-
turn sludge line as set by the operator each day has been used
in LACSD installations for many years. The control system
functions with a minimum of maintenance. However, the main
test of any sludge wasting technique is whether it actually
wastes the correct mass of cells and can maintain desired
MCRT values. Years of operational experience with daily in-
ventories of treatment plant solids indicates that the wasting
technique and associated control equipment function
properly.
Primary Sludge Pump Control
When primary sludge from he Districts' inland renovation
plants is wasted to the existing trunk sewer, precise control is
unnecessary. The only critical observation is to assure that no
sludge blanket accumulates in the tank. Conversely, in the
downstream primary plant (JWPCP) where solids must be pro-
cessed, the maximum attainable sludge concentration is desir-
able to minimize the volume of sludge to be treated.
The Districts utilize radioactive density meters to monitor
and control raw sludge pumping, followed by a control system
which distributes incremental amounts of sludge to each di-
gester until a preset maximum amount is fed to each operating
cell. Seed sludge can be circulated to any digester if required,
but routinely, a completely mixed cell does not require seed
87
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
sludge. All controls are automatic with an operator decision re-
quired at the end of each 24-hour period as to the quantity to
be fed to each of the 31 active digesters currently in operation.
Loading decision is based upon volatile acid concentration,
pH, gas production and previous loading history with some
minor consideration given to alkalinity and gas composition.
Sludge concentration and quantity of flow are critical mea-
surements. Flow measurement is made by two Venturi-type in-
struments in series, one of which is redundant and used as a
check on meter calibration.
Further automation of this process awaits instrumentation
capable of continuous measurement of the critical control
parameters.
Chlorination Control
Control systems in current use at LACSD secondary plants
provide for rate control of chlorine feed, paced by plant flow
and fine-tuned by a chlorine residual feedback. Either free or
combined residual chlorine can be used as the feedback signal.
Liquid chlorine is stored on site and fed through evaporators with
the resultant solution discharged through mixers to the final ef-
fluent. These systems are dependable and easily maintained.
At the primary plant (JWPCP), where chlorination is not
routinely required for deep ocean discharge, liquid chlorine is
mixed with slaked lime, Ca(OH)2, to produce a hypochlorite
solution. When chlorination is required, dosage of about 25 to
30 ppm into an average flow of 350 MGD with peaks up to
550 MGD requires massive amounts of chlorine. The hypo-
chlorite solution is fed into the suction side of effluent pumps
to provide rapid mixing in the pump and downstream
manifold.
Control by chlorine residual is not particularly effective at
present because of the return of centrate from the sludge pro-
cessing plant. Following completion of current construction of
additional solids processing equipment, the centrate will be re-
moved from the plant effluent. This, together with polymer
addition to the primaries, will result in an improved effluent
quality. Control will then be more reliable.
Return Shidge Pump Controls
Control of sludge return rates from the final clarifiers is
normally only critical when a treatment plant approaches de-
sign capacity. Under these conditions, excessively high return
rates can reduce mixed-liquor aeration time below desirable
levels and decrease detention time in the final sedimentation
tanks. Also, during periods of sludge bulking, return rate must
be adjustable to prevent solids accumulation in the final clarifi-
ers with consequent overflow into the plant effluent.
The most effective design utilized by the LACSD is a
blanket detector unit which controls either direct pumping
rate from the sludge hopper or the modulation of a valve con-
trolling withdrawal to a sludge sump. Relatively simple con-
trols can then be used to vary pumping rate from the sump to
maintain a relatively constant water surface elevation in the
sump.
A blanket detector control system has the operating advant-
age of increasing return rate as mixed liquor flow increases and
decreasing the return rate as mixed liquor and plant flow de-
crease.
Manual override of a return sludge system can also be useful
if return rate is to be reduced during periods of low plant flow
to allow maximum compaction of sludge for wasting purposes.
This mode has not been utilized at LACSD plants, however.
since return of activated sludge to the aeration system as
promptly as possible has been a prime goal.
EXISTING USAGE OF COMPUTERS AT CSD PLANTS
Computer usage by the Districts can be divided into two
categories. The first is the classic data processing function,
which includes normal accounting activities, scientific process-
ing and data base systems. The second category of computer
usage by the Districts is the use of computers to respond in a
predefined manner to "real time" signals from remote sensing
devices. These computers are commonly called process control
computers because, in many cases, they actually control the
processing function in a manufacturing plant.
The classical data processing activities are carried out on an
IBM System 370 Model 125 located at the Joint Administra-
tion Office's computer facilities. The IBM 370/125 is a gen-
eral-purpose business and scientific computer. It has multi-
processing capabilities for the running of five jobs concur-
rently and for spooling all input and output for unit record de-
vices to direct-access disk storage. The 370/125 has the capa-
bility, although it is not implemented, for communication and
status inquiry, and control of a MODCOMP mini-computer at
the Districts' primary plant.
There are five major areas of data processing activity at the
computer facilities in the Joint Administration Office.
The first area of activity is batch scientific data processing.
To meet the needs of the Districts' engineering departments,
facilities for the batch submission and execution of scientific
programs have been provided. The programs written either in
the FORTRAN or PL/1 languages are submitted on punched
cards at the data center in the Joint Administrative Office.
Typically, applications consist of problems such as the solu-
tion of traverses, large-scale statistical analyses, and simula-
tions for planning purposes.
A second area of activity is the processing of data from the
batch accounting and personnel departments. Information sub-
mitted by the Accounting Department, Personnel Department,
District landfill sites, etc., is keypunched and submitted at the
data center in the Joint Administration Office. Applications
typically consist of projects such as the Refuse Disposal Billing
System which handles billing for a system handling in excess
of 15,000 tons per day of solid waste.
In addition to the batch processing, there are three areas of
tele-processing activity at the data center.
The Districts' major teleprocessing project is the Water
Renovation Plant Data Base system. The seven larger second-
88
-------
COMPUTER APPLICATIONS IN AUTOMATION
ary sewage treatment facilities located throughout Los Angeles
County dial into the data center once daily using teletypes.
Each plant enters about 100 operational parameters into the
370/125. A series of interactive programs, using the para-
meters supplied by the operator, then sends a detailed re-
sponse to the plant indicating parameters to be used for the
next day's plant operation. This series of programs also pro-
duces daily information as to whether the plant is exceeding
any of the standards of the Water Quality Control Board. If
the plant exceeds these standards, the plant is informed by the
program so that immediate action may be taken. Additionally,
the operational status (total flows, etc.) and Water Quality
Control Board compliance information for all of the plants can
be retrieved from terminals at the Joint Administration Office.
In completing water quality compliance and operational para-
meters, the interactive programs use plant readings for a 30-
day period which are stored by on-line direct-access disk de-
vices. Once monthly, these stored records are used as input for
batch programs which compute a Monthly Operational Status
Report showing long-term trends. A typical report is shown in
Appendix 1.
A second teleprocessing activity at the Districts is the In-
dustrial Waste Project. On-line data inquiry and updating of in-
dustrial waste records for industrial waste surcharge processing
is done by a transaction-oriented data communication/data
base management system (CIVS/VS). The 90 million character
data base consists of sewage flow data, surcharge information,
county assessment figures, and internal accounting informa-
tion for 60.000 industrial properties in Los Angeles County.
Companies with significant waste discharges are charged based
on discharge characteristics and amounts in compliance with
state and federal law. CRT terminals used in this data base/
data communication system are located in the Industrial Waste
Department and the data center at the JAO.
The final major area of activity under development at the
JAO data center is the On-Line Personnel System. On-line in-
quiry and updating of personnel information is done from
CRT's at the JAO Personnel Department. Hard-copy audit
trails of these transactions are made concurrently at the data
center.
The second category of computer usage, the use of comput-
ers to respond to signals from remote sensing devices, is
limited to an alarm monitoring and logging systemat the Joint
Water Pollution Control Plant. The Districts have not yet
found it feasible to install process control computers in sec-
ondary treatment plants. However, the Water Renovation
Plant Data Management System, as previously described,
should allow the Districts to have the process control func-
tions of a secondary treatment plant well defined before instal-
lation of a computerized system.
The Districts have installed a process control computer, a
MODCOMP 11/25, at the Joint Water Pollution Control Plant
(JWPCP). The existing system is primarily an alarm monitoring
and display system. It consists of a central alarm panel and a
data logging system for both remote pumping plant alarm
status and inplant (JWPCP) alarm status.
There are 34 remote pumping plants located at various sites
around the Los Angeles County area. All but one plant are
connected by leased telephone lines to the JWPCP alarm
facility. The most remote pumping plant is periodically con-
nected to the alarm facility by use of an automatic dialer. The
conditions presently being monitored at the pumping plant are
(1) high water level in wet well, (2) high water level in dry
well, (3) power failure, (4) pump control failures, (5) com-
munication failure, (6) A/C power failure to telemetry device.
The inplant alarm system can be broken down into two
types of subsystems. The first is the Methane Detection Sub-
system, which continuously monitors the level of methane gas
at 6 points within the enclosed pipe galleries. The meters are
directly wired to display devices on the central alarm panel.
An alarm is activated if a certain level is exceeded.
The second subsystem is the in-plant telemetry system and is
responsible for the majority of the alarm data transmission.
There are 654 contact-closure type alarm points scattered
throughout the JWPCP. All of these alarms are assigned to and
wired into an existing local annunciator panel at some local-
ized area, i.e., Chlorination Station, Sludge Dewatering Sta-
tion, etc. There are some 14 local annunciator panels through-
out the JWPCP.
The Central Alarm Facility can be best described by tracing
an alarm as it travels through the Facility. Suppose, for ex-
ample, a contact closure occurs at the chlorination facility.
The signal is transmitted to the local annunciator panel where
a window is permanently devoted to displaying the condition
of its alarm point. A hardwired system (in reality a combina-
tion of hardware and a read-only memory computer) polls for
alarms. This polling system passes the alarm to the MODCOMP
computer which in turn lights the proper display window and
logs the alarm on a printer. If the MODCOMP computer is in-
operative, the window will still be lighted by the hardwired
system, but the individual alarm will not be logged. The alarm
windows at the Central Alarm Facility respond to the local an-
nunciator panels within the JWPCP and not to individual
alarms.
The MODCOMP system also has provisions for the employ-
ment of analog monitoring and control functions, depending
on future requirements.
POTENTIAL APPLICATION FOR INSTRUMENTATION
AND COMPUTERS
The full potential of the mini-computer at the JWPCP has
not been explored since the initial decision to purchase the
unit was based upon the unit being less expensive than an
equivalent electronic system using conventional relays and
control wiring, for primary usage as a central alarm station. In-
creased familiarity with the unit should result in reduced fre-
quency of attendance by pumping plant operators to each in-
stallation and the consolidation of surveillance duties by plant
89
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
operators. Improved procedures should be possible for control
of the chlorination system, inlet and effluent pumping, sludge
transfer pumps and the control of the solids processing
centrifuges.
The IBM 370/125 was initially installed at the Joint Admin-
istration Office in the first week in August 1974 to replace a
Univac 9300 plus a number of time-sharing terminals in the
Engineering Department. As promptly as advisable, all separate
functions will be transferred to the new unit. The IBM
370/125 is capable of storing and manipulating vast quantities
of data which would be impossible to collate without the
machine.
As newer applications are tested and proven, CSD wiD have
the hardware and personnel needed to make the necessary
changes. The industrial waste surcharge ordinance will vastly
increase the work involved in computing, billing, and recording
collection of the surcharges from industries using the system.
Coincidental to this effort to charge industries individually for
services rendered, data logging of information from a much
more exacting source control program must be initiated, to
provide information on potential sources of toxic discharges
into the system. Both the Federal Government and the State
of California have instituted limitations on the quantity of
toxic waste material discharged into the sewerage system.
Sludge discharges of industrial waste which upset the treat-
ment processes must be controlled or the Districts will be sub-
ject to fines for non-compliance.
With the advent of unionization of District employees,
ready access must be provided for personnel records. Union
negotiations on wages and fringe benefits result in extremely
complicated settlements which make it more difficult to arrive
at valid comparisons of prevailing wage levels. Computer data
logging for quick information retrieval is urgently needed to
evaluate the status of over 230 class specifications for the
multiplicity of District jobs.
A desk-top study recently completed indicates that serious
evaluation should be given to provision of flow equalization at
the Districts' secondary treatment plants. Certainly, every indi-
cation points towards more complicated, expensive treatment
requirements with far less tolerance of deviation from normal
conditions. The question then must be posed as to whether the
cost of storage and pumping of peak flows is less expensive
than the provision of adequate process equipment and tankage
to handle normal peak flows. Tertiary treatment processes are
almost as expensive as secondary treatment in many cases, and
if consideration to storage of secondary effluent shows it to be
cost-effective, it would be even more economical to store pri-
mary effluent to provide for constant hydraulic flow through
the secondary and tertiary process.
The engineering problems involved in storing primary efflu-
ent are not difficult to resolve and very possibly computer
control of storage capacity versus inflow could assure maxi-
mum utilization of facilities while providing constant hydrau-
lic loading and, to some extent, equalization of organic load.
Primary sedimentation prior to storage could preclude any
solids handling problem. Covered storage with air under the
covers going to aeration blowers could resolve any odor prob-
lems. Modest mixing and aeration could prevent septicity from
increasing air requirements. Computer control of gates and
pumping rates could minimize operator time demands and re-
sult in very effective use of facilities.
A saving in capital cost of 10% to 12% appears to be a con-
servative estimate of cost benefit without regard to rather ob-
vious operator advantages in elimination of peak flow through
CSD treatment plants. The time has probably arrived for im-
plementation of this idea. District experience at the Whittier
Narrows WRP with a reasonably constant hydraulic flow
through the plant has already indicated that a plant originally
designed for 10 MGD can handle 12 MGD even while fully ni-
trifying. Without nitrification the same plant can handle 14 to
18 MGD, depending on sewage temperatures.
CONCLUSION
Increased automation and application of computer tech-
nology in District treatment plants will be implemented by
District management whenever an improvement in operating
efficiency can be demonstrated.
Application of computer technology to management of
new and complex industrial waste regulations is necessary to
implement source control of toxic industrial wastes and to ad-
minister the Districts' new industrial waste surcharge.
Exacting requirements for improved effluent quality and
the high cost of capital improvements now makes considera-
tion of flow equalization in secondary plants a viable alternate
to standard designs which must provide adequate capacity for
peak discharges. Storage of primary effluent to equalize hy-
draulic loading on secondary and tertiary processes could be
controlled by a mini-computer so that effective use of storage
facilities is assured.
90
-------
APPENDIX 1. MONTHLY OPERATIONAL STATUS REPORT.
WHITTIER NARROWS WATER RECLAMATION PLANT
COUNTY SANITATION DISTRICTS
OF LOS ANGELES COUNTY, CALIF.
JOHN D. PARKHURST - CHIEF ENGINEER 6 -GENERAL MANAGrR
*** SUMMARY OF OPERATIONS ***
JUNE 1974
DATE
I
i
J
H
3
o
7
0
V
10
11
U
13
It
1,-j
16
17
Id
Iv
20
21
22
2j
2*
23
2b
27
26
2->
30
MLn.'J
PLANT r.HARCTERISTICS
FLOW
EFFLUENT
MGD
1
11.98
12.02
12. Ob
11.61
11.81
12.40
12.05
11.83
12.07
11.95
11.79
11.73
11.63
11.99
12. CO
11.99
11. bt
12.27
12.15
11.70
11.74
11. b4
11.63
11.43
10.41
12.02
12.22
12.03
12.CE
11.39
11.66
TOTAL
RETURN
ACT
SLUDGE
MGD
2
6.08
6.03
6.05
6.10
5.96
6.00
6.08
5.94
5.85
5.87
5.97
5.93
5.91
5.83
5.70
5.73
5.«3
5.86
5.71
5.55
5.55
5.40
5.36
5.44
5.76
5.51
5.46
5.41
5.53
5.33
5.76
WASTE
ACT
SLUDGE
i-iGD
3
.171
.185
.162
.190
.171
.161
.144
.139
.133
.111
.122
.111
.132
.149
.145
.144
.152
.168
.189
.1V3
.167
.172
.172
.151
.150
.14?
.143
.143
.143
.138
.153
AIR
VCF/DAY
4
29.9
29.9
29.8
29.8
29.1
29.0
29.1
29.1
29.2
29.2
29.2
2 B. 6
2H.4
28.7
28.8
29.1
2<;.i
29.4
2V. 4
2e.6
30.2
29.9
29.6.
29. e
29.5
29.7
29.1
29.5
29.5
29.5
29. ?<
SUSPENDED SOLIDS
RAW
SEWAGE
MG/L
5
408
362
414
364
274
434
418
302
366
414
472
428
M4
396
380
412
394
5':>6
453
476
400
394
434
424
450
466
302
5f-3
403
414.2
PRIMARY
EFFLUENT
MG/L
6
98
74
106
118
102
104
7B
79
106
106
90
86
94
86
88
130
116
148
222
14
96
78
98
96
96
114
114
96
f!2
102.9
RETURN
SLUDGE
MG/L
7
6396
6161
5323
5267
5627
5582
5523
5240
5277
5664
5883
5714
5901
6020
59feO
6126
6434
6798
7372
6871
6690
6355
4691
6579
6218
6383
6322
6206
6236
6036
SEC.
EFFLUENT
MG/L
8
4
4
5
4
7
7
7-
4
5
7
4
6
6
5
5
5
8
5
7
6
6
4
4
5
4
4
4
6
6
4
5.2
; .
CHLOR'Nr
CONTACT
EFFLUENT
MG/L
9
COD
RAW
SEWAGr
TOTAL
MG/L
10
473
716
442
553
646
556
642
568
973
580
549
699
616
PRIMARY
EFFLUENT
TOTAL
MG/L
: i
21'.-
269
239
230
254
224
725
351
414
21-r
234
201
255.8
SECONDARY
EFFLUENT
TOTAL
MG/L
12
22
27
41
29
27
31
26
26
30
27
28
23
28.1
SOL.
MG/L
13
20
23
31
22
25
22
20
23
25
22
24
17
22.8
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
APPENDIX 1. MONTHLY OPERATIONAL STATUS REPORT-(CONTINUED)
WHITTIER NARROWS WATER RECLAMATION PLANT
COUNTY SANITATION DISTRICTS
OF LOS ANGELES COUNTY* CALIF.
JOHN 0. PARKHURST - CHIEF ENGINEER 6 GENERAL MANAGrR
*** SUMMARY OF OPERATIONS ***
JUNE 1974
DATE
1
2
3
A
5
6
7
6
9
10
11
12
13
1-
15
16
17
18
19
20
21
22
23
24
25
26
27
26
29
30
MEAN
PLANT CHARCTERISTICS
NITROGEN
PRIVARY
EFFLUENT
CRG.
MG/L
14
NH3
P.G/L .
15
20
20
19
22
21
21
22
22
20
21
21
22
22
21
19
21
22
22
22
25
18
20
19
20
20
20
20
20
20
20
20.7
SECONDARY
EFFLUENT
NH3
MG/L
16
.1
.2
.2
1.2
.2
. 3
1.1
1.0
.2
.3
.4
.3
.6
.4
.6
.2
1.3
* A
1.0
2.4
.6
.2
.1
. 1
. 1
.3
.6
.4
.'1
. 2
. 5
N02
MG/L
7
.02
.Cl
.01
.04
.02
.03
.06
.02
.02
.02
.04
.04
.04
.05
.04
.04
.09
.04
.07
.06
.04
.02
.02
.02
.03
.03
.03
.03
.02
.02
.03
N03
KG/L
18
18.00
13.50
21.00
19.00
17.50
20.00
20.00
18.5O
19.00
18.00
17.00
20.00
19.00
17.00
16.50
16.50
15.50
10.50
9.50
e.50
13.00
12.00
17.50
16.00
16.00
15.00
12.50
15.00
15.50
15.30
16.24
TOS
ELEC.
CCND. '
jui » /*onwu^
Hi V.KU" rl
/CK
19
M/" /\
nvj/ L
20
683
58'
650
645
593
606
627
577
605
627
567
594
600
633
551
637
566
617
588
651
638
588
654
527
66*
666
703
618
562
6.4.9
92
-------
APPENDIX 1. MONTHLY OPERATIONAL STATUS REPORT-(CONTINUED)
WHITTIER NARROWS WATER RECLAMATION PLANT
COUNTY SANITATION DISTRICTS
OF LOS ANGELES COUNTY, CALIF.
JOHN 0. PARKHURST - CHIEF ENGINEER 6 GENERAL MANAGER
*** SUMMARY CF OPERATIONS ***
1974
DATE
1
<:
3
4
3
6
7
6
y
lo
It
U .
ii
It
la
lb
17
U
it
2o
21
& £.
2s
2t
23
2o
27
2a
2*
3o
.trtN
PLANT CHARCTLRISTICS
EFFLUENT CHAFACTER I ST I CS
SECChI
01SC
DEPTH
FEET
21
9.0
9.0
9.5
9.5
9.0
8.0
6.0
6.0
7.0
7.5
7.0
6.5
7.5
6.0
6.5
7.0
6.5
6.0
7.5
7.5
7.5
7.5
9.0
9.5
9.0
10.0
9.C
e.o
7.0
9.C
7. a
EFFLUENT
Tt>.f».
FAREM.
22
73.0
73.0
74.0
74. C
74.0
74.0
74.0
75.0
74.0
74.0
74.0
74. C
74.0
75.0
74. &
74.0
74. u
74.0
75.0
74.0
74.0
75. C
75.0
75.0
75.0
76.0
77.0
77.0
75.0
74.5
CL2
RESIDUAL
EFFLUcNT
GSAu
KG/L
23
.90
.72
.18
.54
2.40
I.d5
1.90
1.56
0.
.45
.24
.42
.45
2.
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
APPENDIX 1. MONTHLY OPERATIONAL STATUS REPORT-(CONTINUED)
WHITTIER NARROWS WATER RECLAMATION PLANT
COUNTY SANITATION DISTRICTS
OF LOS ANGELES COUNTY, CALIF.
JOHN 0. PARKHURST - CHIEF ENGINEER 6 GENERAL MANAGER
*** SUMMARY OF OPERATIONS ***
JUNE 1974
Date
20, 21
18
All
DATE
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
13
19
?0
21
22
23
24
25
76
27
28
29
30
«rAN
MISCELLANEOUS
34
35
36
37
C.
0.
0.
C.
0.
.01
0.
0.
0.
0.
0.
0.
0.
.02
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
.001
Column
6, 15
8, 12, 13, 16,
17, 18, 26, 37
37
NOTES
Remarks
Primary effluent sampler malfunction
Secondary effluent sampler malfunction
Hexavalent chromium, secondary
effluent, 24-hr composite (mg/L)
94
-------
COMPUTER APPLICATIONS IN AUTOMATION
APPENDIX 1. MONTHLY OPERATIONAL STATUS REPORT-(CONTINUED)
WHITTIER NARROWS WATER REC1 AV.ATION PLANT
COUNTY SANITATION DISTRICTS
OF LOS ANGELES COUNTY, CALIF.
JO"N 0. PARKHURST - CHIEF ENGINEER 6 GENERAL MANAGER
*** SUMMARY OF OPERATIONS ***
JUNE 1974
DATt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
23
29
30
MEAN
AERATION SYSTEM NO. 1
SUSPENDED SOLIDS
TANK
1
MG/L
38
2626
2530
2443
2296
2461
2360
2299
2021
2532
2163
2429
2401
2416
2313
2212
2397
2273
2606
2605
2R3?
2530
2C75
2225
2251
2C9b
23^2
2106
25C3
2278
2373
TANK
2
MG/L
39
1943
1860
1B27
1647
1732
1737
1699
1648
1710
1603
1870
1871
1730
1694
1770
1834
2054
2182
1937
Ib92
1P37
1742
1748
169C
1"04
1844
1926
15)83
1780
ieie
TANK
3
MG/L .
40
1830
1788
1671
1664
1501
1659
1543
1560
1572
1500
1762
1912
1622
1570
1690
1734
1860
2027
1664
1912
1858
1668
1667
1738
1 7 1 :l
1712
1846
1827
1732
1711
TANK
^
Mo/L
41
VOLATILE
SOLIDS
*
V2
75.0
74.0
74.0
75.0
75.0
77.0
7^.o
76.0
7A.O
76.0
77.0
73.0
75.0
74.0
74.0
7^.0
77.0
77.0
76.0
77. C
78.0
77.0
75.0
77.0
77.0
78.0
77.0
77.0
77.0
76.0
7t=..8
RETURN
ACT.
SLUDGE
MGD
43
6.1
6.0
6.1
6.1
6.0
6.0
6.1
5.9.
5.9
5.9
6.0
5.9
5.9
5.8
5.7
5.7
5.9
5.9
5.7
5.6
5.6
5.5
5.4
5.4
5.8
5.5
5.5
5.4
5.5
5.3
5.76
M^XED
L I-^UOP
D.O.
MG/L
MAX
44
WIN
45
RETURN
SLUDGE
AERATION
VOLUME
MG
4(
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.550
.55J
.550
.550
.550
.550
.550
.550
.550
95
-------
APPENDIX 1. MONTHLY OPERATIONAL STATUS REPORT-(CONTINUED)
|
WHITTIER NARROWS WATER RECLAMATION
COUNTY SANITATION DISTRICTS
OF LOS ANGELES COUNTY. CALIF.
JOHN D. PARKHURST - CHIEF ENGINEER 6 GENERAL MANAGPR
*** SUMMAKY OF OPERATIONS ***
JUNE 1974
o
o
n
S
DATt
1
2
3
4
5
b
7
6
9
10
11
12 .
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
26
29
30
?-;EAN
AERATION SYSTEM MO. 1
LOADING PATTERN
TANK
t Ml"i\
I
%
47
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
66
6b
66
65.7
TANK
2
ft
48
34
34
34
34
34
3
-------
APPENDIX I. MONTHLY OPERATIONAL STATUS REPORT-(CONTINUED)
WHITTIE* NARROWS WATER RECLAMATION PLANT
COUNTY SANITATION DISTRICTS
OF LOS ANGCLES CCUNfYt CALIF.
JOHN D. PARKHURST - CHIEF ENGINEER 6 GENERAL MANAGER
*** SUMMARY CF OPERATIONS ***
JUNE 1974
OATt
1
2
3
4
5
fa
7
6
9
10
11 '
12
13
1*
15
16
17
13
19
20
21
22
23
24
25
26
27
28
29
30
tiEArJ
KINETIC PARAMETEkS
AIK PATFS
CUdIC
FEET/
CjMLLCr.
EFFLUENT
9(3
2.50
2.49
2.47
2.57
2.4o
2.34
2.41
2.46
2.4,;
2.44
2.4o
2.44
2.40
2.40
2.4Q
2.43
2.4o
2.39
2.42
2.46
2.57
2.53
2.56
2.CJ
2.6 c. i* i w v c. LV
99
1537
1251
1349
1394
1296
1423
1419
675
758
1600
1359
1689
1329
COD
PENOVAL
«
100
90.7
91.-+
87.0
90. 4
90.2
90.2
91.1
93.4
94.0
89.7
89.7
91,5
90. a
AERATION
SYSTEM
LOAD
(LBS.)
101
21453
26047
24716
23153
24975
22100
22499
35918
40397
20757.
23848
19094
25413
RETURN SLUDGE
AERATION TIMES
AERATION
SYSTEM
1
(H,%S.)
102
2.17
2.19
2.18
2.16
2.21
2.20
2.17
2.22
2.26
2.25
2.21
2.23
2.23
2.26
2.32
2.30
2.24
2.25
2.31
2.38
2.33
2.40
2.46
2.43
2.2S
2.40
2.42
2.44
2.39
2.48
2.29
AERATION
SYSTEM
2
(HRS. )
103
AERATION
SYSTEM
3
(MRS.)
104
MIXED LIQUOR
AERATION TI'-'ES
AEPATION
SYSTEM
r
(MRS.)
103
3.26
3.26
3.24
3.32
3.31
3.20
3.24
3.31
3.28
3.30
3.31
3.33
3.31
3.30
3.32
3.32
3.31
3.24
3.29
3.41
3.40
3.39
3.46
3.48
3.63
3.35
3.33
3.37
3.34
3.52
3.34
AERATION
SYSTEM
2
(HRS. )
106
AERATION
SYSTEM
3
(HRS.)
107
TOTAL AERATION
SOLIDS
AERATION
SYSTEM
1
(LBS.)
103
53384
31691
49548
48430
47488
48005
46254
43610
484K9
43V18
50549
51575
48122
4651:
47304
49748
51600
56837
53426
55344
51916
45912
470T8
47363
46ti71
49023
49023
51816
48289
49279
AERATION
SYSTEM
2
(LBS.)
109
AERATION
SYSTEM
3
(LBS.)
110
co
-------
APPENDIX 1. MONTHLY OPERATIONAL STATUS REPORT-(CONTINUED)
H'HITTIER NARROWS WATER RECLAMATION PLANT
COUNTY SANITATION DISTRICTS
OF LOS ANGELES COUNTY, CALIF.
JOHN 0. PARKHURST - CHIEF ENGINEER 6 GENERAL MANAGER
*** SUMMARY OF OPERATIONS ***
JUNE 1974
5
5
o
DATE
1
2
3
4
5
6
7
0
9
10
11
12
13
1*
lf>
16
17
IB
1*
20
£1
22
23
24
25
26
27
23
29
30
MEAN
KINETIC PARAMETERS
MIXED LICUOR
SL'SPCMGFtj SOLIDS
wwJ~ta!\U/t,L/ J wl» 1 I/ w
H * r
AERATION
SYSTEM
1
LBS.
*» » «
111
41447
4C199
38451
33001
36309
37285
35«10
34430
36V87
34C93
3VS15
40669
37138
36005
37257
3efcfcO
4l01
AERATION
SYSTEM
2
LIS.
112
AERATION
SYSTEM
3
LBS.
113
COP LOADING
L?S.COD/
LtiS.
TPVSf.X
DAY
114
.42
.53
.52
.54
.56
.42
.44
.?8
.76
.45
.43
.39
.519
LES.COD/
LBS-
MLVS5/
DAY
115
.72
.90
.88
.91
.95
.72
.62
1.13
1.29
,76
.80
.66
.88
DAILY
ru i i
V.C.LL
Rt:S.
TH'.E
(DAYS)
1J.6
7.4
6.9
7.4
7.8
7.5
8.6
8.9
9.2
11.3
9.4
11.0
10.2
Q.I
7.0
8.2
7.7
7.2
6.6
5.5
7.2
6.8
6.5
9.6
7.4
8.1
e.i
8.2
8.6
r. .5
e.i
SOLIDS BALAMCE
TOTAL
PtANT
SUSPEND.
SOLIDS
(IPS.)
117
70509
66422
65184
64C01
61534
63529
60739
58208
63199
57955
67037
69466
63300
61203
63119
65974
69004
75805
68997
73^36
69303
61707
62637
63626
62947
65043
66296
68V13
64496
65358
WASTTD
SOL ITS
(LBS.)
118
91?2
9506
6435
7511
7556
6703
6403
5612
4(3^5
5763
5446
62°0
73?3
72PO
71P2
7766
9015
107)5
116C-.6
9570
9SP4
9116
5908
8230
7364
7612
75^-0
74C1
7177
7738
SE".
EFFL"ENT
SUSPEND.
SOLTOS
-------
COMPUTER APPLICATIONS IN AUTOMATION
COMPUTER APPLICATIONS IN AUTOMATION
William E. Dobbins
President, Teetor-Dobbins, P.C., 515 Johnson Avenue, Bohemia, NY 11716
INTRODUCTION
The computer, which has been widely utilized and accepted
in power plants and industrial process control, is now being ap-
plied to wastewater treatment. Most of the initial applications
have been for data logging systems, in which the plant variables
are monitored and the data manipulated so as to present it in
forms and units of greatest use to the operator. The data can
also be stored on magnetic tapes from which it can be re-
trieved for future research. This is a great improvement over
the old systems in which limited historical data can be re-
trieved, but only by being laboriously dug out of numerous re-
corder charts.
While data logging and processing constitutes an important
step forward, the most important advance will be the applica-
tion of the computer to plant operation. The computerized
system has an important advantage over the system of numer-
ous independent analog control loops; control actions can be
based on algorithms which can utilize any or all of the varia-
bles. Thus, it can utilize calculated variables in arriving at deci-
sions regarding process adjustments. The control capability of
the computer goes far beyond that of the human operator,
who, surrounded by dozens of meter indicators and recorders,
cannot possibly react in time to prevent process upsets. The al-
gorithms can be changed and tuned as fundamental process
knowledge and knowledge of the peculiarities of the particular
plant become available.
Two examples of computer algorithms for plant operation
are presented in this paper. Both were developed at the two
highly automated activated sludge plants in Bridgeport, Con-
necticut. One describes a program for the operation of the
main raw sewage pumps and the other the results of a program
which attempts to measure the respiration rates in the aeration
tanks and the BOD of the sewage delivered to them.
PUMP CONTROL PROGRAM
At Bridgeport, two existing primary treatment plants were
modernized and expanded by the provision of secondary treat-
ment by the step-aeration version of the activated sludge pro-
cess. At each plant the sewage is delivered through a large in-
terceptor with its invert about 20 feet below the ground sur-
face. It flows through a bar screen and a trapezoidal-shaped
grit chamber to a wetwell, from which it is pumped to the pri-
mary treatment tanks. The pumping is done by three constant-
speed and two 3-step-speed pumps. Both collection systems
are combined sanitary and storm systems. At the West Side
Plant the flow varies from about 12 mgd to 60 mgd;the East
Side Plant flows are generally about one-half as much. Except
for the magnitudes of the flows, the following discussion ap-
plies to both plants.
No control section was provided for the grit chamber, with
the result that the depth of flow is determined by the level in
the wetwell. In the old plant, the pump selections were acti-
vated by a float-operated control system by which the pump-
ing rate at any given wetwell level was selected so as to provide
an acceptable grit chamber velocity. However, a great deal of
hunting took place, particularly for some of the individual
pumps. Figure 1 shows the operating record in a typical day of
one of the pumps. In such a control system, the combination
of pumps running at any particular wetwell level is always the
same, and the combination is changed when the level changes
by some adjustable increment of about one foot. The basic rea-
son for the excessive hunting is that the system cannot take
the time frame into account. At some times it may require an
hour for the incremental level change to occur; at other times
it may take only five minutes. The computer program resolves
this by continuously calculating the inflow rate and selecting
pump combinations to pace this calculated rate. The algorithm
is very simple:
Inflow rate = Outflow rate + Rate of change of storage
The computer calculates this rate every two minutes, from its
knowledge of the present pumping rate and the relationship
between wetwell storage and water level. From the tables of
flow rates delivered by various pump combinations, which are
stored in its memory, it selects the combinations which most
closely produce the calculated inflow rates. Of course, it is not
quite as simple as this. The program also maintains, within a
designated band, the relationship between water level and inflow
rate which results in an acceptable velocity in the grit
chamber. One of the benefits of this program is that it auto-
matically forces the inceptor to back up during storm flows,
and thus make use of the storage in the collection system to
moderate the inflow rates. That the program produces a very
smooth flow pattern is demonstrated by Figure 2. This is a
typical chart for the Parshall flume meter which measures the
total pumping rate.
RESPIRATION RATE PROGRAM
A knowledge of the strength of the sewage being fed to the
secondary system, and the rate of oxygen utilization in the
aeration tanks, are essential to the implementation of various
control strategies which are currently being studied by numer-
ous investigators. Much of the attention has been on the devel-
opment of on-line instruments for the measurement of TOC,
99
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
Figure 1. Pump Chart Showing Considerable Hunting
(Single Pump)
100
-------
COMPUTER APPLICATIONS IN AUTOMATION
NOON
Figure 2. Chart of ParshaH Flume Meter Showing Smooth Flow Pattern
101
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
COD and other parameters which might correlate with BOD.
An attempt was made at the Bridgeport West Side Plant to de-
velop a computer program which would utilize the operating
condition of the aeration tank itself to evaluate the sewage
strength and the current respiration rate.
The installation consists of three aeration tanks, each equip-
ped with its own final settling tank, sludge return and sludge
wasting facilities. Thus, it is possible to operate each system in-
dependently. A diagram of the Aeration Tanks is shown in Fig-
ure 3. When the program was developed the mode of operation
was to return the sludge to pass No. 1 and to divide the sewage
flow equally between the east and west ends. Thus, each of
passes 2 and 4 received about one-quarter of the total sewage
and pass 3 about one-half of the flow. The return rate was
about 50% of the average sewage flow.
FROM PRIMARY TANKS
TO FINAL TANK N° I
TO FINAL TANK N° 2
TO FINAL TANK N°3
Propeller
Flowmeter
Figure 3. Layout of the Aeration Tanks
102
-------
COMPUTER APPLICATIONS IN AUTOMATION
Program Logic
The program logic is based on the equations expressing the
mass balance of oxygen in the aeration tank. It was actually
applied to the three passes which received the sewage. The
basic equations are :
A = Rate in by flow = Qs DOI + Qr DOR (1)
B = Rate in by Air = Ka Qa (DOS - DOT) (2)
C = Rate out by flow = (Qs + Qr) DOE (3)
D = Rate of change within tank = ADOT V / At (4)
R = Rate of respiration within tank
In these equations:
Qs = Total flow of sewage, MGD
Qr = Flow of return sludge, MGD
Qa = Flow of air, cfm
DOI = Dissolved oxygen in influent sewage , Ib/MG
DOE = Dissolved oxygen in tank effluent, Ib/MG
DOR = Dissolved oxygen in return sludge. Ib/MG
DOT = Dissolved oxygen in tank (average), Ib/MG
DOS = Dissolved oxygen saturation value. Ib/MG
V = Volume of the 3 passes, MG
At = Time increment, days
Kj = Coefficient relating to oxygenation efficiency
The mass balance relationship is:
R=A + B-C-D Ibs/day (5)
All of the factors in this equation, except R, DOS and Ka,
are fixed quantities or measurable variables. DOS is computed
from the temperature of the water and the known depth of
submergence of the air diffusers. Therefore, by determining
either Kg, or R, the other would be determined. The value of
K,, for the particular operating conditions could really only be
estimated and there would be great uncertainty about it. R
could be subject to measurements on samples of the mixed liq-
uor. Once reliable values were established, the respiration rate
could be calculated periodically.
Field Measurements
During September, 1973. various attempts were made to
measure the respiration rates on samples of mixed liquor by
taking them to the laboratory, aerating them and measuring
the DO values with a portable DO probe. As expected, the
rates varied considerably from place to place in the tank and
from time to time at each place. It was decided that it would
require a great deal of work to accurately determine the pro-
files of respiration values that would be necessary for the eval-
uation of Ka. Therefore, a different approach was taken. The
respiration rate for the tank as a whole might be expressed as:
and effluent sewages, Re is the endogenous respiration and Kb
a coefficient. This equation was proposed some years ago by
Eckenfelder and O'Connor (1). The detention time of the sew-
age in the three passes varied from 2 to 4 hours and averaged
about 3 hours. Therefore, the total weight of added BOD, con-
tributed by the sewage present in the tank at any one time,
could be estimated as some factor times the weight added dur-
ing the previous interval plus a lesser factor times the amount
added during the second previous interval, etc. The use of this
equation introduces two new factors, Kb and Re, to be deter-
mined along with K.,.
During a 24-hour period in July, 1973 an attempt was made
to calibrate the system by the use of equations (5) and (6). DO
probes were placed at the influent sewage distribution
chamber, at the exit end of pass 4 and at four other points
within the tank. Every hour, on the hour, a sample of the in-
fluent sewage was taken and three dilutions were set up for
subsequent determinations. Unfortunately, no effluent
samples were taken, because of the lack of sufficient BOD bot-
tles, as well as sufficient time. The computer was programmed
to log out the values of the sewage flows, air flows and DO
readings every 12 minutes. Study of the data showed that a
good value for At was one hour. This eliminated most of the
noise in the readings and showed smooth patterns of change.
Equations (5) and (6) were combined in the form
BODI =
A+B-C-D+E
(7)
R = Kfe (BODI - BODE) Qs +
(6)
where BODI and BODE are the 5-day BOD's of the influent
in which the term E represented the magnitude of Kb . BODE
. Qs - Re. When the lab BOD data were turned out (5 days
later) all of the data were then available to fit to the equations.
The values of BODI and Qs used in the fitting were the
weighted values over the previous three hours and the other
values were the averages for the previous hour. This provided
21 sets of data which could be fitted to the equations, although
there were only 3 unknowns. The fit was made by the method
of least squares, with the following results:
K,, = 0.0901. Kb = 0.314. E = 55.0
Figure 4 shows the computed BOD values compared to the
measured ones, and Figure 5 shows the computed respiration
rates. It is noted that the BOD values are the 3-hour composite
values, whereas the respiration rate is the average for the previ-
ous hour. Although the effluent BOD's were not determined, the
plant effluent values were generally around 25 mg/1 during
that month. Also, the average mixed-liquor volatile suspended
solids were about 800 mg/1. Assuming that these prevailed dur-
ing the test period, and from the value of E = 55 Ibs/day, the
following is derived:
R = Oxygen used up = 0.314 x BOD5 removed + 0.0357 x
lbsofMLVSS(lb/day)
Although they seem to be somewhat low, the coefficients are
in the range of reported values.
103
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
O)
o.
d>
E
in
i
o
o
CD
V)
O
Q.
O
O
ro
120
100
1600 2000
7/12/73
2400 0400 0800
7/13/73
1200
Time Of Day
Legend
-e computer predicted
A lab. results
Figure 4. Comparison of Computer Calculated BOD With Lab Measured BOD
104
-------
COMPUTER APPLICATIONS IN AUTOMATION
1600 2000
7/12/73
2400
Time
0400
7/13/73
Of Day
0800
1200
Figure 5. Calculated Respiration Rates
DISCUSSION
The computer was programmed using the values previously
derived for the coefficients and it printed out the computed
BOD values on the hourly log. It was done individually for
each of the three aeration tanks, to provide a check on the
values. They all fell within the general range of expected
values. The values for Tanks 1 and 3 agreed quite well and
those for Tank 2 were consistently low, a condition which can
be attributed to faulty meter readings.
The values of the three coefficients are influenced by the
temperature of the sewage and will, therefore, be subject to
change. The temperature of the sewage at Bridgeport does not
usually vary greatly within the day. but is subject to a wide
seasonal variation. A program was written whereby the coeffi-
cients can be updated every three days. The previous equations
can be summed to represent the composite BOD for a 24-hour
period. The resulting equations for this calculated composite
contain the three unknown coefficients. The actual composite
BOD for the same 24-hour period is determined in the lab.
When three sets of data are determined, the lab results are
typed into the computer and the computer then solves the
three simultaneous equations to up-date the three coefficients.
They will obviously have a built-in lag of 5 days. However, be-
cause the seasonal temperature changes are quite gradual, the
errors should not be substantial. This program has not yet
been implemented.
It is believed that the results are quite promising and that
this program can be very useful. It is also believed that it can
be improved in several ways. For example, the effluent BOD
should be incorporated into the program. Also, the air flow
readings should be adjusted to standard conditions from the
readings of air temperature and pressure. The endogenous
respiration rate might be expressed as a factor times the mea-
sured concentration of mixed liquor suspended solids. Addi-
tional work should result in the development of a reliable pro-
gram for the evaluation of this very important calculated
variable.
REFERENCES
1. Eckenfelder, W. W., Jr. and O'Connor, D. J., "Biological Waste
Treatment," Pergamon Press. New York, NY (1961).
105
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
DISCUSSION
Joseph f. Roesler:
The installation of a computer can be considered as upgrad-
ing a wastewater treatment plant. Or computers can be con-
sidered as O&M tools, especially if they are used for data ac-
quisition and handling. I would appreciate comments from the
various EPA regions in regards to the region's policy towards
awarding construction grant money for computers. Secondly, I
would appreciate any one else's comments on what EPA's
policy should be and why?
Michael K. Stenstrom:
From a manager's point of view, how much calibration of
Dissolved Oxygen probes can you afford?
Walter W. Schuk:
To provide some guidelines for instrument manufacturers it
would be helpful if treatment plant management would define
the meaning of reasonable maintenance in more detail. Please
give your definition, including both the frequency and man-
hours that would be considered acceptable for maintenance of
an on-line analyzer.
Allen E. Molvar:
Would Mr. Garrison comment on the expected benefits of
flow equalization, particularly in terms of potential process
improvement?
Heinrich O. Buhr:
The development of a method for calculating the respira-
tion rate in an activated sludge system, as in Equation (5) of
Dr. Dobbins' paper, is an important step towards the goal of
controlling the respiration rate per unit mass of sludge, or
specific oxygen utilization rate (SCOUR). It may be shown,
from the classical substrate and sludge mass balance equations.
that a constant value for SCOUR also means a constant spe-
cific growth rate, constant F/M ratio and a constant value for
soluble BOD in the effluent. Keeping these parameters con-
stant further holds out the promise of more consistent sludge
settling characteristics, and it seems likely that SCOUR control
will become one of the most useful concepts in automatic con-
trol of activated sludge plants.
Once a value for R, Equation (S), is available on a regular
basis, the practical implementation of SCOUR control would
be to calculate the sludge concentration required in the aerat-
ors, and then to control the sludge return rate appropriately.
For example, if primary flow rate, Qs, RAS return rate Qr,
and, say, RAS concentration, X,, can be measured, then the
RAS flow rate required will be
Qr = Qs- ^required / C*r - X^,,,,;,.^),
if sludge growth in the aerators is ignored for practical pur-
poses. As Dr. Dobbins pointed out, one of the advantages of a
computer-based system is that it facilitates the implementation
of such multi-input control strategies.
In a system which is subject to strong diurnal load fluctua-
tions, the maintenance of a constant SCOUR will dictate equi-
valent variations in the sludge content of the aeration tanks.
This infers that sludge storage capability will be required, well
beyond what could normally be provided by the secondary
settlers. Apart from the use of storage tanks, one way of tackl-
ing the problem is to employ the step feed process, where a
portion of the aeration basin at the head of the plant is used
essentially for sludge storage during periods of low flow.
Another method, which reduces the variation in sludge re-
quirements, would be to reduce incoming load fluctuations by
providing equalization capacity, as proposed by Mr. Garrison.
Front-end load balancing should certainly be recognized as a
valid control technique, and. in general, an appropriate balance
should be struck between more equalization capacity, on the
one hand, and more complex in-plant control, on the other.
Peter C. Young:
(Editor's note: Dr. Young has combined his comments on the
individual workshop topics into a single document, which is
given under this section because of the particular applicability
of many of his remarks to computer-based systems.)
This Workshop has raised many important questions about
research needs for the automation of wastewater treatment
systems. But the term "automation" has, I feel, been inter-
preted in a rather narrow sense, with much of the discussion
centering on the automation of existing manually operated
processes without considering, in anything but fairly super-
ficial detail, the possibility of improving these processes by the
utilization of advanced control systems analysis and design
methodology.
We have, for example, talked at great length about the limi-
tations of existing sensors and the need for research into the
development of improved or entirely new sensing devices. But
automatic control is not only concerned with the measure-
ment of variables; important as they are. sensors are only one
aspect of control system design. And the modern approach to
systems analysis and control systems synthesis requires an inte-
grated approach to the design problem, with the sensor re-
quirement studies proceeding in parallel with equally import-
ant investigations such as system modelling, simulation and
overall control system design.
It might well be, for instance, that a better understanding
of the dynamic process, obtained by thorough modelling exer-
cises, could lead to the design of new types of control schemes
whose sensor requirements might not necessarily be the same
as existing designs. In other words, there is a need for con-
tinual cross-fertilization of ideas from different component
106
-------
COMPUTER APPLICATIONS IN AUTOMATION
areas of research if the final automated system is to attain its
full potential. To give but one example: there has been a great
deal of discussion on BOD measurement and the need for ob-
taining estimates of BOD more rapidly than at present. And
yet it might be that future research into control systems design
will negate the requirement for BOD measurement, either be-
cause variables other than BOD are found to be more useful
for control purposes or. alternatively, because good statistical
estimates can be obtained in real time from other more easily
obtainable measurements.
My plea is simply that we do not close our eyes to the pos-
sibility of radical innovation at this early stage in the study of
wastewater treatment automation, but rather that we keep all
options open, bearing in mind the kind of developments which
may be possible if an adequate program of research in systems
analysis and control systems design is undertaken. At least it is
incumbent on those who authorize and finance research in the
area to promote studies which encompass all important aspects
of the problem: they should not. it seems to me. allow them-
selves to be unduly influenced by groups who represent only
certain special, albeit important aspects of the problem and
who may not be cognizant of developments in areas outside
those covered by their own expertise and interests.
Of the more detailed topics we have considered at the
Workshop, I feel that there is need for some clarification on
the adequacy of dynamic models. When describing a dynamic
system in mathematical terms, it should be emphasized that
there is not one, all-encompassing dynamic model-some, as it
were, universal panacea to all modelling problems. Rather
there is a whole family of models having different degrees of
complexity depending upon the nature of the problem for
which they are formulated; indeed, it could be argued that one
of the major arts of systems analysis is the choice of a mathe-
matical model which suits the nature of the problem at hand.
In this latter sense, the more complicated and detailed
models, such as those considered by Dr. Andrews, are simula-
tion models aimed first at deriving a better understanding
of the physico-chemical and biological nature of the process,
and second at providing a vehicle for assessing the possible effi-
cacy of different control strategies before these are imple-
mented in practice. Provided such models are developed and
used with care and, in particular, provided the user does not
succumb to the inherent danger of believing that his model is
the system and not merely a mathematical model, then such
modelling exercises can be extremely useful, providing as they
do a natural prelude to systems analysis and control system de-
sign.
But it should be realized that complicated simulation mod-
els are, more often than not, unsuitable bases for analytical
control system design. For these tasks, the systems analyst re-
quires much simpler and more analytically tractable models
which reflect the dominant dynamic modal behaviour (Le., the
behaviour of the dominant modes that characterize the system
response to excitation), without including the extraneous de-
tails (i.e., extraneous to control system design requirements)
that tend to characterize a thorough micro-analysis of the pro-
cess. For example, it has been established recently (1,2) that a
simple second-order differential equation model can explain
the dominant daily variations of DO and BOD in a non-tidal
river system, even though we know that the detailed behaviour
is much more complex. In many examples it would seem that
the multitude of higher-order, nonlinear and stochastic effects
that naturally characterize the detailed operation of the sys-
tem tend to combine in their overall effects and can often be
represented in gross or macro terms as additive stochastic dis-
turbances with fairly simple statistical properties: it is as if
there is a Law of large systems, somewhat analogous to the
well known Law of Large Numbers, so that, while the system
is exceedingly complex, its overall small-perturbation response
characteristics can often be quite simple in form, apparently
dominated by a clearly defined and, more often than not,
linear mode of operation.
It seems to me that the importance of the above points in
relation to the automation of wastewater treatment systems is
that, while excellent work is being carried out in the simula-
tion modelling area and a growing understanding of the prob-
lem is emerging, there appears to be no determined research ef-
fort in the other modelling areas. And this is despite the fact
that analytical techniques such as model identification, para-
meter estimation and evaluation are available and have been
applied with reasonable success to other related problems,
such as the analysis of water resource systems and chemical
processes (3,4).
To consider one example where available analytical tech-
niques may be of use. the question of rainfall prediction has
been raised in connection with wastewater collection systems.
Such problems are closely related to the flow forecasting prob-
lems encountered by hydrologists and recent work on flow
forecasting which uses sophisticated but relatively simple re-
cursive methods of time-series analysis (3,5) may well have po-
tential in the analysis of wastewater collection systems; cer-
tainly initial reference to these techniques could provide im-
portant a priori information for research workers dealing with
such problems.
Another analytical technique developed in the control
systems area which appears to have reasonable potential for
application to wastewater treatment problems is the numerical
filtering of noisy data from dynamic systems-or, as it is often
called in the technical literature, state variable estimation or
Kalman filtering (6). This approach to data processing provides
a method for using the measured variables (which will almost
certainly be corrupted by measurement noise or subject to un-
certainty of some type) to reconstruct, on a firm statistical
basis, estimates of any unobserved or non-measurable variables
that characterize the process dynamic behaviour while, at the
same time, helping to filter off the measurement noise effects
from the measured variables.
A heuristic (Le., intuitive, not based on rigorous analysis or
107
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
derivation) and largely deterministic analog of such a proce-
dure is the estimation of BOD as suggested by Dr. Dobbins. In
effect, the available analytical techniques are able to put such
heuristic procedures on a firm statistical base, so making them
more systematic to design and potentially more reliable. Of
course the analytical techniques, like their heuristic relatives,
require a simple dynamic model of the process which explains
the dominant model behaviour; thus the provision of an ade-
quate mathematical modal is once again seen to be an inherent
requirement of system design.
Another extremely important point raised in the Workshop
discussion is the use of computers in automation. Although
many of the advantages arising from the use of computers have
been mentioned, I feel there has been insufficient emphasis on
the fact that one of the major attractions of having a computer
as part of the operation is that it enables the designer to con-
sider the integrated control of the whole system: no longer
does he only have to consider the local control of unit pro-
cesses, but he can also consider the interaction of such pro-
cesses and the design of control schemes that allow for such in-
teraction. And because the computer is such a powerful tool,
the designer can, at the unit process level, also consider the de-
sign of a control system that acknowledges the multivariable
fie., multi-input, multi-output) nature of many unit processes
(7,8) and takes account, for example, of factors such as the
stochastic nature of the process and the overall objectives.
Of course, the sophisticated use of computers in control
system design holds many problems, most of which still have
to be solved. But there is no doubt that an important research
need is for people in the wastewater treatment field to become
acquainted with recent developments in advanced computer
control so that at least they are able to judge the claims of
over-enthusiastic control system design consultants! There is
no doubt that in the next few years at least some of the tech-
niques which are at present only gleams in the eyes of the re-
searcher will come to fruition.
Finally, lest it be thought that I am over-emphasizing the
need for research on advanced systems analysis, it should be
stressed that I am recommending here the use of techniques
which, for the most part, have proven practical potential in
other areas of application. Indeed, I am a firm opponent of in-
novation for its own sake and would urge great caution in con-
sidering the use of some of the more advanced concepts in
systems design such as self-adaption and optimality.
Such concepts are, of course, extremely tempting because
they seem to offer solutions which, in the one case are able to
counteract the effects of possible changes in the controlled
process, and in the other are "best" in some sense. But it is
only fair to point out that both approaches to control system
design have not been particularly successful when applied to
real (in contrast to simulated) systems. Even in the aerospace
field for example, where the need for adaptive adjustment of
control gains is emphasized by the large and rapid changes in
vehicle dynamic characteristics that occur because of the
rapidly changing environment, the only widely used method of
adaption is the schedule adaptive system, in which the control
gains are pre-programmed to vary as functions of "air data"
variables such as dynamic pressure, altitude. Mach number.
etc.; true self-adaptive control, in which the system adjusts
itself without considerable a priori information, has been used
on relatively few occasions and then mainly for advanced
"one-off" research aircraft.
In the case of dynamic optimal control the picture is much
worse. First of all, it should be realized that the system so
designed is only optimal in the sense that it achieves the
extremum (max. or min.) of some chosen performance
objective or cost function; there is no guarantee that this cost
function is necessarily the best available such function, or even
that a suitable analytical statement of the desired performance
objective can be established at all.
An example of the inappropriate use of optimal control is
the popularity of the "Linear-Quadratic" (L-Q) approach to
optimal control system design which received so much
attention in the period 1960-1970(9). A considerable amount
of time and money was expended on the design of optimal
systems to this kind of specification even though, in its basic
form, the resulting control law does not include integral
action, so that the resulting controlled system can have rather
undesirable steady-state performance characteristics. (In par-
ticular, the resulting controlled system can be extremely
sensitive to the unavoidable uncertainty in the model parame-
ters and can be characterized by considerable steady-state
error to setpoint or constant input commands). While this is
not the only undesirable aspect of L-Q optimal designs (10) it
is, perhaps, one of the principal reasons for the notable lack of
success of such design procedures during the 1960's; certainly
the need for inherent zero steady-state offset to set point
inputs, as provided by integral action, is one of the principal
requirements of most practical control schemes, and the fact
that a considerable period of time passed before this disadvan-
tage was diagnosed and corrected (see for example Young and
Willems (10)) illustrates the dangers introduced by the
uncritical use of theoretical design procedures.
It is because of such mistakes as these in the past that I urge
the support of a strong program of research into advanced
control systems analysis and design for wastewater treatment
processes; I am sure that if an enlightened program of research
is initiated now, such aberrations are less likely to occur. A
healthy body of control systems specialists will become firmly
established; specialists who will be able to judge the utility of
new innovations within the context of wastewater treatment
and so, in the words of the preamble to this Workshop, help to
realize the many potential benefits offered by automation.
108
-------
COMPUTER APPLICATIONS IN AUTOMATION
REFERENCES:
1. Young, P. C. and Beck, M. B-, "The Modelling and Control of
Water Quality in a River System," Automatica (1974) (in press).
2. Beck, M. B. and Young, P. C-, "A Dynamic Model for DO-BOD
Relationships in a Non-tidal stream," Water Res. (to be published).
3. Young, P. C., "Recursive Approaches to Time-series Analysis,"
Bull. Inst. of Math, and its Applications (I.M.A.), 10. No's 5/6,
209-224(1974).
4. Astrom, K. J. and Eykhoff, P., "System Identification: A Survey,"
Automatica, 7.123-162(1971).
5. Young, P. C., Shellswell, S. H. and Neethling, C. G., "A Recursive
Approach to Time-Series Analysis," Tech. Report No.
CUED/B-Control/TR16 (1971), Engineering Department, Univer-
sity of Cambridge, England (1971).
6. Kalman, R. E., "A New Approach to Linear Filtering and
Prediction Theory," Trans, ASME, J. Basic Eng., 82-D, 35-45
(1960).
7. Macfarlane, A. G. J., "A Survey of Some Recent Results in Linear
Multivariable Feedback Theory," Automatica, 8, 455-492 (1972).
8. Macfarlane, A. G- J., "Relationship Between Recent Development
in Linear Control Theory and Classical Design Techniques," Third
IFAC Symp. on Multivariable Technological Systems, Manchester,
16-19 Sept. 1974.
9. Special Issue on Linear, Quadratic Gaussian Problems, IEEE Trans.
on Automatic Control, AC-16, No. 6 (1971).
10. Rosenbrock, H. H. and McMorran, P. D-, "Good, Bad or Optimal,"
IEEE Trans, on Automatic Control. AC-16. No. 6, 552 (1971).
11. Young, P. C. and Willems, J. C., "An Approach to Multivariable
Servo-mechanism Problems," Intl. Jour. Control, 15, 961-979
(1972).
CLOSURE
(Editor's note: In preference to submitting a closing
statement, Messrs. Garrison, Payne and Haug have modified
their paper in the light of the points raised during the
discussion.)
W. E. Dobbins:
The question has been raised as to what is an acceptable
level of maintenance requirement for various water quality
sensors. The experience in Bridgeport, as reported by the
Instrument Engineer, is as follows:
D. O. Probes: The three step-aeration tanks have a total of 18
probesgeneral maintenance, including cleaning, membrane
and electrolyte replacement and calibration, requires about 5
man-hours per week.
Suspended Solids Meters: Six meters of the optical type
require about 2 hours per week, principally for lens cleaning
and calibration. This moderate requirement was after the
installation of a simple automatic flushing system utilizing a
tuner and two small electrically controlled valves (see article in
WPCF "Deeds and Data" October, 1974).
Water Quality Analyzer: An on-line water quality analyzer has
probes for pH, conductivity and temperature, and wet-
chemistry analyzers for chromate, cyanide and copper. A total
of about 9 hours per week are required to keep the system
operating satisfactorily. The work for the most part is cleaning
the probes, and replacing filters in the pretreatment system.
The total effort consists of about 16 man-hours per week.
The Instrument Engineer estimates that one adequately
trained technician can keep the systems in both of the
treatment plants in generally satisfactory operating condition.
In the writer's opinion, this is an acceptable requirement for
these plants in which the total operating personnel numbers
about 80.
Report of Working Party
on
RESEARCH NEEDS
for
COMPUTER APPLICATIONS IN AUTOMATION
Carmen F. Guarino
Commissioner, Philadelphia Water Department,
1160 Municipal Services Building, Philadelphia, PA 19107
John M. Smith
Chief, Municipal Treatment and Reuse,' National Environmental Research Center,
Environmental Protection Agency, Cincinnati, OH 45268
INTRODUCTION
Computers are being widely used to expedite work in the
chemical process industry and to economize the endeavors of
man. To date, computer application in the wastewater
treatment field has been minimal. The use of computers will
solve many problems that presently confront the effective
treatment of domestic and industrial waste.
Without the use of computers, it will be impossible to meet
the requirements of P.L. 92-500 and other related laws
concerning the control and treatment of pollution. Presently,
there is little control of the treatment process and this is more
obvious in the smaller plants than the large. The only way that
we can adequately control the treatment system is through the
development of instrumentation and automation.
We must have some method whereby the process is
monitored continuously, not once every 24 hours or once a
week which is the case in many small plants. Remembering
that there are far more small plants than large plants, this does
have a tremendous impact on national pollution control and
treatment. The development of model systems that can be
applied to both small and large plants will result in effective
109
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
treatment and in a reduced cost of treating wastewater.
There is very little regulation of energy used in treatment
plants today because we do not have adequate control. There
should be no doubt in our minds that there would be a
tremendous saving of energy as well as materials if we only
applied them as necessary, and not on a routine basis as is now
being done. A check of most activated sludge plants will reveal
that they usually use the same number of blowers to supply air
at 3 o'clock in the morning as they do at 3 o'clock in the
afternoon. Yet, conditions usually vary greatly at these hours.
Over a period of time, the prudent use of power nationally
would amount to a great saving.
Instrumentation and automation will furnish the best
treatment at the least cost.
PROBLEMS
1. The application of computers to wastewater treatment
systems is new. Consequently, there is a primary need to
assemble all information from both the national and inter-
national viewpoint. The first problem is the lack of a
complete computer library as applied to wastewater treat-
ment systems.
2. The unavailability of suitable models to simulate the
treatment process.
3. The absence of reliable sensors to monitor important
process variables. This limits the use of models and
computers at this point in time.
4. For the most part, the wastewater treatment system is
treated separately from the collection system. For effective
treatment and the most effective automation, the system
must be considered and made one. The absence of
sufficient information as to the dynamics of sewer collec-
tion as well as the impact of rainfall, infiltration, and means
of controlling flows in the wastewater treatment system
limits the most effective use of the computer.
5. There are many specialists in the wastewater treatment
systems field and there are many specialists in the use of
the computer, but there are very few people that under-
stand both. Consequently, this limits the full application of
the computer in solving the problems of the wastewater
treatment process. There is a need for greater education to
produce personnel who understand both facets of this
work.
Summary
Successful computer application will require educated and
trained personnel, reliable sensors, and time-tested and avail-
able mathematical models.
RESEARCH NEEDS
1. Educationcomplete knowledge of the wastewater treat-
ment process and complete knowledge of computer
applications.
It was obvious at the Workshop that the treatment
specialists and computer specialists were not able to
completely communicate. Before the treatment process
will be successfully computerized, we must have a clear
understanding of the treatment process as well as the
application of the computer.
2. Development of reliable sensors which can, on a real-time
basis, monitor the treatment process. These instruments
would include:
a. B.O.D.
b. C.O.D.
c. Suspended Solidsmixed liquor suspended solids
d. Sludge density
e. Sludge blanket level detectors
f. Ammonia and nitrate
g. Biological activity indication
h. Flow measurement in the collection system
3. Preparation of basic models which can be made available
to those who are attempting to automate the treatment
processes.
There should be two types of models; one that can be
used effectively with the present knowledge of the
treatment process as well as the present knowledge of
sensors; and another to make allowances for the sensors
which are being developed.
Models worthy of development are:
a. Activated sludge process and all its modifications
b. Chemical treatment
c. Aerobic sludge digestion
d. Anaerobic sludge digestion
e. Collector system control
All of the above, of course, have many variations
depending on the particular process chosen; for example.
under 'Chemical Treatment', adsorption could be in-
cluded, etc.
4. Test and evaluate models on a pilot plant basis. There is a
great need for a pilot plant which can be used to both
formulate and test models. This would be the greatest
contribution made towards exploring the use of com-
puters in controlling the treatment process. Proper pilot
plants and proper personnel could develop and de-bug
models which could then be made available for use.
5. Establish some unit that would serve as an information
center for all agencies working towards the automation of
the treatment process. This central office would serve
many purposes, from information dissemination to con-
tinuously updating and improving models and sensors.
6. Integration of treatment system with collection system.
The collection system, through the use of computers,can be
the first step in the treatment process. Control of the flow
to the treatment plant, through effective use of the
collector system, has many advantages from flow equaliza-
tion to being able to treat more flow, particularly during
rainstorms.
7. Inter-relationships in the treatment process should be
110
-------
COMPUTER APPLICATIONS IN AUTOMATION
clearly defined. In order to computerize the system, a
program must be written. A program cannot be written
until the process can be clearly defined.
A complete technological assessment of wastewater treat-
ment via computer applications is required. The assess-
ment should include, but not be limited to. the following:
a) Social and psychological implications
b) Economics
c) Risks, liability, etc.
d) Man/machine interfacing
e) Political and regulatory implications
Determine the role of the computer in wastewater treat-
ment with respect to matching the size and the type of
computer (full-size, mini, micro-processor or analog con-
trol) to the plant size and the planned function (control.
data acquisition, scheduling, accounting, etc.) of the
computer.
10. Determine specifications or guidelines in the selection of a
computer for a particular facility. Preferably, the guide-
lines should be presented so that the function of the
computer is described and related to the size and type of
computer required.
11. Large-sized wastewater treatment systems display a poten-
tial for using a centralized computer for data acquisition,
monitoring by use of the central computer and finally
using the central computer to assist in the preparation of
the monthly state report. Programs should be written for
this application and the concept should be demonstrated.
12. Using a centralized computer and several satellite plants,
each having a terminal, demonstrate the effectiveness of
using a man-in-the-loop to control each plant by commun-
icating with the computer.
111
-------
EVALUATION
OF
THE EFFECTIVENESS OF AUTOMATION
Workshop on Research Needs
Automation of Wastewater Treatment Systems
113
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
Joseph F. Roesler
Pilot & Field Evaluation Section, Advanced Waste Treatment Research Laboratory,
National Environmental Research Center, Cincinnati, OH 45268.
The relative value of automation in the field of wastewater
treatment can only be determined by accurately evaluating the
effectiveness of the various automatic control strategies.
Careful planning of an evaluation is such an important task
that failure can almost be guaranteed if adequate planning is
neglected. Two methods that can be used for evaluating the
effectiveness of automating a wastewater treatment plant are
field evaluation and desk-top computer simulation. Important
and interrelated parameters in any such evaluation are cost-
effectiveness of the control strategy, performance improve-
ments, and energy requirements. This paper describes the
criteria that should be considered in planning an evaluation
and presents field data from several evaluations sponsored by
the U. S. Environmental Protection Agency (EPA). Problems
encountered and some of the final conclusions are also
included.
EVALUATION TECHNIQUES
The usual technique employed in field evaluation is to
compare the performance of a manually operated plant with
an automated plant; however, the standards for manual
operation vary according to the idiosyncrasy of each plant.
Plant layout, piping, type of sewage, etc., may affect the ease
with which the plant is manually operated. There are no
baseline data available that show the performance of a well-run
typical wastewater treatment plant. Therefore each per-
formance evaluation is a comparison of automatic versus
manual operation at only one plant, and each performance
evaluation requires an exacting definition of the manual
operating procedures.
Before beginning an individual plant evaluation to assess
automation performance, the operator-manager attitude must
be carefully evaluated. In some cases, operators believe that
the security of their jobs is threatened by automation. In such
cases, the instruments and automatic control loops will be
placed under the most severe test conditions possible and may
thus "fail." The cause of the failure may be neglect or even
sabotage of the equipment by the operators. On the other
hand, operators and managers may welcome automation;
however, because of their inexperience with the equipment
and lack of proper training, the equipment may still "fail"
because of unintentional neglect or faulty maintenance.
Negative attitudes must be corrected. Assurances must be
given that the operator's job security is not threatened.
Adequate instructions must be given to operators and man-
agers. In plants installing a digital computer, instruction must
be given to the managers so that they comprehend the needs
and requirements of such an installation.
Many plants are symmetrically constructed, giving rise to
duplicate systems. In such cases, the tendency is to compare
the automatic control of one stream with a manually
controlled duplicate stream. In such a comparative evaluation,
the "Hawthorne" effect (1) must be considered. The classic
studies at Hawthorne, Illinois, were conducted by the Western
Electric Company in the late 1920's. The results of this
research indicated that the employees' behavior and attitudes
toward their jobs and toward management are conditioned to
a considerable extent by the values, standards, and expecta-
tions of the work groups to which they belong. Thus there is
an excellent possibility that operators working on the man-
ually operated portion of the plant will work harder and more
diligently to demonstrate that man is superior to machine,
thus invalidating the comparative test results.
COMPARATIVE FIELD EVALUATIONS SPONSORED BY
EPA
Once the group behavior problems are eliminated, simul-
taneous comparative evaluations will yield excellent and
reliable results. An example of such an evaluation (Le.,
without the group behavior problems) is the dissolved oxygen
(DO) study recently carried out at the U. S. EPA's Pilot Plant
at Blue Plains. The objective of the study was to determine the
114
-------
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
effect of DO level on filamentous growth by comparing the
performance of two automated control strategies against one
another. Both systems were operated at a level of 1 mg/1 for a
period of five sludge retention times (SRT's) without incurring
sludge bulking problems. One system (D system) was then
operated at a DO level of 3 mg/1, while the second system (E
system) was maintained at a DO level of 1 mg/1. After five
SRT's, the sludge volume index (SVI) of the D system
increased to 200-300, while the SVI of the E system remained
at a level of about 100. The DO level of each system was then
set at 1 mg/1 for a period of five SRT's, thus causing the SVI
of the D system to decrease to 50. This simple comparative
test makes the dramatic point that DO set point does have an
influence on the sludge handling characteristics.
Two other DO studies in which the comparisons (manual
versus automatic) were sequentially performed are next
described. One of the studies, at Rent on, Washington (2) tends
to confirm the above observations, but no such clear confirma-
tion resulted from the second study, conducted at Palo Alto,
California (3). Though it is too early to come to any
conclusions, it appears that the nature of the comparative tests
(simultaneous versus sequential) does have an effect on the
results. Obviously, the main risk of conducting sequential
comparative tests (Le., manual versus automatic operation) is
that raw sewage entering the plant may differ in nature from
one time frame to the next. This possibility is an inherent
disadvantage that, at best, can be compensated for only
partially.
The Renton plant was operated for about a year (March
1970 to April 1971) under manual control while an automatic
DO control system was being installed in the new aerator. The
following year, the plant was successfully operating with
automatic DO control. Data were collected for comparative
purposes during the months of October, November and
December for years 1970 and 1971. The operators and plant
management had an excellent attitude toward automation.
Also, the manual control policy was well defined and expertly
carried out.
The obvious question, however, is whether the sewage was
identical for both time frames. One partial answer is that the
BOD loading to the plant increased about 50% during the
automatic control period. In spite of this increase, the
performance of the plant did improve. The effluent BOD
decreased from a geometric mean of 11.1 ppm, obtained
during manual operation, to a mean of 3.9 ppm for automatic
operation. Figure 1 shows the effluent BOD data plotted on
logarithmic probability paper to obtain a more normal
distribution of measurements.
Further analysis indicated that the sludge characteristics
may also have been affected by automatic DO control. The
100
90
80
70
60
95 90 BO 70 60 50 40 30 20 10 5 21 0.5 0.2 0.1 005 0.01
a
o »
a to
50
3P
I I
AUTOMATIC CONTROL
IP
0.01
i I I I _ '
I _ I
1 - L
00501 02
0512 5 10 20 30 4O 50 60 70 80 90 95 98 99 995 99.8999
PERCENT OF OBSERVATIONS EQUAL TO OR LESS THAN STATED CLASS MEAN
Figure 1. Frequency Distribution of BOD in the Effluent
115
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
frequency distribution of the sludge volume index (SVI) is
shown in Figure 2. The arithmetic mean for the SVI with
manual control was 332. This was reduced to 86 with
automatic control. Since BOD is a broad analytical term and
since this was a sequential comparative test, the data are
promising but inconclusive. This study, therefore, requires
confirmation with further research. The U. S. EPA Pilot Plant
at Blue Plains has already begun such research, as indicated
previously.
At the Palo Alto Sewage Treatment Plant, four control
strategies were evaluated and compared to manual operation.
The control schemes were DO, air/return sludge (DO/RAS),
respirometry, and mixed liquor suspended solids (MLSS). Each
control strategy was evaluated for about 30 days. There were
three well-defined manual operations to accommodate the
operating requirements associated with wet- and dry-weather
flows.
As compared to manual operation, DO control showed a
13% performance improvement in terms of TOC measure-
ments and an 11% reduction in air use. For MLSS control, the
solids level in the aerators were maintained close to set point,
although a low limit on RAS pumping resulted in uncontrolled
increases in MLSS during periods of low plant flow.
During the evaluation of the MLSS control scheme, proper
sludge management became necessary. A separate aeration
basin was converted to a sludge storage tank to accommodate
high loadings to the plant. Perhaps because an aerator basin
was being used for sludge storage, or perhaps because of sludge
age, a scum appeared in the final clarifier. This scum would
not settle and, as a result, caused a poor quality effluent. When
sludge storage was eliminated, the scum disappeared and the
effluent quality improved.
One technique for complete control of the aeration basin
consists of regulation of food-to-microorganism ratio (F/M).
The technique for measuring the food differs: It may be
accomplished by measuring TOC, COD, or oxygen uptake.
Two different types of suspended solids meters were used to
indicate the microorganism concentration. The readings of
both meters correlated well with each other and with
laboratory suspended solids values.
Suitable automatic TOC and COD analyzers were not
available for on-line control during the Palo Alto experiments.
Therefore, only two F/M control strategies were evaluated.
These were feedback respirometry control and DO/RAS
control. For DO/RAS control, the air flow was used to infer
food, and then the return sludge was adjusted accordingly. In
other words, the entire aeration tank was used as a respir-
ometer to set the return sludge flow. The results of these
experiments were inconclusive, probably for three reasons: (1)
more time was needed for collecting data than was allotted.
IOOO
9OO
00
7OO
6OO
5OO
E
_- no
999998 995 99 98 95 9O »0 70 60 50 4O 30 20 10 5 21 05 02 O.I 0.05 001
I
AUTOMATIC CONTROL
I I
1 I
O-OI 00501 0.2 O.S I J 5 10 2O 3O 40 SO 6O 70 80 9O 95 9« 99 99.5 99.899.9
PERCENT OF OBSERVATIONS EQUAL TO OR LESS THAN STATED CLASS MEAN
Figure 2. Frequency Distribution of SVI
116
-------
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
Table 1. Costs for Various Control Schemes
Control scheme
Mass-proportional
Flow-proportional
Operator control
No control
for
Ibs Alum
per day
16,290
26,850
28.260
35,700
Alum Addition at a 10-mgd
Capital
cost
S
$112,000
128,000
134,000
145,000
Plant (August 1971)
Cost of
alum
c/Kgal
1.66
2.74
2.88
3.64
Total
cost of
plant,
c/K gal
2.22
3.33
3.48
4.27
Annual
cost
Thousands/
year
81
121
127
156
(2) difficulties were encountered in separate sludge storage and
in the ability to supply the proper amounts of RAS at low
flows, and (3) assurance was lacking that the manual tests
could be accurately correlated with their adjacent automatic
tests.
COMPUTER SIMULATIONS
An alternative to field evaluation is computer simulation.
But for effective studies, such simulations must use mathe-
matical models that consider the entire system and have been
demonstrated to be accurate by field evaluations. Most
mathematical models emphasize treatment of liquid waste; as a
result, there is little work on determination of sludge quality,
separation, and subsequent handling. There is evidence that
DO control does affect sludge characteristics, but the mathe-
matical models that exist today do not simulate this. Conse-
quently, accurate cost and performance data must be obtained
from field evaluations.
COST-EFFECTIVE ANALYSIS AND PERFORMANCE IM-
PROVEMENTS
The relationship of cost to performance is sometimes
overlooked in evaluations. To obtain an accurate cost-effective
analysis, all data must be normalized. The cost of improving
the performance of a plant removing about 90% BOD has been
shown (4) to follow an exponential function. Manually
operated plants typically remove 85 to 90% BOD; if automa-
tion were to improve this performance, then such improve-
ments must be considered in computing cost savings to avoid
significant errors in cost-effective determinations.
However, there are certain processes in which the per-
formance appears to be negligible for cost-effective analyses.
Such processes are normally dependent on a chemical addition
in which a minimum amount of chemical achieves the
performance desired. Further addition of a chemical is simply
wasted. An example of such a process is alum addition for
phosphorus removal.
Costs for alum addition at a 10-mgd plant for various
control schemes were calculated by Convery et al. (5) and are
shown in Table 1.
Periodic operator control provided a saving of 529,000 per
year. This saving could be increased by more frequent operator
attention. Similarly, if flow-proportional control were em-
ployed, an additional saving of 55,500 per year would result.
Feedforward mass-proportional control provided an annual
savings of $40,500 compared to flow-proportional control
alone. This amount does not include the significant savings in
sludge-handling costs that would also accrue, yet it is more
than adequate to justify the purchase of a phosphate analyzer.
Similar cost comparisons have been made for methanol
addition for denitrification (6) and for breakpoint chlorination
(7).
ENERGY REQUIREMENTS
The final parameter in evaluating automation is energy
utilization. The Palo Alto study showed that simple DO
control can provide a power saving of 11% and consequently
an economic savings of 55,500 per year. Since air blowers are
one of the largest energy consumers within a treatment plant,
this energy saving is very significant.
If the entire U. S. population were served by activated
sludge treatment plants, it has been estimated that the average
power consumption would be 0.113 kwh per capita per day;
yet the energy available in sludge (8) is 0.154 kwh per capita
per day. We must begin investigating the possibility of
automating sludge-handling so as to achieve cheaper energy for
plant operation. Data obtained from the U. S. EPA's contract
with Raytheon Company indicate that instrumentation and
control schemes for automated sludge processes (those for pH,
temperature, and methane production, for example) have a
pay-back period of 0.94 years for a 10-mgd plant, and 0.65
years for a 100-mgd plant. This is only a beginning.
The problem is serious. Consider, for example, the Min-
neapolis-St. Paul Metropolitan Sewerage District, which has
received an ultimatum from the local utility; this utility
threatens to deny the District any natural gas by 1978. In the
intervening years, the District's gas supply will be progressively
cut back. The District is now seriously looking at all forms of
sludge-handling to optimize energy production and reduce
energy use. With the energy crisis that we face today, there are
117
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
no assurances that other plants will not eventually suffer the
same fate. Automation could play a vital role in this area.
PRESENT NEEDS
In spite of the work that has been done in automation at
the National Environmental Research Center, there are still
many research needs that must be answered. For example, we
still need accurate mathematical models, especially models
describing sludge characteristics and the performance of
sludge-handling and removal processes. New or improved
models for sludge handling, removal, and disposal should ease
the task of formulating and evaluating the corresponding
control strategies for such sludge processes.
The mechanics of implementing automatic control are
reasonably well established, and the Renton, Washington study
demonstrated the benefits of DO control. This technology
must be documented and promulgated to encourage future
plants to install automatic DO control and to encourage
current plants to upgrade their operation by adding DO
control. Meanwhile, research should be conducted to investi-
gate the effects of DO setpoint on effluent quality, sludge-
handling characteristics, optimum DO setpoint, proper loca-
tion, and immersion depth for the DO probe, etc.
We must utilize instruments and automation for energy
conservation and generation. At the Raymond Clayton Plant
in Atlanta there is a methane monitor for manual control of a
digester. The feasibility of automating a control loop to utilize
such a monitor for better and more reliable production of
methane should also be investigated because sludge digesters
that function properly are a valuable energy source. We must
develop monitors and alarms that alert the operator to toxic
substances entering either the plant or the anaerobic digester.
Alternative modes of treatment must also be considered in the
event that toxic materials still manage to enter the plant in
spite of any precautionary measures taken to prevent this from
occurring.
Further energy savings can be made by instrumenting the
dewatering of sludge before incineration. Instrumentation to
measure sludge characteristics is essential to developing an
automatic control loop for polymer addition before vacuum
filtration or centrifugation.
Finally, we must help engineering firms and small munici-
palities by offering typical instrumentation specifications that
plant designers can readily incorporate in their process designs
and for which the city planners could easily solicit bids.
Design manuals for automated control loops should be
published to enhance the state-of-art and encourage imple-
mentation.
The U. S. EPA is doing work in all these areas, but our
resources are meager in comparison to the job that we know
must be done. It is our hope that this workshop will provide
the correct priorities, generate new ideas, and serve as a vehicle
of communication.
REFERENCES
1. Roethlisberger, F. J., and Dickson. W. J., "Management and the
Worker." Harvard University Press. Cambridge, MA (1939).
2. Roesler, J. F., "Plant Performance Using Dissolved Oxygen Con-
trol," Jour. Environ. Eng. Div.. 100. 1069 (1974).
3. Petersack, J. F. and Smith. R. G., "Full-Scale Demonstration of
Advanced Automatic Control Strategies for the Activated Sludge
Process," Final Report submitted to the U. S. Environmental
Protection Agency, July 1974.
4. Smith, R., "Wastewater Treatment Plant Control," Presented at
the Joint Automatic Control Conference, Washington University.
St. Louis, MO (1971).
5. Convery, J.J., Roesler. J.F. and Wise, R. H., "Automation and
Control of Physical-Chemical Treatment for Municipal Waste-
water," Proc. Conf. Applications of New Concepts of Physical-
Chemical Wastewater Treatment. Vanderbilt University (1972).
6. Roesler, J. F., "Factors to Consider in the Selection of a Control
Strategy," Proc. Second U. S.-Japan Conference on Sewage Treat-
ment Technology, Cincinnati. OH and Washington, DC (1972).
7. Roesler, J. F., Internal Memorandum, November 1971.
8. Anon., "Shortages: Water Pollution Control Facilities Face the
Crisis," Jour. Water Poll. Control Fed., 46,621 (1974).
FIELD EVALUATION OF THE EFFECTIVENESS
OF AUTOMATION
Allen E. Molvar
Environmental R&D Director, Raytheon Company, Portsmouth, RI 02871
INTRODUCTION
From successful experiences with instruments and au-
tomatic control devices, the automation of both wet- and
dry-weather wastewater treatment plants offers these well-
known and widely published potential advantages:
Improved process performance
Reduced energy consumption
Reduced chemical consumption
Reduced sludge production
Reduced equipment size
Reduced operating labor
Reduced maintenance requirements
Increased equipment life.
Most of the available literature, however, fails to point out
the potential costs and maintenance requirements that are
characteristic of the proposed instrumentation. Instrument
loops usually include measuring or sensing elements, signal
118
-------
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
transmitting devices, display elements, controllers, and final
control elements. Accordingly, this equipment must be pur-
chased, installed, checked out, and properly maintained.
Although some on-line instrumentation may be essential for
process operation or mandated by regulatory agencies, most
on-line instruments and all automatic control devices used in
wastewater treatment projects are optional; they are installed
to effect savings. In order to select intelligently the instrumen-
tation and automatic control devices that should be installed, a
single touchstone must be met: the potential benefits must
offset the added costs and maintenance burdens. Figure 1
displays pictorially the factors involved in a cost-benefit
analysis that is used as a decision-making aid.
To assess the effectiveness of currently available instru-
ments and automatic control devices under field conditions
and to accumulate the baseline information necessary for
pragmatic cost-benefit analyses, the United States Environ-
mental Protection Agency sponsored a comprehensive study of
the state-of-the-art. These analyses were intended to determine
the type and degree of automation that are best used in
wastewater treatment facilities. As part of this project, a team
of engineers surveyed 50 selected municipal and industrial
wastewater treatment facilities, located throughout the United
States (as shown in Tables 1 and 2). These plants practiced a
wide array of processes. Although the majority of those
surveyed were dry-weather or combined-treatment facilities,
some storm-water treatment plants and control centers were
also examined.
TABLE 1. TYPES OF FACILITIES SURVEYED
Type of Facility
Primary treatment plants
Secondary treatment plants
AWT treatment plants
Storm-water detention facilities
Computer data centers
Industrial waste treatment
Pilot plants
Number Visited
9
25
3
3
5
2
3
REDUCED EQUIPMENT SIZE
1
\
PERFORMANCE
niUCIIAflDTiniVI
LJIMouiyir 1 IUIM
CONSUMPTION
iODUCTION
TSIZE
G LABOR
iNCE REQUIREMENTS
NT LIFE
>
^
~
(
, 1
/
EQUIPMEP
INSTALLA
STARTUP
ENGINEEF
MAINTEN;
LIFE EXPE
INTEREST
\
H-rr+
NET SAVINGS
COST/BENEFIT ANALYSIS
VIA TOTAL ANNUALIZED COSTS
Figure 1. Cost/Benefit Analysis
119
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
TABLE 2.
REGIONAL LOCATIONS
EPA Region
1
2
3
4
5
6
8
9
10
OF PLANTS SURVEYED
Number Visited
2
4
4
5
16
2
1
10
6
METHODS
Prior to the on-site inspections, the survey engineers
attended a 2-day orientation session during which the type of
measuring and sensing devices that might be encountered, as
well as the standardization of all survey reports, question-
naires, and drawings, was discussed. Extensive questionnaires
(Figures 2, 3, and 4), which detailed pertinent background
information, instrument performance and cost, and control
loop experiences, were prepared in advance.
At the start of each facility visit, the survey engineer met
with plant management and those persons responsible for
instrumentation. Plant histories, design flow rates, and opera-
tional characteristics were discussed at these meetings, with
special emphasis placed on the overall benefits or liabilities of
the installed instrumentation. This information was then
documented on the General Survey Questionnaire (Figure 2).
A plant tour (with the facility's instrument engineer usually
functioning as guide) permitted the survey engineer to
examine the operating instruments and control loops item-by-
item. During this tour, measuring devices were inspected, and
pertinent data (including the manufacturer, model number,
maintenance characteristics, accuracy, and application) were
recorded on the Instrument Survey Form (Figure 3). More-
if
i!
i
BC
0
3
|
i <
w °
\\
S£
«
£
i
>-
|
1
s
l
t
i ?
i I
s &
| |§
1 | |
i ^
; £ 1
If! i
: I s =
z ; £ 1
I 111! 1 1!
s
^
1
i
|
a.
l»,l
fill
1
4
s
i
I
j
i
j j
i
1
I
i
I
8
5 i-
§ I
"j
|
-C
j
1
i 1
u
1 1
1
i
II 1 i
i
£
3
1
i
Konp
lollwin DtacHpllon
Camptuat COM:
j
I
Central Control
SupMTtwry Control:
Alarm and Sataty Syiiamr
Automatic Enwrfincy Program!
4
^c
1 ?
i!
Malnlananca and Callbtallon
Spacial Eqvlpmanl:
Spaclal Oparalor Tralnlnl
Tolal In-Flanl Man Howi/Yaar
Tolal Coal of OulUda 5ar-all Icnallu ol Ini
Figure 2. General Survey Questionnaire (Sample Form)
120
-------
INSTRUMENT SURVEY FORM
Instrument
Parameter
Manufacturer
Model Number
Equipment Cost
Operating Experience
In-Plam Maintenance
(mh/yr)
Maintenance Frequency
(no./mo.)
Special Training
Service by Contract
($ or mh/yr)
On-Demand Service
(S or mh/yr)
6
_E
|
f,
I
V
u.
Total Downtime
Downtime Frequency
(no./mo.)
Problems*
n
3
o
o
Equipment
Auxiliary Devices**
Recording Devices***
Comments
"B
t
i
I
VS
c
S1
5
>
o
I
m
m
§
m
m
O
H
O
* Corrosion, fouling, etc.
** Limkers, alarms, ratio relays.
*** Local and central.
-------
LOOP AND raOCESS CONTROL SURVEY FORM
4
I1
s!
Control Techniques
!
]
!
s
1
Type of Controller**
1
*
*
u
1
Estimited Rnfunc
TimediM.)
Remflu
Annul COM Si«lngi
S
j
UlBily (kW-lu/yr)
s
Pmctti Impfovtmtfil
Increase Removal (?)
Parameter Variance
min/max (mf/M
Operating Experience
Maintenance & Calibration
by In-flant Personnel
(Sor mh/yr)
y.
ii
1!
!
M
Service by Contract
(S or mh/yrl
On Demand Service *
< Sor mh/yr)
Downtime (hrs/yr)
Downtime Frequency
(no./mo.)
Comments
' Conlrol mode: relay, proporlionil, propoillonil plui rcxl, tic.
Typei of conlrolleri: iniloi (pnc., hyd. or clec. medli); computer (iiiperviiory, direct dlgllil or «l iniloi).
Finilconlrol element: pne. nluei, virlihli ipeed pump, etc.
-------
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
over, the survey engineer examined the control techniques,
costs, benefits derived, and operating experiences. His observa-
tions were recorded on the Loop and Process Control Survey
Form (Figure 4). In order to coordinate the accumulated
information with respect to in-plant applications, applicable
instrument diagrams were constructed using standard ISA
symbols.
RESULTS AND CONCLUSIONS
Measuring Devices
Unreliable sensors accounted for most of the difficulties
experienced with automatic measurement and control in
wastewater projects. The accumulated instrument operating
experiences, summarized in Table 3, clearly show that waste-
water instruments require more maintenance than their indus-
trial counterparts. Since most measuring devices in wastewater
service interface directly with raw sewage, mixed liquors, or
thickened sludge, these devices are subject to continued
fouling from solids deposition, slime buildup, and precipita-
tion. Accordingly, they need more frequent cleaning and
calibration. Poor mechanical reliability, interferences, and a
lack of established measuring principles are also responsible for
the unsatisfactory state of some analytical sensors.
The distribution of measuring devices (Figure 5) indicates
that flow and level devices account for nearly half the
instrumentation employed in treatment facilities. Analytical
instruments represent approximately a quarter of the instru-
ments observed, and position, speed, weight, and other
mechanical-type measurements add up to about 15%.
The following measuring instruments possess sufficient
reliability for on-line use in wastewater treatment facilities and
are commercially available: level, flow rate, temperature,
pressure, speed, weight, position, conductivity, rainfall, turbid-
VARIABLE
LEVEL
FLOW
DENSITY
ANALYSIS
MISC.
CONTROL
* Estimated
INSTRUMENT
Bubbler
d/p Trans.
Float & Cable
Optical
Flume & Weir
Venturi, etc.
Propellers
Pos. Displace.
Magnetic
Nuclear
Mechanical
pH and ORP
Dissolved O2
Res. Chlor.
Turbidity
Conduct.
Chlorine Gas
Explosive Gas
BOD, TOC, etc.
Temp.
Press.
Speed
Weight
Position
Sampling
Rainfall
Level
Flow
Sludge
Air Flow
Dosage
Res. Chlorine
DO
APPLICATION
Tanks & Wet Wells
Digesters & Sludge
Tanks & Wet Wells
Sludge Blanket
Major Flows
Air and liquids
Clean liquids
Gases
Liq. and Sludge
Med. & Thick Sludge
Med. & Thick Sludge
Aqueous Liquids
Aqueous Liquids
Aqueous Liquids
Fairly Clean Liquid
Aqueous Liquids
Airspace
Airspace
Wastewater
All
AH
Engines, etc.
Sludge or C12
Sloice Gates
Liquid Streams
Storm Waters
Wells & Basins
Al! Fluids
Sludge Separation
Aeration
Chlorination
Aeration
TYPICAL
COST
$200
700
400
IK
2K+
800+
1K+
500+
2K+
5K
2K
2K
5K
3K
IK
3K
3K
300
200
2K
IK
4K
500
TYPICAL MAINTENANCE
FRQ/YR.
12
O.G
24
1.4
4
7
2*
12
48
MH/YR.
STP IND
8 4
5 5
60 5
2
20 6**
10 10
80* 10
12 8
51 40
Excessive
300
100
365
200
24
12
50 29
60 -
140
60
50*
12+ 50
Excessive
1*
5
-
24*
18*
0.5
24*
8* 4
4 4
- -
60*
30*
20 -
50* -
SKILL
1
3
1
2
3
3
4
4
4
(3)
3
4
4
4
4
3
4
3
5
3
3
4
4
3
2
3
3
3
3
3
3
4
RELIAB.
(MTBF)
1-2 yrs.
1-5 yrs.
. 2-2 yrs.
.1-5 yrs.
. 5-5 yrs.
2 mo. -5 yrs.
1 mo.-l yr.
1 mo.-l yr.
.5-10 yrs.
1-3 yrs.
1-6 mos.
1-4 mos.
1-9 mos.
.2-lyr.
1-6 mos.
1-4 mos.
.5-lyr.
.2-1 yr.
. 1-1 mo.
.5-2 yrs.
.1-5 yrs.
.6-5 yrs.
.6-2 yrs.
.1-1 yr.
.1-1 yr.
1-5 yrs.
TYPICAL
USE
5-15 yrs.
5-15 yrs.
2-20 yrs.
2-8 yrs.
5-30 yrs.
5-30 yrs.
1-8 yrs.
1-5 yrs.
5-20 yrs.
8 yrs.
2 yrs.
. 5-5 yrs.
.1-5 yrs.
4 yrs.
4 yrs.
4 yrs.
8 yrs.
8 yrs.
.3-1 yr.
5 yrs.
5 yrs.
5 yrs.
10 yrs.
1 yr.
4 yrs.
12 yrs.
NOTE :
STP = TREATMENT PLANT
IND = INDUSTRIAL.
SEE TEXT
** d/p Converter only
TiHe 3. Instrument Performance
123
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
Figure 5. Distribution of Measuring Instruments Observed During
User Survey
ity, pH, residual chlorine, free chlorine gas, and fire hazardous
(flammable) gas. This is based on actual field experiences in
the surveyed facilities (summarized in Figure 6).
Sludge density meters, sludge blanket level detectors,
on-line respirometers, dissolved oxygen probes, and many
automatic sampling systems use well-established principles that
are suitable for wastewater monitoring and control activities
but require so much maintenance that many users are
dissatisfied with them. These instruments need improved
maintenance characteristics before they will become widely
used.
In spite of the many successful flow-measuring devices
observed in the treatment plants, accurate and reliable flow
rate monitoring for storm water does pose special problems.
Highly transient flows, large operating ranges, high suspended
solids, and frequent collisions with large debris are only some
of the obstacles that an acceptable storm-water in-sewer
flowmeter must overcome. Consequently, a suitable storm-
water flowmeter needs to be developed that will produce the
accurate flow rate data required for sewer regulation.
Automatic Control Loops
As shown in the summary of automatic control devices
(Figure 7), most facilities successfully practice automatic
liquid level, liquid flow rate, and air flow rate control since
fluid regulation is important for proper operation, and
satisfactory flowmeters are readily available. The flow control
systems that are presently available use established designs that
are entirely adequate for wastewater treatment activities.
Process control, however, is used only occasionally in
wastewater treatment. The nationwide survey (summarized in
Figure 7) found that flow-ratio chemical addition, feedback
residual chlorine, and digester temperature control systems
worked well and caused no difficulties. Most plant managers
considered these automatic control systems cost-effective since
they save both energy and chemicals, and improve plant
operation. Automatic feedback dissolved oxygen control
systems effectively reduced oxygenation power consumption,
but some users complained that these systems require con-
siderable probe maintenance. The turbidity and pH control
systems observed during this survey were inadequate and gave
unsatisfactory performance because of faulty system design
and installation. Some of the process control parameters (such
as substrate concentration. MLVSS. and food/microorganism
ratios) that may be the most potentially useful have not been
successfully monitored and controlled in wastewater treatment
plants.
Although a collection of detailed capital costs, operating
improvements, and maintenance data was one of the prime
objectives, fewer than 30% of the surveyed plants had this
information. In all cases, the necessity and cost savings for
automatic flow rate, liquid level, and temperature control
systems were readily apparent. The cost benefits of automatic-
residual chlorine, dissolved oxygen, chemical addition via flow
pacing, and pH, were somewhat more difficult to assess
because of the lack of data. Based on the survey team's
judgement and experiences,the range of potential cost benefits
was estimated for these automatic control systems. Table 4
highlights the control strategies, confidence, advantages, limi-
tations, and recommendations. Most plant superintendents
reported that any operating labor savings due to automatic
control were offset by the added maintenance burdens.
Accordingly, most automatic process control systems should
be evaluated on their chemical and utility savings capabilities.
Central Control
Central control organizes the plant operation so that all
important events, alarms, and treatment information are
displayed, indicated, and recorded in a centralized location
(usually referred to as the Control Room). In addition, most
central facilities practice automatic or remote manual actua-
tion of final control elements. The success of central control is
ensured by the commercial availability of reliable transmitters,
displays, and indicating and recording equipment. Virtually all
the facilities surveyed successfully utilized a high degree of
centralized control. Since centralized control reduces the
number of men required to operate a large treatment plant, it
is one of the few forms of instrumentation that are readily
justifiable on a labor savings basis.
Computers
Modern data-logging systems accumulate, format, record,
and display large quantities of data effectively; consequently,
most new plants have automatic data acquisition systems.
Approximately 20% of the visited facilities used data-logging
computers, and 90% of these users were satisfied with their
automatic data acquisition systems.
124
-------
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
1C
.:
20
25
NO. OF CASES
30 35
40
BUBBLER TYPE LEVEL DETECTORS
DIFFERENTIAL PRESSURE, LEVEL DETECTOR
FLOATS
ALL OTHER LEVEL DETECTORS
WEIRS AND FLUMES
VENTURIS, ORIFICES, NOZZLES, ETC.
MAGNETIC FLOW RATE
[""] UNSATISFACTORY
OTHER FLOW RATE METERS
NUCLEAR RADIATION DENSITY METERS
TRANSMITTING RAIN GAUGES
TEMPERATURE
PRESSURE
ROTATIONAL SPEED
WEIGHT
POSITION
TURBIDITY
CONDUCTIVITY
pH ANDORP
THALLIUM DO PROBE
MEMBRANE DO PROBE
RESIDUAL CHLORINE
OTHER ANALYTICAL SENSORS
^] GAS MONITORS
SAMPLING SYSTEMS
FAIR
n
SATISFACTORY
Figure 6. Performance Summary for Measuring Devices in Wastewater Treatment Service
125
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
50 FACILITIES
1972-3
10
45
20
25
NO. OF CASES
30
35
LIQUID LEVEL CONTROL
LIQUID FLOW RATE CONTROL
SLUDGE PUMPING
AIR FLOW RATE
CHEMICAL ADDITION
RESIDUAL CHLORINE
L3
DISSOLVED OXYGEN
PH
TURBIDITY
AUTOMATIC SCUM REMOVAL
J AUTOMATIC DATA,
SUPERVISORY COMPUTER CONTROL
DIRECT PROCESS CONTROL BY DIGITAL COMPUTER
n
UNSATISFACTORY
FAIR
| _ |
SATISFACTORY
Figure 7. Performance Summary for Automatic Control Loops in Wastewater
Treatment Service
Although process and supervisory control computers have
demonstrated their merits in many industries, they are not
well established in dry-weather treatment plants. Only two of
the surveyed facilities had process control computers, whereas
three storm-water control centers used computerized super-
visory control. All of these computer systems worked well.
Computerization of dry-weather facilities is still in its infancy.
and not enough operating experience has been accumulated to
assess its desirability in wastewater treatment projects.
Computerized supervisory control of large storm and
combined sewer systems is cost-effective because the vast
number of variables and control points exceeds human
computational and decision-making ability within corrective
time limits.
Maintenance
In spite of the low instrument usage rates, wastewater
treatment personnel exhibited a good attitude toward instru-
mentation, as measured by their willingness to use and
maintain the installed instruments. The survey team found
that the treatment plants supplied approximately 90rf of the
maintenance resources needed. Small abandonment rates also
attest to the favorable attitude of wastewater treatment
personnel. Individual plant managers' disposition toward in-
strumentation, however, ranged from poor to excellent.
As a group, satellite storm-water treatment facilities sup-
plied less than adequate maintenance. Possibly because of its
newness, storm-water instrument maintenance is not well
understood. Since none of the satellite facilities started up or
126
-------
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
Table 4. Automatic Control Benefits
Description
feedback compound
residual chlorine
control
Feedback dissolved
oxygen control
pH
Chemical addition via
flow pacing
Settled sludge removal
from clarifiers based
on feedback density
measurements
Potential Savings
25 to 50% savings in Cl
10 to 40% savings in
aeration power
consumption
Essential for pH adjust-
ment processes
15 to 30% chemical
savings
20 to 50% reduction of
sludge volume
pumped
Confidence
High
Moderate
Fair
High
Fair
Basis
Field observations and
engineering judgement
Field observations
Engineering judgement
Engineering judgement
Advantages
Produces good residual
chlorine control;
responds rapidly to flow
changes; highly stable
Energy -effective control
system; responds to
actual oxygen demand
Keeps pH with a reason-
able range of desired
point
Simple, reliable, cost-
effective control sys-
tem; responds to flow
rate variations; stable
control strategy
Produces a dense sludge
with minimum amount
of wate r
Limitation
Residual chlorine ana-
lyzer requires con-
siderable maintenance
DO probes require a
moderate amount of
maintenance; may be
difficult to apply to
some plug flow systems
pH probes tend to foul
in wustewiiter service
Does not respond to
strength variations
Density analyzers
require frequent repair
and maintenance
Recommendations
Suitable for most medium
and larger sized plants
Suitable for most plants,
especially completely
mixed aeration systems
Advisable to use ultra-
sonic cleaning
Suitable for service
wherever demand per
unit volume remains
constant
Useful in large plants
where downstream
sludge processes are
sensitive to water
content
shut down automatically, it would behoove individuals con-
cerned with storm-water treatment facilities to direct more
attention to instruments and automatic devices and to their
maintenance. On the other hand, storm-water control centers.
which typically receive storm-water and combined sewer
network information, were well maintained and operated
satisfactorily.
Although most plants have reasonably well-qualified instru-
ment maintenance staffs, any plans that call for the addition
of sophisticated instruments and automatic control devices
must provide for upgrading the staff's qualifications.
Instrument Budgets
The nationwide survey of 50 wastewater facilities found
that most of the treatment plants used fewer instruments and
automatic control devices than the closely related water
supply and chemical processing plants. The amassed cost data
shows that the average secondary wastewater treatment plant
spends about 3% of its construction costs for installed
instruments, whereas water supply and chemical processing
plants allocate about 6% and 8% respectively, for installed
instruments. Remote satellite wet-weather treatment plants,
which in theory should operate unattended or with just a
minimal amount of operating manpower, budgeted only about
2% for instrumentation and automation.
In all fairness, it must be pointed out that the larger
instrument budgets of the chemical and water supply indus-
tries are due principally to the higher usage rates of physical-
type measurement and control systems (that is, temperature,
tevel, pressure, and flow). Most of the analytical sensors that
the wastewater treatment field needs must detect trace
amounts of a specific substance from a multi-substance
mixture. These analyzers are challenged by a difficult assign-
ment and should be viewed in some respects as being similar to
the early stages of the on-line gas chromatographs, which took
10 years and hundreds of thousands of dollars to develop.
RECOMMENDATIONS
In summary, the field survey team assessed the instrument
utilization rates and performances, and estimated the special
manpower skills, training, and equipment necessary to operate
and maintain instruments and automatic control devices in
wastewater treatment service. Wherever available, the total
control system costs, as well as the economic and performance
benefits obtained, were also noted. With this information a
cost-benefit analysis can be performed to decide whether or
not to use a particular instrumentation or control system.
Moreover, the commercially available on-line analyzers and
automatic control devices were classified, on the basis of the
field experiences, as being: 1) unacceptable, 2) fair, and
3) satisfactory.
The small number of automatic loops observed in the plant
survey attests to the low level of automation that is charac-
teristic of most wastewater treatment plants. The survey's
observations indicate that a lack of sufficiently reliable
analytical sensors for automatic control has impeded process
control efforts. Other commercially available process control
components (such as transmitters, display devices, controllers,
and final control elements) have proved their ability to provide
reliable service in wastewater treatment plants. To help assess
an automatic control loop's desirability, a uniform and easily
practiced record keeping system is badly needed. Also, much
misunderstanding and confusion can be avoided in the future
by using standard instrument symbols and drawings.
An intensive application of elaborate and novel logic
schemes, computers, displays, and recorders will not improve
wastewater treatment effectiveness. Instead, well-documented
field evaluation programs are needed to help ferret out
desirable control systems from the numerous potentially viable
ones.
127
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
DISCUSSION
Robert A. Ryder:
DO control has been utilized at the Reno WWTP since
1966, and at the present time is set to maintain DO = 0.5 mg/1
at the end of the aeration tanks. This has been found to be
highly effective in maintaining a low SVI, preventing bulking,
and producing a good effluent. There is implicit indication
that this is due to ORP suppression of the sludge, inhibiting
certain strict-aerobic filamentous organisms.
Recent experience with a pilot plant of Phostrip with 10-14
hours of quiescent sludge settling, further indicates great
stability of the sludge, low SVI's and excellent BOD removal.
Scum has been noted at many activated sludge plants, and
seems to be a function of very long sludge age or SRT. More
research is needed on this problem.
N. J. Biscan:
Mr. Roesler mentioned the need for monitors and alarms
which alert the operator to toxic substances entering the
treatment facility. I'd like to point out that we at Dow have
developed such an instrument under EPA Grant No. S800 766,
"Optimizing a Petrochemical Waste Bio-oxidation System
Through Automation."
The instrument operates on a one-hour time cycle and is
based on a measure of the oxygen consumed in the oxidation
of the organic substrate in the samples added. Briefly, the
instrument operates by first determining the area under an
oxygen-uptake curve of a sample of standard biodegradable
substrate, then a feed sample, and finally the standard sample
again. The ratio of the standard areas gives the indication of
toxicity. Questions that remain to be answered, however, are
the following:
1. Is there in fact a significant problem with toxicity in feeds
to municipal activated sludge plants, and what is the
potential demand for a toxicity indicator?
2. How can instrument companies market an instrument,
which was developed under an EPA grant, without the
company obtaining patents or licensing rights?
Edmond P. Lomasney:
I should like to comment on the subject of utilizing gaseous
products that are generated in the anaerobic process.
Recently completed research has indicated that the
dynamics of the anaerobic process in sewage treatment have
been incorrectly interpreted. The basic concepts of the be-
havior of the bacteria involved indicates that those bacteria
which participate also manufacture combustible gases. These
gases, methane and hydrogen, are products generated in this
process, and the efficiency of the process is indirectly related
to the ratios of these gases (excess hydrogen acting as an
inhibitor to the performance of the methanogenic bacteria).
As a consequence, a gaseous monitoring procedure for
detecting and measuring these gases is a necessity for the
continuous efficient and effective performance of the
anaerobic process.
Ronald N. Doty:
Much work has been done to evaluate available instruments
and to determine what is needed. However, we seem to be
experiencing a dichotomy which will negate much ot the R &
D and evaluation work which has, and is continuing to be
done: the construction grants division of EPA. and the States.
force us to write open specifications, thus precluding us from
specifying that equipment which has been proven superior.
L. A. Schafer:
Although setting standards for wastewater instrumentation
can be restrictive, some standards (similar, for instance, to fire
or electrical regulations) could well be applied, subject to the
following rules:
1. The standard-setting body should include both theoretical
and practical people.
2. Standards should be firm, but with an autonomous applica-
tion assessing procedure for evaluation of results and
provisions for exceptions.
3. Standards should be proposed and discussed by a cross-
section of users before adoption.
Some standards which might profitably be adopted are:
standard symbols, standard classification of facility plans by
disciplines (Civil, Structural, Architectural. Mechanical, Elec-
trical, Instrument (or Control) and General), a standard
selection of instrument signal ranges (4-20 ma. 10-50 ma.
315 psi, 2080 inches vac., etc.). a requirement that all
facilities above a certain size contain a signal interface area (Le.
terminal strips) suitable for connection of temporary com-
puters or data loggers, etc.
Richard R. Keppler:
In view of the large amount of discussion that has
concerned itself with the proper maintenance of monitoring
equipment utilized for wastewater treatment systems, my
experience in the petroleum refining industry might be
helpful, since a similar problem existed in this industry about
15 or 20 years ago. It was the custom at that time to utilize,
for maintenance purposes, separate groups, such as the
insrument department, and to have these various maintenance
groups respond to requests from the operators for main-
tenance required. This was ineffective in that the insrument
people, for example, would consider that some of the
problems which resulted in breakdown of their equipment
were due to poor operator performance and likewise the
operators said that an upset on the equipment was a result of
poor maintenance on the part of the instrument people.
Therefore, there was a continuous discussion between the
128
-------
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
maintenance people and the operating people concerning the
responsibility for the proper operation of the equipment.
Later, in an effort to overcome this problem, it became the
industry's pattern to contract to outside organizations for
maintenance of equipment in the operation of a refinery.
Although possibly more expert and skilled maintenance
became available by this means, it did not eliminate the
fundamental problem of one man blaming another for poor
operation.
In the past several years there has developed a so-called
"refinery technician"this man begins as an operator and
learns additional mechanical skills, receiving an increase in pay
for each additional skill that he masters. In this way the
operator is also a machinist and instrument repair man, and
therefore takes great interest in proper maintenance of the
equipment which he is required to operate. It is my suggestion
that this kind of philosophy be adopted by the municipalities
operating wastewater treatment systems, in an effort to
improve the level of maintenance and performance of those
systems. This combined "refinery technician" has resulted in a
significant improvement in efficiency of refinery operations.
J. F. Andrews:
The discusser would like to suggest that in order to obtain
maximum benefit from the automation of wastewater treat-
ment plants it will be necessary to significantly upgrade the
quality of personnel in plant operations. One method of doing
this would be to involve more professional engineers directly
in plant operations. Unfortunately, with the exception of our
large cities, the major portion of environmental engineering
profession has frequently been guilty of considering plant
operations as a subprofessional area not worthy of significant
engineering attention.
Most environmental engineering programs in universities
have considered wastewater treatment plant operations to be a
low-level task which is not suited for education or research in
universities. Those few universities which have become in-
volved in plant operations have usually done so at the
technician or "operator training" level by offering "short
courses" or correspondence courses. These courses have
provided a valuable public service; however, they may also be
partially responsible for the lack of recognition of the
importance of the engineer in treatment plant operations. The
major portion of the material taught in these courses is at the
technician and not the engineering level and would best be
taught at the vocational high school or two-year technical
school level. There is a strong trend in this direction at the
present time and it is expected that this trend will continue.
What, then, would be the major differences between the
traditional environmental engineering curriculum and a cur-
riculum emphasizing plant operations? There would, perhaps,
be as many different opinions on this point as there are
environmental engineering professors; however, in the discus-
ser's opinion the major differences would be an increased
emphasis on the dynamic behavior of plants and techniques,
such as the application of control systems, to improve
dynamic behavior. Most current design criteria, both as taught
in universities and practised in the field, are based on average
inputs or, at best, maximum and minimum inputs and have
not directly considered that inputs to the plant are highly
variable with respect to time. These variations are the primary
reason why control is needed and they must be considered in
design as well as in operation. It should be noted that a course
in Process Dynamics and Control is commonly found in most
chemical engineering curricula and, in the discusser's opinion.
we would be well advised to include a course in "Dynamics
and Control of Wastewater Treatment Systems" in environ-
mental engineering curricula.
In addition to providing educational programs for plant
operations engineers, universities should also become more
involved in research on plant operations. Most environmental
engineering programs have concentrated on design-oriented
research and have neglected operations-oriented research,
perhaps because of the assumption that this is not worthy of
Ph.D.-level studies with which most university research is
associated. However, the discusser's experience has been that
this type of research is more difficult, from both theoretical
and experimental points of view, than the usual scientific or
design-oriented research. The discusser has had a total of nine
Ph.D. students who have conducted their research in either the
dynamics or control of wastewater treatment processes.
Although much is still to be learned, since the theory put forth
in these Ph.D. dissertations has for the most part yet to be
tested in the field, a beginning has been made toward a
quantitative description of the dynamic behavior of treatment
processes, and control systems for improvement of this
behavior.
129
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
Report of Working Party
On
RESEARCH NEEDS
FOR
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
Walter G. Gilbert
Chief, Municipal Operations Branch,
Environmental Protection Agency, Washington, DC 20460
James A. Mueller
Associate Professor, Environmental Engineering & Science Program,
Manhattan College, Bronx, NY 10471
The success of utilizing instrumentation and automation in
the operation of wastewater treatment facilities will be
measured to a large degree by the capability to produce better
treatment in a more effective and efficient manner. It is
important to note that the discussions of this workshop
recognized the fact that automation can mean many different
things, from the simple and reliable instrumentation needed for
the smallest plant to the complex, computer-assisted operation
of our largest facilities. There will be many variations in
between, depending upon the size, complexity, and many
other variables of the facility involved. Measuring the effec-
tiveness of automation, therefore, involves many different
factors.
PROBLEMS
In attempts to evaluate the effectiveness of automation for
wastewater treatment, problems are encountered in three
broad areas as follows: (1) effluent quality improvement, (2)
cost-effectiveness, and (3) human-machine interface. Specific
statements and relative priorities of these problem areas are
discussed more fully below.
1. Determine the extent of effluent quality improvement,
lower contaminant concentrations and reduced variability
attainable through the use of instrumentation and automa-
tion.
This problem area should receive the highest priority in
future automation research in keeping with the goals of the
Federal Water Quality Act Amendment of 1972 (PL
92-500) for improvement of our Nation's water quality.
Regardless of other constraints, if increased automation of
wastewater treatment systems reduces effluent concentra-
tions and variability, its effectiveness will have been
demonstrated and future utilization insured.
In applying automation to attain the above goals, the
interaction of the collection system and treatment plant
must be considered. Improvement of existing systems by
increased utilization of automation with appropriate con-
trol strategies and required system modifications must be
evaluated. New system development incorporating process
and system theory with required automation should be
encouraged.
2. Evaluate the cost-effectiveness of incorporating automation
into wastewater systems which are required to meet a
specified performance objective.
Existing and future wastewater systems typically are
required to meet performance specifications due to either
administrative or legislative policy or local allocation based
on receiving water quality objectives. To meet this specified
performance objective the type and degree of automation
that should be employed in a wastewater system to yield
reduced system costs must be evaluated. Unless this is
accomplished, a manager will rarely consider employing
automation in his system if it is already meeting per-
formance objectives with an existing degree of manual
operation.
To accurately evaluate the cost-effectiveness of increased
automation, a uniform set of criteria incorporating all
economical factors must be developed. Systems must also
be evaluated to determine the most cost-effective location
of instrumentation as well as type.
3. Evaluate the effectiveness of the human-machine interface.
If treatment plant automation is to have the desired
degree of effectiveness, interactions of the wastewater
system operators, designers, and managers must be con-
sidered. Due to its importance, this problem area has the
same priority as the above cost-effectiveness area. If a plant
operator lacks the training or desire to effectively utilize and
maintain instrumentation, the automation system will fail.
If a designer does not obtain operator feedback with
respect to automation effectiveness, or long-term data
feedback in a readily utilizable fashion, the effectiveness of
automation will not be properly evaluated and future
process and automation improvements will not result. If a
manager does not allocate sufficient funds nor personnel to
adequately maintain and utilize the automated facilities, a
significant portion of the capital outlay for the plant will be
wasted and it will not meet treatment objectives.
Therefore, to insure the effectiveness of automation,
adequate operator training with data feedback to both
designer and managers must be employed.
130
-------
EVALUATION OF THE EFFECTIVENESS OF AUTOMATION
RESEARCH NEEDS
To respond to the problem areas discussed above, research
needs were identified which must be satisfied if we are to be
able to fully evaluate the effectiveness of automated systems
for plant operations. These needs are addressed below in
priority order.
1. Develop control strategies to utilize instrumentation and
automation in achieving maximum process reliability.
Control strategies are needed that .can fully utilize the
information and response capabilities derived from vary-
ing degrees of instrumentation and automation to provide
the process control data necessary for good plant opera-
tion. Such strategies must utilize the short response times
that would be available through such automation to react to
the widely varying extremes in influent quality, so as to
produce a more consistent effluent quality. The develop-
ment of such strategies must also recognize the need for
varying degrees of automation for different sizes and
types of plants. Achieving process reliability is an inherent
part of this need.
2. Investigate the full utilization of surge capacity or flow
equalization in maximizing plant performance. Consider
the use of in-line or off-line storage or control, with
determination of the optimal location of such control.
Determine to what extent treatment facilities can accept
varying loads and produce acceptable effluent quality.
Conduct field studies required for verification.
It is essential that feasible means of control over
influent quantity and quality be evaluated, along with
related impacts on plant operations. Determination must
be made on the degree of control needed to achieve the
desired end result. Means of automating such control
systems, which provide feedback to treatment process
control, are needed.
3. Establish a data collection system with appropriate
feedback mechanisms to design engineers for continued
evaluation of plant performance and use in improving
future plant design. The system should be assembled with
adequate correlation of related data elements for ease of
retrieval.
Although this need satisfies essentially a long-range
objective, it is necessary to start it now so as to fully
utilize much of the historical data now being collected.
4. Develop uniform criteria for evaluation of the cost-effec-
tiveness of varying degrees and types of facility automa-
tion.
Cost-effectiveness is determined on a comparative
basis between alternative systems. Uniformity of evalua-
tion is essential to allow ease of comparison and allow for
final selection of instrumentation and control systems.
Inherent to this research need is the identification and
quantification in economic terms of the benefits which
accrue from use of instrumentation or automation in
wastewater collection, treatment, and disposal systems.
5. Identify areas of potential cost saving in plant operations
to guide priorities in developing instrumentation and
control systems. Determine and investigate in detail the
most cost-effective points of application of instrumenta-
tion in wastewater treatment processes.
Such efforts are needed to provide a guide in
determining the most effective course of action in
developing instrumentation and automation for any given
treatment facility.
6. Evaluate the effectiveness of existing automated systems
based on operator feedback.
The operator must be recognized as a key element of
the total control system. Such systems must be evaluated
with this in mind. Optimization of automated systems
must consider the role of the operator in the control loop.
7. Investigate upstream monitoring for predictive purposes
with adequate lead times for process control.
As well as providing information for immediate
reaction to influent quality variations, such monitoring
would provide historical data needed to develop effective
process control strategies needed to respond to such
variations. Efforts should also include a determination of
the types of monitoring required, including both quality
and quantity parameters.
8. Identify and evaluate available sensors and instruments as
applied in wastewater treatment processes in terms of
their performance and reliability.
This relates directly to the need for standard tests and
specifications identified in earlier workshop groups.
9. Develop the research input to a training program to
permit full utilization of instrumentation and automation.
A good training program is essential to develop the
proper operator attitude and encourage operator ac-
ceptance of automated systems. Such training must also
address the technical details of maintenance and calibra-
tion needs of automation as well as the use of the control
system to improve performance.
10. Develop new and innovative treatment processes utilizing
process theory in conjunction with the capabilities of
instrumentation and automation.
This is conceived as a need to develop new treatment
processes and methods based on a "no holds barred"
approach. It is essentially pure research free of the
constraints of "traditional" design of treatment facilities.
11. Develop improved models to more closely simulate
physical, chemical, and biological processes using measura-
ble parameters. Utilize these models for process develop-
ment, process control, and research requirements.
Satisfaction of this need will help accelerate the
process of evaluating the effectiveness of alternative
automation strategies.
PRIORITIES
Priorities for the research needs identified above were
assigned primarily on the basis of the need for immediate,
131
-------
AUTOMATION OF WASTEWATER TREATMENT SYSTEMS
short-term answers to certain pressing questions regarding the can proceed in a more logical sequence. Lower priority was
role of instrumentation and automation in plant operations. The assigned to needs with longer-range benefits or where the
framework for procedures to evaluate the effectiveness of such beneficial effects of such efforts may be largely unknown at
systems must be established now so that development work this time.
132
-------
LIST OF PARTICIPANTS
ANDERSON, James J., President
Watermation, Inc.
1404 East 9th Street
Cleveland, OH 44114
ANDREWS, John F., Professor
Environmental Systems Engineering
Clemson University
(present address:
Dept. of Civil Engineering
University of Houston
Houston, TX 7 7004)
AUSTIN, John H.. Department Head
Environmental Systems Engineering
Clemson University
Clemson, SC 29631
BABCOCK, Russell H., Chief Engineer
C. E. Maguire Inc.
60 First Avenue
Walt ham, MA 02154
BALLOTT1, Elmer F., Partner
Greeley and Hansen Engineers
222 South Riverside Plaza
Chicago, 1L 60606
BARNES, George D.
Assistant Dirctor of Public Works
303 City Hall
Atlanta, GA 30303
BERTHOUEX, Paul Mac, Associate Professor
Civil and Environmental Engineering
University of Wisconsin
Madison, WI 53706
BERTRAM, William H.
Supervisory Sanitary Engineer
Environmental Protection Agency
1860 Lincoln Street
Denver, CO 80203
BISCAN, N. J., Research Specialist
Dow Chemical U.S.A.
A-l 127 Building
Freeport,TX 77541
BLAKELY, Christopher P., Market Manager
Water Management
Honeywell, Inc.
1100 Virginia Drive
Ft. Washington, PA 19034
BREIDENBACH, Andrew W., Director
National Environmental Research Center
Environmental Protection Agency
Cincinnati, OH 45 268
BRUBAKER, Jay H., Section Leader
Union Carbide Corp.
P.O. Box 8361
South Charleston, WV 25303
BRYANT, James O.
Municipal Operations Branch
Environmental Protection Agency
Springfield. VA 22153
BUHR. Heinrich O.
Visiting Associate Professor
Environmental Systems Engineering
Clemson University
(on leave from:
Dept. of Chemical Engineering
University of Cape Town
Rondebosch. CP 7700
Republic of South Africa)
CARKEEK, John G.? Associate
Consoer Townsend & Associates
360 East Grand Avenue
Chicago, IL 60611
CARROLL, Leo J., Manager
Environmental Division
Fischer & Porter Co.
Warminstei. PA 18974
COHEN, Al, President
Astro Ecology Corp.
801 Link Road
League City. TX 77573
CONVERY. John, Director
Advanced Waste Treatment
Research Laboratory
National Environmental Research Center
Environmental Protection Agency
Cincinnati, OH 45268
CROW, Merle
Stevens T. Mason Building
Lansing, MI 48926
DAY, Robert E., Project Manager
Black, Crow & Eidsness, Inc.
Penn Towers-Suite 434
1819 John F. Kennedy Blvd.
Philadelphia, PA 19103
DENIT, Jeffery D., Chief
Impact Analysis Section
Environmental Protection Agency
Washington, DC 20402
DICK, Richard I., Professor
Department of Civil Engineering
University of Delaware
Newark, DE 19711
DOBBINS, William E., President
Teetor-Dobbins, P.C.
515 Johnson Avenue
Bohemia, NY 11716
DOTY, Ronald N.
Supt. Water Quality Control
250 Hamilton Street
Palo Alto, C'A 94301
DU CROS, Michael. Marketing Manager
Technicon
Benedict Avenue
Tanytown, NY 10591
ECKENFELDER, Wesley W.. Jr.. Professor
Environmental & Resources Engr.
Vanderbilt University
Nashville, TN 37235
ELSAHRAGTY, Mohammed
Quirk. Lawler & Matusky Engineers
415 Route 303
Tappan. NY 10983
FARRELL, J.B., Chief
Ultimate Disposal Section
Advanced Waste Treatment Research Laboratory
National Environmental Research Center
Environmental Protection Agency
Cincinnati, OH 45268
FERTIK, Harry A., Senior Scientist
Leeds & Northrup Company
Technical Center
North Wales, PA 19454
FETCH, John J.
Capital Controls Co.
Div. Dart Ind.
Advance Lane
Colmar, PA 18915
FLANAGAN, Michael J.
Brown & Caldwell Consulting Engineers
66 Mint Street
San Francisco. CA 94103
FREEMAN, Mark P., Senior Scientist
Dorr-Oliver Incorporated
77 Havemeyer Lane
Stamford, CT 06904
GARRISON, Walter E., Assistant Chief Engineer
Sanitation Districts of L. A. County
1955 Workman Mill Road
Whittier, CA 90601
GILBERT, Walter G., Chief
Municipal Operations Branch
Environmental Protection Agency
Washington, DC 20460
OILMAN, Harold D.
Greeley and Hansen Engineers
Six Penn Center Plaza
Philadelphia, PA 19103
133
-------
GLENN, Don
Space Division
General Electric Company
P. O. Box 8555
Philadelphia, PA 19101
GRAEF, Stephen P.,
Principal Sanitary Engineer
Metropolitan Sanitary District of
Greater Chicago
100 East Erie
Chicago, IL 60611
GUARINO, Carmen F., Commissioner
Philadelphia Water Department
1160 Municipal Services Building
Philadelphia, PA 19107
GULEVICH, Whdimir, Chief
Water Quality Engineering Division
U. S. Army Environmental Hygiene Agency
Aberdeen Proving Ground, MD 21010
HAYES, Charles H.
Assistant General Manager
Gulf Coast Waste Disposal Authority
910 Bay Area Blvd.
Houston, TX 77058
HUMPHREY, Marshall F.
Jet Propulsion Laboratory
4800 Oak Grove Drive
Pasadena, CA 91103
JACOBSON, Kenneth J.
Dept. of Chemical Engineering
University of Pennsylvania
Philadelphia, PA 19104
JOYCE, R. J., Production Manager
Dohrmann-Envirotech
3240 Scott Blvd.
Santa Clara, CA 95050
KEINATH, Thomas M., Associate Professor
Environmental Systems Engineering
Clemson University
Ciemson, SC 29631
KEPPLER, Richard R.
Research and Monitoring
Environmental Protection Agency
Boston, MA 02203
KEY, Phi
FMC Corporation
Denver, CO 80210
KUGELMAN, Irwin J., Chief
Pilot Plant Field Investigation Program
National Environmental Research Center
Environmental Protection Agency
Cincinnati, OH 45268
KURLAND, Paul
Delta Scientific Corporation
1172 Route 109
Lindenhurst, NY 11757
LAGER, John A., Vice President
MetcalfA Eddy, Inc.
1029 Corporation Way
Palo Alto, CA 94303
LEISER, Curtis P.
Manager of Computer Services
Seattle Metro
410 West Harrison Street
Seattle, WA 98119
LEJEUNE, T., Manager
R. M. Clayton WPC Plant
2240 Bolton Road, NW
Atlanta, GA 30318
LOMASNEY, Edmond P.
R & D Program Director
Environmental Protection Agency
1421 Peachtree Street, NE
Atlanta, GA 30309
MANKES, Bob
Vice President, Sales
Delta Scientific Corporation
1172 Rt. 109
Lindenhurst, NY 11757
MASTERS, Hugh E.
Storm & Combined Sewer Section
Environmental Protection Agency
Woodbridge Avenue
Edison, NJ 08817
MATHEWS, Michael B.
Sanitation Superintendent
City of Ventura
P. O. Box 99
Ventura, CA 93001
MATSON, Jack, Assistant Professor
Civil Engineering Department
University of Houston
Houston, TX 77004
McKNIGHT, M. Dolan, Plant Superintendent
Fort Worth Water Dept.
P. O. Box 870
Fort Worth, TX 76101
MEYER, John M.
Vice President, Marketing
Systems Control, Inc.
1810 Page Mifl Road
Pato Alto, CA 94304
MILLER, G. Wake, Project Director
Public Technology. Inc.
1140 Connecticut Ave., NW
Suite 804
Washington, DC 20036
MOLVAR, Alten E.
Environmental R & D Director
Rat neon Company
P. O. Box 360
West Main Road
Portsmouth, RI 02871
MONTAGUE, Albert, Director
Office of Research & Development
Region III
Environmental Protection Agency
6th and Walnut Streets
Philadelphia, PA 19106
MOSS, J. E.
Union Carbide
P.O. Box 8361
South Charleston, WV 25 303
MUELLER, James A., Associate Professor
Environmental Engineering & Science Program
Manhattan College
Bronx, NY 10471
NELSON, John K.
Assistant Director Planning Operations
Metro Denver
3100 East 60th Avenue
Commerce City, CO 80022
NORKIS, Charles M., Sanitary Engineer
Philadelphia Water Department
Broad and Arch Streets
Philadelphia, PA 19106
OLSSON, Gustaf, Assistant Professor
Division of Automatic Control
Lund University
Lund 7, Sweden
OPATKEN, Edward
Municipal Pollution Control Division
Environmental Protection Agency
Washington, DC 20402
PALLESEN, Lars
Dept. of Statistics
University of Wisconsin
Madison, WI 5 3706
PEIL, Kefly M.
CPT/Engineering Branch Chief
USA Med Bioengineering R & D Laboratory
Ft. Detrick
Frederick, MD 21701
PETERSEN, W.Carl
Associate-Instrumentation
Consoer Townsend & Associates
360 Grand Avenue
Chicago, IL 60611
POLTA, Robert C., Research Engineer
Metropolitan Sewer Board
350 Metro Square Building
7th and Robert Streets
St. Paul, MN 55101
QUIGLEY.John
Assistant Professor of Engineering
University of Wisconsin
Madison, WI 5 3706
RADEMACHER, John
Department of Natural Resources
State of Maryland
Annapolis, MD 21401
RICHARD, Dan
Technology Applications Programs
NASA Headquarters
Washington, DC 20007
RISLEY, Clifford
Region V
Environmental Protection Agency
Chicago, IL 60606
134
-------
ROESLER, Joseph F.
Advanced Waste Treatment Research Laboratory
National Environmental Research Center
Environmental Protection Agency
Cincinnati, OH 45 268
ROSENKRANZ, William, A., Director
Municipal Pollution Control Division
Environmental Protection Agency
Washington, DC 20460
ROTH, John A., Chairman
Division of Chemical, Fluid and
Thermal Sciences
Vanderbilt University
Nashville, TN 37235
RYCKMAN, D. W., President
Ryckman, Edgerley, Tomlinson & Associates
12161 Lackland Road
St. Louis, MO 63141
RYDER, Robert A., Vice President
Kennedy Engineers, Inc.
657 Howard Street
San Francisco, CA 94105
SAKO, Frank F., Staff Engineer
FMC Corporation
1185 Coleman Avenue
Santa Clara, CA 9505 2
SALLOUM, J. Duane, Director
Wastewater Technology Centre
Environment Canada
Ottawa, Ontario KIA OH3
CANADA
SCHAFER, Lawrence A.
Principal Engineer
C. E. Maguire, Inc.
60 First Avenue
Waltham, MA02154
SCHUK, Walter W.
Pilot Plant
Environmental Protection Agency
5000 Overlook Avenue
Washington, DC 20032
SMITH, John M., Chief
Municipal Treatment and Reuse
National Environmental Research Center
Environmental Protection Agency
Cincinnati, OH 45268
SMITH, Wayne
Process Control Branch
National Field Investigations Center
Environmental Protection Agency
Denver, CO 80225
SORENSEN, Poul Erik, Chemical Engineer
Water Quality Research Institute
Academy of Technical Sciences
Soborg, Denmark
STAMBERG.J.
Office of Research & Monitoring
Environmental Protection Agency
Washington, DC 20460
STENSTROM, Michael K.
Environmental Systems Engineering
Clemson University
Clemson, SC 29631
STEPHENS, Ed, Associate
Clinton Bogert Associates
2125 Center Avenue
Fort Lee, NJ 07024
SWEENEY, Robert F.
Associate Professor
Dept. of Chemical Engineering
Villanova University
Villanova, PA 19085
TAYLOR, Reuben
Water & Wastewater Systems
Urban Systems Project Office
NASA, Johnson Space Center
Houston, TX 7705 8
TOBIN, Patrick
Municipal Pollution Control Division
Environmental Protection Agency
Washington, DC 20460
TOMCZYK, Harry
Senior Electrical Engineer
Metropolitan Sanitary District of Chicago
100 East Erie
Chicago, IL60611
TORNO, Harry
Office of Research & Monitoring
Environmental Protection Agency
Washington, DC 20460
TRAX, John
Office of Water Programs
Environmental Protection Agency
Washington, DC 20460
TREUPEL.HansW.
Senior Physicist
IBM Corporation
18100 Frederick Pike
Gaithersburg, MD 20760
WEST, A. W.
Office of Enforcement and General Counsel
National Field Investigations Center
Environmental Protection Agency
Cincinnati, OH 45 268
WILKINS, Judd R., Microbiologist
NASA
Langley Research Center
Hampton, VA 23665
WISE, Robert H., Research Chemist
National Environmental Research Center
Environmental Protection Agency
Cincinnati, OH 45 268
WOODRUFF, Paul H., President
Roy F. Weston, Inc.
WestonWay
West Chester, PA 19380
WRIGHT, Darwin R.
Municipal Pollution Control Division
Environmental Protection Agency
Washington, DC 20460
YOUNG, Pet erC.
Centre for Resource and Environmental Studies
Australian National University
Canberra, Australia
Z1CKEFOOSE, Charles S.
Operation Consultant
Stevens Thompson & Runyan, Inc.
5505 S. E. Milwaukie
Portland, OR 97202
135
------- |