U.S. DEPARTMENT OF COMMERCE
National Technical Information Service
PB-249 049
FUTURE DIRECTION OF URBAN WATER MODELS
WATER RESOURCES ENGINEERS, INCORPORATED
PREPARED FOR
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
)
FEBRUARY 1976
-------
KEEP UP TO DATE
Between the time you ordered this report—
which is only one of the hundreds of 'thou-
sands in the NTIS information collection avail-
able to you—and the time you are reading
this message, sjeveral new reports relevant to
your interests probably have entered the col-
lection.
Subscribe to the Weekly Government
Abstracts series that will bring you sum-
maries of new reports as soon as they are
received by NTIS from the originators of the
research. The WGA's are an NTIS weekly
newsletter service covering the most recent
research findings in 25 areas of industrial,
technological, and sociological interest—
invaluable information for executives and
professionals who must keep up to date.
The executive and professional informa-
tion service provided by NTIS in the Weekly
Government Abstracts newsletters will give
you thorough and comprehensive coverage
of government-conducted or sponsored re-
search activities. And you'll get this impor-
tant information within two weeks of the time
it's released by originating agencies.
WGA newsletters are computer produced
and electronically photocomposed to slash
the time gap between the release of a report
and its availability. You can learn about
technical innovations immediately—and use
them in the most meaningful and productive
ways possible for your organization. Please
request NTIS-PR-205/PCW for more infor-
mation.
The weekly newsletter series will keep you '
current. But learn what you have missed in
the past by ordering a computer NTISearch
of all the research reports in your area of
interest, dating as far back as 1964, if you
wish. Please request NTIS-PR-186/PCN for
more information.
WRITE: Managing Editor
5285 Port Royal Road
Springfield, VA 22161
Keep Up To Date With SRIM
SRIM (Selected Research in Microfiche)
provides you with regular, automatic distri-
bution of the complete texts of NTIS research
reports only in the subject areas you select.
SRIM covers almost all Government re-
search reports by subject area and/or the
originating Federal or local government
agency. You may subscribe by any category
or subcategory of our WGA (Weekly Govern-
ment Abstracts) or Government Reports
Announcements and Index categories, or to
the reports issued by a particular agency
such as the Department of Defense, Federal
Energy Administration, or Environmental
Protection Agency. Other options that will
give you greater selectivity are available on
request.
The cost of SRIM service is only 45£
domestic (60£ foreign) for each complete
microfiched report. Your SRIM service begins
as soon as your order is received arid proc-
essed and you will receive biweekly ship-
ments thereafter. If you wish, your service
will be backdated to furnish you microfiche
of reports issued, earlier.
Because of contractual arrangements'with
several Special Technology Groups, not all
NTIS reports are distributed in the SRIM
program. You will receive a notice in your
microfiche shipments identifying the excep-
tionally priced reports not available through
SRIM.
A deposit account with NTIS is required
before this service can be initiated. If you
have specific questions concerning this serv-
ice, please call (703) 451-1558, or write NTIS,
attention SRIM Product Manager.
This information product distributed by
-------
EPA-600/2-76-058
February 1976
FUTURE DIRECTION OF URBAN WATER MODELS
by
Michael B. Sonnen
Larry A. Roesner
Robert P. Shubinski
Water Resources Engineers, Inc.
Walnut Creek, California 94596
Contract No. 68-03-0499
Project Officer
Chi-Yuan Fan
Storm and Combined Sewer Section (Edison, N. J.)
Municipal Environmental Research Laboratory
Cincinnati, Ohio 45268
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
L..
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
" --i.-CflTNO. 2.
L HPA-600/2-76-058
M _» AND SUBTITLE'
\ FUTURE DIRECTION OF URBAN WATER MODELS
J
\ ...,.-...,,.
r. ,VJTHOR(S) Michael B. bonnen
Larry A. Roesner
' Robert P. Shubinski
- ^--RFORMING ORG'VNIZATION NAME AND ADDRESS
Water Resources Engineers , Inc.
710 South Broadway, Suite 200
Walnut Creek, California 94596
| ' J. SPONSORING AGENCY NAME AND ADDRESS
1 Municipal Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
3. RECIPIENT'S ACCESSIONING. . J
1
5. REPORT DATE
February 1976 !
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
12490
10. PROGRAM ELEMENT NO.
1BB034;ROAP 21-ATA;Task 027
1 1 . CONTRACT/GRANT NO. |
68-03-0499 . j
13. TYPE OF REPORT AND PERIOD COVERED j
Final j
14. SPONSORING AGENCY CODE ;
EPA-ORD I
(5, SUPPLEMENTARY NOTES \
<
a. ABSTRACT • • ',
The state-of-the-art of urban water modeling since 1968 was reviewed. Urban ,
water supply; water distribution facilities; water use; waste collection and
conveyance; waste treatment; receiving \vaters; and water reuse. Future urban
water models were suggested, from 1) the review and, 2) demonstrated needs for
additional problem solving capabilities. A phased model implementation program for
the EPA's Storm and Combined Sewer Section was suggested. Contains 161 references.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
' 'nhematical models, Urban areas,
, Isderal budgets, Reviews, Water supply,
I Waste disposal, Planning, Operations,
• Maintenance, Systems engineering
State-of-the-art review,
Urban water management, '
Runoff quality, Water
resources economics
Water management, Storm
runoff, Water reuse
Optimum design
13-B-
-liTRIBUTION STATEMENT
release to public
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
SPA Form 2220-1 (9-73)
-------
FOREWORD
Man and his environment must be protected from the adverse
effects of pesticides, radiation, noise and other forms of pollution,
and the unwise management of solid waste. Efforts to protect the
enyironment require a focus that recognizes the interplay between
the components of our physical environment--air, water, and land.
The Municipal Environmental Research Laboratory provides this
multidisciplinary focus through programs engaged in
studies on the effects of environmental contaminants
on man and the biosphere, and
a search for ways to prevent contamination and to
recycle valuable resources.
In this study, a review was made for the Storm and Combined
Sewer Section concerning existing urban water mathematical modeling
capability. From this review, gaps in needed modeling technology were
identified, and a philosophical approach to filling those gaps was
developed. Finally, a phased implementation program for developing
the needed models was suggested.
Louis W. Lefke
Acting Director
Municipal Environmental
Research Laboratory
iii
-------
ABSTRACT
A literature review was made to assess the state-of-the-art of
urban water mathematical modeling since 1968. Particular urban water
phenomena or subsystems addressed included: rainfall, runoff, and
snowmelt; urban watersheds; water supply; water distribution systems;
water use; waste collection and conveyance; waste treatment; receiving
waters; and water reuse.
Current model development work and uses of models, as well as
the problems being addressed with models, were also reviewed.
Future urban water models were suggested based on 1) the state-
of-the-art review and 2) the demonstrated need for ability to treat
certain existing planning and design problems.
A phased implementation program for model development was
suggested for the Storm and Combined Sewer Section of the U. S.
Environmental Protection Agency. This program took cognizance of
both the Section's .current and historical programs and the needs
identified earlier in the report.
This report was submitted in fulfillment of Contract Number .
68-03-0499 by "Water Resources Engineers, Inc. under the sponsorship
of the Environmental Protection Agency. Work was completed as-of
June 1975.
IV
-------
CONTENTS
Page
Foreword iii
Abstract iv
List of Figures vi
List of Tables vi
Acknowledgments vii
Sections
I Conclusions 1
II Recommendations 3
III Background of Study 5
IV Modeling Philosophy 9
V State-of-the-Art 29
VI Future Urban Water Models 57
VII Phased Implementation Program 68
VIII References 72
-------
_ i i'ie jjitcnsity oi Ui'ban 11
-------
ACKNOWLEDGMENTS
Mr. Murray B. McPherson served as an advisor to this project.
He is the Director of the Urban Water Resources Research Program
of the American Society of Civil Engineers. His review of literature
on water distribution systems, which was used here substantially as
he prepared it, was supported by Contract No. 14-31-0001-4224,
Office of Water Research and Technology, U. S. Department of the
. Interior. The writers wish to thank him not only for this section
of the report, but also for his continued helpful suggestions throughout
the project, and for his review of the draft final report.
Additionally, during the course of this work, many current users
of urban water models were contacted and asked their opinion about
future needs in this area. Others who are "front-line" water and
wastewater managers -and who constantly face the problems needing
earliest attention were also contacted, whether they were currently
using models or not. Among this group, there were many who took
special and lengthy efforts to respond. These included John J. Bailey,
Jr., Dan Brock, Neil Grigg, Brendan Harley, Stifel Jens, D. Earl
Jones, Robert McWhinnie, George Smoot, L. Scott Tucker, and
Harry Wenzel. These gentlemen are thanked for their help, as are
the numerous unnamed persons with whom this topic was discussed
generally over a period of many months.
The Project Officer for EPA was Chi-Yuan Fan, who provided
technical guidance and product review throughout the project. Richard
Field also contributed product review for EPA; and the improvements
made through their knowledgeable backgrounds and experiences are
acknowledged with gratitude.
Will B. Betchart of WRE reviewed many of the background
literature citations and prepared a written synthesis of that work.
Vll
-------
SECTION I
CONCLUSIONS
This project included a review of literature and numerous
discussions with model developers and model users who are
currently improving the state-of-the-art of urban water modeling.
From analysis of information gathered from these sources, the
following conclusions are drawn.
1) Emphasis in urban water modeling should remain for the
immediate future with simulation of discrete processes
occurring in the overall urban setting. That is, we should
continue to simulate subsystems before attempting simu-
lation or formal optimization of connected subsystems.
2) In the past, rather serious communications problems have
arisen between model developers and subsequent users of
those models who know little or nothing of the program's
coding, assumptions, data requirements, or limitations.
Even documentation reports and user's manuals intended
to avoid these problems have not been adequate. Users
of programs of all types developed by second parties have
inadvertently attempted to use the programs for problems
outside the limits of the codes' applicability. There is
no simple way to avoid this continuing problem. There are
two tedious ways: 1) Users should be trained and constantly
informed of progress and direction during the development
process, and 2) Developers should include redundancies
and hyper simplifications both within the computer program
and its output messages and within their documentation
and user's manuals.
3) The Storm and Combined Sewer Section of EPA should
become involved in development of analysis tools for other
urban water subsystems besides urban runoff, sewer
systems, and receiving waters. But the expansion into
other areas should occur systematically and as a result
-------
of developments of more refined techniques in its tradi-
tional area of expertise. The potential reuse of urban storm-
water and even wastewater or mixtures of both provides the
logical first step into expansion into other subsystems, such
as water supply, water use, and waste disposal.
-------
SECTION II
RECOMMENDATIONS
It is recommended that the Storm and Combined Sewer Section
consider fairly immediate support of model developments and demon-
strations in the-areas of 1) runoff transport modeling, 2) real-time
control technology for combined sewer systems, and 3) receiving
water quality-economics models.
Particularly important items worthy of slightly later support
include 1) a reclamation/reuse routing model for quantities and
qualities of urban water, 2) a long-period ecologic model for receiving
waters, and 3) an improved runoff quality simulation model.
The overall program for model development recommended to be
undertaken by 1980 includes development of the following capabilities.
For planning of urban water facilities:
1) A new and better watershed quality model.
2) A transport simulation capability in a planning model for
storage, treatment, and overflow evaluations.
3) Capability to simulate control or treatment processes in the
same model mentioned in 2 above.
4) A long-period year (10-30) receiving water ecologic model.
5) An urban water users economic effects model.
6) A downstream water users (receiving water or reuse)
economic effects model.
'*•-•,
For design or analysis of specific facilities, the following models
are needed.
1) A solids deposition and scour simulation capability in a
hydraulically sound sewer transport model.
-------
2) Improved dry-weather waste treatment simulation capability
in existing combined sewer design/analysis models.
3) Reclamation or reuse routing capability in a transport/
treatment model.
4) Nonstructural runoff control simulation capability in existing
urban runoff analysis models.
For operation and control of urban water facilities, the following
capabilities should be developed.
1) Real-time control software for sewer systems.
2) Real-time spatially varied rainfall prediction capability.
-------
SECTION III
BACKGROUND OF STUDY
SCOPE AND OBJECTIVES
The Storm and Combined Sewer Section of the U. S. Environmental
Protection Agency (EPA) wished to have an independent evaluation of
the future direction that it should take in urban water management
modeling. Accordingly, the purpose of this study was to review the
current state-of-the-art of models that have potential for becoming
integral parts of a larger model package and to define points of possible
interface between hydrologic models and models of treatment or control
facilities. The resulting report was to contain a phased plan of model
development which the Section could use as a guide for planning its
overall, modeling program.
EPA for several years has been sponsoring the development of
the Storm Water Management Model (SWMM). This model is a complex
simulation model for the dynamic events of rainfall and runoff, sewer
transport of combined wastes, and behavior of receiving waters to
•which such wastewaters are discharged. This model and several
variants have been extensively applied. Several EPA projects have
demonstrated the usefulness of these models in analyses of abatement
alternatives and determination of their environmental effects.
It appears, as our ability to develop and use such models
increases, that a requirement exists to define a next "generation"
of models incorporating the entire urban water management system.
The increasing emphasis on joint control of water supply and waste-
water systems, the development of refined storage, treatment, and
control systems for combined sewers, the requirement to assess
the water reuse potential of such systems, and the increasing need
of planners to assess the impacts of alternative land use plans also
argue for such models. Indeed, it seems clear that urban water
management must be considered in the largest context.
Future urban water models should recognize both the heirarchies
and categories of use to which they may be applied. Hierarchies of
-------
use vary from the closely defined needs of an engineer solving a
specific, single overflow problem to the loosely defined needs of
a regional planner attempting to evaluate various land use alternatives
to arrive at zoning decisions. While it might be hoped that a single
program of computer cards could address each of these problems
and intermediate ones as well, it is recognized at the outset that this
will very likely not be completely feasible. Categories of use may
be: description of the behavior of an existing sewer system subjected
to a specified input, design or planning of a new system, or operation
or control of an existing system.
Models to be developed should emphasize the physical system
and its operation, but the same or other computer programs should
also evaluate the costs of the alternative systems simulated. The first
step, however, should not be to develop a comprehensive economics
model in the larger context; social and political questions such as
pricing schemes would not be given early consideration and are beyond
the scope of this study.
Both previous studies and current model development work were
reviewed, and a definition of the next generation of needed urban
water management models was forthcoming from that review. In
particular, workers involved in the following previous studies were
contacted:
1) Urban Water Resources Research Program of American
Society of Civil Engineers, 23 Watson Street, Marblehead,
Massachusetts 01945.
2) Colorado State University's Metropolitan Water Intelligence ,
Systems Study, Colorado State University, Civil Engineering
Department, Fort Collins, Colorado 80521.
3) University of Florida's Storm and Combined Sewer Modeling
Program (SWMM development), University of Florida, College
of Engineering, Department of Environmental Engineering
Science, Gainesville, Florida 32601.
4) EPA Municipal Environmental Research Laboratory,
Systems « Economic Analysis Section, Cincinnati, OH
45268
5) Storm and Combined Sewer Section, EPA, Edison, New Jersey
08817.
6) Resource Analyses, Inc., Cambridge, Massachusetts 02139.
Current developments in progress, about which EPA furnished
the contractor with information, included work in progress at:
-------
1) Battelle Memorial Institute, Pacific Northwest Laboratories
(EPA Contract No. 68-03-0251).
2) University of Illinois, Department of Civil Engineering (El? A
Contract No. 68-03-0302).
3) American Public "Works Association Research Foundation,
and University of Florida (EPA Contract No. 68-03-0283).
4) City and County of San Francisco, Division of Sanitary
Engineering.
The study was to define a phased development approach, since it
may not be possible to fund the work recommended in a single year.
Such phasing was to be ordered as follows:
1) Wet and dry weather facilities considered together.
2) Addition of systems control considerations.
3) Addition of water reuse considerations.
The definition produced was to be in "broad general terms" and
was to include consideration of:
1) A basic philosophical approach to such modeling.
2) Specific components of such a model package.
3) Reliable treatment effectiveness models or routines.
4) Ability of each model to assess receiving water impacts.
5) Costs associated with using each model.
6) Selection of "best" model to optimize 4 and 5 above,
7) Costs associated with developing each model.
8) Relationship of each model to system operation or control, •
if any.
9) Data needs and availability for each model.
ORGANIZATION OF THIS REPORT
Following this section, Section IV addresses the first and most
important item for consideration, the general modeling philosophy.
-------
Section V reviews the state-of-the-art in urban water modeling, both
as reflected in the literature and as the study team was able to
discern progress underway through discussion with investigators and
model users around the country. Section VI outlines the urban water
management models that should be developed in the near future.
Section VII gives a phased program of model development support
whereby the Storm and Combined Sewer Section can most effectively
help in assuring that the model needs specified in Section VI are
met.
8
-------
SECTION IV
MODELING PHILOSOPHY
GENERAL QUESTIONS
Before specifying needed future urban water models, we must
establish 1) the problems of the immediate future to be addressed, and
2) the current capability for dealing with those problems in terms of
available models. Then we will be able to define, by a subjective sort
of subtraction, which models remain to be developed. But we can see
straightaway that such an X - Y = Z approach is too simplistic. The
X's, or problem definitions, and the Y's, or available model capa-
bilities, are each functions of parameters yet undefined or whose
size is yet unresolved. Consequently, we must first provide the limit-
ations on X and Y through a philosophical process of definition.
Some general questions could be raised to draw into perspective
the modeling work remaining to be done.
1) Which urban water problems require early attention of model
developers, given the base of knowledge and card decks
available today on which to draw?
2) Who are the likely users of the capability(ies) to be developed,
and what are likely to be their requirements in terms of
ease of use, comprehensiveness and time and space detail?
3) Should emphasis remain with storm and combined sewer
modeling, should it be placed somewhere else, or should
it be removed to give more Attention to each Of the
various portions of the urban water resources system?
4) Should the planning, design, and operation hierarchies of
analysis, which heretofore have been related primarily to
the temporal details of models, be stressed or downplayed?
-------
5) What realistic opportunities exist for combining separate
programs through interface programming to build more
comprehensive urban water models dealing with more than
one subsystem?
6) What, if any, should be the formalized computerized degree
of data management capability? Is a "Data Management
System" a necessity for using the eventual capability to be
developed? Is there a need for a nation-wide repository of
model-ready urban water data, as some have suggested?
Some of these questions will not be answered in this report. The
remainder of this chapter will address them all, however, and hopefully
the early portions of the phased implementation program outlined in
Section VII will answer those left unresolved here.
HISTORICAL PERSPECTIVE
The somewhat strange diagrams of Figure 1 are intended to address
historically the nation's perception of the size or importance of its urban
water problems, as well as to show the relative success that mathe-
matical modelers have had in developing tools that significantly aided
solutions. Such a diagram is open to considerable argument, oir course,
but it is presented merely as an illustration of the writers' view of
recent history, not necessarily as incontrovertible fact. The point
of it is to illustrate that for many urban water subsystems the
importance of "the problem" has recently become or remained greater
than the capability of models available or verified, despite the progress
in modeling in the last decade.
Without citing reference to some obvious works, we might dwell
for a moment on that history. A "blue book" was published in 1962
wherein the first comprehensive work on systems analysis in water
engineering was directed to river basin planning. Also unfortunately
systems analysis was oversold both by the book's authors and by the
sycophants who clung to their coattails. This work was produced during
the Kennedy administration which included Secretary McNamara, an
advocate of systems analysis in its larger but more business oriented
sense. It was logical, perhaps, that computer using engineers would
suggest that their tools would be directly and importantly of service
to government, but routine usage has been long in coming, and many
individuals and governmental bodies have become discouraged and
disgusted with systems analysis in the interim. Nonetheless, river
basin systems analysis blossomed and then waned, until the Texas
Water Development Board began analyzing the Texas Water Plan;
and, as time went on, more and more basins were eyed as potential
sources of water for urban areas hundreds of miles away. The
10
-------
8
I
O) <
II
,-J
I!
RIVER
BASINS
1.0
0.5
0.0
URBAN
HYDROLOGY
STORM &
WATER COMBINED
DISTRIBUTION SEWERS
WASTE RECEIVING
TREATMENT WATER
-I 1
-I 1 1
-I 1-
-t 1
H 1 1
•H H
TOTAL
URBAN
WATER
SYSTEM
f \
60 65 70 75 60 65 70 75 60 65 70 75 60 65 70 75 60 65 70 75 60 65 70 75 60 65 70 75
YEAR
Intensity of Problems
Model Development Successes
Figure 1 Historical relationships between the intensity of urban water
subsystem problems and successes with modeled solutions
-------
importance of river basin planning itself also waned then when the
concept of using river basins as planning units fell into some disfavor,
but it grew yet again as water became scarcer.
Urban hydrology received considerable attention from modelers
as soon as computers became routinely available to them. This
resulted in part because urban flooding and drainage problems were
acute; damages were high and frequent. Moreover, analysts knew
intuitively that the rational method for designing runoff facilities was
weak theoretically and yet tedious in complex applications. So hydro-
graphs, unit hydrographs, instantaneous unit hydrographs, systems
of linear reservoirs, infiltration equations, Markov chains, and
numerous other pieces of the puzzle were fed into the computer.
Urban watershed models of quantity and quality have been the latest
result. The recent attention to "nonpoint" sources of pollution has .
raised the importance of the urban runoff problem, while the ability
to model the phenomena occurring, particularly the quality phenomena,
has culminated for the moment with "dirt and dust" linkages that
are theoretically weak, if empirically capable of calibration.
Water distribution systems have been analyzed with computer
methods for years. Numerous utilities and private consultants have
more than adequate versions of programs that balance heads and flows
in these closed systems. Some if not most of the programs deal with
numerically complicating system paraphernalia such as pressure
reducing valvts, variable speed pumps, and the like. The quality of
water in these systems has not been included, however, and recent
discussions of lead poisoning and carcinogenic substances in water
supplies may draw more analysis attention to this important piece
of the urban water system.
Storm and combined sewers were perceived to be much more
important than previously thought when published results'showed
that stormwater can contain as much BOD and other pollutants as
raw sewage. The Federal Water Pollution Control Act Amendments
of 1972, calling ultimately for zero discharge and relatively advanced
treatment almost immediately, heightened interest in this problem
even more. From about 1968 forward, modeling attention has been
concentrated in this area. Now the EPA Storm Water Management
Model (SWMM, pronounced "swim") is the most available of the
numerous versions of runoff, sewer or conduit transport, and
receiving water packages that exist. Several private versions exist
in the United States, and there is at least one in Germany and another
in France. Another model, much less sophisticated, and hence
much easier and more economical to use, is the planning tool called
STORM (Storage, Treatment, and Overflow Runoff Model) which was
developed for the City of San Francisco. This is a planning model
that analyzes long records of historical rainfall to determine
frequency of combined sewer overflows, given storage and treatment
12
-------
availability under alternative plans. Both of these models, SWMM
and STORM, include quality simulation capacity in which BOD arid
suspended solids concentrations are derived from correlations with
accumulated "dirt and dust" on the watershed. The transferability
of the correlations from available local data has been shown to be
fairly weak; and a better, more defensible quality model or analogy
is needed.
Waste treatment received a flurry of modeling attention from
early systems analysts. Most of this work was directed at optimizing
the amount of waste treatment at various points along a stream,
given a dissolved oxygen standard and the Streeter-Phelps equation.
A later tack was taken to simulate waste treatment processes
themselves. The majority of this work to date has been an exercise
in programming the ."CUClimentS of sanitary engineering design, but
some elucidation of process variables has resulted. Waste treatment
importance was given a big boost by the National Environmental
Policy Act of 1969 and its requirements related to environmental
impact reports, but it was driven to its pinnacle of importance by
the 1972 Act.
Receiving waters, principally estuaries and freshwater streams,
have received considerable modeling attention. Many, many programs
to solve the Streeter-Phelps equation along streams were developed
in the early 1960's. Link-node models for estuaries with dynamic
hydraulic solutions were developed by 1965 for the Delaware estuary
and for San Francisco Bay and the Sacramento-San Joaquin Delta
Lake and reservoir temperature models and groundwater models
followed by 1967-1969. In 1969-1971 the receiving water model called
RECEIV was incorporated in the EPA SWMM model. During this
period, the feasibility was shown for modeling several aquatic trophic
levels and their, interrelated responses to ambient water quality as
indicated by a host of chemical constituents. This philosophy was
eventually demonstrated for San Francisco Bay and Lake Washington.
Since then many stream, estuary, and lake models have been
developed and updated to include the "ecologic model" inter-
relationships. Numerous basin modeling projects sponsored by EPA
have epitomized this effort during the 1970-1974 period. Throughout
the roughly 15-year history of modeling of receiving waters, the
capabilities of the developed models have lagged behind the
importance of the problems being faced by urban water management
decision makers. Current problems specified for attention by the
1972 Act include derivation of "wasteload allocations" for waters
designated as "water quality class segments", which are those for
which secondary treatment at all point discharges still would not
raise quality to levels satisfactory for all designated beneficial uses.
Current ecologic models applied to this problem have proved helpful
but less than completely satisfactory.
13
-------
Modeling of the total urban water system has not been extensive,
although it has been attempted. In 1968 the ASCE Urban Hydrology
Research Council, with OWRR support, coordinated a series of
feasibility studies to determine the requirements for performing such
a task. The common answer was that a formidable R and D effort
would be required. In 1970 an attempt had been made, but the step-
wise, explicit simulation model developed was only applicable to
limited facilities configurations. Other workers developed an opti-
mization model to move water through an urban set of facilities'and
users at least cost, but its use was also limited, and no major
decisions on urban water policy have been reported to be based on
the outputs of such tools.
One application of systems analysis methods to many subsystems
of the urban water scene has been performed in the Santa Ana River
Basin of Southern California over a period of many years, since 1967.
This effort has involved several models or programs, one to simulate
the extensive groundwater aquifer that underlies the entire basin,
one to organize the large data base describing water import quantities
and water uses on the ground surface, and a third to estimate the
numerous costs associated with importation, pumping, treatment,
delivery, and use of water, as well as the collection, treatment,
and disposal of wastes.
It is interesting to note that stormwater collection and disposal
facilities have not been included in these otherwise extensive modeling
efforts, since there are relatively few facilities provided in the basin
for storm runoff management. Moreover, these facilities are not the
responsibility of the agencies who have sponsored the work performed.
Pictorially, the urban water resources system was shown by WRE
in 1968 to look like Figure 2. The storm sewer was indicated there
to be a rather separate piece of equipment which kept its contents
apart from municipal wastewaters, although combined systems were
recognized to exist. In 1971, the SWMM project was completed :
wherein EPA contractors had concentrated most heavily on the missing
combined sewer part of the picture. The SWMM picture is shown in
Figure 3. The result of attempting to be absolutely complete in
pictorially representing the urban water system is shown in Figure
4. If that appears to be overcomplicated, exaggerated detail, which
WRE tried to avoid in its 1968 schematic, consider the picture of
Denver, Colorado, shown in Figure 5. (Note that SWMM would hardly
apply to that picture.) Now consider any other real urban water scene
and how it is more or less detailed, but certainly different.
The point of the pictures is to show that combined sewer models,
waste treatment process models, water distribution system models,
or lake temperature models, regardless of their validity, represent
only a small fraction of the total urban water picture. Moreover,
14
-------
(Jl
Figure 2 The urban water resources system
-------
REUSE - RETURN
j
N
G
W.
Figure 3 The storm and combined sewer subsystem
16
-------
ATHOSI'HCRE
{MECIFITATION)
(SUIFME WATER
UUIKU«I~
IWCIITEO
swir
(DIRECT EFFLUENT
— DISCHARGES)— «
imnmol
I supfir |
1 (E>PO»TEO
L, INOUSTRIAL
CONSUMPTIVE
USES)
(SUBSURFACE
— GROUNDWATER— •
OUTFLOW)
Figure 4 Schematic of the urban hydrologic system
(from McPherson, October 1973)
17
-------
SOUTHPLATTE RIVER
00
KEY
O
1 RAPID FILTER IANTHAFILT)
2 RAPID SAND FILTERS
3 SLOW SAND FILTERS
4 SAND FILTERS, MICRO STRAINERS
5 CHLORINATION, SAND FILTERS
6 SAND PRESSURE FILTERS
7 SAND FILTERS
8 ZEOLITE FILTERS
9 MICRO FLOC
10 CHLORINATION
1 ACTIVATED SLUDGE
2 TRICKLING FILTER
3 EXTENDED AERATION
Figure 5 Schematic of Denver's urban water resources system
(from McPherson, October 1973, after Smith, 1972)
-------
because the pictures correctly show flow both forward and backward,
the flow and quality at any point, say at the entrance to the combined
sewer system, may well be functions of the water leaving'the other
end of the subsystem. In short the urban water system is nonlinear,
_i. e., many subsystem models are probably given incorrectly estimated
inputs every time they are used.
As McPherson [October 1973] pointed out, "a total resource
systems approach is necessary if subsystem phenomena truly are
to be identified because of the complex linkages involved ... the
initial stage of a comprehensive systems analyses of the water
resources of a metropolitan area is essentially nothing more than
the attainment of a suitable metropolitan water-balance inventory".
Whipple, etal. [1974] suggested that "unrecorded pollution"
from urban areas, which includes all the runoff-carried pollutants not
now controlled by collection and treatment agencies or pollutants never
before measured by anybody, can be as large or larger than those
that are controlled. The implication is that universal secondary
treatment will not significantly improve the quality of receiving waters
in many places, particularly immediately following storms. The basic
thesis is like McPherson1 s, that metropolitan water balance inventories
are not yet complete; and they need to be complete, or uneconomical water
quality management plans are likely to result. Recently recognized
"nonpoint" sources of pollution merely highlight that the characterizations
of stormwater in pipes alone (like SWMM) are incomplete.
To simplify and synthesize what all this appears to mean, the
urban water balance inventory may be represented as shown in Figure
6 and it can be concluded that we will misrepresent the effects of
perfectly represented alternative plans if all the flows and qualities
indicated are not well documented as inputs and correctly mixed as
outputs of the model. Urban water representations must include
the man-made physical pipe and plant networks (including illicitly
connected pipes), plus the water falling and flowing freely, almost
glibly wherever it wishes, across the natural or altered land surface.
CATEGORIES OF USE
The historical review has indicated that models have been
developed and used to analyze various parts of the urban water system
and at several levels of time and space detail. The preponderance
of model "users" has included the developers of the models--the
programmers--but not the front-line, decision-making, operating
water managers. So several categories of model use appear to exist.
19
-------
WASTE
MANAGEMENT
Figure 6 Simplified urban water resources schematic
20
-------
The "Users" of Models
Researchers-model developers-scientists interested in simulation
or optimization of various processes with the intent of better
understanding the basic physics, chemistry, or economics of some
water subsystem have apparently been the developers of most computer
programs dealing with urban water. City engineers-urban planners-
sewer maintenance personnel in municipal government positions with
responsibility for planning, design, and operation of physical facilities
have only recently become involved with models, and they have tended
to hire the scientist-model developer to model the city's system
(subsystem) for them.
Not so strangely perhaps, but unfortunately, the scientist, model
developer has started from hydraulic and chemical first principles and
simplified where numerically necessary to get available computers to
approximate exact solutions, while for years the front-line manager
has concerned himself more with the hydraulically or chemically
complicated special cases he has encountered and has relegated the
simpler parts of his problem to handbook, nomograph solutions.
There has resulted a considerable amount of garbled communication
and missed points between them. Interestingly, it has been the
scientists, the purists, who have argued that the approximations are
good enough, while the engineers, the pragmatists, have held that
greater detail and more knowledge are required.
It is essential that model developers, whoever they are, cease
claiming utility for their products until the utility has been demon-
strated in a real situation. This will probably mean that models
will be developed with a real use in mind, and the "user" should
be identified as the person who gains practically applied assistance
from the exercise. The "user" must be an urban facilities planning,
design, or operations unit, and not its consultant. If the model
developed has transferable utility for another urban area, that will
have to be demonstrated before It is claimed; and the very safe bet
is that some modifications will be required for every transference.
A possible exception, and there may be some more, is a water
distribution system analysis program; but even these encounter
occasional situations that require further modification.
So programs developed for research or background purposes
should be held suspect, and models used in one application for
practical purposes should not be expected to be applicable straightaway
in another use. The "user" of any model should identify clearly both
himself and his purposes when reporting his model's results.
21
-------
The Functions of Models
Table 1 shows the relatively entrenched hierarchies of urban water
functions: planning, design, and operation, as well as examples of
model applications that have been attempted in each case. The table
is rather sparsely entered. What this means is that in terms of the
traditional engineering sense of designing something and getting it
built, models have not been much help. They have been rather
extensively used at a planning level and on a single subsystem at a
time (e. g. stream DO models, treatment plant sizing models, or
STORM). The examples of applications to connected subsystems
(e._g. sewers, treatment plants, and receiving waters) and to whole
urban water systems are not very numerous. The Santa Ana
Watershed Planning Agency's use of comprehensive urban water
modeling [WRE, 1969] is the only example we know, and that work
did not treat every subsystem possible.
Many of the historical model applications indicated in the table
are documented in the state-of-the-art literature review presented in
Section V. But the design and operation entries indicate the following
applications that have been drawn in part from our own, sometimes
unreported, experiences, and in part from the unpublished but
significant contributions of others.
The design applications related to "river basins" have been flood
control channel designs done for the Corps of Engineers. The
Tennessee Valley Authority (TVA) and the Corps of Engineers have
used certain models to answer design questions related to dam and
reservoir spillway and outlet configurations. (Whether other models
were or were not used for structural analyses, we do not know.)
Additions to water distribution systems have been made based on
computations with distribution system analysis programs. Sanitary
sewer systems that have been "designed" with computer models have
also subsequently been constructed.
In the operation phase it is known that TVA and the Bonneville
Power Authority use models to predict operating schedules for their
systems of rivers and reservoirs. Experience exists in Minneapolis,
Seattle, and Cleveland with operation of storm and combined sewer,
systems by computer-aided means. Also certain isolated instances of
computer-aided operation of waste treatment plants have been reported.
The economics "models" shown in the operation phase are merely
statistical processors of use data and billing schedules with which water
and waste departments and many utilities prepare bills.
22
-------
Table 1. FUNCTIONAL CATEGORIES OF MODEL USE
Function
River
' Basins
Storage
Tanks or
Reservoi rs
Urban
Hydrology-
Runoff
Water
Distribution
Storm,
Combined, or
Sanitary
Sewers
Waste
Treatment
Receiving
Waters
PLANNING
1) Subsystem Analyses
Physics
Chemistry or Biology
Economi cs
2) Connected Subsystem Analyses
a) Physics
b) Chemistry or Biology
c) Economics
3) Urban Water System Analyses
a) Physics
b) Chemistry or Biology
c) Economics
X X
X
X
X
X
X
X
X
X
X
X
X
DESIGN
1) Subsystem Analyses
a) Physics
b) Chemistry or Biology
c) Economics
2) Connected Subsystem Analyses
a) Physics
D) Chemistry or Biology
c) Economics
3) Urban Water System Analyses
a) Physics
b) Chemistry or Biology
c) Economics
OPERATION
1) Subsystem Analyses
a) Physics
b) Chemistry or Biology
c) Economics
2) Connected Subsystem Analyses
a) Physics
b) Chemistry or Biology
c) Economics
3) Urban Water System Analyses
a) Physics
b) Chemistry or Biology
c) Economics
x = subsystems modeled alone
o—o = subsystems modeled as connected, interdependent entities
23
-------
USER NEEDS OR PROGRAM OBJECTIVES
Given the historical perspective just reviewed and the computer
programs that exist today to deal with urban water problems, the
following list of questions that are now being asked could probably
be answered with the aid of models that could be developed fairly
easily.
1) What effects do our water supply sources, water treatment
processes, and water use patterns have on our waste
treatment and receiving water subsystems? --on reuse
potential?
2) In addition to urban water data we are now collecting, what
additional data should we be collecting to allow us to plan,
design, and operate future facilities?
3) What are the magnitudes and relative magnitudes of industrial
water use and waste discharge in our city, and what are the
magnitudes of their effects on receiving waters?
4) How can we predict the effects of urban point source
discharges on aquatic ecosystems this year, tomorrow, 30
years from now?
5) What are the magnitude, character, and origin of nonpoint
sources of pollution in our city? How do they affect nearby
receiving waters and downstream users both in the short
term and in the long term? How can we control them if they
are significant?
6) How can we or should we describe the chemistry and physics
of urban runoff for planning, design, and operations purposes?
7) How much treatment and storage should we provide in our
city for controlled runoff, currently uncontrolled runoff,
or sanitary sewage--this year or ever? •
8) How can we or should we predict rainfall in our urban area
this year, today, next month, over the city, over subdivision
C34? What is the planning "event", the design "event", a
single operational "event"?
9) What combination of upstream watershed detention, on-line
storage, or downstream off-line storage and/or treatment
should we provide for storm flows in our city this year, next
year, ever? Is the least expensive combination within our
city limits a defensible answer? If it is, is it the best
answer?
24
-------
10) In what magnitude and how often can we expect flooding
to occur in our city, study area, subdivision, downtown area?
11) How should we design and operate our combined sewers,
given our topography and land uses, to minimize the
deleterious effects of "first flush" pollutional loads following
the onset of rainfall and sewered runoff?
12) Which treatment process should we install for our storm
and combined sewer flows at outfall R-2, at outfall M-14,
at existing treatment plant T-25?
Few if any of these questions, which are fairly basic and are being
asked almost daily, can be answered adequately with currently existing
models. But these are all questions pertaining to the storm and
combined sewer portion of the urban water scene which SWMM and
STORM were developed to address. With some amount of additional
development and application, separate versions of those models could
be made to address each of these questions. More extensive model
development would be necessary to deal with a host of yet unasked
questions that may be raised pertaining to water supply, water
treatment, or water use. The writers are reluctant to suggest model
developments in these other areas until honest-to-goodness "users"
raise the questions to be answered. More academically contrived
problems are hardly needed. Emphasis should stay on the questions
already asked by front-line problem solvers who say they have certain
problems now.
The user needs for additional urban water modeling appear to
be, then, primarily in:
1) The planning realm;
2) The storm, combined, and sanitary sewer areas; and
3) The form of expansions to existing models required to
encompass urban water subsystems most nearly connected
to wastewater conveyance subsystems; namely, water use
(dry weather flow generation), urban hydrology (quantity
and quality of runoff over unsewered areas), waste
treatment, reuse, and receiving streams.
In the design and operations areas the modeling needs appear
to be specialized and again in the storm and combined sewer
subsystem. A problem for design of combined sewers (which SWMM
type models have highlighted) is the high pollutant loads flushed from
sewers during early portions of storm runoff events. This
phenomenon, known as "first flush", results when pollutant laden
solids that have been deposited during dry weather are scoured from
25
-------
paved surfaces and sewers and whooshed into receiving waters by
the stormflow, which rapidly increases in both flow and velocity
following the onset of rainfall. Although SWMM contains statements to
represent this phenomenon, hydraulically more advanced versions of
the model do not, and no model contains the ability to simulate various
abatement mea'sures. Improvements in this area are needed, since the
first flush pollution loads are significant and since methods to control
their particular load are neither mechanically trivial nor inexpensive.
Yet first flush control emphasis appears to offer the most frugal
answer to stormwater pollution abatement, since flows are lower than
the peak and pollutant concentrations and masses are highest during
that period.
An operations problem in the combined sewer subsystem to be
addressed with models is, just that: operation. A few cities,
notably Seattle, Minneapolis, Cleveland, and San Francisco, have
attempted'remote operation or are contemplating real-time automatic
operational control of storm or combined sewers. 'In addition to
hardware for actuation in response to telemetered signals for valves
or other appurtenances, this will require software for predicting
sewer system and/or storage and treatment plant behavior both
hydraulically and pollutionally. Present systems respond only to
hydraulics based on the premise that to minimize the quantity of
overflow is also to minimize the pollutant load to receiving waters.
Several cities already have installed equipment for remote
manual control whereby operators can monitor performance at various
locations by telemetry and then signal back commands, based on
certain operating rules, to valves, inflatable dams, gates, or other
gear. This mode of operation is advanced and sophisticated in some
respects, but control is nonetheless limited and in some particular
situations too slow to effect optimal control of the water and the
pollutants whose peak loads do not coincide.
Of particular note to the writers in this discussion of program
objectives and user needs is a haunting concept from the past, which
the writers conjured in 1968, and which apparently should now be put
to final rest. .Figure 7, which several authors have reproduced to
make a similar point, attempts to show that planning models are
long-period, large scale models, while design and operations models
are short-period, small scale, highly detailed models. This concept
is so general that it appears to have more exceptions than proofs
of the rule. For example, simulation models are formulated to operate
in either steady-state or dynamic mode. In. steady-state mode, time
steps are not relevant or at least they are not explicitly set. In
dynamic hydraulic models, stability of several sorts requires that
computational time steps be roughly equivalent to travel times, which
may be as short as 10 seconds. Dynamic quality models use time
steps short enough to allow simulation of events that occur diurnally
26-
-------
1
I
1
YEAR -
\ MONTH -
\ WEEK -
\ DAY -
\ HOUR -
Normal Modeling
Sequence
Desirable Urban
Modeling Sequence
Planning and
Administration
I960
v Design and
Construction
^-Operation and
Maintenance
1980
2000
2020
TODAY
DATE
Figure 7 Stages of development and levels of detail to be encountered
in systems analysis of the urban water resources system
-------
or hourly, while they are often made as long as possible to save
computer time. These quality models are usually operated with
hydraulic input data that have been averaged in an hydraulic model
over a number of hydraulic time steps equal to one quality model time
step.
The point of this is that the time and space scales used have
little or no relationship to the planning, design, or operational
functions being performed. In our own experience, estuary models
and SWMM have been used for planning with 30-second time steps,
STORM has been used for planning with 1 -hour time steps (but no
length or space scale for the sewer transport system); groundwater
models have been used for planning with 1-year time steps; and design
has been performed with steady-state models of several sorts and
with dynamic models with time steps of 30 seconds to 1 month. Link
sizes have varied from several hundred feet to many miles.
It is the general rule that planning can be performed for longer
periods of real future time and over larger areas than for design
or operation. But these large geographic scales and long "planning
horizons" are really not germane to the way in which the physics
or chemistry of the plan elements are represented in mathematical
models. A model that is sorely needed, whose characteristics will
typify the quandary of long period planning and short model cycles,
is a 30-year planning model for receiving waters that will properly
estimate ecological effects of increased, decreased, or transient
waste discharge conditions. We all know that phytoplankton, which
form a basic element of trophic level successions, function on a
diurnal cycle of sunlight and dark periods. It remains to be determined
whether an averaged seasonal or even annual model of the phenomenon
can be constructed so that very long periods of time can be modeled
for planning purposes without modeling, say, 30 years in 30-minute
time steps. As expensive as that sounds in terms of computer time,
however, it is also true that computers are becoming less expensive
to use and their cost is becoming a smaller portion of the total planning
cost. So if the short time step, long duration simulation cannot be
avoided, it should not be decided straightaway that such planning
analysis is infeasible.
28
-------
SECTION V
STATE-OF-THE-ART
This chapter is primarily comprised of a review of literature on
modeling of urban water subsystems reported since 1968. -This date was
picked because WRE [1968] reported a review of literature on the same
subject in that year, which covered the previous literature.
The last portion of this chapter summarizes the developmental
work still underway by various investigators and potential model users
who were contacted during the course of this investigation.
The literature reviews are presented by urban water subsystem
for each category.
RAINFALL/RUNOFF/SNOWMELT
Rainfall
Little literature was found that reported models of precipitation,
particularly computer models of storm events on a scale that would be
relevant to urban areas. Raudkivi and Lawgun [1970] developed a
model that generates times between storms by sampling frequency
distributions. Storm duration is based on a Markov model and a
frequency distribution, and storm precipitation is based on a joint
frequency distribution with duration. These same authors [Raudkivi
and Lawgun, 1972] subsequently refined their model for generating
storm durations. Their model was developed and calibrated based
on data for Auckland, New Zealand.
Duckstein, et al. [1972] have also developed a probabilistic
model for summer event rainfalls. Their model samples from a
Poisson distribution to generate a number of events and from either
a geometric or a negative binomial distribution to generate either
point or regional rainfall amounts. The model does not consider the
temporal spacing of the storms. None of the above papers considers
the temporal or spatial distributions of rainfall within a particular
29
-------
event. Other authors, notably Huff [February 1970; April 1970], have
presented data on this subject but do not develop rainfall generation
models.
A fairly recent report by Gregg, Labadie, and Wenzel [1974]
concluded that a major obstacle to automatic real-time control of
stormwater facilities is the lack of a dependable, short-term rainfall
forecasting model. Subsequent work by Trotta [1975] has centered on
stochastic prediction of runoff, skipping rainfall prediction, in part
because it is difficult to do, and his results indicate that this may be
adequate. The point should be proved in several locations.
Runoff
Given information on precipitation and snowmelt in a particular
urban basin, an obvious concern is with describing the nature of the
runoff that will result. Models that characterize this transformation
have been recently reviewed by Linsley [1971], Brandstetter [1975],
Papadakis and Preul [1973], Keeps and Mein [1974], and McPherson
[1975], Thus a detailed review is unnecessary here.
It is helpful, however, to recognize three different types of
potential concerns in characterizing basin runoff from a storm event:
1) the total volume of runoff, 2) the maximum rate of runoff at a
downstream point, and 3) the temporal distribution of runoff. Any
particular model may be oriented toward accomplishing one or more
of these characterizations.
Considering the total volume of a storm's runoff, Miller and
Viessman [1972] presented two simple, empirical relationships.
When the total rainfall was less than 1. 5 inches, pervious areas were
assumed not to contribute, and total runoff from impervious areas was
computed as a linear function of the rainfall. For total rainfall of more
than approximately 1.5 inches a Soil Conservation Service method was
suggested for determining total runoff. In another study, Viessman,
Keating and Srinivasa [1970] used three alternative means for
estimating the portion of rainfall that runs off: 1) the (j> index charac-
terization of infiltration, 2) 0.15 inches of initial abstractions together
with the $ index, and 3) 0.10 inches of initial abstractions together
with an exponential loss function.
Other more sophisticated and comprehensive models use slightly
more involved representations, at least of infiltration. Horton's,
Holtan's, and Philip's infiltration equations are three that are
commonly used to represent infiltration through urban pervious
surfaces.
Mein and Larson [1973] have recently presented an infiltration
model that depends entirely on measurable soil parameters. Although
30
-------
this new model has been experimentally and numerically verified, it
has not yet been incorporated into models of either rural or urban
runoff events. Of course, after any method has been used to estimate
initial abstractions and total event infiltration, the total storm runoff
is easily determined by subtracting these total losses from the total
rainfall.
Characterizing a runoff event with the peak flow rate has most
frequently been accomplished with the rational "formula" (Q = CIA).
DaCosta [1970] has explored the use of this equation for estimating the
effect of urbanization on peak flows. Others (see Linsley's 1971
review) have concentrated on frequency analysis in studying runoff
peaks; however, this data-intensive approach has the main weakness
of not indicating changes due to urbanization. Thus, most characteri-
zations of peak flows are obtained in the context of more compre-
hensive representations of the temporal distribution of runoff. The
estimation of the total storm hydrograph has been accomplished by
two primary methods: 1) hydrograph synthesis based on empirical,
"black-box" watershed response analyzers, or 2) use of a compre-
hensive, deterministic, cause-effect model of the rainfall-runoff
process.
Linear hydrograph theory [Brater and Sherrill, 1975] is the
simplest example of "black-box" hydrograph synthesis. More sophis-
ticated approaches recently reported include the nonlinear approaches
of Bidwell [1971] and of Amorocho and Brandstetter [1971], the time
varying approach of Mandeville and O'Donnell [1973], and the
nonlinear, time varying approach of Chiu and Huang [1970]. Since
these approaches require calibration with data from the basin being
represented, it is difficult to use them for predicting the effects
of particular water resources control projects in a "new" basin.
Comprehensive, deterministic, cause-effect models in the form
of computer programs are, therefore, the most common and
interesting type of model currently being used. The most widely
known of these models include the Hydrocomp Simulation Program
(HSP) [Linsley, 1971], and EPA's Storm Water Management Model
(SWMM) [Metcalf and Eddy, Inc., et al.', 1971; Chen and Shubinski,
.1971; and Lager, Shubinski, and Russell, 1971], The five previously
cited reviews describe others. The most recently presented models
are Chicago's Flaw Simulation System [Lanyon and Jackson, 1974];
the Linearized Subhydrograph Method [Chien and Saigal, 1974]; a
partial area technique [Engman arid Rogowski, 1974]; STORM, a
coarse Storage, Treatment, Overflow, and Runoff Model developed
by WRE [Roesner et al. , 1 974] and documented by the U. S. Army
Corps of Engineers [T973]; and the San Francisco Stormwater Model
[Kibler, Monser, and Roesner, 1975]. Each such model incorporates
significant differences in approach from the others, thus the previously
cited reviev/s probably provide the best (though usually biased) first
31
-------
step in assessing the model's applicabilities to particular problems.
It is significant that in some cases, differences among the results
produced by the models may be due more to the relative accuracy
of input data or model parameters than to the relative fidelity of
their respective methods for representing the rainfall-runoff process
[Papadakis and Preul, 1973],
Snowmelt
A few of the comprehensive models for computing runoff include
capabilities for computing runoff contributions due to snow accumu-
lation and melting. Brandstetter [1976] indicates four such models,
including the Hydrocomp Simulation Program, STORM, and the
Chicago Flow Simulation System. The respective methods used are
a Corps of Engineers snowmelt prediction procedure, the degree-day
method, and a Bureau of Reclamation method based on precipitation
and daily averages of air temperatures and wind velocity. In general,
however, modeling of snowmelt has been postponed while more
concerted attention is devoted to more generally occurring phenomena.
Some field studies of snowmelt have been conducted [Dunne and Black,
1971], but mathematical models were not synthesized.
Quality , ,
Models that characterize the quality of urban runoff are less well
developed than those for water quantity. Early relationships were
simple graphical correlations of concentrations and flow rates. This
approach was then extended by Cleveland, Reid, and Harp [1970],
Cleveland, etal. [1970], and AVCO Economic Systems [1970] who
present several linear regressions, mostly between land use variables
and impurity concentrations. However, the regression approach has
been largely unsuccessful, because 1) a large amount of data is
required, 2) poor quality data were gathered, and 3) transferability
of such relationships has not been documented conclusively. A recent
study employing regression analysis for urban pollutant loads was
reported by Colston [1974]. The equations derived had high coefficients
of correlation for a Durham, North Carolina, data set. But no
indication of transferability to other cities was given.
A basically different type of model has been suggested as part
of the EPA SWMM package by Metcalf and Eddy, Inc. et al. [1971]
based on a simple linear, first-order differential equation with a
coefficient proportional to water discharge. This type of repre-
sentation has subsequently been used and improved in other models
by such workers as Preul and Papadakis [1970] and by Sartor and
Boyd [1972]. Huff and Kruger [1970] have used a similar repre-
sentation of rural water quality in association with the Stanford
Watershed Model.
32
-------
"Without question urban runoff quality modeling still needs
refinement. Nearly every applier of SWMM has questioned the
transfer ability of the surface pollutant loading factors. Recent
testing by Colston [1974] indicated that these factors are indeed not
transferable. See the discussion below on Urban Watershed
Management for details of this particular art's status.
URBAN WATERSHED MANAGEMENT
Since 1971 the SWMM model [EPA, 1971] has contained provision
for estimating the quality of urban runoff (BOD, suspended solids, and
coliforms). The contributing sources of pollutants that are considered
include wash-off of accumulated dust and dirt on the watershed and
accumulated pollutants in catchbasins. The relationships included in
the model were derived from data collected in Chicago [APWA, 1969]
and Cincinnati [Wiebel, et al., 1964], and these were tested against
data from San FranciscDBEngineering Science, Inc. 1967]. Verification,
as might have been expected, was not completely satisfactory; although
discrepancies could be explained, and caveats about the insufficiency
of data and the incomplete accounting for all sources of urban runoff
pollution were given in the SWMM report [EPA, 1971].
In 1975 an updated version of SWMM was reported [Huber, et al. ,
1975]. It now includes COD, settleable solids, nitrogen, phosphate,
and grease in addition to BOD, suspended solids, and coliforms.
All of these have been related by mg/gram linkages to accumulated
.dust and dirt. Also a comprehensive but yet unverified addition of
the Universal Soil Loss Equation was made [Huber, et al. , 1975]
to estimate soil erosion, principally from construction sites. Another
model using the Universal Soil Loss Equation has been reported by
C. N. Chen [1975], Heinemann and Piest [1975] reviewed the current
(circa 1972) research on soil erosion, sediment yield, and modeling
of these phenomena; but they reported no applications in predominantly
urban watersheds. The state-of-the-art here is clearly directed to
areas of predominantly open lands of credible soils, and definitely
not to paved or roofed areas that accumulate artificially introduced
loose materials.
WATER SUPPLY
Some major cities, such as Philadelphia, withdraw water supplies
at their door. As a result, their simulation scale covers entire
tributary river basins, and they are merely one among several parties
involved. Alternatively, cities like New York and Denver have gone
long distances to develop supplies exclusively for their own urban
needs. Yet again, satisfaction of both urban and agricultural needs
are addressed in-such statewide supply programs as in Texas and
California. A still larger multi-state example of multiple need planning
is the Northeast Water Supply study by the Corps of Engineers. Thus,
water supply modeling has ranged from planning'multiple river basin
impoundment systems to drought analyses of mainstem flows.
33
-------
Operation models are frequently concerned only with existing
facilities; design models deal with facilities to build in the near future;
and planning models consider several future stages of development. An
excellent example of an operations model was presented by Joeres,
Liebman, and Revelle [1971]. They used linear programming and
simulated streamflows to explore conjuctive operation of Baltimore's
gravity flow reservoirs and a large-scale pumped source. They
demonstrated substantial savings by using a linear decision rule
although there was no indication that this type of rule is optimal.
Deininger [1970] also considered optimal operation, in this case
operation of an existing well field. Two formulations were presented,
one that maximized yield (linear programming), and another that
minimized costs for a given yield (quadratic programming). Other
examples could undoubtedly be found, since operations studies
typically consider well-defined systems and are thus most easily
subjected to mathematical programming formulations.
Design models in water supply are more likely to be used for
selecting and sizing new facilities in single stages of development.
Deininger [1970] also presented such an example where several
potential sources of surface wate_r_and their costs were being
considered. Linear programming was used to minimize costs while
obtaining a required supply. Weddle, et al. [1970] also suggested
a linear programming formulation; however, they used a network
analogy formulated with integer variables. Rather than considering
a group of potential surface water reservoirs, they illustrated the
technique relative to surface water supply, desalination, conventional
water treatment, and reuse. Bishop, et al. [1971] presented a similar
formulation. This approach is applicable for optimizing the
development of any potential configuration of treatment and water
transport facilities as long as the overall size of the problem remains
tractable and as long as adequate unit cost data for treatment and
transport are available. Weddle, et al. [1970] indicated an iterative-
application of the optimization technique to allow for nonlinearities in
the cost (versus capacity) functions. Additional examples of one-shot
facilities development are available; optimization techniques are ideal
for their solution, provided suitable cost data are available and an
applicable, scalar objective function can be defined.
Exercises in water supply planning must generally consider
multiple stages of development, a factor that may severely complicate
both the models used and the decision making that occurs.. However,
when a projected demand is to be met and when the sizes of various
facilities are predetermined, appropriate timing and sequencing is
relatively simple to discover. For example, Hoppel and Viessman
[1972] considered seven possible well fields developed to sustained
yield, their associated pumping and pipeline costs, and the possibility
34
-------
of recharging one field. Linear programming was used to determine
optimal delivery from each well field at various demand levels and
thus identified the sequence of development. Deininger [1970] used a
similar approach in a surface water sources example.
A more complex situation exists when sizing has not been
prescribed. Scarato [1969] presented a general procedure for
determining optimal sizes and timing of water treatment expansions
based on 1) linearly increasing demand, 2) constant economies of
scale, 3) a constant discount rate, 4) the requirement that all demands
be satisfied, and 5) an infinite planning horizon. Singh and Lonnquist
[1972] have relaxed some of these assumptions and then explored the
sensitivity, due to various factors, of monetary savings from optimal
stagings. Savings appear to be most sensitive to the rate of growth
of demand and the discount rate but are also affected by the efficiency
with which capacity expansion can be made, the initial water demand,
and the plant's useful life. Hinomoto [1972] formulated this same
problem with dynamic programming, concluding that two important
factors are the maximum daily demand and the fire-fighting demand,
both of which he satisfied with booster pumping rather than overall
increased capacity. Riordan [April 1971; June 1971] also extended
Scarato1 s work by relaxing the requirements that water demands be
met. He used dynamic programming to incorporate the price
sensitivity of water demand in finding optimal investment (staging)
and pricing decisions in water supply. He concluded that severe
assumptions had been made and that a 10-20 percent increase in
net benefits might result from applying the method. However he also
concluded that the empirical data needed are not currently available.
Becker and Yeh [1974] considered jointly the staging and sizing
of water supply development. They used dynamic programming,
together with an emphasis that a reservoir's benefits depend on its
contribution to overall firm yield, which in turn depends on the
reservoirs that are already available in other basin developments,
A similar problem was formulated by Mulvihill and Dracup [1974]
who used nonlinear programming.
35
-------
By far the most complex considerations of water supply involve
a joint consideration of operating rules, facility sizing, and staging of
developments. Blood, Clyde, and Peterson [1971] considered least
cost means for conjunctive development and operation of a freshwater
source and desalination to meet a given, long-term demand function.
They 1) established the ultimate demand, 2) estimated the desalting .
plant capacity needed to meet it, 3) found appropriate operating rules
for that plant and a sample hydrology, 4) repeated this operations
analysis for several other potential plant sizes, 5) selected the optimal
ultimate plant size, and 6) explored various staging strategies to
select the minimum cost overall plan for development. Clearly their
trial and error approach and selection from only a few itemized
alternatives, lacks the power of optimization by mathematical
programming, but it does result in a joint consideration of all three
significant aspects.
Previously, Evenson and Moseley [1970] reported joint consider-
ation of water supply operation, facility choice and sizing, and program
staging through an iterative application of mathematical programming
techniques and simulation models. Their analysis was oriented to
the Texas Water System and, for a sample problem, involved planning
for a system of 18 reservoirs and 42 canals over a 36 month period.
The size of this problem required development of screening procedures
to eliminate less attractive alternatives; however, mathematical
programming was used extensively in the screening process. Although
this approach has been improved and extended [Texas Water
Development Board, 1971; Montgomery-WRE, 1974], much
additional work remains to develop generally useful methods for
jointly considering operation, facility selection, sizing, sequencing,
and timing of water resources developments, especially with the
complexity involved in planning for urban regions. It might be noted
that water quality in large scale transfer schemes has not been
simulated, much less optimized, at all.
36
-------
WATER DISTRIBUTION SYSTEMS
Introduction
Included in a distribution system are all water works components
for the delivery to users of finished or potable water by means of
gravity storage or pumps through conveyance piping networks,
together with distribution equalizing storage.
Water distribution systems are typically divided into inter-
connected service districts according to delivery pressure
requirements, which are related to topography, in a manner roughly
analogous to urban drainage catchments. Planning/analysis/design
and operational control are usually approached on a district-by-
district basis.
Planning/Analysis / De sign
Very rarely is there an opportunity to design a total distribution
system. Typically, planning/analysis/design is concerned with the
extension and improvement of existing systems. Because system
enlargement is usually piecemeal, algorithms for wholesale
deployment of network piping or optimal allocation of water have
found limited application. More useful is the extensive theoretical
attention given optimization of network branches for planning
extensions or reinforcements to distribution systems, such as by
Jacoby [1969], including least-cost analysis with linear or dynamic
programming.
Considerable advances have been made in simulation techniques
for planning/analysis/design. Some of the larger water works have
developed their own computer programs [Brock, 1970], A compre-
hensive computer program [Epp, et al. , 1970], available on request,
has been adapted for use in several large and medium sized systems.
The various mathematical simulation techniques for system balancing
have been reviewed in detail [Shamir, 1973], and their different
characteristics have subsequently been summarized [Shamir, 1974].
Even topologic properties of networks have received mathematical
attention [Enger and Feng, 1971]. "In sum, literally dozens of
technical papers have been published over the last few years dealing
with mathematical aspects of distribution system simulation, seemingly
approaching a point of saturation. In fact, although shortcomings still
exist in available analytic tools, techniques and data processing
methods have developed to a level of sophistication that is often well
above that of the input basic data" [AWWA, June 1974].
Automatic Operational Control
The trend toward automation in water treatment, transmission
and distribution is strong and rising [Radziul, 1971]. Literally hundreds
37
-------
of water works, including the smaller ones, have some degree of
distribution system automatic operational control capability, although
this includes only monitoring in most instances. In the most advanced
cases [Van Dyke, 1970; Frenz, 1971; Juknievich and Richardson,
1973; Brock, 1973], all pumping and pressure-regulating functions,
but not the piping network, are under computer-directed control. The
critical, but incompletely developed, element for using a computer
as a full-fledged surrogate for the control system's human supervisor
is the required control logic or software. Suitable hardware is
available. Although water works lead the urban water resources field
in automation [McPherson, 1971], complete distribution system auto-
mation is still a long way off [McPherson, June 1973],
Required control logic includes capabilities for determination
of the piezometric surface, recognition and definition of any off-normal
conditions inferred thereby, resolution of applicable best service and
least cost criteria to be met, actuation of field regulation devices for
satisfaction of those criteria, and confirmation of success of actuation.
As opposed to "normal" operation that involves the satisfaction of
service demand variations that recur in a more or less similar daily
pattern almost all year long, "off-normal" operation refers to unusual,
•infrequent, sometimes catastrophic, occurrences such a major fires,
main breaks, power outages, and other facilities failures. Costs for
normal operation can be reduced by using automation to allocate source
inputs and prescribe pumping station schedules more efficiently. Among
the approaches under consideration in control logic development is
simultaneous network simulation-fitting and option-testing [Brock,
1963], whereas another calls for the sequencing of these steps [Oilman,
etal., 1973]. ,
Because people not familiar with the problem are prone to regard
control computations merely as extensions of the network balancing
process, it is important to note the important differentiation between
them. Geohydrologists have recognized this fundamental difference
in groundwater computations [Emsellem and Demarsily, 1971], where
aquifer hydraulic capacity characteristics must be deduced from the
performance of a limited number of test wells (the operations
equivalent), as opposed to water table mapping where all parameters
are known or can be specified (the planning/analysis/design equivalent).
For network planning/analysis/design, because adequacy of field
pressures is the central service criterion, the computational objective
is to define the piezometric surface (or field) of the network. Because
nonlinear terms exist in the energy relations involved, a direct solution
of equations is not presently achievable and a near-solution or "network
balance" is determined instead by means of iterative, successive
approximation techniques. The equations involved describe continuity .
of mass (flows at each node) and continuity of energy (pressure
attrition between each pair of nodes). That is, flows external to those
38
-------
within the network branches are givens, and determination of the
piezometric surface is the objective. Fixed piezometric heads can be
admitted, such as storage water levels. A near-solution is possible
simply because, and only because, as many equations are involved as
there are unknowns, a condition made possible by creating an identity
of number of unknowns and number of equations, the result being a
snapshot definition of the system's piezometric surface for the instant
of system operation simulated. An identity status is accomplished
through employment of certain approximations and simplifying
assumptions, the collective validity of which is indirectly reflected
by the proximity of simulation results to observed prototype
performance. (Uniqueness of "solution" can be obtained for nearly
all mathematical techniques now in use.)
. In striking contrast, the operating mode has precisely the opposite
requirements: with a measure of the piezometric surface given, what
are the flow conditions in the network? "A measure of" the piezometric
surface is an appropriate phrase because economic reality requires that
only a limited number of pressure sensors can be installed in any
service district, perhaps at no more than about a tenth to a fifth
of the major network nodes, and therefore it is feasible only to sample
the piezometric surface. Furthermore, most flows external to those
within the network are usually unknowns, not givens, and the sum
result is that the total number of unknowns greatly exceeds the number
of "equations" available.
Status of Automatic Operational Control Capability
Maximization of service (flows) and minimization of operating
costs are the basic objectives in automatic operational control.
Research on control capability has centered in three projects. The
first [McPherson and Prasad, 1966; McPherson and Heidari, 1966],
supported by the U. S. Public Health Service, demonstrated among
other things that the hour-by-hour operation of simple pumped-input
systems with a single station sendout and a single equalizing storage
site can be continuously simulated for a full year of demands. The
second [Oilman, j2tai. , 1973], supported by OWRR, explored field-
sensor tracking, system performance simulation, demand projections,
and control criteria. The third project [Rao, efc al., 1974] also
supported by OWRR, now the Office of Water Research and Technology
(OWRT), will continue through 1975.
The basic tasks involved in achieving full, hands-off automatic
control are as follows [Rao and Bree, 1975]:
1) Network simulation sufficiently rapid to keep pace with
changing field conditions, over time intervals as short as
a quarter of an hour.
39
-------
2) Reduced hydraulically-equivalent networks, for use in state
estimation and prediction.
3) Load forecasting.
4) Optimization of pump selection.
5) Real-time estimation of system behavior, or system state,
based on telemetered information.
6) Sensitivity analyses of system response to control tactics and
strategies.
7) Integration of the above six items into a compatible total
capability.
Of the seven items, only the first has received adequate investi-
gation. While the others have all been studied [Oilman, et al, 1973;
Rao and Bree, 1975; Bree, 1974], considerably more work could be
and should be done.
Transferability of Results
The 1975 follow-on to the most recent work [Rao and Bree, 1975]
will quite likely carry the overall complete automatic operational
control concept about as far as it can be taken and still yield national
transferability of findings. However, eminently transferable would
be additional work on the first four items listed above. Still, the
work on items 3 through 6 should be regarded as only partially
transferable, because conditions of actual water works districts have
been used in their historical development, and additional work would
be required to adapt the findings to other water works.
Because the best planning, analysis, and design include simulation
of system performance to demonstrate the virtues and flexibility of
expected operating improvements, advances in automatic control
capability will accordingly yield spillover benefits to these other
functions.
There is general agreement that the central requirement for full
national transferability of a total software package for complete auto-
matic control would be the characterization of the hydraulic behavior
of distribution systems in general terms--the development of a
universal transfer function, as it were. Although a preliminary
generalization for relatively simple systems has been made by means
of an empirical network representation [McPherson, 1961], which
was later extended [Wiseman and McPherson, 1965], and in which
a range of general pump and storage characteristics were incorporated
40
-------
[McPherson, 1 966; McPherson and Prasad, November 1966], the
absence of suitable applied mathematical tools for direct solution
of application equations precludes much further generalization.
Work on load forecasting, item 3 above, would contribute toward
a better understanding of short-term distribution system demand
variations, a priority research need identified by AWWA [September
1974]. Techniques for analyzing seasonal demand variations [Salas-L,a
Cruz and Yevjevich, 1972; Gracie, 1966] also deserve greater study.
A hierarchical approach for automatic control of combined sewer
overflow pollution has been advocated, with slave minicomputer
surveillance o£ individual catchments and master overview of the total
jurisdiction by a central computer [Wenzel and Bradford, 1974]. This
is the current "building-block approach to automation" in vogue in
manufacturing [Anon. , 1973], and may be appropriate for distribution
system control, especially for service districts vis-a-vis an entire
water works.
While use of statistical tools has a tendency to obscure the working
nature of physical processes, some of the newer cybernetic develop^
ments deserve watching. Progress has been made in manufacturing
process-control where the computer programs describing a process
are internally adjusted as information from process sensors
accumulates, in a "self-learning" or "self-programming" mode,
because "many of today's processes are so incredibly complex as to
defy reasoning from first principles [Anon., 1974],
Lastly, as a stress-reduction tactic, it has been suggested that
automation of distribution systems could be used to restrict peak
demands when a system cannot meet them [AWWA, September 1974],
obviating investment in extra facilities for satisfaction of infrequent,
extreme loads.
WATER USE
To the writers' knowledge, the urban water use subsystem has
never been rigorously simulated, in a cause-and-effect sense. The
most elaborate model constructed appears to be MAIN-II, developed
by Hittman Associates [1969]. This model either accepts projections
or makes its own for "independent" variables such as population
density, values of dwelling units, and numbers of dwelling units in
each value range. Including residential, commercial-institutional,
industrial, and public-unaccounted sectors of the community, 150
separate water use categories can be projected.
WRE [March 1969] and Montgomery-WRE [1974] have developed
similar but less ambitious "Data Management Systems" that calculate
41
-------
annual demands through time for several categories of urban and rural
•water users. These models use population or land use in each water
user category, multiplied by appropriate unit factors, to arrive at water
demands. "Waste flow projections are derived similarly. Unit factors
for increases in quality constituent concentrations can also be
incorporated to calculate waste qualities.
Several authors have suggested other approaches for projecting
municipal and industrial water use. Whitford [1972] focused on per
capita residential water demand by projecting the trend in this variable
and incorporating various subjective adjustments to obtain a proba-
bilistic characterization of per capita demand which had a standard
deviation of approximately 10 percent. Burke [1970] used multiple
linear regression and the U.S. Public Health Service inventory data
on U. S. water supplies to project total demand for any city. Schaake
and Major [1972] reported water use projections in the North Atlantic
Region relating water usage with population served and per capita
income for municipal use, population and per capita uses for rural
use, and input-output table technical coefficients for industrial uses.
All of the approaches to water use projection mentioned above,
however, depend on prior projections of independent variables, for
example, population, per capita income, or water pricing policy. As
such, they are all computerizations of effects and their trends, rather
than models of water demand causes that simulate resultant effects.
Other authors have considered other types of water use, primarily
recreational use. McCuen [1973] suggested a sequential decision
approach for estimating recreational visitor-days for prospective
facilities, an approach which incorporated information on the use
of currently available facilities, empirical equations calibrated to
describe usage in other locations, and market survey. Dietz [1973]
postulated access as the principal limitation to boater use of Lake
Michigan and developed a model to predict the need for refuge and
transient facilities at various lake shore locations if access were
improved in the Chicago area.
Andrews and "Weyrick [1973] assumed the scarcity of water and
studied water allocations by linear programming under alternative
objective functions to indicate most plausible potential future uses.
The water use subsystem is the most critical of all urban water
subsystems, because it sets the quantity and quality demands for all
upstream subsystems, plus it is the source of the quantity and quality
loads imposed on all subsystems downstream. But historically it has
not been within the jurisdiction of the engineer to control (limit) the
•water use subsystem; thus there has been no apparent incentive to
describe it mathematically. Hence, there are no cause and effect
models of actual usage of water (much less quality deterioration)
in industries, parks> or single-family or mutliple family homes.
'•42
-------
Tihansky [1974] and Sonnen [1973] have each developed some quality-
use-consumer cost programs that calculate the added costs to home-
owners or industries of excessive hardness or TDS in their supplies,
but these accept demands as given and do not account for any diminution
in projected unit demands if quality deteriorates or increases in use
if quality is improved. In short, much more work could be done
in simulation and economic modeling analysis of urban water use.
COLLECTION AND CONVEYANCE
Models dealing with pipe systems for collection and conveyance
of storm and combined sewer flows can be divided into four distinct
categories: performance analysis, control, design, and planning.
Analysis models describe the performance of a given collection
and conveyance system under given inflow conditions. Model output is
usually in terms of flow rates, and possibly in terms of impurity
concentrations over time at various points including at the system
outfall. Brandstetter [1975] has conducted a comprehensive review
of the more sophisticated of these models, and hence such a review
will not be repeated here. However, the initially developed hydraulic
transport routine for EPA's SWMM model [1971] is typical. Depending
on the level of resolution needed to represent temporal variables and
the pipe network, relatively coarse to highly sophisticated analytical
models are available. SWMM is one of the more sophisticated (and
expensive) of these models.
Computed descriptions of hydrographs and pollutographs are not
the ends in themselves. To be helpful these results must aid identi-
fication and solution of operation, design, and planning problems.
Of the currently available analytical models, few have been applied
to the operational control of drainage systems. Indeed, the Battelle
model [Brandstetter, et aL. , 1973] is the only one integrated with
a computation procedure for making real-time control decisions. A
less sophisticated model [Minneapolis-St. Paul Sanitary District, 1971]
was developed for the expressed purpose of real-time control; however,
it provided only an analysis of the results (hydrographs) of trial control
decisions and was discarded as inefficient. Rapid prediction or even
automatic decision making are the most important criteria for models
used in operation. In Seattle [Gibbs, et al., 1972], operating decisions
are made automatically with rule curves and real-time data without
use of explicitly predicted effects on the system. (The rule curves
presumably reflect predictable responses.)
There is also a noticeable disparity between the availability of
analytical models and their use in design, especially since the level of
resolution incorporated by most models is suitable for design purposes.
Heaney, et al., [1973] have added a modest design capability to SWMM:
43
-------
The program checks for surcharges; and, if it is found, the pipe
diameter is increased by a standard increment retaining the originally
given slope and network configuration. From Brandstetter's-review
[1975], five other models apparently have similar capabilities thus
allowing consideration of the dynamics of flow when sizing pipes. No
variation of network configuration or pipe slopes for design analysis
appears to have been accomplished, except possibly through trial
and error applications of the above mentioned models. No article was
found reporting the use of such a model in solving an actual design
problem that led to construction.
Design problems have also been approached with models that
adopt the steady-state "design flow" concept, which obviously makes
them more applicable to sanitary sewers than to storm sewers. Mathe-
matical programming techniques have been used to discover optimal
sizes, slopes, and--in rare cases--configurations of drainage
networks. Walsh and Brown [1973] and Merritt and Bogan [1973]
have developed slightly different dynamic programming formulations
for determining optimal pipe sizes and depths (and thus slopes) for
a given design flow and network configuration. Their approaches
appear extremely promising. Dajani, e_t al. , [1972] and Dajani and
Hasit [1973] explored linear programming to accomplish the same
purpose. Initial assumptions of full flow in the pipes and a continuous
(rather than discrete) range of sewer diameters were unfortunate
and led to a separable convex mixed integer formulation that required
excessive computer time. A two-step sequence for solving the problem
was finally recommended which first assumes full flow of pipes and
then uses this initial solution in the time consuming algorithm.
Fisher, et al. [1971] presented an integer programming formulation
for the diameter-slope problem. The example they considered was an
interceptor in the Chicago area, a significantly larger sewer than other
optimization studies have typically considered. In spite of finding a 10
percent cost savings over a traditional design method, the authors
concluded that uncertainties in excavation costs, the dynamic nature of
actual flows, and the arbitrary nature of velocity constraints detract,
considerably from the significance of the indicated saving. Argaman,
et al. [1973] have also used programming techniques for sewer design,
but they have considered optimal network configuration as well as pipe
sizes and slopes. They found their dynamic programming approach to
require amounts of computer time that severely limit the size of the-"
sewer network that can be considered. The development of programming
techniques for sewer design is relatively recent, and their application
to real problems has not been documented. However, it appears that
Walsh is actively implementing his approach [Walsh and Brown, 1973].
By far the most neglected category of collection and conveyance
modeling is planning. Chen and Saxton [1973] have reported a planning
level study in which triangular hydrographs and runoffvolum.es propor-
tional to rainfall have been used with storm frequency data to develop
44
-------
curves showing 1) the number, duration, and volume of overflows for
Spokane, Washington given intercepting capacity, and 2) the capacities
of storage or treatment facilities required to reduce or mitigate
Spokane's overflows. Unfortunately, the resulting curves are location
specific, and no information was presented regarding the sewer
networks needed or their costs at the various levels of storage or
treatment.
Water Resources Engineers has developed [Roesner, £t al. , 1974]
and the Corps of Engineers has documented [U. S. Corps of Engineers,
1974] a planning level Storage, Treatment, Overflow and Runoff Model
(STORM) in which continuous computer simulation (at hourly intervals)
is used to predict the effects of various capacities for treatment and
storage on overflow quantities and quality. No consideration is given
to the collection and conveyance system, however; and no cost
relationships or optimizing algorithms have been included. Therefore,
the model requires considerable further development to include the
collection and conveyance network.
.Although the configuration and cost of storm and combined sewer
pipe systems have not received significant attention in planning models,
other conveyance facilities have been modeled and evaluated for relative
costs. Montgomery-Water Resources Engineers [1974] developed
"Water, Wastewater and Flood Control Facilities Models" for. the
San Diego Comprehensive Planning Organization. Based on various
planning criteria and stipulated evolutions of land uses, the models
identify cost, and optimize (within a given type) the water supply,
flood control, or sanitary sewer facilities needed. Stormwater facilities
could be considered in a manner analogous to that used for sanitary
sewers and flood control channels.
In summary, existing models of collection and conveyance sys.tems
for storm and combined sewer flows are primarily analytical models.
Some possess a limited design capability. Most incorporate levels of
resolution appropriate for either design or control. Not nearly enough
effort has been devoted to planning models.
45
-------
WASTE TREATMENT
*
One of the first, serious attempts to simulate waste treatment
processes was that of Smith [1968], The stated purposes of the model
were "preliminary design and simulation". Preliminary design is
not exactly the same as planning in our hierarchial staircase used
here. Traditionally this has meant the analyses needed to determine
dimensions and estimated costs of various tanks and equipment, which
stops considerably short of specifying types and amounts of reinforcing
steel to be used in tank walls or other truly design answers.
By way of example, consider the design of a primary sedimentation
tank. The objective of the design is to determine the dimensions of the
tank. The objective of the tank is to remove settleable solids and BOD.
The objective of treatment simulation models built to date has been
to estimate the removal percentages of suspended solids and BOD
from a tank having a given overflow rate in gallons per day per square
foot.r
Mathematical models for design of treatment processes built to
date have been predicated largely on correlations from data and not on
predictions of the dependent variables from theoretical interaction
of the independent variables. In short they have served to derive
general dimensions in a hurry that can quickly be used in generalized
cost equations. As such they represent tentative planning tools but
do not materially assist rational design.
Additional papers have been published in which treatment simu-
lation or optimization models have been described. Virtually all of
these have used similar correlations from general data or even
removal percentages that must be estimated straightaway to "simulate"
performance, and those that deal with costs have all used cost equations
derived from generally available data. While they have elucidated
some general sensitivities of cost to various process types and sizes,
they remain unproven and probably unprovable as computer aids for
decision making in specific instances. Of more value to designers
have been the theoretical and experimental developments of such authors
as Lawrence and McCarty [1970], and Jenkins and Garrison [1968],
46
-------
who related efficiency of organic removals in secondary treatment
to mean cell residence time, and Dick [1970] and Dick and Javahefi
[1971], who showed the interdependence of secondary clarifier operation
with the preceding oxidation step in determining overall performance
of biological treatment. These authors, while referring to their work
as "mathematical models", do not report that their "models" have
been computerized. The computerized models that have been reported
for process and cost evaluation were included in a comprehensive
bibliography by Tihansky in 1974 [May 1974].
A beginning has been made in computer-assisted operational control
of waste treatment systems. It would seem that no prototype plants are
operating in this mode as yet, but the works of Guarino e_t al. [1972] in
Philadelphia and Lacroix and Bloodgood [September 1972; December
1972] in Fort Wayne indicate that activated" sludge and digester control
algorithms that give process adjustment data to operators are now
developed and near implementation. Klei and Sundstrom [1974] have
achieved automatic control of a laboratory model activated sludge
plant (without computer assistance) by automatic sampling of total
carbon and COo iru the field and sending an electronic signal to the
sludge recycle pump to increase the rate of returning sludge in
proportion to the carbon concentration in the influent.
It is interesting to note that the algorithms used in these studies
differ from those in other design simulation studies in two important
ways: 1) they are much simpler, _i._e, , less complex algebraically,
and 2) they are solutions for parameters that were fixed by rule-of-
thumb assumptions in the design process, such as the recycle rate.
So in operation the variables sought earlier in design are the givens
from actual measurements one time step ago and the design givens
are the operational variables being sought for one time step ahead.
RECEIVING WATERS
- There is such a plethora of models for receiving water quality and
quantity now that it would be futile to even attempt a listing or a
comparison. Some of the latest compendia on the subject include
47
-------
those of Systems Control, Inc. [1974] and Brown, efcal. [1974],
It is notable from those reviews of available models that many are
similar, most are quite usable and helpful for some purposes, all
have some weaknesses, and none is perfect.
These two particular reviews are so objective that a reader can
find no reported facts or even opinions to lead him to choose one model
over another. Perhaps they are that similar. Potential users are
probably best advised to use versions whose documentation is most
complete and readable. If user talent is to be acquired as well, the
client would probably be smart to get the services of the consultant
who understands water bodies best and to consider that person's
modeling techniques as added gravy.
To summarize available programs succinctly, there are workable
programs available for
1) estuaries,
2) streams,
3) lakes and reservoirs , and
4) groundwaters.
The receiving water model in SWMM is a version of the old EPA
Dynamic Estuary Model [Feigner and Harris, 1970]. Virtually all
estuary models of this type have a hydrodynamics subprogram that ~
computes heads at junctions and velocities and flows in one-dimensional
links of a two-dimensional network. The quality subprogram is operated
with quantity information supplied as taped output from the hydro-
dynamic subprogram. Recent versions [Genet, Smith, and Sonnen,
1974; WRE, (September 1974] include over 20 constituents as well
as phytoplankton and their growth/death relationships with chemical
constituents, light energy, and temperature. One version, reported
by Chen and Orlob [1972] and applied to San Francisco Bay, includes
zooplankton, fish, and benthic animals.
The Texas Water Development Board [1971] produced a stream
model (QUAL-I) in 1971 that works on a discrete element network. .
That model simulated temperature, BOD-DO and conservative
constituents, but additions have been made recently to include numerous
constituents and phytoplankton, just as included in the estuary models
[Roesner, Monser, and Evenson, 1973]. Another version simulates,
"temperature" either, dvnanrica1.lv or in steady-state (Norton, et al.§ 1974).
whereas the first version simulates temperature dynamically ifHesired,
or otherwise it must be read from cards as input if a steady-state
simulation is being made for the remaining properties.
48
-------
Lake models available tend to be one-dimensional in the vertical,
stressing the heat budget and attendant stratification which determines
the disposition of all other dissolved constituents. Storm or waste loads
are specified as input and the model mixes these in the lake or
reservoir slice having the same density. The outflows from the
reservoir are specified as well, thus the hydraulic solution is given
by solution of the continuity equation alone. Versions exist [Chen and
Orlob, 1972; Chen and Orlob, 1973; Glanz and Orlob, 1973] that
simulate numerous constituents and phytoplankton, zooplankton, fish,
and benthic animals.
Groundwater models are less numerous. The only one used
vigorously for water quality planning purposes that the writers know is
their own [Water Resources Engineers, 1969] which simulates total
dissolved solids movement in a junction-link network for the Santa Ana
River Basin aquifer in Southern California. It also has a hydraulics
subprogram that simulates water movement through both the
unsaturated and saturated zones, and a quality program that computes
mass transfers given surface mass inputs of stormwater, rainfall,
and irrigation percolation, and waste loads. (This program, for obvious
reasons, works on annual time steps, whereas the models of surface
waters deal with time steps of minutes, hours, or days.)
In summary, there are numerous available models for receiving
water simulation. The writers obviously know their own versions best,
so we have cited documentations that we know to be descriptive of
applied, verified versions. Other developers with similar but assuredly
slightly different versions of their own include Hydroscience,
Hydrocomp, Harleman and co-workers at MIT, Dysart at Clemson,
Deininger at Michigan, Systems Control, Inc., Loucks and co-workers
at Cornell, Environmental Dynamics, Inc., Battelle, SOGREAH in
France, and many more. Many versions of receiving water models
and their documentation are in the public domain and may be acquired,
perhaps with payment of a nominal fee, from the Office of Water
Research and Technology, EPA, the Corps of Engineers (Hydrologic
Engineering Center) or through state or local governments.
WATER REUSE
As has been demonstrated by Weddle, et aL [1970], and by
Mulvihill and Draeup [1974], models representing the integration of
water supply and wastewater operations in capital expansion programs
are completely feasible. Procedures are analogous to those used in
analyzing multiple-unit water supply projects when a cost minimization
objective is established together with constraints on water supply
(demand) and wastewater quality (effluent of receiving water). It is
only necessary mathematically to characterize the connection between
the two systems in terms of volume and quality degradation in use.
49
-------
Of course, the same types of data on costs and technology are needed'
as when analyzing each system separately. In spite of the feasibility
of constructing such models, little effort has been devoted to their
development and demonstration.
In explicitly considering water reuse, Mallory [1973] compared
various alternative schemes for renovating stormwater. He found that
for general renovation and use it is more economical to treat water
and distribute it through the regular water distribution system because
a second (e_. g. lawn watering) distribution system would be prohibitively
expensive. Distribution to special users of renovated water may be <•
feasible. Of course, the primary deterrent to water reuse is the lack of
knowledge on the presence, fate, and ultimate effects of the tremendous
variety of impurities that might occur in recycled wastewater. Clearly
such impurities are likely to build up within a simply conceived
recycling system. Thus, a basic need in considering water reuse,
prior to selecting model parameters, is soundly developed constraints
related to health or other considerations. Such constraints then would
motivate a search for treatment techniques that insure satisfaction
of such standards for the least, but probably higher than usual cost.
in spite of the availability of modeling techniques for representing
the occurrence and build up of water impurities, the construction and
demonstration of such models has been only superficially attempted in
urban water resources planning [WRE, 1970; Kugelman, 1974]. For
example, Bishop and Hendricks [1971] considered only the relevance
of dissolved solids in their study of water reuse versus desalination.
CURRENT MODEL DEVELOPMENTS AND USES
During the week of August 11, 1974, the writers attended a
research conference at Franklin Pierce College, Rindge, New
Hampshire, entitled "Urban Runoff: Quantity and Quality". The
Proceedings of that conference [ASCE, 1975] have now been published.
During the conference, the writers presented a cursory statement ,
of the state-of-the-art of urban water modeling, which is published
in the Proceedings. Also all the attendees and certain model developers
and users in particular were asked to react to the written statements ;
and to send summaries of their current activities to WRE so their '
current work could be included in this state-of-the-art review.
.Subsequently, ten model developers and/or potential users wrote
their replies. Their responses are reviewed here.
The respondents were:
1) John J. Bailey, Jr., Vice President
Reitz and Jens, Inc., St. Louis
50
-------
2) Dan Brock, Dallas Water Utilities, Dallas
3) Neil S. Grigg, Associate Professor
Colorado State University, Ft. Collins
4) Brendan Harley, Resource Analysis, Inc.,
Cambridge, Massachusetts
5) Stifel W. Jens, Senior Vice President
Reitz and Jens, Inc., St. Louis
6) D. Earl Jones, Jr., Department of Housing and
Urban Development, Washington, D. C.
7) Robert C. McWhinnie, Board of Water
Commissioners, Denver
8) George F. Smoot, U. S. Geological Survey,
Reston, Virginia
9) L. Scott Tucker, Urban Drainage and Flood
Control District, Denver
10) H. G. Wenzel, Associate Professor, University
of Illinois, Urbana
Some of the comments received, particularly those addressed to
future directions that modelers should take, were candid and very
helpful. Nonetheless, since these comments may be controversial
even within the respondent's own organization, some of them will be
reported here anonymously. . •
Perhaps the most telling remark in the entire set of responses
came from a person engaged in a local water supply agency. In
answering a philosophical question related to communications problems
between model developers and model users, he said, simply, "we
seem to be most successful where the developer is a (the) user. "
Many if not all of the replies expanded on this same theme. There
is apparently a serious and vexing problem that arises again and again
when a new user tries to use a program ("model") developed by
somebody else. Invariably, the program contains bugs, solves a slightly
different and usually much simpler problem than the one(s) advertised,
or simply will not function or execute.
There are many variations of the same communications problem.
Often the delivered card deck and documentation reports do not clearly
annotate the options available or assumptions implict in the programs.
Sometimes the mathematical statement of the general problem is far
more precise than the data used to "verify" the model, and hence the
51
-------
program takes inordinate amounts of time and money to generate its
highly approximated and questionable results. Saddest of all are the ,
cases where the model developer and the ultimate user of the model's
results, often the fellow who paid for the development, never communi-
cated from the start; and the model developed addresses a problem
the user never had. But his real problem is still there, unsolved.
Every developer of a model who gave his program to someone
else to use has heard these complaints. Ironically, he knew he would,
and he let the program out of his hands anyway. Usually, he lets it go
because the user bought it from him. But he knows, and the user cannot
believe, that there will be problems with the very next application.
There is no excuse for this phenomenon, just as there is no excuse
for somebody else's meatloaf not tasting like your mom's. They just
are not the same; they were made "differently even though they were
called the same thing. Another reason it occurs is because the modeler
knows from the start that he is setting out to approximate a solution
to a theoretical problem with both an approximation of the theory and
an approximation of the prototype water body. The model user or
the user of the model's results views his problem, and the theoretical.
statement, as precise and infinitesimal. Almost invariably, the first
application of a handed-over program is made to a problem that either
1) lies outside the range of applicability of the equations simplified
in the program, or 2) requires a time step shorter than the model
£r its "theory" can accept. Highly qualified and experienced
programmers make these mistakes just like neophytes do. The
nondeveloper-user almost always expects a new program to be both
more exact and more flexible than it is or was ever intended to be.
Without question, improvements in model documentation and
preparation of user's manuals can be made. The communications
problems between modelers and subsequent users of their products
are too numerous and well documented for simple sloppiness of
explanation to continue. Responsibilities lie with both parties, however,
and the tedious method of constant re - explanation between the developer
and the subsequent user is the only fail-safe procedure. A user who
picks up a program deck cold is going to have problems with it. Period.
A second viewpoint that appeared in one form or another in almost
every response was that today's models generally outstrip the data
available to calibrate them, much less verify them. Perhaps predictably
from this group, the most important data need was thought to be time-
coordinated data on quantity and quality of urban runoff--on watersheds,
in pipes and open channels, and in receiving waters. There is
apparently an undercurrent of feeling that a better data base might '
suggest improved methods and even improved theoretical/mathematical
bases for modeling the quality of urban runoff. (The writers tend
to agree.)
52
-------
A third comment that appeared in several of the responses was
that the models should continue to be addressed to singular, particular
subsystem problems because, regardless of modeling capabilities to
deal with several or even'all urban water subsystems simultaneously,
the functional responsibilities of various political/water agencies are
likely to remain splintered. Flood control or water supply agencies
are not likely to face operational problems with waste treatment, and
even municipal public works departments tend to be split into "water
departments" and "sewer departments" with special duties, their own
budgets, and their own problems. It is likely that only where wastewater
reclamation, including municipal reuse, becomes a viable alternative
will a single agency require comprehensive urban water modeling
approaches.
Specific comments of merit that we are happy to assign to specific
individual respondents include the following.
John Bailey, Jr. presented a lengthy and thoughtful discussion
of specific programming particulars that would be helpful to any
program developer whose program is to be used later by other parties.
Some of his major points are these:
1) "The user should be informed which sections of the program
coding were executed during the particular analyses that was
made. [The results] should also specify which sections of
coding were not executed ... where successive approxi-
mations or iterative solutions are made in the course of the
analyses, the report should specify whether the maximum
numbers of iterations built into the coding to avoid endless
loops were reached and whether a result of less than the
desired accuracy was finally adopted. "
2) ". . . at least 1/3 of the coding should be devoted to checking
the reasonableness and consistency of the input data ... it
is not necessarily a program defect if certain of the data
are required to be specified more than once or in different
forms, and if the computer then checks these data for consis-
tency before printing large volumes of . . .'numbers to be
used in making important decisions . . . often the user is
never certain that the problem being solved by the computer
is the same one that the user thinks he has specified. "
3) [Usually it is much more difficult to get data for a model
than it is to produce the computer program itself.] "For this
reason, it should be [made] rather difficult for the user to_
get a solution from the system model. He should be forced
to_ acknowledge, in the input stream, that he is_ using default
options, that certain data appear to_ be inconsistent or even
incorrect and that intermediate computed values are
53
-------
considerably larger or smaller than the values ordinarily to
be expected in the kind of engineering computations the model
is designed to make if such is the case. Statements in the
output stream like "rainfall depth is 12 inches in one hour',
or 'velocity in channel is 55 feet per second', or ,'pipe diameter
is 25 feet1, should be prominently displayed so that the user
is aware that values outside the normal range were encountered
during execution of the program. " [Emphasis is Bailey's.]
Acknowledging that these precautions.imply the use of extra
computer time for the many checks he advocates, Bailey points out
that the addition in computation time is small compared to the saved
headache and mistake time the user must spend deciphering an unanno-
tated piece of output. Lastly Bailey warned that programming for
others requires preknowledge of other hardware and operating software
that might subsequently be used with the program, and that the original
programming effort should be carefully executed to avoid statements
or operations likely to be incompatible with other users' machinery.
Bailey is obviously one of the fairly unique individuals who has both
programmed for others and has used other developers' programs. His
insights to the nitty-gritty programming and documenting safety checks,
therefore, are especially incisive, and the writers want to thank him.
particularly for his response.
Earl Jones is a hydrologist with HUD, highly experienced in flood
control analyses. He would tell anybody that he is not a modeler
himself. Nonetheless, he carries with him a body of experience with
the "resource out of place" called a flood and a patience, both of
which modelers occasionally test. Jones is especially concerned with
lack of data and experience in the hands of modelers who tend to rely
on their singular product, the program. "This has been a particular
problem with regard to use of the [named] model, which is excellent .
of itself, but which is dependent upon input data . . . that in turn are
often questionable, or at least possessed of undefined interdependence. "
"We find it interesting that Jones, the engineer-hydrologist, would
also conclude, "There remains a tendency to model quantifiable factors
and to disregard the seemingly intangible considerations. Soft sciences
inputs to water resources problems are greatly in need of further
refinements. "
Bob McWhinnie is the Director of Planning and Water Resources
for the Denver Board of "Water Commissioners. He also is not himself
a modeler. But the Board has a uniquely qualified modeling staff and
access to not one but several computers. Use of models is routine in
this forward-looking local agency [Hobbs and Britton, 1974], McWhinnie
suggests that modelers should stress:
54
-------
1) "the need to verify any model with local data,
2) the concept that the model is only a tool, and
3) that analysis of the results of the model run are the real
key to application of any model. "
This last point is highly significant and may be the most helpful
concept in ameliorating the developer /user communication problem.
It is worth amplifying that computer printout rarely if ever contains
"the" answer to the "the" problem. The significant analysis leading to
solution of a real problem starts when a successful run or set of runs
ends. A model user has to be a qualified results analyst, or he is not
a user at all. George Smoot highlights this distinction when he says,
"We have noted no exceptionable and unresolvable communication
problems with others in the field of modeling. Perhaps our biggest
concern is in reporting the accuracy of model results. " Getting a
program to execute with data given in the format described in a user's
manual is one problem, but interpreting the results is quite another
and more important problem.
Grigg, Tucker, and Wenzel have all cautioned that automatic
operational control software is going to be a complex and difficult
development problem. The writers had indicated that the programming
technology is now available or soon would be to permit the development
of the needed software in the near future. These respondents, parti-
cularly Grigg and Wenzel, point out that while the techniques are
available, the dog-work of checking all the possible control options,
programming the numerous but otherwise simple on-off commands
in the proper sequences, and general debugging and verification steps
will be tedious and time consuming. Several hundred years went by
between the time that man figured out how to get to the moon and
the time when he actually went. This is a fitting last restatement
of the warning against overpromising on the part of model developers.
Several of the respondents listed "characteristics" of some specific
programs they have developed, are now developing, or have acquired
from others and are now using. Some of the more interesting develop-
ments now underway are
1) A river system operation model (700 reaches) for quantity
and quality being developed by the Dallas Water Utilities.
This model, which will operate on hourly time steps, was
not completed by September 1974, but it was already being -
verified against field data.
2) A multiple impounding reservoir operations model (reservoirs
and 2 distribution centers) also being developed by the Dallas
Water Utilities. This will be a dynamic programming,
-------
optimization model operating in monthly time steps. It will
produce the optimal operation pattern in terms of cost and
final storage volume.
3) A reservoir operation simulation (flood routing) model being
developed by the Denver Board of Water Commissioners.
This model will operate on hourly time steps. It will be
an empirical model, but it was preliminarily (September 1974)
operating on a teletype terminal to a Honeywell 635 machine
in approximately 1 second for a 15 day simulation.
It is interesting to note that the Denver Board of Water
Commissioners is able to access five separate computers: IBM/370-195,
Honeywell/440, Honeywell/635, Univac/1108, and GE/430. The Dallas
Water Utilities uses an IBM/370-145. Resource Analysis, Inc. has
access to IBM/370-155 and IBM/370-195 equipment. The USGS model
reported by Smoot for peak flow synthesis for rainfall/runoff events
is operated on an IBM/370-155 machine. So routine users of scientific
software in the urban water field are generally using large storage,
rapid computers. Denver's access to such varied hardware is probably
unique, but entities interested in beginning routine water modeling
will probably find it advisable and economic to use at least one model
of first rate scientific computing equipment.
56
-------
SECTION VI
FUTURE URBAN WATER MODELS.
There is a variety of criteria that can be used to judge what
should be developed or what is likely to be developed next in the urban
water modeling field. In this chapter the writers' suggestions are based
on two criteria:
1) the state-of-the-art as just described in Section V, and
2) The writers' collective view of what is a) most likely of early
success, b) most required in terms of pressing needs, and
c) most feasible in terms of EPA's and the Storm and Combined
Sewer Section's responsibilities and budgetary wherewithal
as we understand them.
Right at the outset it might be helpful to summarize the models
believed to be needed and possible of early development. Following
that, each model type will be discussed in some detail. Urban water
models that should be developed by 1980 include the following
Planning
1) A new and better watershed quality model.
2) A transport simulation capability in a planning model for
storage/treatment/overflow evaluations (STORM-II).
3) Capability to simulate quality control or treatment processes
in STORM-II.
4) A long-period (10-30 year) receiving water ecologic model.
5) An urban water users economic effects model.
6) A downstream water users (receiving water or reuse) economic
effects model.
57
-------
Design/Analysis
1) A solids deposition and scour simulation capability in an
hydraulically sound sewer transport model.
2) Dry-weather waste treatment simulation capability in a SWMM-
type model.
3) Reclamation or reuse routing capability in a transport/
treatment model.
4) Nonstructural runoff control simulation capability "in SWMM-
type runoff module.
Operation/ Control
1) Real-time control software for sewer systems.
2) Real-time spatially varied rainfall prediction capability.
SIMULATION MODELS
Planning
Runoff Quality Model--
Quality simulation capability now existing in SWMM-type runoff
models is adequate for some planning applications. However, the dirt-
and-dust linkage inherent in this model can be and has been assailed
as difficult to calibrate and deficient in its statistical nature which
requires a considerable data base in each new calibration/application.
Particularly weak is the assumption that dissolved pollutants are in
some way related to the fractions of the same materials attached to '
suspended particles and that the attached pollutants are directly related
to the amount of suspendable dirt or dust being carried in the runoff.
There is no present capability to simulate subsequent dissolution or
dissociation of attached pollutants to the dissolved state.
What is needed is a still fairly simple, but more first-principles
derived model of the accumulation, motion, and fate of dissolvable and
solids-associated pollutants. The model should include relationships
and accounting procedures for the following states and change
processes:
Accumulation on watersheds:
Dissolvable materials (BOD, nutrients, heavy metals)
Solids-associated materials (BOD, metals)
58
-------
Motion during runoff
Dissolvable materials (fractions of accumulated
pollutants dissolved and carried with runoff)
Solids-associated materials (fraction of accumulated
solids scoured into motion and carried by runoff)
Fate of pollutants
Dissolvable materials (decay, precipitation, adsorption,
conservative transport)
Solids-associated materials (decay, dissolution,
deposition, resuspension)
Runoff Transport Model--
The storage-treatment-overflow and runoff model known as STORM
has received widespread application. Nonetheless, as useful as the
model is, virtually every user who has tried it has wished that the
sewer transport subsystem were represented in some fashion. The
users quickly point out that SWMM-type models are probably too
detailed and require too much data for the planning questions at hand,
but STORM has no transport routing capability at all. What is needed
is something in between--a continuity-balancing routing model for
major trunk sewers, either storm or combined ones. The model
probably would become too cumbersome if it had more than 20 routing
elements in it. Moreover, there need be no provision for hydraulic
solutions for weirs or other appurtenances, but one might wish to
specify flow diversions either as a constant value, a variable with
time, or a variable with head at another point. More refinement than
that would probably make the model unwieldy for planning applications.
The question to be answered is simply, how does the physical layout
of the city and its drainage network affect what should be done next
for storage, treatment, and overflow management? The answer can
be very significant, and incorporation of some routing capability in
STORM would probably be the simplest way to answer the question.
Quality Control Process Model--
We hesitate to call this simply a "treatment model" because what
is intended is less than a process simulation model for various unit
process strings and more than removal percentages for treatment
plants alone. What might well be done is to add to the transport
simulation just recommended some removal capability, variable by
location and constituent. Let's call the variable RMOV(N,I), where
N is the location identifier, and I is the constituent counter. This
would operate on the concentrations of various constituents at a node
to represent either treatment reductions or upstream land treatments,
retention basins, or other devices that reduce pollutant loads to
59
-------
downstream points. Refinements might include having one variable
for treatment processes for which, default removal efficiencies are
programmed, while upstream removals would have to be specified
each time.
Long-period Receiving Water Ecologic Model--
While there have been significant improvements beyond Streeter-
Phelps, DO sag models in recent years with regard to the simulation
of water quality effects on aquatic communities, the comprehensive
ecologic models developed have become temporally very detailed. They
generally use time-steps of 10 minutes to 3 hours (a day has also been
used) depending on hydraulic limitations and other factors, but in any
event they are usually operated to simulate the diurnal sunlight-
darkness phenomena associated with photosynthesis and temperature
changes. And, while they have been used, at some considerable
computer expense, to simulate as much as one year in these small
time steps, they appear to have an inherent tendency for long time
instability such that the investigator often finds himself at the end
of a simulated "year with a pond-full of fish, but no algae, zooplankton
or benthic animals.
There are important long-term, planning questions that need to be
asked and cannot be answered economically with the models currently
available. These include: 1) How much beneficial effect in the long
term future can we expect from instituting more advanced waste
treatment in, say, 1977, 1983, and 1985? 2) How long will it take
for a water body, say Lake Erie, to respond beneficially to removal
of waste loads by 25, 50, 75, or 100 percent? 3) What are the
long-term implications of increased population and changed land uses
on receiving water bodies to which, increased waste flows will be
discharged following secondary or more advanced treatment? 4) In
what year should nutrient removal for either nitrogen or phosphorus
or both be required for municipal or industrial discharges to a specific
receiving water?
It would seem that a long-period aquatic ecologic model would
be of considerable value. Some fundamental conceptualization and
experimentation will be necessary, however, before such a model
becomes a routinely usable tool. It will be necessary, for example,
to determine how big a time step may be used (days, months, seasons,
or years) while maintaining chemical, physical, and biological credi-
bility. Can the diurnal variabilities be averaged or assumed away
and still permit development of a model that faithfully reproduces
explainable long-term behavior? Will the representations of the more
stable aspects of a receiving water, such as the benthic community
and the fish population, not have to be more exact and better understood
as the time step is increased, particularly if it is increased beyond
the hydraulic residence time? Can such a model be built as a
60
-------
representation of physical, chemical and biological behavior, or must
it be a statistical analyzer and predictor operating on historical data?
These questions would have to be addressed in a feasibility study
before such a model could be constructed; but the uses of such a model
are compelling enough that the feasibility study, at least, should be
undertaken, and fairly soon.
Design/Analysis
Solids Modeling Capability- -
Into either the SWMM or San Francisco combined sewer transport
models, capability should be added to simulate the "first flush" of
deposited solids and associated pollutants, and the accumulation of
these materials during dry(er) weather.
This capability should include not only the physics and the timing
of the accumulation and scouring phenomena but also the simulation of
processes or devices to minimize their deleterious effects. For example,
cunettes in the bottom of large pipes or channels, although they present
certain clogging and maintenance problems, still work well if properly
designed. Another method of control that could be designed with model
assistance is the swirl concentrator. A third possible mechanism for
dealing with first flush solids is a conventional grit chamber, which
might have to be made larger than one designed for nominal dry weather
flows alone, to accommodate both higher flows and higher solids loadings
during storms. Simulation of all these devices to aid design of combined
sewer systems should be undertaken immediately.
New models should also address the implications of improved
solids management methods upstream on sludge handling at treatment
plants. Factors to consider include volumes generated plus the size
and cost of facilities for transmission, upstream entrapment, downstream
treatment and disposal. Consideration should also be given to the
treatability of stormwater solids by various processes, especially the
biological ones.
Dry Weather Waste Treatment--
In SWMM, treatment is represented as BOD, suspended solids, and
coliform removals that were based on data from specific field examples
and on general rules of thumb for a variety of treatment processes
applicable to stormwater. Treatment processes "selected" by SWMM
for construction are sized on the basis of the storm, hydrograph going
through during the simulation, and operating costs are computed on
the basis of power and chemical costs only during the storm.
What is needed is capability to simulate either existing or
planned additional treatment plants which may include primary,
61
-------
secondary, or advanced waste treament processes. Sizing should
be either specified (if it has nothing to do with the storm event currently
being simulated), or the option should exist to size the (new) plant
on the basis of a specified percentage, up to 100 percent, of the
maximum storm event flow that enters the plant during the simulated
period.
Costs should be calculated optionally as either 1) the annualized
capital and operation expenses regardless of the storm being simulated,
or 2) as the annual total cost times 1/365, times the number of days
in the simulated event. Options should also be developed whereby it is
possible to use existing capital and operating cost and flow-efficiency
data from the plants being simulated or to use generalized functions
from the literature to estimate these values.
The object should be to permit the calculation of effluent qualities
for various constituents and the cost to attain them for all the dry
weather and wet weather facilities available or to be added.
Reclamation/Reuse Routing Capability--
SWMM-type transport models should be expanded to include simu-
lation capability for routing storm or combined sewer flows (treated,
partially treated, or untreated) to points of further use wherein the
conveyance facility is not the receiving water. Second uses could
include industrial nonpotable uses, agricultural irrigation, and storage
for either recreational or eventual municipal reuse. The quantity
and quality of delivered water should be calculated; and if the water
enters storage, the resulting quality in the storage reservoir should
be calculated. If water must be discharged to the receiving water
net of that demanded for reuse, the resulting quantity and quality of
the discharge should be calculated as well. One refinement of this
capability that might be added is the ability to send water to the
point of reuse according to a specified demand schedule that varies
by season, month, day of the week, or hour.
The advent of zero discharge as a goal will make this capability
an item of ever-increasing priority.
Watershed Management Simulation Capability--
There are several runoff management practices that can be
instituted to lower the flow rate, total volume or pollutional load of
runoff at downstream points. These include 1) relatively small, but
perhaps numerous detention basins on upstream parts of urban
watersheds, 2) porous pavement materials, 3) roof-top storage,
4) street sweeping, litter pick-up, and other means of land surface
cleansing, and 5) erosion control practices at construction sites or
elsewhere.
62
-------
In current models of runoff, distinction is made between the runoff
produced from pervious surfaces and that from impervious, surfaces.
Rainfall occurring on pervious surfaces is converted to runoff following
two abstraction processes, infiltration first and detention storage
second. Rainfall on impervious surfaces is converted to runoff after
detention storage alone is satisfied. Pollutant loads are generated
from a dirt-and-dust accumulation linkage.
Consequently, to represent the runoff management practices listed
above would require rather simple modifications to existing programs in
terms of 1) increasing infiltration rates, 2) increasing detention storage
volumes, 3) defining new storage nodes in the model network, or
4) altering solids accumulation rates or schedules of accumulation
and cleaning..
SYSTEM CONTROL SOFTWARE -
During automatic operational control of urban water systems,
simulation will also take place; but a major distinction will be that the
time step and the duration of the simulated event will be real. That is,
the software will not be working with discrete and complete hyetographs
or hydrographs and be asked to simulate the entire storm. Quite
differently, it will be given partial records of rainfall at discrete
points on the ground and partial records of flow at discrete points
in the sewer system and asked to predict behavior ahead in real time
from right now. Such software is likely to be totally different, certainly
quite different, from SWMM or its cousins.
There are two operational software development projects that
should be initiated as soon as possible. These both are directed to real-
time operational control of combined sewer systems.
Spatially Varied Rainfall Prediction Capability
There are two possible ways that real-time automatic-control
of sewer systems can work: 1) in reaction mode, or 2) in predictive
mode. Now hardware and software can be linked together to operate in
reaction mode and the operation will go. This would entail real-time
statistical analyses of what had just occurred at several raingages
and/or at several pipe junctions or storage basins. However, it is not
difficult to see that conservative over-de sign would probably result if
historical data only were used to make real-time decisions. If it were
possible to look ahead ten minutes to an hour, better operational
decisions would almost certainly result.
Statistical analyses of very recent rainfall and head or storage
data (last 20 minutes or so) would still be used, but through "storm
pattern recognition" analyses the software could predict ahead whether
63.
-------
the rainfall, and hence runoff, to come would be greater, less, or the
same as that in the recent past. The control software would "be able
to "decide", then, that since rainfall is predicted to get even heavier,
say, that retention basin B should continue to empty; whereas even
though rainfall were very heavy at the current time, if the prediction
was for rainfall to cease imminently, the retention basin's outflow valve
could be closed. It is possible that quality considerations, such as
whether "first flush" had occurred yet or not, might als'o be
incorporated in the overall control model, but here we recommend
merely that the ability to predict rainfall ahead in real-time from
real-time, recent rainfall records be developed.
Real-Time Sewer Simulation and Control Models
In addition to the model just described for predicting rainfall inputs
in real time, it will be necessary to predict what will happen downstream
in the sewer system, given those inputs. Then it will be necessary to
open or close valves, divert or store flows, or otherwise change or
maintain operation on the basis o£ the predicted response.
The first model necessary is a real-time simulation model for the
sewer system. The current SWMM-type transport models could provide
the base from which such a model could be developed. However, to
minimize execution time and core storage, the operational simulation
model would have to be much reduced in scale. While it would have to
perform some routing from the point of control through downstream
check points, it would probably not have to simulate the pipe system
throughout the watershed. A network of pipes containing 10 or so pipes
and junctions would likely suffice. It is possible that this model or
variations of it would be used for several locations throughout a metro-
politan area, but in each real-time application only a small portion of
the system would be simulated and even then in a simplified manner.
Secondly, the software to actuate controls at various points such
as retention basins, inflatable dams, or valves must be developed.
Grigg and his'co-workers at Colorado State University are currently
developing such software and contemplating its testing with a laboratory-
scale retention basin. It remains to adapt their products to a prototype,
such as the retention basin(s) now planned for construction and operation
in the City of San Francisco. The approach to be used in development
of this software is to provide two or three levels or "heirarchies"
of control. One would regulate runoff in the trunk lines of a given
catchment area through use of local retention basins responding to
local rainfall and flows in local trunk sewers. The second level of
control would be superior to the first, in which local control could
be overridden according to available sewer or retention basins or
treatment capacity in another catchment area, to which flows might
be routed from a second catchment, regardless of current status in
the second area.
64
-------
While a third and still higher level of control is possible, it is
recommended that two level control be developed first. Additionally,
control models that direct responses based on quality should be
considered for future development as well. Optimal control could
be made dependent in part on suspended solids concentrations at various
points, either as monitored or as predicted by the simulation software.
The first task, however, should be the development of control software
that responds to rainfall inputs and downstream flow and storage inform-
ation monitored or simulated in real time.
COSTS AND BENEFITS
It may seem inappropriate to suggest that the Storm and Combined
Sewer Section of EPA develop models of benefits and costs related
to overall urban water management alternatives. However, the advent
of municipal reuse of storm and other waste-waters makes it imperative
that decision analytics be developed to help decide the worth of either
reusing water in the urban area from which the wastewaters are
generated or discharging the wastewaters, treated or otherwise to
permit beneficial or damaging impacts downstream.
Two basic models are recommended. One would assess the cost
for urban users of supplying and using waters of various qualities,
including reclaimed wastewaters from storm or combined sewers. The
second would assess the treatment costs and downstream damages, if
any, resulting from discharge of partially or "completely" treated
wastewater. These models are described below.
Urban Water User Economics Model
Many of the benefits, or foregone costs, of water supplies treated
to acceptable levels have yet to be counted. Indeed methods or equations
for relating concentrations of various constituents in water to what those
concentrations are worth have not been developed for constituents other
than TDS and hardness, and those benefits have not all been quantified.
In a social-acceptability context it has been argued that "benefits"
per se should not be sought or even discussed because they are likely to
be numerically much smaller than the costs to attain them. Instead, it.
has been suggested, "costs" should be counted including the costs users
must bear for slightly or considerably degraded quality water, and then
the objective should become minimization of total cost. So be it. The
argument is not compelling from an academic, how-to-do-it view.
The point is that damages are suffered, they are likely sizeable in
dollars and cents, and considerable arithmetic must be performed to
account for them. Therefore, a computerized accounting procedure
(model) should be built. Sufficient quality-use-cost data exist that such
a model could be built.
65
-------
The first step should be to derive explicit relationships between
costs (or savings) and concentrations (or reductions in concentrations)
for various constituents. Then a computer program can-slowly be built
to incorporate each new constituent, water use group, or treatment
process.
The model should be applicable to specific water supply, treatment,
and use planning alternatives, not a general model for the United States.
It should operate on time steps of not less than one year and probably of
five years. It should be able to deal with water supply alternatives
including blending of water from several sources either before or
after joint treatment, or before or after separate treatment including
none. Changes in quality within the distribution system resulting from
operational changes in supplying water from several sources (reservoirs
or treatment plants) can probably be ignored. Constituents for early
inclusion should be TDS, hardness, iron (and manganese), taste and
odor, and turbidity. Later inclusions might be bacteria and viruses,
organic compounds, sodium, and pH.
Receiving Water Users Economic Effects Model
A similar, but nonetheless different, model should be constructed
to assess damages to users of receiving waters who are located
downstream of single or multiple waste discharge points. The users
included should be 1) fish and wildlife, 2) recreation and esthetics,
3) industrial water users, 4) agriculture, and 5) municipal water users.
Constituents of interest, which should be related to downstream costs,
include dissolved oxygen, TDS, hardness, coliform organisms, heavy
metals, temperature, nitrogen (ammonia and nitrate), phosphorus,
and turbidity. Later additions might include pesticides, trace organics,
pH, color, sodium, and boron.
The input, qualities to be given to this model should be the concen-
trations at the point(s) of use. This implies that a transformation
function of some type will exist to perform the dilutioh-dispersion-
decay changes downstream from the points of discharge. Existing quality
models could be used or modified and then used for this purpose or the
functions could be solved by hand if need be.
Even though such a model should operate on long time steps of
one year or more, nonetheless it should be capable of reflecting at least
seasonal quality degradation caused by nonpoint runoff. Whether this
would be reflected by using stream qualities at the levels following
.average storms or the average annual quality including storm effects or
some combination of both could be left to the model user's discretion.
The model should calculate the annualized capital and operations
costs of an alternative plan of waste treatment and disposal facilities
and the costs to downstream users of accepting the resulting receiving
66
-------
water quality. The model could compare these costs for each alternative
with the total costs incurred in the "no action" alternative, presumably
evaluated first.
Lastly, it might be noted that "downstream" users could also
include reusers or reclaimers of stormwater or municipal or industrial
wastewaters.
67
-------
SECTION VII
PHASED IMPLEMENTATION PROGRAM
Quite naturally, the schedule and budget suggested below reflect
the writer's opinions of 1) priorities for earlier or later development,
and 2) the relative magnitude of the development tasks involved. The
funding levels suggested for each fiscal year may not be completely
realistic with respect to the budgetary future of the Storm and Combined
Sewer Section. Indeed the estimates for later years are higher than
the currently estimated figures we inferred during discussions with
Section personnel. This merely reflects another of our opinions: that
the current estimates are too low for an effective but ambitious
program. The model development tasks suggested are only those
described in Section VI, and considerably more work might be
deserving of attention. We have to presume that the other tasks
will be left to other agencies, sections, or years.
The suggested schedule and project budgets are shown in Table 2.
Some discussion of the major task headings is given below.
PLANNING--SIMULATION MODELS .
First of all, every model listed in Table 2 and described previously
in Section VI could be developed starting virtually immediately. The
earlier development tasks are so indicated because 1) there is a
pressing need for their results, or 2) their products are already
partially developed now.
The Runoff-Transport Model that is suggested for early development
in the planning tasks would be an addition of capability to the existing
STORM model. This could be accomplished by borrowing from the SWMM
.transport model or by developing a new, planning level representation of
runoff transport. But this task is assigned an early position primarily
because it represents an addition to currently developed tools.
The Improved Runoff Quality Model and the Long-period Ecologic
Model could both be used today. They will represent important new
68
-------
Table 2. PHASED IMPLEMENTATION PROGRAM
AND SUGGESTED MAN-EFFORT*
(estimates in man-years)
Task
Objectives
PLANNING- -SIMULATION
1) Improved Runoff Quality Model
2) Runoff Transport Model (STORM-II)
3) Quality Control Process Model
4) Long -Period Ecologic Model
DESIGN/ ANALYSIS- -SIMULATION
1) Solids Deposition-Scour Model
2) Solids Handling Model
3) Dry-Weather Treatment Model
4) Reclamation/Reuse Routing Model
Years
76 77 78 79 80
1 1 2
1
1 1
1 2
2
1 0.5
1 2
2226
(UWMM)
5) Watershed Management Controls
Model
OPERATION/CONTROL SOFTWARE
1) Real-Time Control Software
2) Real-Time Control Predictor
0.5
1.5 2
1.5
0.5
1.5 1.5
ECONOMIC PLANNING MODELS
1) Urban Water Use Economic Model
2) Receiving Water Economics Model
Annual Totals
7.5 8.5 8.5 9.5 10.5
^Administrative effort for the Storm and Combined Sewer Section is not
included.
69
-------
developments in planning capability. It is suggested that their develop-
ment be deferred until 1977 only because there are other developments
already planned for 1976, and these model projects will have to be
deferred one year.
Development of the Quality Control Process Model has been
deferred for several reasons. First, the capability represented by this
model can be crudely but easily approximated now in a SWMM runoff
model, but perhaps even better in the improved STORM model. So
partial capability can be introduced now or not later than 1977. Still
better, however, would be inclusion of this capability in the Improved
Runoff Quality Model not scheduled to be completed until 1979.
Additionally, the suggested budget for this task is relatively high
because some fundamental changes in the runoff model structure are
likely to be needed to represent upstream watershed controls
accurately. Specifically it may be necessary to develop a runoff
quantity-quality model that accommodates nodal locations within
subwatersheds, so a retention basin or other control measure can
be simulated to be at the upstream, middle, or downstream portion
of the watershed in question.
DESIGN/ANALYSIS--SIMULATION MODELS
The Storm and Combined Sewer Section is currently contemplating
development of a Solids Deposition and Scour Model for combined sewer
simulation. Consequently, its development has been scheduled early in
the program. Moveover, the work will likely entail incorporating
necessary statements into existing SWMM-type transport models; so
this task also represents an extension of existing capability,, and hence
it can begin immediately.
The development of a Solids Handling Model should begin almost
immediately. This problem represents the last major subarea of
modeling capability that is required to address the entire stormwater
pollution management picture.
The Dry-Weather Treatment Model has been deferred primarily
because more rule-of-thumb models simply are not needed, and cause-
and-effect models must await more fundamental research on kinetics
of individual processes and waste materials.
The Reclamation/Reuse Routing Model is destined to become the
Urban Water Management Model (UWMM). This model should be capable
of routing water and quality consitutents back through the urban water
use subsystem from the waste disposal subsystem. In short, it will
include feedbacks and hence be nonlinear. Optimization techniques
should be considered as possible ways of solving the resulting set of
equations to be solved simultaneously. Considerable effort and money
70
-------
will be required for the development, calibration, and demonstration
of this model in it final form. The demonstration effort will likely
take until beyond 1980.
The. Watershed Management Controls Model should be deferred
until more field data have been collected on the behavior of such
alternatives as porous pavements and erosion control practices. This
model will require inclusion of the Runoff Quality Model that properly
treats soluble and sediment-attached pollutants.
OPERATION/CONTROL, SOFTWARE
The two operation-level models identified are both geared to real-
time automatic operational control of combined sewer systems. We
have identified the Real-Time Control Software and the Rainfall
Predictor as the needed elements because work is going foward now
in the control area, because a demonstration grant to develop control
software is being contemplated, because the results are needed almost
immediately, and because rainfall prediction in real time will perhaps
be the single most important, central factor in controlling sewer
systems automatically.
ECONOMIC PLANNING MODELS
The Receiving Water Economics Model has been scheduled before
the Urban Water Use Economics Model for several reasons. First,
receiving waters are being polluted and damages are being suffered in
yet unquantified amounts right now. Moreover, agencies such as the
National Commission on Water Quality are faced already with answering
the question of whether meeting 'the PL 92-500 abatement levels will
be worth the cost. There is no analytical economic method today to
answer that question. The writers believe that sufficient data exist to
develop an accounting procedure relating concentrations of numerous
constituents, to the utility of water in streams, lakes, estuaries, and
the oceans. But formidable tasks remain to derive the relationships
from the literature and to encode these in a model, probably most
expeditiously tied to receiving water quality models.
The Urban Water Use Economics Model has been deferred because
some existing models already treat the economic effects of TDS and
hardness in urban water supplies, and because regardless of the
amounts of economic damages suffered from slightly degraded quality,
the tendency has been to supply whatever amounts of water are
demanded in at least adequate condition with respect to quality. As
water shortages increase, however, such a model is likely to be
very useful in allocating limited supplies of increasingly less desirable
waters.
71
-------
REFERENCES
American Public Works Association, Water Pollution Aspects of Urban
Runoff, Water Pollution Control Research Series, WP-20-.15,
Federal Water Pollution Control Administration, Washington,
D. C., January 1969, 272 p.
American Society of Civil Engineers and Water Pollution Control
Federation, Sewage Treatment Plant Design, ASCE Manual of
Practice No. 36, WPCF Manual of Practice No. 8, 1959, 357 p.
American Society of Civil Engineers, Urban Runoff Quantity and
Quality, Proceedings of a Research Conference, William Whipple,
Jr., editor, ASCE, New York, 1975, 272 p.
Amorocho, J. , and A. Brandstetter, "Determination of Nonlinear
Functional Response Functions in Rainfall-Runoff Processes, "
Water Resources Research, AGU, Vol. 7, No. 5, October 1971,
pp. 1087-1101.
Andrews, R. A. , and R. R. Weyrick, "Linear Programming Use for
Evaluating Water Resources and Cost and Benefit Allocation, "
Water Resources Bulletin, AWRA, Vol. 9, No. 2, April 1973,
pp. 258-272.
Anonymous, "Controls That Learn to Make Their Own Decisions,
Business Week, April 6, 1974, pp. 30D-30H.
Anonymous, "Minicomputers That Run the Factory, " Special Report,
Business Week, December 8, 1973, pp. 68-78.
Argaman, Y. , U. Shamir, and E. Spivak., "Design of Optimal Sewerage
Systems, " Journal of_ the Environmental Engineering Division,
ASCE, Vol. 99, No. EE5, October 1973, pp0 703-716,
AVCO Economic Systems Corporation, Storm Water Pollution from
Urban Land Activity, AVCO Economic Systems Corporation,
Washington, D. C. , 1970, 325 pc
AWWA Research Committee on Distribution Systems, "Priorities
in Distribution Research and Applied Development Needs, "
Journal of the American Water Works Association, Vol. 66,
No. 9, September 1974, pp. 507-509 =
AWWA Research Committee on Distribution Systems, "Water Distribution
Research and Applied Development Needs, " Journal of the
American Water Works Association, Vol0 66, No, 6, June 1974,
pp. 385-390.
72
-------
Becker, L., and W. W-G. Yeh, "Optimal Timing, Sequencing, and
Sizing of Multiple Reservoir Surface Water Supply Facilities,
Water Resources Research, AGU, Vol. 10, No. 1, February
1974, pp. 57-62.
Berthouex, P. M. , and L. B. Polkowski, "Optimum Waste Treatment
Plant Design Under Uncertainty, " Journal of the Water Pollution
Control Federation, Vol. 42, No. 9, September 1970, pp.
1589-1613.
Bidwell, V. J., "Regression Analysis of Nonlinear Catchment Systems, "
Water Resources Research, AGU, Vol. 7, No. 5, October 1971,
pp. 1118-1126.
Bishop, A. B., D. W. Hendricks, and J. H. Milligan, "Assessment
Analysis for Water Supply Alternatives, " Water Resources
Bulletin, AWRA, Vol. 7, No. 3, June 1971, pp. 542-553.
Blood, W. H., C. G. Clyde, and D. F. Peterson, Jr., "A Computer
Simulation Program for Optimizing Conjunctive Operation
of Desalting Plants, " Water Resources Bulletin, Vol. 7,
No. 4, August 1971, pp. 785-801.
Brandstetter, A., Assessment of Mathematical Models for Storm and Combined
Sewer Management? £pA, Municipal Environmental Research Laboratory.
Cincinnati, OH, 1975, 540 p.
Brandstetter, A. , R. L. Engel, and D. B. Cearlock, "A Mathematical
Model for Optimum Design and Control of Metropolitan'Wastewater
Management Systems, " Water Resources Bulletin,
AWRA, Vol. 9, No. 6, December 1973, pp. 1188-1200.
Brater, E. F. , and J. D. Sherrill, Rainfall-Runoff Relations on Urban
and Rural Areas, EPA-670/2-75-046, National Environmental
Research Center, EPA, Cincinnati, May 1975, 97 p.
Bree, D. W. , Jr., Reduced Representation of Demand-Serving Portion
of Water Distribution Networks, Technical Memorandum No. 2,
System Control Inc., Palo Alto, California, 1974, 29 p.
Brock, D. A., "Closed-Loop Automatic Control of Water System
Operations, " Journal of the American Water Works Association,
Vol. 55, No. 4, April 1963., pp. 467-480.
Brock, D. A, Metropolitan Water System Operation Subsequent to Nuclear
Attack or Natural Disaster, Dallas Water Utilities, Dallas, Texas,
May 1970, 364 p. (NTISAD711 956).
Brock, D. , "Simulation; Here's How," Water and Wastes Engineering,
Vol. 10, No. 7, July 1973, pp. 29-32.
73
-------
Brown, J. W. , M. R. Walsh, R. W. McCarley, A. J. Green, Jr.,
and H. W. West, Models and Methods Applicable to Corps of
Engineers Urban Studies, U. S. Army Waterways Experiment
Station Miscellaneous Paper H-74-8, Hydraulics Laboratory,
Vicksburg, August 1974, 34 p. and 382 p. appendix.
Burke, T. R., "A Municipal Water Demand Model for the Coterminous
United States, " Water Resources Bulletin, AWRA, Vol. 6,
No. 4, August 1970, pp. 661-681.
»
Carey, G. W. , L. Zobler, M. R. Greenber, and R. M. Hordon,
Urbanization, Water Pollution, and Public Policy, Center for
Urban Policy Research, Rutgers University, New Brunswick,
New Jersey, April 1972, 214 p.
Chen, C. K. , and W. W. Saxton, "Combined Wastewater Overflows, "
Journal of the Water Pollution Control Federation, Vol. 45, No. 3,
March 1973, pp. 434-448.
Chen, C. N. , "Planning Tools for Erosion Control in Urbanizing
Watershed, " Urban Runoff; Quantity and Quality, Proceedings
of a Research conference held at Franklin Pierce College, Rindge,
New Hampshire, August 11-16, 1974, American Society of Civil
Engineers, New York, 1975, pp. 159-165.
Chen, C. W. and G. T. Orlob, Ecologic Simulation for Aquatic
Environments, Final Report to the Office of Water Resources
Research, Water Resources Engineers, Walnut Creek, California,
December 1972, 156 p.
Chen, C. W. and G. T. Orlob, Ecologic Study of Lake Koocanusa--Libby
Dam, prepared for the Department of the Army, Seattle District,
Corps of Engineers, Water Resources Engineers, Walnut Creek,
California, January 1973, 63 p. and appendices.
Chen, C. W. and R. P. Shubinski, "Computer Simulation of Urban Storm
Water Runoff, " Journal of the Hydraulics Division, ASCE, Vol. 97,
No. HY2, February 1971, pp. 289-301,,
Chen, Gilbert K, L. T. Fan, and Larry F. Erickson, "Computer
Software for Wastewater Treatment Plant Design, " Journal of the
Water Pollution Control Federation, VoL 44, No. 5. May 1972,
pp. 746-762.
Chien, J., and K. K. Saigal, "Urban Runoff by Linearized Subhydrograph
Method, " Journal of the Hydraulics Division, ASCE, Vol. 100,
No. HY8, August 1974, pp. 1141-1157,
74
-------
Chiu, C., and J. T. Huang, "Nonlinear Time Varying Model of Rainfall-
Runoff Relation, " Water Resources Research, AGU, Vol. 6,
No. 5, October 1970, pp. 1277-1286.
Cleveland, J. G. , G. W. Reid, and J. F. Harp, Evalution o_f Dispersed
Pollutional Loads from Urban Areas, Bureau of Water Resources
Research, University of Oklahoma, Normal, 1970, 213 p.
Cleveland, J. G,, R. H. Ramsey, P. R. Walters, and G. L. Miessler,
A Multi-Phasic Component Study to Predict Storm Water
Pollution from Urban Areas, AVCO Economic Systems
Corporation, Washington, B.C., 1970, 248 p.
Colston, N. V., Jr., Characterization and Treatment of Urban Land
Runoff, EPA-670/2-74-096, National Environmental Research
Center, EPA, Cincinnati, Ohio, December 1974, 157 p.
DaCosta, P. C. C., "Effect of Urbanization on Storm Water Peak
Flows, " Journal of the Sanitary Engineering Division, ASCE,
Vol. 96, No. SA2, April 1970, pp. 187-193.
Dajani, J. S., R. S. Gemmell, and E. K. Morlok, "Optimal Design
of Urban Wastewater Collection Networks, " Journal of the Sanitary
Engineering Division, ASCE, Vol. 98, No. SA6, December 1972,
pp. 853-867.
Dajani, J. S. , and Y. Hasit, "Capital Cost Minimization of Drainage
Networks, Journal of the Environmental Engineering Division,
ASCE, Vol. 100, No. EE2, April 1974, pp. 325-337.
Deininger, R. A., "Systems Analysis of Water Supply Systems, " Water
Resources Bulletin, AWRA, Vol. 6, No. 4, August 1970,
pp. 573-579.
Dick, Richard I. , "Role of Activated Sludge Final Settling Tanks, "
Journal of the Sanitary Engineering Division, ASCE, Vol. 96,
No. SA2, April 1970, pp. 423-436.
Dick, Richard I. , and Ali R. Javakeri, Discussion of "Unified Basis for
Biological Treatment Design and Operation, " by Lawerence and
McCarty, Journal of the Sanitary Engineering Division, ASCE,
Vol. 97, No. SA2, April 1971, pp. 234-238.
Dietz, A. L.., "Computer Simulation of Recreational Boating Activities, "
Water Resources Bulletin, AWRA, Vol. 9, No. 3, June 1973,
pp. 494-498.
Dracup, J. A., and T. J. Fogarty, "Optimal Planning for a Thermal
Discharge Treatment System," Water Resources Research,
Vol. 10, No. 1, February 1974, pp. 67-71.
75
-------
Duckstein, L., M. M. Fogel, and C. C. Kiesel, "A Stochastic Model
of Runoff-Producing Rainfall for Summer Type Storms, " Water
Resources Research, AGU, Vol. 8, No. 2, April 1972,
pp. 410-421.
Dunne, T. , and R. D. Black, "Runoff Processes During Snowmelt, "
Water Resources Research, AGU, Vol. 7, No. 5, October 1971,
pp. 1160-1172.
Ecker, J. G. , and J. R. McNamara, "Geometric Programming and
the Preliminary Design of Industrial Waste Treatment Plants,
Water Resources Research, Vol. 7, No. 1, February 1971,
pp. 18-22.
Emsellem, Y. , and G. Demarsily, "An Automatic Solution for the
Inverse Problem," Water Resources Research, AGU, Vol. 7,
No. 5, October 1971, pp. 1264-1283.
Enger, T., and C. C. Feng, "Flowgraph Analysis of Pipeline Networks,"
Journal of the Hydraulics Division, ASCE, Vol. 97, No. HY10,
October 1971, pp. 1607-1624.
Engineering-Science, Inc., Characterization and Treatment of Combined
Sewer Overflows, submitted by San Francisco Department of
Public Works to the Federal Water Pollution Control Administration,
November 1967, 139 p. and figures.
Engman, E. T. , and A. S. Rogowski, "A Partial Area Model for Storm
Flow Synthesis, " Water Resources Research, AGU, Vol. 10,
No. 3, June 1974, pp. 464-472.
Environmental Protection Agency, Storm Water Management Model,
4 Volumes, Washington, D. C. , 1971. "Volume I-Final Report",
"Volume II-Verification and Testing", "Volume Ill-User's Manual",
"Volume IV-Program Listing".
Epp, R. , and A. G. Fowler, "Efficient Code for Steady-State Flows
in Networks, " Journal of the Hydraulics Division, ASCE,
Vol. 96, No. HY1, January 1970, pp. 43-56. Closure, Vol. 97,
No. HY7, July 1971, pp. 1131-1132.
Evenson, D. E., and J. C. Moseley, "Simulation/Optimization Techniques
for Multi-Basin Water Resources Planning, " Water Resources
Bulletin, Vol. 6, No. 5, October 1970, pp. 725-736.
Fan, L. T. , L. E. Erickson, J. C. Baltes, and P. S. Shah, "Analysis
and Optimisation of Two-Stage Digestion, " Journal of the Water
Pollution Control Federation, Vol. 45, No. 4, April 1973, pp.
591-610.
76
-------
Fair, G. M., and J. C. Geyer, Water Supply and Waste-Water Disposal,
John Wiley & Sons, Inc., New York, 1954, 973.
Feigner, K. D. , and H. S. Harris, Documentation Report--FWQA
Dynamic Estuary Model, Federal Water Quality Administration,
U.S. Department of the Interior, Washington, D. C., July 1970,
248 p.
Fisher, J. M., G. M. Karadi, and W. W. McVinnie, "Design of Sewer.
Systems, " Water Resources Bulletin, AWRA, Vol. 7, No. 2
'April 1971, pp. 294-302.
Frank D. Masch and Associates and the Texas Water Development Board,
Simulation of Water Quality in Streams and Canals, Theory and
Description of the QUAL-I Mathematical Model System, Texas
Water Development Board Report No. 128, Austin, Texas,
May 1971, 64 p.
Frenz, C. M., "Automated System Control, " Journal of the American
Water Works Association, Vol. 63, No. 8, August 1971,
pp. 308-312.
Genet, L. A., D. J. Smith, and M. B. Sonnen, Computer Program.
Documentation for the Dynamic Estuary Model, prepared for
the Environmental Protection Agency, Systems Development
Branch, Water Resources Engineers, Walnut Creek, California,
May 1974, 226 p.
Gibbs, C. V. , S. M. Alexander, and C. P. Leiser, "System for
Regulation of Combined Sewage Flows, " Journal of the Sanitary
Engineering Division, ASCE, Vol. 98, No. SA6, December 1972,
pp. 951-972.
Gilman, H. D. , M. Y. Goodman, and R. DeMoyer, Jr., "Replication
Modeling for Water Distribution Control, " Journal of the
American Water Works Association, Vol. 65, No. 4, April
1973, pp. 255-260.
Glanz, D. J. and G. T. Orlob, Lincoln Lake Ecologic Study, prepared
for U.S. Army Corps of Engineers, Louisville District, Water
Resources Engineers, Springfield, Virginia, December 1973,
98 p.
Gotaas, H. B., and W. S. Caller, "Design Optimization for Biological
Filter Models, " Journal £f the Environmenta 1 Engineering
Division, ASCE, Vol. 99, No. EE6, December 1973,
pp. 831-850.
77
-------
Grade, G. , "Analysis of Distribution Demand Variations, " Journal of
the American Water Works Association, Vol. 58, No. 1,
January 1966, pp. 51-67. '
Grigg, N. S. , J. W. Labadie, and H. G. Wenzel, Metropolitan Water
Intelligence Systems, Completion Report - Phase III, Submitted
to Office of Water Resources Research, Colorado State University,
Fort Collins, June 1974, 258 p.
Guarino, Carmen, F., Harold D. Gilman, Michael D. Nelson, and
Carl M. Koch, "Computer Control of Wastewater Treatment, "
Journal of the Water Pollution Control Federation, Vol. 44, No.
9, September 1972, pp. 1718-1729.
Heaney, J. P. , W. C. Huber, H. Sheikh, J. R. Doyle, J. E. Darling,
and W. A. Peltz, Final Report, Storm Water Management
Model; Refinements, Testing and Decision-Making, for the
City of Lancaster, Pennsylvania and EPA, University of Florida,
Gainesville, June 1973, 249 p.
Heeps, D. P., and R. G. Mein, "Independent Comparison of Three
Urban Runoff Models, " Journal of the Hydraulics Division, ASCE,
Vol. 100, No. HY7, July 1974, pp. 995-1009.
Heinemann, H. G. , and R. F. Piest, "Soil Erosion--Sediment Yield
Research in Progress," EOS, Transactions, American Geophysical
Union, Vol. 56, No. 3, March 1975, pp. 149-159.
Hinomoto, H., "Dynamic Programming of Capacity Expansion of Municipal
Water Treatment System, " Water Resources Research, AGU,
Vol. 8, No. 5, October 1972, pp. 1178-1187.
Hittman Associates, Inc., Forecasting Municipal Water Requirements,
Report HIT-413, Hittman Associates, Inc., Columbia, Maryland
September 1969; Vol. I, "The MAIN II System," 208 p.; Vol. II,
"The MAIN II System Users Manual, " 425 p, (NTIS PB 190 275
and PB 190 276).
Hobbs, N. , and J. D. Britton, Computer Modeling Applications in
Urban Water Planning, ASCE Urban Water Resources Research
Program Technical Memorandum No, 22, American Society of
Civil Engineers, New York, March 1974, 31 p.
Hoffman, T. W. , D. R. Woods, K. L. Murphy, and J. D. Norman,
"Simulation of a Petroleum Refinery Waste Treatment Process, "
Journal of the Water Pollution Control Federation, Vol. 45, No.
11, November 1973, pp, 2321-2334,
78
-------
Hoppel, S. K., and W. Viessman, Jr., "A Linear Analysis of an Urban
Water Supply System, " Water Resources Bulletin, AWRA, Vol. 8,
No. 2, April 1972, pp. 304-311.
Huber, W. C. , J. P. Heany, M. A. Medina, W. A. Peltz, H. Sheikh,
and G. F. Smith, Storm Water Management Model User's Manual;
Version II, National Environmental Research Center, Environment
Protection Technology Series, No. EPA-670/2-75-017, Environ-
mental Protection Agency, Cincinnati, March 1975, 350 p.
Huff, D. D., and P. Kruger, "Simulation of the Hydrologic Transport
of Radioactive Aerosols," Advances in Chemistry Series No. 93,
American Chemical Society, Washington, D. C., 1970,
pp. 487-505.
Huff, F. A., "Spatial Distribution of Rainfall Rates, " Water Resources
Research, AGU, Vol. 6, No. 1, February 1970, pp. 254-260.
Huff, F. A., "Time Distribution Characteristics of Rainfall Rates, "t
Water Resources Research, AGU, Vol. 6, No. 2, April 1970,
pp. 447-454.
Jacoby, S. L. S., "Design of Optimal Hydraulic Networks, " Journal
of the Hydraulics Division, ASCE, Vol. 94, No. HY3, May 1968,
pp. 641-661. Closure, Vol. 95, No. HY5, September 1969,
pp. 1732-1735.
Jenkins, David, and Walter E. Garrison, "Control of Activated Sludge
by Mean Cell Residence Time," Journal of the Water Pollution
Pollution Control Federation, Vol. 40, No. 11, Part 1, November
1968, pp. 1905-1919.
Joeres, E. F., J. C. Liebman, and C. S. Revelle, "Operating Rules
for Joint Operation of Raw Water Sources, " Water Resources
Research, AGU, Vol. 7, No. 2, April 1971, pp. 225-235.
Juknievich, J. , and P. Richardson, "It's a Big Jump, " Water and
Wastes Engineering, Vol. 10, No. 6, June 1973, pp. 38-39.
Klei, H. E., and D. W. Sundstrom, "Automatic Control of an Activated
Sludge Reactor, " Journal jof the Water Pollution Control Federation,
Vol. 46, No. 5, May 1974, pp. 993-998.
Kugelman, I. J., "Water Reclamation and Reuse, " Journal of the
Water Pollution Control Federation, Vol. 46, No. 6, June 1974,
pp. 1195-1199.
79
-------
Lacroix, P. G. , and D. G. Bloodgood, "Computer Simulation of
Activated Sludge Plant Operation, " Journal of the Water Pollution
Control Federation, Vol. 44, No. 9, September 1972, pp.
1778-1788.
Lacroix, P. G. , and D. G. Bloodgood, "Computer-Assisted Activated
Sludge Plant Operation, " Journal £f the Water Pollution Control
Federation. Vol. 44, No. 12, December 1972, pp. 2205-2217.
Lager, J. A. , R. P. Shubinski, and L. W. Russell, "Development of
a Simulation Model for Stormwater Management, " Journal of the
Water Pollution Control Federation, Vol. 43, No. 12, December
1971, pp. 2424-2435.
Lanyon, R. F. , and J. Jackson, "Flow Simulation System, " Journal of
the Hydraulics Division, ASCE, Vol. 100, No. HY8, August 1974,
pp. 1089-1105.
Lawrence, Alonzo W., and Perry L. McCarty, "Unified Basis for
Biological Treatment Design and Operation, " Journal of the
Sanitary Engineering Division, ASCE, Vol. 96, No. SA3,
June 1970, pp. 757-778.
Linsley, R. K. , A Critical Review of Currently Available Hydrologic
Models for Analysis of Urban Stormwater Runoff, Hydrocomp
International, Palo Alto, California, August 1971, 83 p.
Mallory, C. W. , The Beneficial Use of Storm Water, Final Report to
U. S. Environmental Protection Agency, Environmental Protection
Technology Series EPA-R2-73-139, ffittman Associates, Inc.,
Columbia, Maryland, January 1973, 266 p.
Mandeville, A. N. , and T. O'Donnell, "Introduction of Time Variance to
Linear Conceptual Catchment Models," Water Resources Research,
AGU, Vol. 9, No. 2, April 1973, pp. 298-310.
McCuen, R. H. , "A Sequential Decision Approach in Recreational
Analysis," Water Resources Bulletin, AWRA, Vol. 9, No. 2,
April 1973, pp. 219-230.
McPherson, M. B. , and M. Heidari, "Power Consumption with Elevated
Storage Compared to Direct and Booster Pumping,"' Journal of
the American Water Works Association, Vol. 58, No. 12,
December 1966, pp. 1585-1594,
McPherson, M. B. , and R. Prasad, "Distribution System Equalizing
Storage Hydraulics, " Journal of_ the Hydraulics Division, ASCE,
Vol. 92, No. HY6, November 1966, pp. 151-178. Closure, Vol. 94,
No. HY2, March 1968, pp. 558-559.
80
-------
McPherson, M. B. , and R. Prasad, "Power Consumption for
Equalizing-Storage Operating Options, " Journal of the American
Water Works Association, Vol. 58, No. 1, January 1966, pp.
67-90. Corrections, Vol. 58, No. 4, April 1966, p. 454.
McPherson, M. B. , "Complete Automation--It's Still a Long .Ways
Off, " Water and Wastes Engineering. Vol. 10, No. 6, June 1973,
pp. 36-38.
McPherson, M. B. , Feasibility of the Metropolitan Water Intelligence
System Concept (Integrated Automatic Operational Control),
ASCE Urban Water Resources Research Program, Technical
Memorandum No. 15, New York, December 1971, 110 p.
(NTIS PB 207 301).
McPherson, M. B. , "Generalized Distribution Head Loss Charac-
teristics, " ^Transactions^ ASCE, Vol. 126, Parti, 1961, pp.
1190-1234.
McPherson, M. B., "Ground Storage Booster Pumping Hydraulics,"
Journal of the American Water Works Association, Vol. 58,
No. 3, March 1966, pp. 311-325.
McPherson, M. B. , "Need for Metropolitan Water Balance Inventories",
Journal of the Hydraulics Division, ASCE, Vol. 99, No. HY10,
October 1973, pp. 1837-1848.
Mein, R. G. , and C. L. Larson, "Modeling Infiltration During a Steady
Rain, " Water Resources Research, AGU, Vol. 9, No. 2, April
1973, pp. 384-394.
Merritt, L. B., and R. H. Bogan, "Computer-Based Optimal Design
of Sewer Systems, " Journal of the Environmental Engineering
Division, ASCE, Vol. 99, No. EE1, February 1973, pp. 35-53.
Miller, C. R. , and W. Viessman, Jr., "Runoff Volumes of Small
Urban Watersheds," Water Resources Research, AGU, Vol. 8,
No. 2, April 1972, pp. 429-434.
Minneapolis-St. Paul Sanitary District, Dispatching System for Control
of Combined Sewer Losses, EPA Water Pollution Control Research
Series 11020 FAQ 03/71, March 1971, 209 p.
Mishra, Prakash N., Liang-tseng Fan, and Larry E, Erickson,
"Optimal Capacity Expansion of a Secondary Treatment System, "
Journal of the Water Pollution Control Federation, Vol. 46, No.
12, December 1974, pp. 2704-2718.
81
-------
Montgomery-Water Resources Engineers, Water Waste-water and Flood
Control Facilities Planning Model, submitted to San Diego
Comprehensive Planning Organization, January 1974, 189 p.
Mulvihill, M. E. , and J. A. Dracup, "Optimal Timing and Sizing of
a Conjunctive Urban Water Supply and Waste Water System
with Nonlinear Programming, Water Resources Research,
Vol. 10, No. 2, April 1974, pp. 170-175.
Norton, W. R., L.. A. Roegner, J. R, Monser and D. E. Evenson,
Computer Program Documentation for the Stream Quality Model,
QUAL-II, prepared for the Environmental Protection Agency,
Systems Development Branch, Water Resources Engineers,
Walnut Creek, California, May 1974, 84 p. plus listings and
example.
Papadakis, C. N. , and H. C. Preul, "Testing of Methods for Deter-
mination of Urban Runoff, " Journal of the Hydraulics Division,
ASCE, Vol. 9.9, No. HI9, September 1973, pp. 1319-1335.
Parkin, G. F., and R. R. Dague, "Optimal Design of Wastewater
Treatment Systems by Enumeration, " Journal of the Sanitary
Engineering Division, ASCE, Vol. 98, Nof SA6, December 1972,
pp. 833-851.
Pitt, R. E., and R. Field, "Water Quality Effects from Urban Runoff, "
Presented to the American Water Works Association 94th Annual
Conference, Boston, Massachusetts, June 16-21, 1974, Session
Paper No. 4-4, Copyright June 1974 by AWWA, 14 p.
Preul, H. C., and C. Papadakis, Urban Runoff Characteristics, for
the U.S. Environmental Protection Agency, University of
Cincinnati, Cincinnati, Ohio, 1970, 340 p.
Radziul, J. V., "Automation and Instrumentation Committee's Purpose, "
Journal of the American Water Works Association, Vol. 63,
No. 8, August 1971, pp. 467-469.
Rao, H. S., and D. W. Bree, Jr., Improved Methodology for Planning
and Operation of Water Distribution Systems, Report to OWRT,
Systems Control, Inc., Palo Alto, California (in press).
Rao, H. S. , D. W. Bree, Jr., and R. Benzvi, Extended Period
Simulation of Water Distribution Networks, Systems Control,
Inc., Palo Alto, California, February 1974, 133 p.
(NTIS PB 230 148).
Raudkivi, A. J. , and N. Lawgun, "Generation of Serially Correlated
Nonnormally Distributed Rainfall Durations, " Water Resources
Research, AGU, Vol. 8, No. 2, April 1972, pp. 398-409.
82
-------
Raudkivi, A. J., and N. Lawgun, "Synthesis of Urban Rainfall, " Water
. Resources Research, AGU, Vol. 6, No. 2, April 1970,
pp. 455-464.
Riordan, C., "General Multistage Marginal Cost Dynamic Programming
Model for the Optimization of a Class of Investment-Pricing
Decisions, " Water Resources Research, Vol. 7, No. 2, April
1971, pp. 245-253.
Riordan, C., "Multistage Marginal Cost Model of Investment-Pricing
Decisions: Application to Urban Water Supply Treatment Facilities, "
Water Resources Research, AGU, Vol. 7, No. 3, June 1974,
pp. 463-478.
Roesler, J. F., R. Smith, and R. G. Eilers, "Simulation of Ammonia
Stripping from Wastewater, " Journal of the Sanitary Engineering
Division, ASCE, Vol. 97, No. SA3, June 1971, pp. 269-286.
Roesner, L. A., J. R. Monser and D. E. Evenson, Computer Program
Documentation for the Stream Quality Model QUAL-II, prepared
for the Environmental Protection Agency, Systems Development
Branch, Water Resources Engineers, Walnut Creek, California,
May 1973, 84 p. and listings and example.
Roesner, L. A. , H. M. Nichandros, R. P. Shubinski, A. D. Feldman,
J. W. Abbott, and A. O. Friedland, A Model for Evaluating
Runoff-Quality in Metropolitan Master Planning, ASCE Urban
Water Resources Research Program Technical Memorandum
No. 23, American Society of Civil Engineers, New York,
pril 1974, 73 p.
Salas-LaCruz, J. D., and V. Yevjevich, Stochastic Structure of Water
Use Time Series, Colorado State University, Hydrology Paper
No. 52, June 1972, 71 p.
Sartor, J. D., and G. B. Boyd, Water Pollution Aspects of Street
Surface Contaminants, EPA-R2-72-081, Prepared for Office of
Research and Monitoring, EPA, Washington, D. C., November
1972, 236 p.
Scarato, R. F,, "Time-Capacity Expansion of Urban Water Systems, "
Water Resources Research, AGU, Vol. 5, No. 5, October 1969,
pp. 929-936.
Schaake, J. C., Jr., and D. C. Major, "Model for Estimating Regional
Water Needs," Water Resources Research, AGU, Vol. 8, No. 3,
June 1972, pp. 755-759.
83
-------
Shamir, U., "Optimal Design and Operation of Water Distribution
Networks," Water Resources Research, AGU, Vol. 10, No. 1,
February 1974, pp. 27-36.
Shamir, U., Water Distribution Systems Analysis, Report RC 4389,
IBM, Thomas J. Watson Research Center, Yorktown Heights,
New York, June 13, 1973, over 200 p.
Shih, C. S., and J. A. DeFilippi, "System Optimization of Waste
Treatment Plant Process Design, " Journal of the Sanitary
Engineering Division, ASCE, Vol. 96, No. SA2, April 1970,
pp 409-421.
Singh, K. P. and C. G. Lonnquist, "Water Treatment Plant Staging
Policy, " Water Resources Bulletin, AWRA, Vol. 8, No. 2,
April 1972, pp. 239-249.
Smith, D. W., "Computer Design of CMAS Systems," Journal of the
Sanitary Engineering Division, ASCE, Vol. 96, No. SA4, August
1970, pp. 977-990.
Smith, F. L,., Jr., "The Urban Water System--A Comprehensive
Analysis, " Thesis presented to the University of Denver, at
Denver, Colora'do, 1972, in partial fulfillment of the require-
ments for the degree of Master of Science.
Smith, Robert, Preliminary Design and Simulation of Conventional
Wastewater Renovation Systems Using the Digital Computer,
Water Pollution Control Research Series Report WP-20-9,
Federal Water Pollution Control Administration, Cincinnati,
Ohio, March 1968, 64 p. and appendix.
Sonnen, M. B., "Quality Related Costs of Regional Water Users, "
Journal of the Hydraulics Division, ASCE, Vol. 99, No. HY10,
October 1973, pp. 1849-1864. Closure, Vol. 101, No. HY4,
April 1975, pp. 407-408.
Sullivan, Richard H. , "Letters, " Journal of the Water Pollution
Control Federation, Vol. 46, No. 8, August 1974, p. 537a.
Systems Control, Inc., Use of Mathematical Models for Water Quality
Planning, Water Resources Information System Technical Bulletin
No. 3, State of Washington, Department of Ecology, June 1974,
212 p.
Texas Water Development Board, Stochastic Optimization and Simulation
Techniques for Management p_f Regional Water Resource Systems',
TWDB Report No. 131, Austin, Texas, July 1971, 129 p.
84
-------
Tihansky, Dennis P., "Historical Development of Water Pollution
Control Cost Functions," Journal of the Water Pollution Control
Federation, Vol. 46, No. 5, May 1974, pp. 813-833.
Tihansky, D. P., "Damage Assessment of Household Water Quality, "
Journal of the Environmental Engineering Division, ASCE,
Vol. 100, No. EE4, August 1974, pp. 905-918.
Trotta, P. D., "On-line Adaptive Control for Combined Sewer
Systems," thesis-presented to Colorado State University, at Fort
Collins, Colorado, 1975, in partial fulfillment of the requirements
for the degree of Doctor of Philosophy.
U.S. Army Corps of Engineers, Urban Runoff; Storage, Treatment,
and Overflow Model "STORM", Draft Generalized Computer
Program 723-58-2250, Hydrologic Engineering Center, Davis,
California, September 1973, 62 p.
Van Dyke, R. P. , "Computerized Operation of Distribution Systems, "
Journal of the American Water Works Association, Vol. 62,
No. 4, April 1970, pp. 234-240.
Viessman, W. , Jr., W. R. Keating, and K. N. Srinivasa, "Urban
Storm Runoff Relations, " Water Resources Research, AGU
Vol. 6, No. 1, February 1970, pp. 275-279.
Vitale, A. M., and P. M. Sprey, Total Urban Water Pollution Loads:
The Impact of Storm Water, Prepared for the Council on
Environmental Quality, Enviro-Control, Inc., Rbckville,
Maryland, 1974, 183 p.
Voshel, Doris, and J. G. Sak, "Effect of Primary Effluent Suspended
Solids and BOD on Activated Sludge Production, " Journal of the
Water Pollution Control Federation, Vol. 40, No. 5, Part 2, May
1968, pp. R203-R212.
Walsh, S., and L. C. Brown, "Least Cost Method for Sewer Design, "
Journal of the Environmental Engineering Division, ASCE,
Vol. 99, No. EE3, June 1973, pp. 333-345.
Water Resources Engineers, An Investigation of Salt Balance in the Upper
Santa Ana River Basin, presented to the State Water Resources
Control Board and the Santa Ana River Basin Regional Water
Quality Control Board, March 1969, 198 p.
Water Resources Engineers, Comprehensive System Engineering
Analysis of All Aspects of Urban Water . . . A Prefeasibility
Study, Prepared for the ASCE Urban Water Resources Research
Program, Water Resources Engineers, Walnut Creek, California,
July 1968, 121 p.
85
-------
Water Resources Engineers, Computer Program Documentation for the
Dynamic Estuary Model, prepared for the Florida Department
of Pollution Control, Water Resources Engineers, Austin, Texas,
September 1974, 234 p.
Water Resources Engineers, Systems Analysis for Urban Water
Management, Prepared for the Office of Water Resources
Research, Through the ASCE Urban Water Resources Research
Program, Water Resources Engineers, Walnut Creek, California,
September 1970, 78 p.
Weddle, C. L,., S. K. Mukherjee, J. W. Porter, and H. P. Skarheim,
"Mathematical Model- for Water-Wastewater Systems, " Journal
of the American Water Works Association, Vol. 62, No. 12,
December 1970, pp. 769-775.
Wenzel, H. G., and B. H. Bradford, Control Strategy Development
Study, San Francisco Master Plan, Colorado State University,
Ft. Collins, August 1974, 138 p.
Whipple, W., J. V. Hunter, and S. L. Yu, "Unrecorded Pollution
from Urban Runoff, " Journal of_ the Water Pollution Control
Federation, Vol. 46, No. 5, May 1974, pp. 873-885.
Whipple, William, Jr., "Letters, " Journal of the Water Pollution
Control Federation, Vol. 46, No. 11, November 1974, p. 809a.
Whitford, P. W. , "Residential Water Demand Forecasting, " Water
Resources Research, AGU, Vol. 8, No. 4, August 1972,
pp. 829-839.
Wiebel, S. R. , R. J. Anderson, and R. L. Woodward, "Urban Land
Runoff as a Factor in Stream Pollution, Journal of the Water
Pollution Control Federation, Vol. 36, No. 7, July 1964 pp.
914-924.
Winiarski, L. D. , and B. A. Tichenor, "Model of Natural Draft
Cooling Tower Performance," Journal of the Sanitary Engineering
Division, ASCE, Vol. 96, No. SA4, August 1970, pp. 927-943.
Wiseman, R. A., and M. B. McPherson, A Study of the Applicability
of Generalized Distribution Network Head Loss Characteristics,
Civil Engineering Studies, Hydraulic Engineering Series No. 7,
University of Illinois, Urbana, June 1965, 42 p. and appendices.
86
------- |