?,EPA
United States
Environmental Protection
Agency
Municipal Environmental Research
Laboratory
Cincinnati OH 45268
EPA-600/8-80-048
November 1980
Research and Development
User's Guide
Methodology for
Evaluating the
Impact and
Abatement of
Combined Sewer
Overflows
A Case Study of
Onondaga Lake,
New York
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established Jo facilitate further development and application of en-
vironmental technology. Elimination of traditional- grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology ,
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the "SPECIAL" REPORTS series. This series is
reserved for reports targeted to meet the technical information needs of specific
user groups. The series includes problem-oriented reports, research application
reports, and executive summary documents. Examples include state-of-the-art
analyses, technology assessments, design manuals, user manuals, and reports
on the results of major research and development efforts.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/8-80-048
November 1980
METHODOLOGY FOR EVALUATING THE IMPACT
AND ABATEMENT OP COMBINED SEWER OVERFLOWS
A Case Study of Onondaga Lake, New York
Peter E. Moffa
John C. Byron
Steven D. Freedman
Stearns & Wheler, Civil and Sanitary Engineers
Cazenovia, New York 13035
and
John M. Karanik
Randy Ott
Department of Drainage and Sanitation
Onondaga County, New York 13215
Grant No. R805096
Project Officer
Richard Field
Storm and Combined Sewer Section
Wastewater Research Division
Municipal Environmental Research Laboratory (Cincinnati)
Edison, New Jersey 08817
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Municipal Environmental
Research Laboratory, U. S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents neces-
sarily reflect the views and policies of the U. S. Environmental Protection
Agency, nor does mention of trade names or commercial products consti-
tute endorsement or recommendation for use.
11
-------
FOREWORD
The U.S. Environmental Protection Agency was created because
of increasing public and government concern about the dangers of
pollution to the health and welfare of the American people. Nox-
ious air, foul water, and spoiled land are tragic testimonies to
the deterioration of our natural environment. The complexity of
that environment and the interplay of its components require a
concentrated and integrated attack on the problem.
Research and development is that necessary first step in
problem solution; it involves defining the problem, measuring its
impact, and searching for solutions. The Municipal Environmental
Research Laboratory develops new and improved technology and
systems to prevent, treat, and manage wastewater and solid and
hazardous waste pollutant discharges from municipal and community
sources, to preserve and treat public drinking water supplies,
and to minimize the adverse economic, social, health, and aes-
thetic effects of pollution. This publication is one of the
products of that research and provides a most vital communications
link between the researcher and the user community.
This Methodology can be used by municipalities, consulting
engineers and planners for the evaluation of the impact and
abatement of combined sewer overflows on receiving waters. In
this report the relationships between rainfall and overflow and
between overflow and water quality are developed and an approach
for selection of the most cost-effective level of treatment is
presented.
Francis T. Mayo
Director
Municipal Environmental Research
Laboratory
111
-------
ABSTRACT
A general methodology is presented for the evaluation of the impact
and abatement of combined sewer overflows on receiving waters. It was
developed from experience with Onondaga Lake, an urban lake in Central
New York that receives combined sewer overflows from the City of
Syracuse via three tributary streams.
Field investigations of the combined sewer system and the receiving
water must first be undertaken. The field work includes flow measurement
and water-quality sampling of the sewer overflows and the receiving water
during several different storms. Use of a computerized data bank has been
found virtually essential for the storage and manipulation of the large
quantity of data resulting from the sampling and analysis.
Mathematical modeling of the receiving water is undertaken to evalu-
ate water quality as a function of pollutant load; the storm sewer system is
modeled to determine the quantities of pollutants discharged during various
storm conditions. Prior to the modeling effort, analysis of local rainfall
records is necessary to develop the classical intensity-duration-frequency
relationships. After assessing the water-quality impact of dry-weather
pollutants from wastewater treatment plants and other sources, the results
of the two models can be combined to express the reduction in stormwater
pollutants needed to achieve a particular water-quality goal as a function of
storm frequency or storm recurrence interval.
Abatement alternatives, and their respective costs, for the reduction
of pollutants from wet-weather sources, particularly combined sewer over-
flows, are next investigated. Using engineering judgment of the most effec-
tive and economic abatement measures, a relationship is then developed
between abatement cost and storm condition for each of several water-
quality criteria or goals. From the cost-benefit relationships thus devel-
oped, a graphical determination can be made of the "general optimum solu-
tion" (GOS) for reduction or treatment of combined sewer overflows.
It is recognized that the quality of the receiving water resulting from
the GOS may not be acceptable to the general public or regulatory agencies.
In that case, a decision to provide greater (or lesser) pollution abatement
iv
-------
will be based upon social or political considerations, but the governmental
body making the decision will be cognizant of its economic implications.
In the study for Onondaga County, New York, from which the method-
ology was developed, 35 overflows from the combined sewers of the City
of Syracuse, which serve an area of about eight square miles, were moni-
tored for a period of one year. Onondaga Lake, the principal receiving
water, is approximately four and one-half square miles in surface area;
it was sampled at ten surface locations, each .at two distinct depths, for
the period of influence of each of six storms. The Storm Water Manage-
ment Model (SWMM) was applied to the City's sewer system. A 27-segment,
three-dimensional, dynamic water-quality model of the lake, with capa-
bility of predicting enteric bacteria, dissolved oxygen, nutrients, and
toxic materials, was developed.
From the models, it was determined that the impact of CSO's on
dissolved-oxygen concentrations in Onondaga Lake will not be critical after
tertiary treatment facilities for dry-weather wastewaters are placed in
operation; a maximum DO deficit of 2. 8 milligrams per liter was predicted
for a 10-year, two-hour storm. Combined sewer overflow contributions of
phosphorus will be negligible in comparison to those from other sources.
In an average rainfall year, 38 violations of the fecal coliform stan»
dard will occur in the area of the lake intended for contact recreation. If
abatement of CSO pollution were to follow the "general optimum solution" of
this methodology, there would still be 13 annual violations, ten of which
would occur from June through September. Inasmuch as each violation
persists for about three days, more extensive CSO abatement will be
required if the projected recreational usage of Onondaga Lake is to be
realized.
This report was prepared in fulfillment of Grant No. R805096
by Stearns & Wheler, Civil and Sanitary Engineers, and the Onondaga
County Department of Drainage and Sanitation, under sponsorship of the
U. S. Environmental Protection Agency.
-------
-------
CONTENTS
Foreword .......... . .......................................... iu
Abstract [[[ iv
Figures [[[ x
Tables ......................... . ..... ......................... x*-11
Acknowledgments .............................................. ^
1. Introduction .................. ...................... • • • • 1
2. Summary ............................................... 3
Combined Sewer Overflow System ..................... 3
Onondaga County ................................ 4
Receiving Water Sampling ............................ 5
Onondaga County ................................ 6
Data Bank .......... . ..... . ............ ............. 6
Onondaga County ................................ 7
Mathematical Modeling of Combined Sewer System ...... 7
Onondaga Co'unty ................................ 8
Mathematical Modeling of Receiving Water ............. 8
Onondaga County ................................ 9
Model Projections ................................... 10
Onondaga County .......... . ..................... 10
Cost -Effectiveness Analyses .................. •. ...... H
Onondaga County ................................ 12
3. Recommendations ....................................... 14
4. Combined Sewer Overflow System . . . 0 ..................... 15
Selection of CSO Monitoring Sites ..................... 15
Onondaga County .......... „ ..................... I6
Equipment Installation ......... . ..................... 16
Overflow Sampling Program .... ..................... 22
Flow Monitoring ................................ 22
Onondaga County ...... . ..................... 23
Quality Sampling ................................ 23
Parameters ................................ 23
Location ................................... 25
Duration .... ............................... 25
Frequency .................................. 25
-------
CONTENTS (continued)
5. Receiving Water Sampling 27
General 27
Water Quality Concerns 28
Sampling Locations .. 3^
Duration of Sampling 33
Frequency of Sampling 33
Automation 35
6. Data Bank 3g
Onondaga Lake 33
7. Mathematical Modeling of Combined Sewer System 39
Objective of Modeling 39
Types of Models Available 40
Levels of Analysis 40
EPA SWMM 42
EPA Simplified SWMM (SSWMMi '..'.'.'. 43
Rainfall Characterization 44
Storage-Treatment Balance 45
Overflow Quality Assessment 45
Receiving Water Response 45
Comparison of Detailed Single-Event SWMM and
Simplified SWMM (SSWMM) 45
Complementary Model Usage 45
Design Considerations for Quantity Projections 47
Continuous Simulation 47
Single-Event Design Storm 48
Other Techniques „ 50
Design Considerations for Quality Projections 50
Pollutant Characteristics 52
CSO Modeling for Onondaga County 53
8. Mathematical Modeling of Receiving Water 59
Preliminary Approach 59
Onondaga County 63
Detailed Model 64
Fecal Coliform Model Kinetics 67
Total Phosphorus Model Kinetics 68
Dissolved Oxygen Model Kinetics 69
Removal Rates for Organic Components ....... 69
Biological Deoxygenation 70
Benthal Oxygen Demand 71
Reaeration 7 j
Temperature Corrections 73
Net Productivity 73
Vlll
-------
CONTENTS (continued)
Model Calibration and Verification 75
Hydrology, Influent Loads, Circulation,
Dispersive Transport 15
Reaction Coefficients 76
Environmental Conditions 79
Sensitivity Analysis 79
9. Model Projections , 90
Critical Input Conditions 90
Onondaga County 94
Projected Pollutant Load „ 95
Onondaga County 9 7
Lake Model Projections 9 8
Results - Onondaga Lake Q8
10. Cost-Effectiveness Analysis 1°4
CSO Relationships , 104
Rainfall Patterns 104
Receiving Water Impact 106
CSO Discharges 1°6
Abatement Costs 108
Cost-Effective Solution 11°
Onondaga County 112
References
,117
ix
-------
FIGURES
Number
1
2
Page
11
12
13
14
Typical control section and flow monitoring structure 17
CSO modeling tasks and data requirements 18
3 Typical layout of combined sewer system at the point of
overflow 19
4 Plots typical of CSO data presentation 21
5 Relationship between accuracy of measured runoff coefficient
and number of storms monitored 24
6 General plan and sampling locations for Onondaga Lake 32
7 Model segmentation 34
8 Summary of data review process 37
9 Major components of EPA Storm Water Management Model . 42
10 Interrelationship of tasks in the Simplified Stormwater
Management Model 44
Synthetic hyetograph for Syracuse, New York 49
Typical correlation between total rainfall and total overflow
for an individual drainage area 54
Comparison of SWMM and SSWMM output to measured data
for Onondaga County, New York 55
Total BOD5 load discharged from the combined sewer
system to Onondaga Creek for storms of various
return periods 58
x
-------
FIGURES (continued)
Number
15
16
18
19
20
21
22
23
24
25
26
27
28
29
30
Time frame of concern for various water-quality
parameters
Page
60
Development of predictive water quality model 61
17 Comparison of the effects on concentration of advective
and advective - dispersive transport 65
Forces acting on £ pollutant mass discharged to a
two-dimensional advective - dispersive system
66
Reaeration rate as a function of wind speed 72
Individual current measurements 78
Development of calibrated fecal coliform model 80
Development of calibrated dissolved-oxygen model 81
Comparison of measured fecal coliform concentrations
with model output 82
Comparison of measured CBOD concentrations with
model output 83
Comparison of measured NBOD concentrations with
model output 84
Comparison of measured DO concentrations with
model output 85
Comparison of measured Total Phosphorus concentration
with model output 86
A typical loading curve relating pollutant load to water
quality response 91
General procedure for developing loading curve and
determining allowable load 92
Fecal coliform loading curves for Onondaga Lake 102
xi
-------
FIGURES (continued)
Number
31 Summary of monthly rainfall data 105
32 Intensity-duration-frequency curves, Syracuse, New York... 107
33 Procedure for establishing the most cost effective
treatment level to meet a given water quality goal 109
34 Procedure for determining the cost effective water
quality goal Ill
35 Number of water quality violations occurring annually in
Onondaga Lake as a result of treating various intensity
storms „ 113
36 Costs associated with treatment of combined sewer over-
flows to reduce the frequency of annual water-quality
violations 114
37 Average number of violations of fecal coliform standards
during the recreation season resulting from treating
various intensity storms, Onondaga Lake, New York .... 116
xix
-------
TABLES
Number Page
1 Water Quality Parameters 30
2 Categories of Stormwater Models 40
3 Stormwater Analysis Level and Model Complexity 41
4 Comparison of Single-Event SWMM and SSWMM 46
5 Factors Affecting CSO Quality 51
6 Geometric Mean Concentrations of Pollutants Entering
Onondaga Creek and Ley Creek from Combined
Sewer Overflows 57
7 Temperature Correction Factors 73
8 Dispersion Coefficients 77
9 Reaction Coefficients and Sinking Rates Verified Through
Model Calibration 78
10 Dissolved Oxygen Reaction Coefficients 78
11 Sensitivity Analysis, Reaction Rates 88
12 Sensitivity Analysis of Productivity on In-Lake
Dissolved Oxygen 89
13 Critical Environmental Conditions for Model Projections ... 93
14 Dry-Weather Pollutant Loads "
15 Wet-Weather Pollutant Loads 100
X13.1
-------
TABLES (continued)
Number Page
16 Projected Impact of Wet-Weather Loads on
Onondaga Lake 101
17 Average Annual Phosphorus Concentrations in
Onondaga Lake 103
xiv
-------
ACKNOWLEDGMENTS
John M. Karanik, Deputy Commissioner, and Randy Ott, Project
Engineer, for the Onondaga County Department of Drainage and Sanitation
served effectively to coordinate the related projects performed by O'Brien
& Gere and Stearns & Wheler. Facilities planning and previous EPA dem-
onstration projects conducted through Onondaga County, New York, served
as the primary basis for this methodology.
O'Brien & Gere Engineers, Inc., is presently conducting facilities
planning for the combined sewer overflow system. The cooperation of a
number of engineers, and in particular, Mr. Dwight MacArthur, Managing
Engineer, and Mr. Thomas Jordan, Project Engineer, of O'Brien & Gere,
is gratefully acknowledged.
Stearns & Wheler conducted the receiving water investigations as
part of this facilities planning. Dr. Raymond P. Canale, President, and
Mr. Paul Freedman, Vice President, of Limno-Tech, Inc., developed
much of the receiving water modeling.
Project Officer Richard Field and Messrs. Rick Traver and Douglas
Ammon of EPA provided guidance and review throughout the project.
Dr. William O. Lynch of Stearns & Wheler, who was the partner in
charge, closely reviewed this report and offered many valuable sugges-
tions.
-------
-------
SECTION 1
INTRODUCTION
The impetus provided by the Federal laws concerning water pollution
(1, 2) has resulted in great strides in wastewater treatment. The prevailing
attitude leading to the 1972 Amendments to the Water Pollution Control Act
(1) was that there had been a general lack of progress in abating the
country's pollution. Consequently, secondary treatment plants throughout
the country were approved and initiated on the basis of the obvious benefit to
receiving water quality. The latter benefit was provided through water
quality impact studies as part of Section 201 of PL, 92-500.
However, since 1972, the required "201" facilities planning has pro-
gressively expanded to include analyses and evaluations of separate sewer
systems, and most recently, combined sewer systems. The combined
sewer overflow (CSO) investigations have had to address an old but rela-
tively undefined component of municipal sewage; namely, stormwater. The
highly variable nature of this element and its double-edged impact on
receiving water of providing dilution as well as adding pollutants has
resulted in a more deliberate approach to cost-benefit relationships than
had previously been the case for "201" facilities planning.
The common "dry weather" receiving water standard based on con-
secutive seven-day low flow conditions (MA7CD10) occurring once every
ten years for conventional treatment has gradually given way to the con-
sideration of allowing deviation from "dry weather" for certain stormwater
events. Recently, attention has been devoted to the concept of the "design
storm" as being that storm for which abatement facilities are planned (3).
However, the elusiveness of this approach has led to the consideration of
cumulative storm effects over annual cycles. A defined approach is not yet
available, but actual ongoing experiences are now providing the basis for
practical approaches to the problem.
The methodology outlined herein is based largely on the successful
experience in identifying the discharges from a combined sewer system
and their impact on the water quality of an urban lake on Onondaga County
and just north of Syracuse, New York. This methodology is not limited to
-------
Onondaga County, but draws from other experiences as well and to a large
extent is applicable to receiving waters in general.
The methodology is broken down separately for both the combined
sewer system and the receiving water aspects into: (a) equipment installa-
tion; (b) field sampling; (c) data analyses and mathematical modeling; and
(d) impact projections. The extent to which sewer system investigations
are related to receiving water investigations and the coordination required
to arrive at an overall abatement strategy will be described.
-------
SECTION 2
SUMMARY
The experiences of Onondaga County, New York, in defining the pol-
lutants discharged from the City of Syracuse's combined sewer system and
their impact on the water quality of Onondaga Lake served as the basis for
developing a general methodology applicable to other communities. Com-
bined sewer overflow (CSO) investigations were conducted by one firm of
consulting engineers, while receiving water investigations were conducted
by a second firm; both efforts were funded under a PL 92-500 Facilities
"201" Grant to Onondaga County for the Syracuse Metropolitan Sewage
Treatment Plant.
In the summary that follows, each major topic is initially discussed in
broad, general terms, then illustrated by specific reference to the proce-
dure followed in Onondaga County.
COMBINED SEWER OVERFLOW SYSTEM
The first step in a CSO study is to identify the pollutants through field
sampling and laboratory analysis of overflows at selected sampling sites.
The number and location of sampling sites should be based on the extent to
which they represent the system as a whole. Resulting data can then be
projected to portray-the entire system. Other practical considerations
should be taken into account, such as accessibility of manholes and the
hydraulic characteristics of the various structures in relation to accuracy
of flow measurement.
Local rainfall data and the physical characteristics of the sewer
system are analyzed and used to develop a series of mathematical relation-
ships that express the operation of the system under various rainfall con-
ditions. The number and location of rain gages is important to being able
to relate the rainfall in a particular area to its associated runoff. As a
general rule, a density of one to five rain gages per square mile is desir-
able. Care should be exercised in locating these gages so as to minimize
any interference from nearby structures or trees.
The measurement of quantity at an overflow may involve the measure-
ment of the flow reaching the interceptor as well as the flow discharged to
the receiving water, depending upon the capacity of the interceptor. The
number of individual points that have to be measured should be minimized
-------
to reduce equipment maintenance and data coordination. It also is desirable
to install the flow metering device at the same location as the automatic
sampler which measures quality so that operation of the sampler will be
triggered by an actual overflow occurrence. Such time coordination
becomes critical for determining pollutant loads, which are calculated as
the product of flow quantity and pollutant concentration. Thus, the simplest
and most direct measurement is at the overflow to a receiving water. Under
surcharge conditions, it is particularly necessary to estimate the intercep-
tor flow as well as the receiving water overflow in order to determine the
true quantity of runoff originating from a tributary area.
The quality parameters selected for a monitoring program are depen-
dent upon the intended best usage of the receiving water. The frequency and
duration of sampling should be based on an adequate description of the peak
period of pollutant discharge, often called the first-flush period, and the
subsequent subsidence of this period to relatively insignificant quantities. It
has been found that in the period of five minutes following the onset of run-
off, significant concentrations of pollutants can occur. Thereafter,
7-minute intervals have proven to be adequate to define an overflow dis-
charge. In general, the minimum duration of sampling should be ample to
capture the runoff from a six-hour storm.
Qnondaga County
The combined sewer system of the City of Syracuse covers approxi-
mately eight square miles and has 87 active overflow sites, of which 35
were monitored for flow quantity and pollutant quality. The monitored sites
encompassed 85 percent of the total defined runoff area and hence were
representative of the overall system. To monitor rainfall, six rain gages
were located throughout the City, each representing an area of approx-
mately two square miles.
The main interceptor system has little or no capacity for wet-weather
flow and thus the quantity of flow discharged to the stream, was assumed to
be the total quaritity entering the diversion structure. This assumption may
not be applicable in many cases and must be verified by field data.
Overflows were monitored through the use of ultra-sonic flow meters
in combination with sampling units capable of taking 24 separate samples.
Samplers were initiated by a signal from the flow meter and were operated
at 7-1/2-minute intervals thereafter.
-------
RECEIVING WATER SAMPLING
Adequate information on receiving water quality is generally lacking.
Such information is probably more critical for CSO projects than for conven-
tional waste-water treatment because water-quality impact is more difficult to
predict and the risk of providing unneeded abatement facilities is greater.
Although combined sewer overflows can introduce a multitude of pollu-
tants into a receiving water, the primary concerns can be categorized into
(a) pathogens, because of public health, and (b) dissolved oxygen, nutrients
and toxics, all because of their direct effect on the aquatic system and indi-
rect effect on public health,, The fecal coliform and fecal streptococcus
groups can be used as pathogenic indicators. Dissolved oxygen, phosphorus,
and, in some cases, nitrogen, are critical nutrients or indices to be evalua-
ted for their impact on aquatic life. Initially, evaluation of toxics should be
limited to the heavy metals, unless specific information indicates the need
to investigate toxicity from organic chemicals, e.g., pesticides, PCB's.
Organic analyses represent a relatively high level of complexity. Toxicity
observed in the BOD test can be used as an indicator for proceeding with
any additional analyses. One exception to this dictum would be the case
where bottom deposits that may be subject to toxic organic contaminants are
scoured frequently and play a significant role in the year-round quality of
overlying waters.
Decisions as to what storms should trigger receiving water sampling
should be based on the anticipated storm intensity. Close coordination
should be maintained with those sampling the sewer system so as to maxi-
mize the number of overflows being monitored during receiving water sam-
pling. Sampling locations and the frequency and duration of sampling are
dependent upon the changes a particular parameter may undergo during a
storm period. Thus, prior knowledge from other studies or background
data on the receiving water is helpful. In the absence of such information,
the general guidelines outlined herein can be used.
The location of sampling points should reflect the intended usage
within the body of the receiving water. In general, areas where contact
recreation is anticipated should be sampled more intensely than areas where
fishing or boating are anticipated. The circulation pattern of the receiving
water should be considered in determining surface locations and depths of
sampling. Surface locations should include tributaries before entering the
lake as well as in-lake locations reflecting the outer boundary of any such
tributary's mixing zone. In this way, the impact of a particular tributary
can be estimated. If a tributary receives combined sewer overflows, it
should be sampled so as to bracket these discharges, either individually
or collectively.
-------
Qnondaga County
Water-quality data from dry-weather periods were available for
Onondaga Lake and were used to estimate the pollutant levels expected
during storms. In actuality, these estimated levels were often exceeded.
Anticipated storm intensity determined whether or not receiving water
sampling was to be conducted for any particular storm. On each positive
occasion, the crews in charge of sampling the sewer system were alerted.
The frequency and duration of sampling during a storm were estimated at
the onset of the project, but were later adjusted on the basis of the first two
storms. The water-quality parameters generally measured were bacteria,
BOD, DO, solids, nutrients, and heavy metals.
All major tributaries were sampled and measured at their points of
entrance into Onondaga Lake. In addition, two of the tributaries were sam-
pled upstream so as to bracket the- collective input of the combined sewer
system. Preliminary sampling was performed to determine the validity of
sampling techniques for each of the tributary locations. To be representa-
tive, tributaries were sampled in cross section and in depth, depending
upon their dimensions and the degree of mixing in each. The lake was sam-
pled at ten surface stations; each station was sampled at two depths to
reflect the epilimnetic (upper) and hypolimnetic (lower) waters. One sur-
face station was located at each of the lake's two deepest points, the
remainder being located to reflect tributary inputs or shoreline areas
expected to be used for contact recreation.
DATA BANK
Whenever CSO studies yield large amounts of data, it is important
that they be handled in a form that is easily retrieved and analyzed. For
any kind of system, every sample must be adequately identified, e. g.,
storm number, location, date, time and parameter. A data file may take
the form of summary sheets or can be computerized, depending upon the
size of title project. A computerized file will permit the rapid retrieval of
data in various forms, which is particularly helpful in CSO projects where
dilution volumes of samples for certain parameters may have to be adjusted
following the first storm sampling. These and other adjustments, such as
monitoring locations and possible elimination of certain stations, require
that initial data be evaluated as soon as possible. Provision should be made
to plot the concentration levels of the various parameters versus time for a
given combined sewer overflow and, when combined with flow, calculate a
plot of pounds of material versus time. Other sub-routines could be
developed to determine the pounds of material discharged during an entire
storm.
-------
Onondaga County
A computerized file system was established for the combined sewer
system and a separate file was set up for the receiving water. Since the
combined sewer system investigation and the receiving water investigations
were conducted by independent organizations, separate data systems were
developed; this presented no problem since, with the exception of storm
sampling, the data analyses were relatively independent of each other. The
output requirements and mathematical modeling for the sewer system were
markedly different from those of the receiving water.
MATHEMATICAL MODELING OF COMBINED SEWER SYSTEM
The complexity of-combined sewer systems and the variety of use of
tributary lands generally requires a series of mathematical expressions or
a mathematical model to project the pollutant discharges associated with
various storm conditions. There are no guidelines for the storm conditions
for which abatement facilities should be designed. Only on the basis of
actual field measurements and the subsequent development of cost-benefit
relationships can reasonable design decisions be made. The capturing of
storm conditions for which facilities should be designed would be pure hap-
penchance; therefore, mathematical modeling of the sewer system as well
as the receiving water becomes an integral part of a CSO study.
The major objective of a field sampling program is to measure actual
discharges and their impact on water quality sufficiently to develop the
cause-and-effect relationships between the sewer system and the receiving
water. A mathematical model can then be developed to describe these rela-
tionships and used to project combined sewer overflows for various storm
conditions. There are a number of mathematical models available to per-
form this service, depending on the complexity of the combined sewer sys-
tem. In some cases, preliminary simplified modeling should be performed
to establish guidelines for the application of more extensive modeling.
Simplified modeling can be particularly useful in estimating the long-term
rainfall picture -or performing continuous, simulation,, whereas more
detailed modeling is required for single-storm evaluations.
A major limitation of the models that are available is the level of
accuracy with which they predict pollutant concentrations. The use of sta-
tistical analyses of quality data to develop coefficients for use in conjunction
with quantity data has been an approach. Thus, mathematical models can
be used to estimate CSO quantities and statistical analyses qan be used to
associate quality with the models' outputs.
-------
Onondaga County
The Simplified Stormwater Management Model (SSWMM) was used to
estimate the total volume of overflow from the Syracuse system for various
rainfall conditions so that preliminary storage and treatment schemes could
be developed and screened. The SSWMM was calibrated against 35 actual
overflow points; however, surcharge conditions, which existed in all but
very minor storms, limited application of the simplified model. Accord-
ingly, the CSO system was modeled using the detailed Storm Water Manage-
ment Model (SWMM) with the EXTRAN option, which provides the means for
calculating surcharge conditions. Data from the individual overflows were
also used to calibrate the detailed SWMM.
For the purpose of defining rainfall conditions, a statistically random
two-year data set of the rainfall record for the Syracuse area was developed
and found to be equivalent to the 25 years of record available. The two-year
record was then used to develop rainfall hyetographs representing various
recurrence intervals. Pollutant loadings were determined by combining
statistical analysis of the field-measured quality parameters with associated
flows. Quantity and quality responses were simulated for a number of storm
conditions. The pollutant loadings were then entered in the receiving water
model to assess water-quality impact.
MATHEMATICAL MODELING OP RECEIVING WATER
Because of the great range and variability of weather patterns, com-
bined sewer system characteristics, and receiving water usage, it is
unlikely that any pre-determined criteria of receiving water quality, quan-
tity and recurrence interval can be as universally adopted to set the treat-
ment of CSO pollutants as they have been adopted to set the treatment of
dry-weather pollutants in conventional plants. Consequently, it is neces-
sary and desirable to develop the relationships required to determine the
water quality impact of CSO's under a variety of conditions. Mathematical
modeling of the receiving water greatly facilitates this need. The type of
model employed is dependent on intended best usage of the receiving water,
time frame, confidence level, and project scope. Simplified modeling
should be attempted initially to determine if more detailed modeling is
needed. If the results of the simplified approach border on contravention
or violation of some established standard, then more detailed modeling is
warranted.
_!"<•
A receiving water model should include certain fundamental functions
as well as those required to meet specific needs. The model should con-
sider circulation as fundamental to the distribution of pollutants in the
receiving water. The circulation model should include advective and dis-
persive transport and, in addition, mechanisms for settling and scour. The
8
-------
modeling of pollutants can be done in a modular fashion to minimize com-
puter costs and, in general, should include (1) bacteria, (2) DO balance,
(3) nutrients, and (4) distribution of conservative materials, e. g. , heavy
metals.
The success of a mathematical model is the degree to which it repre-
sents real conditions; hence, the receiving water model should be calibrated
against several storms and verified accordingly. If a number of constants
were employed in the model because of practical limitations in the office or
conditions experienced in the field, these values can be tested through sensi-
tivity analyses. Through such analyses, the concern for assumed values
can often be minimized.
Onondaga County
Simplified modeling of a one-year storm showed contravention of bac-
terial standards throughout the recreational zones of Onondaga Lake. In
order to verify these results, more detailed modeling was undertaken.
Simplified modeling was attempted to determine the dissolved-oxygen rela-
tionships in the lake; owing to the complexity and intensity of algal growths
and their significant influence on the DO level, a more detailed model was
required. A simplified approach, on an annual basis, was taken to deter-
mine the significance of phosphorus loadings from storm runoff. This
approximation was considered adequate to assess what turned out to be the
relatively insignificant contribution of stormwater to the phosphorus content
of the lake.
Each CSO was identified as entering one of two tributaries to the lake.
In the detailed receiving water model, the CSO's entering each tributary
were treated as single, collective inputs. The lake was divided into 27
distinct segments for the purpose of defining circulation patterns. Twenty-
one (21) of the segments were located in the epilimnetic (upper) waters and
six of the segments were located in the hypolimnetic (bottom) waters. The
model utilized steady-state circulation patterns and hydraulic loads in com-
bination with time-variable pollutant loads to simulate the lake's water-
quality response to stormwater discharges. Bacterial modeling was under-
taken specifically for fecal coliforms and accounted for loss of organisms
due to both settling and die-off, including terms for protozoan predation
and sunlight disinfection as well as respiration and natural death. The
dissolved-oxygen function included expressions for algal photosynthesis and
respiration. Total phosphorus was considered to be a relatively conserva-
tive substance in which net changes in concentration were limited to settling
and sediment release.
The Onondaga Lake model was calibrated with data from four
distinct storms. The resulting model was then used to predict two of the
-------
same storms and, in both cases, resulted in good fits. In addition,
sensitivity analyses were performed for the various rate coefficients. In
general, 25 to 50 percent changes, plus and minus, in the coefficients ori-
ginally employed were used to test the model's sensitivity to such values in
terms of water-quality impact.
MODEL PROJECTIONS
Prior to projecting abatement requirements, the critical environ-
mental conditions must be determined for use in the water-quality model.
These conditions include wind speed, background water quality, light con-
ditions, and temperature. Input conditions must be carefully selected so as
to be critical for the parameter in question. For example, the effect of
fecal coliforms on water quality is most severe under conditions of minimal
die-off (low temperatures) and rapid transport (high wind speeds) to a sen-
sitive area of a lake, such as a bathing beach; the dissolved oxygen para-
meter, on the other hand, is most adversely affected when biochemical
rates are maximum (high temperatures) and reaeration is minimum (low
wind speeds). Since historical records generally are not available for lake
circulation patterns, the flows used in the predictive water-quality model
must be determined through lake current measurements in conjunction with
recordings of wind speed and direction. Bacterial die-off rates should be
determined through laboratory-con trolled tests on the receiving water. In
predicting the dissolved-oxygen content of productive lakes, light conditions
and the algal standing crop can represent the dominant effect on the DO con-
centration.
The projection of pollutant loads requires considerable judgment to
avoid a combination of events that may have a very infrequent reoccurrence
interval. If, for example, an MA7CD10 dry-weather stream flow is com-
bined with -storm runoff having a recurrence interval of one year, the pro-
bability of both conditions occurring at the same time can exceed once in
30 years. The selection of dry-weather flow conditions for a large lake is
less critical than for a stream because circulation is more influenced by
lake elevations and lunar effects than incoming tributaries. In addition to
flow considerations, pollutant loadings in a model must include dry-weather
components representing both point and background sources. Point source
information often can be obtained through the NPDES or State permit sys-
tem, but background source information may require field measurements
during dry periods.
Qnondaga County
Current measurements taken on Onondaga Lake during the storm
sampling periods served as the basis for determining the in-lake circula-
tion patterns assumed in the mathematical model. Background water-quality
10
-------
data was available from prior dry-weather sampling. Light conditions were
determined from regional data in conjunction with percent-cloud-cover
measurements at the local weather station.
Dry-weather data from prior years served as the basis for deter-
mining point-source pollutants. The data used were limited to the months
of June through September and represented average values of two to six
years of data, depending upon whether or not the data reflected contempo-
rary operations in the tributary. In some cases, modification of the sewer
system prevented the use of all six years of data.
Projections of the water-quality impact of storms with recurrence
intervals of one (1), two (2) and ten (10) years were made. It was found that
the one-year, two-hour storm would result in bacterial numbers in excess
of the standard of 200 cells/100ml in all "upper" segments of the lake, and a
maximum concentration of greater than 11, 000 fecal coliform cells/100ml in
the northern basin where contact recreation is anticipated. It was predicted
that a storm of this magnitude would cause contravention of coliform stan-
dards for a period in excess of three days. On the other hand, the dissolved
oxygen demand from even a 10-year, two-hour storm had only minimal
effect on the dissolved-oxygen balance of the lake; the maximum DO deficit
for that storm, under the most critical conditions, was determined to be
2. 8 mg/1. It was determined that the algal standing crop and abrupt algal
die-off, following treatment of the domestic and industrial wastes, will
become the principal factor affecting the lake's dissolved oxygen. Annual
storm-related contributions of total phosphorus are negligible when com-
pared to the projected dry-weather loads.
COST- EFFECTIVENESS .ANA LYSES
A common economic approach to stormwater treatment has been to
develop relationships between the impact of stormwater on receiving water
quality and the cost of treatment for various storm conditions. This cost-
benefit information has then been used by local governments and regulatory
agencies to make reasonable economic decisions.
Through the use of a receiving water model, water quality response
can be expressed as a function of total pollutant load, including both dry-
weather loads and stormwater inputs, especially CSO's. An independent
combined sewer system model can be used to relate the pollutant load dis-
charged to various storm conditions which can be expressed in terms of
recurrence interval. If the receiving water can assimilate the dry-weather
component of the pollutant load after suitable treatment, the two relationships
previously developed can be combined to express the required removal of
pollutants from the wet-weather components, for any water-quality goal,
as a function of storm recurrence interval.
11
-------
A series of cost estimates must then be prepared for the removal of
various quantities of pollutants from the combined sewer system, using
engineering judgment as to the best method of accomplishing each level of
abatement. Abatement can be classified as either source control or treat™
ment. Source controls or "best management practices" (BMP) include con-
trol before the stormwater enters the combined sewer system or control
within the sewer system itself. Treatment of combined sewer overflows
can take place either at the individual diversion chambers within the systen
or at one or more locations along each interceptor.
The abatement costs can then be combined with the previously devel-
oped relationships between required pollutant removal and storm condition
to yield a graphical presentation of the cost of meeting a particular water-
quality goal versus storm recurrence interval or & measureable benefit
such as number of water-quality violations. Rainfall records can be
employed to determine the frequency of violation over the period of record
available. The plotted curve of this relationship may show a pronounced
knee ; this knee may be interpreted as representing the highest recurrence
interval (lowest storm frequency) or least number of water-quality viola-
tions at which a water-quality goal can be met at £ relatively low marginal
cost. It is the optimal or cost-effective solution for that particular water-
quality standard. At all points beyond the knee (the point of "diminishing
returns"), the additional benefits derived from incremental increases in
expenditure decrease rapidly and can be justified only for reasons other than
economic ones.
Similar curves can be developed for different water-quality standards
and the entire family of such curves analyzed to weigh the merits of more
stringent or less stringent standards. In the absence of national guidelines
and in consideration of the many and varied factors involved in stormwater
abatement, economic relationships such as those just described become
essential to the formulation of reasonable plans for the abatement of pollu-
tion from stormwaters, especially combined sewer overflows.
Ctetondaga County
Analysis of the Qnondaga Lake data showed that only the fecal coli-
form parameter was significantly and adversely affected by CSO discharges.
Of the 65 annual overflow-producing events, 38 result in discharges of sig-
nificant magnitude to cause violation of the bacterial standards in the con-
tact recreation zone.
For a maximum bacterial concentration (or standard) of 200 cells/
100 ml, various abatement alternatives were plotted against the number of
violations of this standard that would occur in the lake in a typical rainfall
year. The "knee" of the curve occurred at a point where there would still
12
-------
be 13 annual violations, ten of which would occur during the recreation
season from June through September. Since ten violations would essentially
close the lake to contact recreation, higher levels of pollution abatement
are now being considered, the goal being a maximum of one violation.
13
-------
SECTION 3
RECOMMENDATIONS
On the basis of the experiences of Onondaga County, the following
recommendations are made for communities facing similar problems of
evaluating and abating pollution from combined sewer overflows:
1. The water-quality impact of the combined sewer over-
flows should be determined by measurement and analysis
of representative overflows and the receiving water under
real storm conditions.
2. Mathematical relationships should be developed from these
data for the purpose of projecting the impact of overflows
on water quality under a wide range of storm conditions.
3. Costs should be developed for abatement plans represen-
tative of a range of water-quality goals associated with
various storm conditions.
4. Cost-benefit relationships should serve as the basis for
selecting the water-quality goals, associated stormwater
conditions, and abatement methods that best fit both the
local needs and the national interest.
14
-------
SECTION 4
COMBINED SEWER OVERFLOW SYSTEM
Prior to evaluating the impact of the overflows on the water quality of
a receiving water, the pollutional load contributed from this source must be
identified. Since for most communities it is financially and physically
impractical to monitor all overflow sites, it is necessary to select for moni-
toring some representative locations with drainage basin characteristics
similar to the unmonitored areas. The observed characteristics may then
be applied to the unmonitored areas for projecting overflow quantities.
SELECTION OF CSO MONITORING SITES
Major factors to be considered in the selection of monitoring stations
are size of drainage area, uniformity of land uses within that drainage area,
slope and percent imperviousness of the land, and overall applicability of
the particular drainage basin characteristics to other areas within the
system.
The preliminary selection of these sites can be made with the aid of
detailed sewer maps, land use maps, aerial photographs and topographic
maps, in conjunction with physical inspection. This type of information is
generally available for most cities throughout the country. Once an adequate
number of sites have been selected by "desk-top" studies, it becomes
necessary to use field investigations to verify the appropriateness and ease
of monitoring of these sites.
The ability to utilize a selected site is dependent on accessibility, the
hydraulic condition at the point of flow measurement, and the slope within
the contiguous transport system. Since samples will be collected during
inclement weather, the accessibility of the site is extremely important,
primarily from a safety standpoint. The hydraulic condition at the point of
measurement is critical to accurate flow measurement and will determine
the method of measurement employed. The accurate measurement of flow
requires the use of a control section for which a relationship between head
and flow exists (3). This often requires installation of a control section,
such as a weir. The need for such construction may severely reduce the
number of suitable monitoring sites.
15
-------
Weirs, when used, should be constructed as high above the invert of
the incoming sewer as possible (Figure 1) in order to minimize the approach
velocity at the point of measurement. This minimizes the velocity head,
which enables the use of simplified weir formulas which require only a
measure of depth of flow over the weir (H) to calculate flow. Elevating the
weir crest also decreases the chances of weir submergence (h,>hw); how-
ever, it increases the chances of basement flooding upstream of the weir.
Prior to altering the hydraulics, the slope of the upstream sewer system
should be checked to insure that it is sufficient to permit this construction
without risking severe backup.
An alternate to the construction of a control section is to utilize the
sewer pipe itself as a control section and monitor flow either in the pipe up-
stream of the diversion (total flow) or in the outfall pipe itself (overflow).
No matter where monitoring is accomplished, the selected structure or
manhole must also be of sufficient size and depth to permit installation of
monitoring equipment if costs are to be minimized.
Onondaga County
In the specific case of Onondaga County, there exists a total of 8 7
active overflow sites. Of these 87 sites, O'Brien & Gere Engineers, Inc.
initially monitored 25 sites for both quantity and quality. These 25 sites
represented 79 percent of the total combined sewered area of the system.
An additional 10 sites were ultimately added to further define the system
characteristics. Since the 35 sites monitored comprised such alarge portion
(85 percent) of the total defined runoff area and represented the entire range
of land uses within the combined system, these sites were considered rep-
resentative of the overall system with characteristics applicable to the
unmonitored sites.
EQUIPMENT INSTALLATION
In order to determine the cause-and-effect relationships involving
pollutants being discharged from the given combined sewer overflow area,
a measure of quantity and quality of the overflow event and a measure of the
rainfall producing that overflow is required first (Figure 2). Quantity
measurements can be made either on total runoff or total overflow from a
given drainage area or both (Figure 3); however, to measure both becomes
expensive and generally unnecessary. While both of these measures are
useful, the collection of sufficient.data to perform a flow balance about the
diversion structure is generally sufficient. The collection of data is dis-
cussed in greater detail in the flow monitoring section.
In addition to quantity measurements, collection of water quality
samples with time is required in order to adequately define the pollutant
16
-------
E.6.L.
/
Vo
Ill
tc
MANHOLE OR OTHER
MONITORING STRUCTURE
VELOCITY HEAD (Vo/2g)
EFFLUENT PIPE
^
NFLUENT PIPE
hw - WEIR HEIGHT
H
Vo
- DEPTH OF FLOW OVER WEIR
- VELOCITY OF APPROACH
hd - DEPTH OF FLOW DOWNSTREAM
OF WEIR
E6.L- ENERGY GRADE LINE
Figure I. Typical control section and flow monitoring structure.
17
-------
CO
CO
-------
RECEIVING WATER-^S~*-
-BACKFLOW GATE
WEIR
COMBINED
SEWER —
-DIVERSION
MANHOLE
°TOTAL » INTERCEPTOR 4- °ovE8FL0w
Figure 3. Typical layout of combined sewer system at the point of
overflow.
19
-------
load (pollutograph) for a given drainage area and given storm. A polluto-
graph is defined as a plot of the mass rate of pollutant discharge (i. e.,
pounds per day of suspended solids)versus time (Figure 4).
In order to assess the relationship between rainfall and runoff, the
quantity of rainfall falling in a given area must also be determined for each
storm. Due to variations in rainfall patterns throughout a community, rain
gage placement and density (number per square miles) are critical to a
successful monitoring program. In order to limit the standard deviation of
the measured coefficient of runoff (MC" factor) to between 10 and 25 percent
of mean, it has been estimated that rain gage density on the order of 1 to 5
gages per square mile is desirable (4). Rooftops or open spaces proximate
to treatment plants, fire and police stations, and other patrolled facilities
are desirable locations for gages since vandalism can be minimized and
access is possible 24 hours per day. Care must be taken to install rain
gages in areas where they are not sheltered by surrounding trees or struc-
tures. For ease of operation and to minimize maintenance, a location with
on-site power is desirable but power should be available at all of the above-
mentioned facilities.
It is not the function of this methodology to discuss the flow metering
and sampling equipment currently available. This subject is addressed in
greater detail in two EPA publications which provide an assessment of
CSO sampling and measuring devices (3, 5). While many manufacturers
produce both flow meters and samplers, few provide a system which inte-
grates the two devices specifically for overflow monitoring. Since portable
flow meters are not equipped with time clocks to indicate when samples
were collected, an extremely helpful feature of the integrated systems
available is that they record this information on the flow meter chart.
The actual type of equipment applicable to a particular study will
depend on the individual monitoring sites selected. The site specific factors
requiring evaluation prior to equipment procurement are:
1. Location of flow measurement (i. e., total flow in channel,
overflow over weir, or pump station wet well). For these
different locations, the range of liquid level and the ease
of mounting a level sensor will vary considerably.
2. Space available for equipment installation, which will determine
size of equipment to be used.
3. Accessibility to continuous power source.
4. Anticipated frequency of checking operation of monitoring sites,
which will determine type of recording chart desirable.
20
-------
TIME
aHYETOGRAPH
£
«e^
u
(9
5
TIME
b. HYDRQGRAPH
TIME
d.OVERFLOW POLLUTOORAPH
TIME
C.GQNCETRAT80N
TIME
e. LOA006RAPH
Figure 4. Plots typical of CSO data presentation.
21
-------
5.
Atmosphere in which equipment will be installed, that will
determine type of equipment housing required.
Once these factors are determined and evaluated, the above-mentioned
references may be consulted for types of equipment available.
An important aspect of the flow monitoring program is the installation
and calibration of the control sections. Whether an open channel or weir is
used, the flow versus gage height must be calibrated over a wide range to
span the fluctuations in flow encountered during storm events. All control
sections should be located in areas where turbulence can be minimized.
The accuracy of measurements in turbulent zones is extremely questionable
and should be avoided whenever possible.
OVERFLOW SAMPLING PROGRAM
Flow Monitoring
The overall combined sewer study involves the development of
relationships between rainfall and sewer system response and between
pollutional load and receiving water response. Each of these relationships
requires that a different portion of the total flow stream be known. The
sewer system model is best calibrated versus total flow from the drainage
area, whereas the receiving water model and National Pollutant Discharge
Elimination System (NPDES) Permits require that the overflow be known.
Depending on the particular drainage basin, it may be more advantageous to
measure one or the other of these flows; therefore, the most logical
approach involves obtaining enough information through monitoring to per-
form a flow balance about the overflow structure. To have an accurate
measure of each of these components requires making a measure of either
total flow or overflow, while at the same time obtaining a record of sur-
charge condition in the interceptor. This enables an estimate to be made of
the flow to the interceptor using hydraulic calculations (see Figure 3). If
total flow is to be measured, flow monitoring equipment should be set up in
a manhole upstream of the diversion structure, whereas if QOverflow is to
be monitored, equipment can be installed in the diversion structure itself.
This latter setup is generally more convenient as far as access since these
manholes are generally located near the receiving water and away from
streets and busy areas. Monitoring flow in the diversion manhole also
enables the flow metering device and sampler to be installed in the same
structure, allowing the sampler to be triggered by the flow meter. This
provides for a greater confidence in the time coordination of the two instru-
ments. Since a pollutograph is the product of flow and concentration at a
given point in time, this is extremely critical to determination of both total
and instantaneous pollutant loads.
22
-------
Onondaga County—
In the case of the Onondaga County study, Qoverflow was measured
and equated to the total wet weather component of flow (6). Measuring
Qoverflow was considered sufficient since the interceptor had nearly zero
ability to handle wet-weather flows. Since most combined sewer systems
have wet-weather capacity f.n the interceptor, equating QOverflow to ^total
is generally not accurate. It should be noted that to obtain a true picture of
total flow, there is a need for an indicator of interceptor surcharging. While
it is possible to establish maximum surcharge height using chalk lines on
manhole walls or test tubes mounted at various elevations, the most desir-
able measure is made with a continuous recorder. Surcharge information
enables the instantaneous flow to be computed for the entire storm event and
provides all information required for performing a mass balance at the
diversion manhole.
The total duration of the overflow sampling program, including both
quantity and quality measurements, is dependent on the type of weather
encountered. The success of the program is dependent on observing suffi-
cient storms to characterize the system response with an acceptable level of
confidence. Based on statistical analysis (7), it appears that between seven
and ten storms should be monitored to achieve such confidence; seven to ten
storms represent points beyond the "knee" of the curve shown in Figure 5.
For example, if ten storms are monitored, the observer can be 90 percent
confident that the average runoff coefficient measure for the ten storms will
fall within 40 percent plus or minus of its "true" value; if a 68 percent con-
fidence level is acceptable, the estimated percentage error in the observed
mean runoff coefficient is reduced to 25 percent for the same number of
storms. As may be seen from Figure 5, the percent error can be mini-
mized by increasing the number of storms monitored.
Since the monitoring program relies on the occurrence of at least
seven overflow-producing events, the program is at the mercy of nature. In
the northeast where a majority of .the combined sewers in the United States
exist, it is impractical to expect to obtain meaningful data between the
months of December and April. During these months, snowfall and snow-
melt mask the precipitation patterns and the total runoff precipitation
relationship. Also, during periods of snowfall and extreme cold, operation
and maintenance of equipment become extremely difficult and sometimes
dangerous.
Quality Sampling
Parameters--
The parameters selected to assess the quality of combined sewer
overflows are dependent upon the intended best usage of the receiving water.
This best usage has been largely defined on a state level. However,
23
-------
100
-o 9O
•ft
£ 80
UJ
o
t 70
UJ
8
t 6
o
50
iti
40
o:
§ 30
u.
o 20
UJ
F 10
0
NOTE: CURVES ARE BASED ON A
SELECTED COEFFICIENT OF
VARIATION (STANDARD
DEVIATION/ MEAN) OF 0.75.
90% CONFIDENCE LEVEL
68% CONFIDENCE LEVEL
10 20 30
NUMBER OF STORMS MONITORED
40
50
Figure 5. Relationship between accuracy of measured runoff
coefficient and number of storms monitored.
24
-------
practically and realistically, receiving waters should reflect the local
interests, particularly in the cases where the receiving water is primarily
used by local residents. Section 208 of PL 92-500 provides the means
whereby local interests and concerns are channeled through state-designated
planning entities for the purposes of establishing classes of use. Conse-
quently, many states are now considering modified classifications for those
receiving waters exposed to combined sewer overflows in order to accom-
modate the water quality variations brought about by overflows. A list of
basic parameters for both the system and the receiving waters appears later
in this section.
Location—
In the investigation of any overflow system, it is generally recognized
that limiting the system monitoring to a certain number of "representative"
overflows should be adequate to provide the basis upon which projections for
the entire system can be made. The extent to which this approach can be
taken is dependent upon the typical or atypical nature of drainage areas and
their associated sewer networks. Representativeness of monitoring sites is
dependent on: (a) relative land uses in the form of residential, commercial
or industrial areas; and (b) the sewer network, including locations of catch
basins and the structural condition of the diversion chamber prior to over-
flow. Other practical considerations include accessibility, physical
features, and the size of the monitoring site in question.
Duration—
Duration of sampling should be adequate to capture the peak concen-
trations of pollutants which normally occur in the initial runoff period. The
pollution peak is oftentimes in the vicinity of the "time of concentration" of
the associated runoff volume. The total duration of the sampling should
also be adequate to capture the decline of pollutant concentrations as well
as runoff quantities. Consequently, first estimates could be based on the
"time of concentration" associated with a particular drainage area. Real
data will ultimately serve as the basis for any needed adjustments to the
sampling duration. While the duration must be long enough to capture the
"first flush" periods, intervals of sampling must be small enough to
adequately define rate changes of pollutant concentrations as well as runoff
volume. Figure 4 illustrates typical plots depicting characteristics of an
overflow discharge.
Frequency—
The frequency of sampling is important in order to adequately capture
the significant changes that occur in both pollutant concentration and runoff
volume. Generally it has been found that in as little as five minutes following
the onset of runoff volume, significant concentrations can occur. There-
after, 15-minute intervals have proven to be adequate to define the "first
flush" period and the gradual decline of pollution in a discharge. These
25
-------
time intervals have been largely adopted by EPA for the purposes of pro-
viding the required NPDES information. A more thorough definition of the
variation of pollutant concentrations with time can be obtained by decreasing
the size of sample intervals. However, the sampling frequency required to
define the quality response of an area is a function of the time of concentra-
tion and should be selected on an individual basis.
Samples can be classified as either discrete or composite (8). While
composite samples are useful in characterizing pollutant loads which
approximate steady state (e.g., wastewater treatment plant influent), they
are of little value in depicting the dynamic nature of storm-related sources.
In order to adequately define the variations described above, discrete
sampling is necessary at some prescribed time intervals. The results of
such sampling, when combined with the continuous recording of flow, pro-
vide the necessary data for the construction of discharge pollutographs and
loadographs, as shown in Figure 4.
Automation--
Since the measurement of representative overflows for a given
sewerage system will normally require a significant percentage of the total
overflows, and rainfall events can occur during the more inconvenient hours
of the day, automatic sampling has been commonly recognized as being
more cost-effective than manual sampling. Equipment is available which
allows such sampling to be coordinated with continuous recordings of sewer
flows. Many NPDES permits require the measurement of 25 percent of
Type 1 discharges, which are those that are known to overflow in excess of
20 times a year for as many as four separate and independent rainfall
events. Thus, for the purposes of meeting such requirements and at the
same time to obtain the needed data for overflow investigations, the need for
automation in large systems becomes obvious.
Manual Sampling—
It is, however, possible in smaller systems where manpower is
available that the necessary data could be collected manually. Manual
sampling can be done so as to represent a waste stream and in addition can
be done at such time intervals as to more accurately coincide with flow
variations. In addition, malfunction of sampling devices can be eliminated.
The primary drawback of manual sampling is that manpower must be
deployed at the first hint of rain or risk missing the all important "first
flush" period. On the basis of this alone, manual sampling for the sewer
system is generally discouraged.
26
-------
SECTION 5
RECEIVING WATER SAMPLING
GENERAL
The governing criterion for the abatement of any pollution is the
impact on water quality. Despite the lack of data in the past, it has been
generally recognized that future programs will require a heavy emphasis on
impact data. In its March 1976 report, the National Commission on Water
Quality (9) stated:
"We also find that there is still a major lack of adequate
information. We simply do not know enough. There are
not sufficient data to tell us exactly how bad the water
was, or how touch better it is getting.. The measuring
and analytical techniques and predictive methodologies
are not good enough in many instances to tell us the scope
and value of incremental water quality improvements. If
billions of dollars are to be invested wisely, we must
have more and better data collected over an adequate
term of years. "
This condition still prevails today. Programs must be initiated now
and conducted for some years to come in order to have the long-term data
base necessary for rational decisions. A "Report to Congress" by the.
Comptroller General's Office issued December 1976 (10) concurred with the
Commission's concern and recommended that EPA:
"Determine whether existing resources at the State level
are adequate to implement effective comprehensive water
quality planning and data collection programs. If
existing resources are inadequate, additional resources
should be requested from the Congress. "
In the case of combined sewer overflows, the need for receiving water
data is particularly important since there is the risk that large amounts of
dollars may be needed to remove relatively small amounts of pollutants.
27
-------
-Also, the water quality impact may be of such a transient nature that the
benefits derived may not be cost effective.
WATER QUALITY CONCERNS
Parameters for overflow discharges should reflect the water quality
criteria and hence should be the same as those measured for the receiving
water. The exceptions are those biological parameters that reflect the
aquatic food chain in a receiving water. For the purposes of this methodo-
logy* "the primary concerns of stormwater impact on a lake are:
(a) pathogens; (b) dissolved oxygen; (c) nutrients; and (d) acute toxicity.
Several bacterial groups, including the total coliform (TC), fecal
coliform (PC) and fecal streptococcus (FS) groups have been used to
indicate the probable presence of pathogens. Indicator groups are used
since their presence is more easily and rapidly detected, and their corre-
lation to the presence of pathogens is widely documented (11, 12). The fecal
coliform group is generally recognized as the single most reliable indicator.
The fecal streptotoccus group is useful in that a ratio between this group
and those of the fecal coliform group can indicate the source of the fecal
contamination. The ratios of fecal coliforms to fecal streptococci have been
interpreted as follows (13):
FC/FS
>4
0.7-4.0
<0.7
Source of Pollution
Human
Human and Animal
Animal
The dissolved oxygen (DO) concentration of a lake is influenced by a
variety of physical, chemical and biological processes. The velocity and
depth of the water and its altitude and temperature are all physical con-
straints on the amount of DO in a body of water. Oxygen is consumed during
chemical or biochemical reactions in the water column or in the sediments.
In the absence of oxygen, anaerobic processes predominate, leading to
odors and other septic conditions. .Oxidizable waste products discharged
into municipal and industrial wastewaters may reduce DO to very low con-
centrations, even to zero. Because aquatic organisms consume oxygen
during respiration and because aquatic plants, particularly algae, produce
oxygen during photosynthesis, DO concentrations may vary significantly
during the course of a day. DO standards are often established to insure the
sustenance and well being of a certain desired fish population. For
example, in many states, DO standards are based on the sustenance and
weU being of trout.
28
-------
Nutrients, in addition to DO, play the major role in determining the
trophic status of the lake. Eutrophication of a lake occurs when there is an
excess of nutrients, causing an enriched water system. Eutrophication of a
lake can have direct as well as indirect effects on public health. Toxics
from certain algae species and parasitic organisms have been suspected as
the cause of certain ear, nose and throat infections. Intense algae blooms
can be a nuisance to recreational uses such as swimming and boating.
Nitrogen and phosphorus have been identified as major elements which, in
excess concentrations, can cause eutrophication. Phosphorus is an essen-
tial plant nutrient that is often critical to plant productivity because of its
marginal concentrations in many receiving waters. Addition of this
material associated with cultural activities often results in stimulation of
plant productivity. The artificially-induced increased productivity is
referred to as "cultural eutrophication". Major sources of phosphorus in
waterways include industrial wastewater discharges, municipal wastewater
discharges, and agricultural and urban stormwater runoff.
Toxics are a major concern in that they can have the most immediate,
severe and direct effect on either the aquatic system or public health. Un-
fortunately, toxics are by their nature specific and not detectable by conven-
tional laboratory equipment. Thus, either prior knowledge of a toxic
material must exist or a periodic monitoring- program must provide for a
scan of suspect materials. In consideration of the new, extensive and
always increasing list of toxics, as defined by EPA under provisions of the
Toxic Substances Control Act, projects can rarely accommodate analyses
for toxics; the only exceptions are those heavy metals that have been
commonly associated with i-unoff and can be analyzed relatively easily.
Since a material is toxic by virtue of the adverse effect it has on the water
system in question, a reasonable alternative would be to utilize the BOD
test as an indicator of the existence of any such material. If a decreasing
trend in BOD is consistently noted with a decreasing dilution ratio (ml
dilution water used/ml sample used), the sample probably contains an
inhibitory or acute toxic substance. It would then warrant further investi-
gation into the source and nature of the material.
Table 1 lists the parameters to be considered for analysis in both the
CSO system and the receiving water. Certain of these parameters require
field measurement, as indicated by the table. Some analyses can be con-
sidered optional in that they can be measured in place of other parameters.
Chemical oxygen demand and total organic carbon have both been considered
as substitutes for the BOD test. However, in terms of impact on the aquatic
system, the BOD test still represents a more direct measure of actual
demand as it occurs in the receiving water and, as described above, can
indicate inhibition or toxicity. Certain biological parameters are direct
measures of the food chain within the receiving water. The extent to which
measurements are required depends largely on the necessity of having to
29
-------
measure the balance of the food chain. In most overflow impact studies,
such detail would not be warranted unless toxins produced by one or more of
these organisms are so in evidence that control procedures would be
required.
TABLE 1. WATER QUALITY PARAMETERS
Discharges & Tributaries
Lake
Field
Measurements
Biological
Parameters
Laboratory
Analyses
Flow
Dissolved oxygen
Temperature
PH
Lake currents
Dissolved oxygen
Temperature
pH
Transparency
Plankton
Periphyton
Macrophyton
Macroinvertebrates
Fish bioassays
ALL LOCATIONS
Fecal coliform
Fecal streptococcus
BOD5
BOD2?
Chemical oxygen demand*
Total organic carbon*
Total solids
Dissolved
Suspended
Volatile and fixed
Settleable
Nitrogen compounds
Organic
Nitrite
Nitrate
Total phosphorus
Ortho-phosphorus
Poly-phosphorus
Chlorides
Sulfates
Sulfides
Alkalinity
Hardness
Heavy metals
Lead
Copper
Zinc
Chromium
Mercury
^Optional
30
-------
SAMPLING LOCATIONS
The objective of receiving water sampling is to define pollutant con-
centrations at all significant discharges and tributaries to the lake and in the
various zones within the lake prior to, during and following storms. All
major inputs should be identified in order to establish a reasonable
materials balance throughout the storm. Any tributaries to the lake should
be sampled so as to bracket the segments that receive combined sewer
overflows. In this manner, upstream non-point pollutant contributions can
be identified and compared to those coming from the sewer system. In some
cases it may also be desirable to isolate those areas served by a separate
sewer system, as opposed to a combined system, in order to determine the
significance of urban stormwater. Minor tributaries need not be monitored
if they are known to contain a relatively insignificant quantity of pollutants.
Using Onondaga Lake as an example, Fiffure 6 illustrates the major pollu-
tant sources that should be identified in any sampling program.
In stratified lakes there exist two major zones which should be
sampled in most limnological investigations. These zones are the
epilimnion (upper water) and hypolimnion (lower water). They are separated
by a zone of rapidly changing density (metalimnion), resulting in two
relatively distinct and independent volumes of water. In shallow lakes,
where such a separation does not exist, measurement at one depth may be
adequate. Preliminary investigations can be conducted to determine the
variation of temperature and dissolved oxygen to determine if stratification
exists and if multi-depth sampling is required. Depth profiles should be
taken throughout the study to chart movement of the metalimnion so that
sampling depths can be adjusted.
Transverse movement in a stream is often ignored in simple one-
dimensional modeling. In the case of a lake, the transverse direction is not
distinguishable from the longitudinal direction. Generally, stream sampling
emphasizes time of travel in the longitudinal or downstream direction for
the purpose of tracking both the dissolved oxygen sag and the recovery zone
of the stream. Within a lake, "downstream" can be in any compass direc-
tion, depending upon such external influences as tributary currents, wind,
and the consequent internal oscillations referred to as the seiche effect; the
lunar influence is not significant in the smaller lakes.
The essential difference between a stream and a lake emphasizes the
need for a greater knowledge of lake currents as they affect circulation
patterns of a lake. Based on these patterns, zones within a lake can then be
separated, sampled and identified. Segmentation of a lake into zones is
particularly useful in order to distinguish certain tributaries and shoreline
areas of the lake, where contact recreation will take place, from the deeper
areas of the lake. The latter areas are more apt to represent the major
31
-------
aoanos
NON.
a
a
o>
a
o
6
a
o
0»
_c
"5.
I
•o
o
_o
Q.
a>
a>
CO
§
32
-------
lake water quality, whereas areas directly exposed to shoreline activity will
reflect more temporal effects. Figure 7 illustrates an example of lake seg-
mentation. The relationships considered within each segment are presented
in Section 8 of this report.
DURATION OF SAMPLING
The duration and frequency of sampling must be adequate to profile
pollutant loads in much the same manner as outlined for overflow dis-
charges.
The number of storms for which the receiving water should be sampled
is dependent upon the adequacy with which the resulting information can
relate water quality of the lake to storm pollutants. One approach accom-
plishing this objective is to conduct the sampling until the impact of the
"design" storm is actually measured. The "design" storm has come to
mean that storm for which abatement facilities are planned. While this
approach would yield the most directly usable impact data, sampling of the
design storm is aimed at an elusive target and can result in extensive
expenditures for field work.
Another approach to accomplishing the objective is to conduct only
enough sampling to define the essential functions within the lake and use
mathematical projections to calculate the impact of a variety of storms. In
the absence of suitable and practical guidelines for the selection of the
design storm, the use of mathematical projections has become necessary in
order to evaluate the wide variety of conditions often imposed on the
receiving water. It is likely, moreover, that sufficient field data will have
been collected to develop essential relationships within the lake before any
preselected design storm is encountered. In addition, mathematical
modeling techniques provide a great amount of flexibility to evaluate the
combined sewer system response to a variety of rainfall patterns and sewer
system modifications. Modeling techniques available for this purpose are
covered in Section 7 of this report.
FREQUENCY OF SAMPLING
Frequency of sampling of the tributaries and lake must be adequate
to establish the altered character of the ecosystem associated with a storm
occurrence. It is important, therefore, to measure stream and lake condi-
tions before, during and after each storm. The following sampling schedule
was adopted initially to document the impact of storms monitored on
Onondaga Lake during the period of June through September 1976:
33
-------
o
"c
I
o>
0)
-o
o
§
o>
34
-------
Sample Period
Frequency of Sampling
12 hours prior to storm runoff
During 12-hour storm
3-day period following storm
Lake Stations
1
3
6
Tributary Stations
1
6
9
As the sampling phase of the study progressed, some alterations to the
frequency were made, reflecting specific storm durations, weather, night-
fall and evaluated data. The pre-storm sample was particularly difficult to
obtain since long-range weather forecasting is often inaccurate. Prolonged
storms required modification of the sampling schedules to obtain sufficient
post-storm measurements.
AUTOMATION
Unlike the overflow discharges, receiving water locations, particu-
larly in a lake, do not lend themselves easily to automatic sampling.
Navigational restrictions and equipment costs generally rule out the installa-
tion of automatic equipment in a lake. Depending on the size of a tributary,
these restrictions may also apply to stream measurements.
A major concern in any sampling is the representativeness of the
sample. A lake or stream is more likely to show variations of flow and
pollutant concentration through the cross section of flow than is a piped dis-
charge. Thus, multiple sampling ports may be required if automatic
sampling is to be considered. Although, in theory, multiple sampling points
may appear reasonable, such installations are generally subject to loss and
destruction from the exposure to the elements common to receiving water
locations; such installations are generally to be discouraged. If "cross
sectional" variation can be considered negligible, then automatic sampling
may be feasible for discharges and tributaries. Automation notwithstanding
mechanical failures, would assure the capture of "first flush" periods in
the stream regardless of the time of day or night. However, there often is
a lag period between the start of a storm and the time when the impact of
the storm is noticeable in a lake, manual sampling of the lake can
commence the following morning, if the storm occurs at night.
35
-------
SECTION 6
DATA BANK
An important aspect of any study is the convenient storage of all
data in a form that is easily reviewed and analyzed. For studies involving
water quality sampling with a large volume of analyses, it is of utmost
importance to be able to store, process and retrieve the data as rapidly as
possible. As shown on the flow chart (Figure 8), the ideal method of
reviewing data proceeds from sampling and analysis to computer input and
output, to laboratory for review, and back to the computer for corrections.
Once the data has been checked by laboratory personnel for accuracy
it should then be reviewed by the engineering or planning staff for overall
applicability. Since large sums of money are being spent on data collection
and laboratory analysis related to CSO studies, it is extremely important,
as the project progresses, to review the data prior to establishing the para-
meters to be analyzed during subsequent storms. Data review ensures that
dilutions for certain laboratory analyses have been properly selected, and if
not, provides time to correct these inaccuracies before the next sampling
event. The comparison of trends in both the receiving water and the system
discharges will focus the reviewer's attention on areas of critical concern
and maximum impact. This detailed review of data collected will highlight
monitoring locations where more or less intensive studies are dictated,
where monitoring stations should be added, and, possibly, where sampling
can be eliminated. It also will reveal parameters that fail to reflect storm
impact and enable these analyses to be eliminated. While ideally these data
should be stored and retrieved from the computer and reviewed prior to
embarking on the next sampling event, this procedure is not always
practical.
The file structure should include sufficient identifiers so that data
collected can be recalled from the data bank by location, storm, date and
time, and parameter name. For location it is desirable to have two
identifiers available, one for location in plan and the other for the depth
from which the sample was collected. The data base should be established
so that any of the primary identifiers, such as all BOD5 data associated with
Location #7 on a given date, or all dissolved oxygen measurements
36
-------
0>
u
&
5
0>
o
-a
o
E
3
CO
CO
a>
o»
37
-------
associated with a given location. Identifiers enable all significant opera-
tions to be readily performed on the data.
The most useful method of presenting tributary data is in the form of
a pollutograph or pollutant load (pounds per day) versus time plot for a given
location (Figure 4). For lake stations, a time-varying plot of concentration
for each station is the most convenient and usable method of presentation.
Another useful measure for comparing tributary contributions is the area
under the pollutograph (Figure 4), or the total pounds of pollutant discharged
per storm. A further refinement of this calculation would involve total
contribution per day or even total contribution per hour; however, rarely is
sampling accomplished frequently enough to justify this detail.
In addition to providing for easy perusal of data, an organized data
base makes statistical analyses much easier to perform, particularly in the
critical case of data collected from the combined sewer system. Statistical
analyses performed to obtain both the relationship of the quantity of runoff
to rainfall and the quality of overflow to time or runoff are useful in deter-
mining the quantity and quality of predicted overflows.
ONONDAGA LAKE
The Onondaga Lake data base was set up as described in the preceding
paragraphs. This level of identification permitted the retrieval of data in
all the output forms found necessary throughout the project.
For Onondaga Lake, portions of the data were available for review in
handwritten form, while at no time were data from the computer available
prior to the next sampling event. In retrospect, this could have been a
major time- and money-saving tool. Perusal of the entire data base "after
the fact" shows that many of the parameters could have been eliminated or
the frequency of analysis reduced at several of the sites. Examples of sites
that could have been totally ignored were Bloody Brook and one of the indus-
trial discharges (Figure 6); parameters that could have been measured on a
reduced frequency included all metals, chlorides at all but NineMile Creek,
and elimination of all analyses at the East Flume with the exception of total
Kjeldahl nitrogen.
38
-------
SECTION 7
MATHEMATICAL MODELING OF COMBINED SEWER SYSTEM
OBJECTIVE OF MODELING
The objective of modeling the sewer system in CSO and stormwater
analyses is to develop a mathematical representation of the hydrologic,
hydraulic and water quality aspects of the drainage area and wastewater
collection network. The model, once calibrated, can be used for predicting
both the quantity and quality of CSO at specified design storm conditions.
Modeling also reduces the amount of data gathering required. Once an
adequate number of quantity and quality measurements are made over a
range of storm types, the model enables the prediction of results for any
reasonable set of storm conditions. It is not feasible in most instances to
measure rainfall, CSO or stormwater quantities and quality, and receiving
water quality for the period of time necessary to ensure that appropriate
design conditions will be captured. Thus, the proper balance between actual
field monitoring and mathematical modeling must be determined for each
particular study.
All stormwater models, with the exception of simple "desk-top"
models, require calibration and verification of the predicted to the
measured data, if results are to have credibility. Calibration of the model
parameters and coefficients is an integral part of overall modeling proce-
dure in that the finer the model is calibrated, the more confidence can be
placed in its predictions. Calibration and verification require sets of
measured field data. The measured data should be taken over a range of
storm conditions if at all possible. Some data, such as drainage area
characteristics, dry-weather flow information, and collection system con-
figuration, do not have to be obtained in the field. These data can often be
found in aerial photographs, topographic maps, sewer plans, planning and
engineering reports, and other sources. While it is not necessary to field-
measure these characteristics, each area should be field-checked to insure
the accuracy of these sources, to point out any discrepancies, and to give
the modeler a visual perception of the area. Rainfall and runoff quality and
quantity, however, must be measured in the field in a time-coordinated
manner. Rainfall data can be obtained from the National Oceanic and At-
mospheric Administration (NOA.A), National Climatic Center, Federal
39
-------
Building, Asheville, North Carolina. This information is available on com-
puter tapes for all continuously-operating NOAA weather stations.
In general, the overall objectives and budget constraints of the storm-
water study will dictate the level of both the field sampling and the mathe-
matical modeling. If the study is to result in the preliminary, or actual,
design of CSO or stormwater abatement facilities, the level of effort asso-
ciated with the modeling and data gathering will be great. On the other hand,
if the results of the study are to be used as a preliminary assessment of
pollutant loadings and their impact on a particular waterway, then the moni-
toring and modeling requirements will be relatively small, with most of the
effort placed on relatively simple modeling techniques.
TYPES OF MODELS AVAILABLE
Levels of Analysis
There are a number of mathematical models currently available for
CSO and stormwater analyses. They range from the "desk-top", hand-
calculated models, such as the simple "rational formula", to the sophisti-
cated EPA Storm Water Management Model (SWMM) (15, 16). Three
categories have been suggested for all stormwater assessment models,
based on their applications, as shown in Table 2 below.
TABLE 2. CATEGORIES OF STORMWATER MODELS
Level Application
I Problem Assessment
II Planning
III Event Analysis
The mathematical models used in stormwater analyses increase in
complexity from Level I to Level III. Also increasing are the data require-
ments associated with the model calibration and actual operation. Table 3
further describes the three categories of stormwater assessment in terms
of model complexity and characteristics.
Some models are capable of being used for more than one application.
For example, the EPA SWMM model is considered a single event, or Level
III model, but it can also be operated continuously, or as a Level II model.
Those models which do not require the use of computers, such as the EPA-
Level I SWMM; the URS, Inc. procedure; and the EPA Areawide Assessment
Procedures Manual method, are considered Level I, or problem assessment
models (7, 17, 20, 21). In addition to the continuously-operated EPA
SWMM, other planning, or Level II, models include the U. S. Army Corps
40
-------
TABLE 3. STORMWATER ANALYSIS LEVEL
AND MODEL COMPLEXITY (17)
Analysis
Level
I
II
III
Model
Type
Desk-top
Continuous
simulation
Single
event
simulation
Model
Complexity
Low to
medium
Low to
medium
Medium
to high
Purpose
of Model
Problem assess-
ment, prelimi-
nary planning,
alternative
screening.
Problem assess-
ment, planning,
preliminary
sizing of facili-
ties (particularly
storage), alterna-
tive screening.
Assess long-
term impacts of
designs.
Analysis for
design, detailed
planning.
Model
Characteristics
No computers.
Equations, nomo-
graphs based on
statistical analyses
of many years of
records.
Program of few
hundred to few
thousand statements.
Uses many years of
rainfall records
with daily time
steps, or worst two
years with hourly
time steps. May
include flow routing
and continuous
receiving water
analysis.
Program to cover
10, 000 statements.
Higher modeling
precision, from
rainfall through
sewers, possiblyto
receiving waters.
Short -time steps
and simulation
times. Fewer
alternatives to be
evaluated.
41
-------
of Engineers' STORM and EPA Simplified SWMM (SSWMM) (21, 22).
A number of models of all levels of complexity are capable of predic-
ting only quantity. Others are only applicable for separate stormwater
systems since dry-weather flows are not readily considered. Many of the
available models are continually being updated and modified to reflect user-
supplied suggestions or better knowledge of actual system responses. A
number of models are presently being revised to include quality and/or dry-
weather flows to be applicable for CSO studies. EPA has recently published
four reports which analyze most of the currently-available stormwater
models (7, 17, 23, 24). It is recommended that these reports be reviewed
by those planning to use a mathematical model for CSO or stormwater
analyses since the advantages and disadvantages of the models are presented
with respect to various applications. An important consideration should be
that the model selected for any study should be the simplest that can
adequately perform the necessary analyses.
The two models that were used in the Onondaga County CSO Study,
SWMM and SSWMM, are discussed in greater detail below.
EPA SWMM
The EPA SWMM is one of the most comprehensive yet flexible storm-
water models currently available. Dry-weather flows can be considered,
runoff quality as well as quantity can be simulated, sewer routing is
included, and it can be operated in either a single event or continuous mode.
SWMM, being relatively complex, requires a large amount of input data and
a relatively large computer. First introduced in 1971, the model has been
updated on several occasions to reflect both changes in the state-of-the-art
of stormwater modeling as well as to correct errors in the proffram itself.
The model consists of five basic components: the RUNOFF, TRANS-
PORT, EXTENDED TRANSPORT (EXTRAN), STORAGE-TREATMENT
(S/T), and RECEIVING WATER (RECEIV) BLOCKS. Figure 9 shows the
relationship of the blocks to the model's central component, or the
EXECUTIVE BLOCK.
Figure 9. Major components of EPA Storm Water Management Model.
42
-------
The RUNOFF BLOCK simulates the rainfall and subsequent runoff
processes including overland, gutter and/or pipe flow. Drainage basin
characteristics such as size, slope, imperviousness, roughness and other
factors are taken into account. Land use and pollutant accumulation of
street surfaces are also considered when runoff quality is simulated. Also
taken into account is the surface storage on the impervious and pervious
areas, the soil infiltration capacity, and flow routing through the gutter and/
or pipe network. More detailed sewer routing usually takes place in one of
the transport blocks, although this is not always warranted. The results of
the RUNOFF BLOCK simulation for both quantity and quality can be the input
to other blocks of SWMM or can be used by themselves.
The TRANSPORT BLOCK is used to route the input hydrographs from
the RUNOFF BLOCK throusli the assigned sewer network to either treat-
ment or storage facilities, direct discharge, regulated discharge, or to
additional transport networks. TRANSPORT does not simulate sip-charging,
flow reversals nor looping as it can only simulate one-directional flow.
EXTRAN is used instead of TRANSPORT where surcharging or back-
flow problems are encountered. This block is more costly to run than
TRANSPORT due to the relatively small time-step required to account for
model instability problems.
Both transport blocks route quality along with the quantity, taking into
account dry-weather flow and pollutant scouring in the sewers in addition to
the pollutant washoff from RUNOFF.
S/T is used to simulate storage and treatment alternatives. Input
hydrographs can come 'either directly from RUNOFF or from one of the
transport blocks. A number of treatment processes are included in the
block, each with assigned pollutant removal efficiencies.
RECEIV is used to simulate the hydro-logic and water quality impact
of actual or predicted discharges on the receiving water. RECEIV can accept
input from any of the other blocks or it can be used independently. It is a
dynamic water quality model in that it can simulate the effects of time-
varying waste loads, such as CSO or stormwater discharges. The block is
similar in composition to EXTRAN in that backwater effects, such as
encountered with tide cycles, are taken into consideration.
EPA Simplified SWMM .(SSWMM)
Due to the complexities of Level IH SWMM, its use is not recommen-
ded for problem assessment or for the preliminary sizing of storage and
treatment facilities. SSWMM was developed to fill the needs of those
43
-------
requiring a model with both problem assessment and planning capabilities,
yet requiring a less complex, more economical tool than a Level III model.
SSWMM, a Level II model, has greater detail than the Level I desk-top
model. Field data is required for model calibration and model operation,
but not to the same order of magnitude as for detailed SWMM or other
similar models.
SSWMM consists of a number of small independent computer programs
for its five major modeling tasks. Figure 10 shows these tasks along with
their interrelationships to each other. Only three tasks -- rainfall charac-
terization, storage-treatment balance, and receiving water response —
require the use of a computer.
-1. DATA PREPARATION
2. RAINFALL CHARACTERIZATION
>3. STORAGE-TREATMENT BALANCE-
-4. OVERFLOW QUALITY ASSESSMENT-
5. RECEIVING WATER RESPONSE4
Figure 10.
Interrelationship of tasks in the Simplified
Storm Water Management Model
The model differs in a number of respects from detailed SWMM,
Unlike SWMM, which can be operated in a range of time-steps to correspond
to the particular rainfall hyetographs, or drainage or transport network
characteristics, SSWMM can only be run on a one-hour or daily time-step
to correspond to computerized rain tapes. As with most planning models,
SSWMM can only be run in a continuous mode. Runoff is calculated in a
rather simple fashion using a runoff coefficient "K" (the gross runoff co-
efficient) similar to that of the "rational formula". However, the runoff
calculations are more detailed than the "rational formula" in that runoff
volumes are computed for each rainfall. The major tasks involved with the
use of SSWMM are discussed in the four paragraphs that follow.
Rainfall Characterization—
Five small computer programs are included in SSWMM to analyze the
data on the rain tapes from the National Climatic Center. These include
programs which define storm events, sequence storms and rank storms by
assigned parameters. Once the data is analyzed the rainfall is subjected to
runoff, storage, treatment and/or overflows in the storage-treatment
program.
44
-------
Storage-Treatment Balance—
The storage-treatment program of SSWMM can be operated either
hourly or daily. Interceptor capacities, storage rates and treatment rates
are set with overflows occurring when the various rates are expected. All of
the rates can be modified independently to compare a number of alternatives
in handling and treating the runoff.
Overflow Quality Assessment—
Overflow quality is not determined within the actual model, but is cal-
culated manually either using literature pollutant loading functions or statis-
tically manipulating actual, measured quality data. The volume of overflow,
as obtained in the storage-treatment program, is paired with the calculated
quality data obtained through the above analysis. Dry weather flows and
quality not included in the model should be included with these hand calcula-
tions.
Eeceiving Water Response—
The response of the receiving water to the stormwater discharges is
simulated through the receiving water program. The use of an existing
receiving-water quality model, if any, is recommended in preference to the
one provided in the model. The receiving-water quality model included with
SSWMM is general in nature and requires calibration prior to use as a
predictive tool; it is also a steady-state model and does not simulate dynamic
loadings adequately for all applications.
Comparison of Detailed Single-Event SWMM and Simplified SWMM (SSWMM)
Both SWMM and SSWMM have their advantages and disadvantages in
range of application, level of complexity, cost of operation, and data
requirements. Thus, a direct comparison of the two models is not totally
applicable since both models would not necessarily be considered for the
same specific situations. Table 4 highlights some of the major features of
each.
Complementary Model Usage
As in the case of the Onondaga County CSO studies, it may be
desirable to employ both the simplified and detailed models. SSWMM can
be used initially to describe some basic rainfall-runoff or rainfall-overflow
relationships for the key drainage areas. These relationships can be used
preliminarily to locate storage and treatment facilities in addition to
assessing water quality impacts in the receiving waters. A further use of
such relationships is in determining the design storm or storms. Although
the rainfall data for SSWMM is in hourly increments, the results of the
model will give a reasonable picture of the sensitivity of a drainage area to
rainfall and overflow conditions spanning long periods of record.
45
-------
TABLE 4. COMPARISON OF SINGLE-EVENT SWMM AND SSWMM
Single-Event SWMM
SSWMM
Model complexity
Storage core
requirements
Cost of operation
Data requirements
Rainfall data
Dry weather flow
and quality
Flow routing
Sur char ging-b ackflo w
Receiving water
quality model
High
350K
High
High
Multi-hyetographs can be
used along with a range
of time-steps in the order
of minutes; 1-hour rain
data for multi-years can
also be used.
Included
Included in either one
of three blocks
Included
Dynamic model included
Low
40K
Low
Moderate
Single hyetograph
with 1-hour time-
step; 10-20 years of
data preferred.
Not included
Not included; only
interceptor capacity
at overflow point
Not included
Steady state model
included
46
-------
In this same light, the continuous simulations of the simplified model
can be used to predict the number or frequency of overflow events that may
occur, either under existing conditions or employing a number of abatement
alternatives. Overflow occurrences can also be used as a calibration tool
in comparing the number of overflows monitored versus those predicted
over a period of months or years.
SSWMM can also be used to assist in choosing field monitoring loca-
tions. This use may arise where a number of drainage or catchment areas
appear equally suited for monitoring based on existing data and field recon-
naissance. Some preliminary rainfall-runoff analyses of SSWMM may
reveal some obvious catchments worthy of monitoring due to their sensitivity
to rainfall,, interceptor capacity, or other factors.
Once calibration of SSWMM proves successful, the model can be used
to screen the various storage and treatment schemes. Detailed SWMM can
then be used to determine gross dimensions of the various abatement facili-
ties by simulating an actual design storm or storms. Detailed SWMM, with
its better-defined rain events, will produce more accurate peak flow rates
and overflow volumes than result from SSWMM.
If a good correlation is found between the predicted outputs of both the
detailed and simplified models, more emphasis can be placed on the less
complex model for additional screening of alternatives. Detailed SWMM can
then be used for final sizing of facilities and for interceptor analyses.
Analysis of interceptor sewer systems can be a very important aspect of the
total CSO investigation, particularly when dry-weather wastewater treatment
facilities are being considered for treating additional wet-weather volumes
or when the amount of undiverted flow reaching the interceptor system is
affected by surcharging.
DESIGN CONSIDERATIONS FOR QUANTITY PROJECTIONS
The choosing1 of a design storm or storms is an important considera-
tion in determining the quantity of overflow in the sewer system modeling
program. There are several approaches to choosing design storms and
there is no one method best suited for all applications. The two common
approaches are: (a)long-term, continuous simulation; and (b) the single-
event design storm. Both approaches have certain advantages and dis-
advantages as discussed below.
Continuous Simulation
Continuous simulation offers a major advantage by utilizing actual
rainfall data in the CSO analysis for both projecting pollutant loadings and
47
-------
testing abatement alternatives.
fall simulation.
Level II models all employ continuous rain-
Continuous simulation, however, is often more costly to run than a
single-event model, particularly if the model employed is moderately com-
plex and several years of data are used. However, to estimate annual or
seasonal loadings, long-term rain data employed in conjunction with a cali-
brated model will give reasonably accurate results. Also, continuous rain-
fall simulation is an excellent method for estimating the seasonal or annual
pollutant reductions from various storage-treatment alternatives. When a
lake or slow-moving receiving water is under study, seasonal or annual
loadings are important considerations, in addition to the loadings for a
Single design storm.
Single-Event Design Storm
The single-event design storm has been used for a number of years in
stormwater analyses, particularly in the design of storm drainage facilities.
Data on rainfall-frequency-duration is usually obtained from the Rainfall
Frequency Atlas of the United States (25). This atlas includes rainfall for a
number of durations and frequencies, ranging from 5 minutes to 24 hours
duration and recurrence frequencies of 1 to 100 years. However, neither
the atlas nor hourly rainfall data gives the detailed temporal distribution of
rainfall during the course of a rainstorm. If such distribution is needed or
desired, as is the case for Level III single-event modeling, synthetic hyeto-
graphs are employed.
The synthetic hyetograph has the advantage of applying a storm at the
desired duration, return period frequency and average intensity, while
incorporating the statistically-determined intensity distribution. Synthetic
hyetographs can be developed for any area where sufficient rainfall records
are available. The concept is based on the theory that rain falling in a
geographical area will fit into a pattern that can be described mathemati-
cally. The key elements of a synthetic hyetograph are the volumes of rain
falling before the peak, after the peak, and the location of the peak intensity
period itself. One of the first papers on the concept was published in 1957
by two City of Chicago engineers (26). An example of a synthetic hyetograph
for Syracuse, New York, developed for a number of return periods is
shown in Figure 11. The storm duration used with the synthetic hyetograph,
or any similar design, storm, is usually equal to the time of concentration
of the drainage area under consideration.
There have been some recent studies and criticisms of the synthetic
hyetograph approach for design storms (27, 28). Some apparent drawbacks
stem from the potential for over-predicting of peak flow rates because of
48
-------
O 2 o — —
>».-.
*I|?S8
£?«***
E S .
5 g. i fe k
Z* -•«
o>
V)
Q.
o
0»
o
15
49
-------
the high peak intensities resulting from the development of the synthetic
hyetograph. Care must therefore be taken to ensure that the hyetograph is
simplified enough, through averaging of time increments, so as not to over-
predict peak overflow rates. Sensitivity analyses of various time incre-
ments are recommended when employing the synthetic hyetograph for design
purposes.
Other Techniques
Routines for the statistical analysis of rainfall data are sometimes
incorporated in stormwater models. Such is the case for SSWMM, the
Illinois Urban Storm Runoff Model (IUSRM), and the Level I Areawide
Assessment Procedures Manual (7, 22, 24). These statistical routines,
which can be used separately or in conjunction with the models, allow the
user to analyze and select rainfall volumes, durations, intensities and
antecedent dry-weather conditions that are either representative or occur at
a frequency of return best suited for the particular CSO analysis.
Another method is to reduce long-term rainfall data to a statistically
representative period of time. This approach, which is referred to as
"annual averaging", was used in the Onondaga County CSO study where over
20 years of rainfall data were reduced through random selection techniques
to a two-year period of record for use in SSWMM.
Other possibilities include using a synthetic hyetograph for actual
design of a facility, with continuous rainfall simulation employed to test the
pollutant-removal effectiveness over the course of a season or a year. This
complementary usage offers the advantages of both continuous simulation
and the single-event design storm.
When an analysis has multiple objectives, such as for CSO pollution
abatement and flood protection, it may be desirable to apply a variety of
design storms in conjunction with continuous simulation. For example,
design considerations for water quality objectives require an annual loading
determination whereas flood protection usually is based on an infrequently
occurring storm. For situations where water quality is the prime objective,
it is desirable to project loadings for a number of storms with various
return periods. The results of these projections can be used to assess the
cost effectiveness of the return period selected for the design storm, as
will be discussed later in greater detail.
DESIGN CONSIDERATIONS FOR QUALITY PROJECTIONS
There are a number of factors affecting the quality of stormwater
runoff and combined sewer overflows. Eight of the more important factors
are presented in Table 5.
50
-------
TABLE 5. FACTORS AFFECTING CSO QUALITY
Antecedent Dry Weather
Land Use of Catchment Area
Pollutant Accumulation
and Washoff
Street Sweeping Practices
Soil Erodibility and Erosion
Control Practices
Dry Weather Waste water
Solids Deposition and Scour-
in Sewer System
Sewer System Characteristics
Affects amount of pollutants available on
surfaces and in sewers, and sets initial
infiltration capacities of previous
surfaces.
The type of land use can dramatically
affect the pollutant accumulation rates
in an area.
Accumulation is dependent on antecedent
dry weather, land use, street sweeping
practices and rainfall characteristics.
Pollutants are often modeled as fractions
of the amount of dust and dirt.
Affects the amount of solids on the
streets which is available for washoff.
Erosion is a source of solids in combined
sewers.
Determines the background quantity and
quality within an area.
Dependent on antecedent dry weather,
sewer system characteristics and rain-
fall characteristics.
Slope, shape and size affect solids
transport and deposition.
51
-------
Historically, quality models have been more difficult to calibrate than
quantity models, for a number of reasons. First, many of the components
of quality listed in Table 5 are not fully understood, or at least are not yet
able to be modeled mathematically in an effective manner. Pollutant
accumulation and washoff, erosion and erosion control, solids deposition
and scour, effects of antecedent dry weather, and the mechanisms of the
"first flush" are all complex phenomena that do not lend themselves readily
to either simple field measurements or mathematical relationships. Second,
a relatively large amount of monitoring is required to develop the necessary
quantity-quality or rainfall-overflow relationships because of the large
variability of CSO and stormwater quality data. Although, not simple tasks,
the hydrologic aspects of rainfall and runoff and the hydraulic aspects of
sewer networks are more susceptible to quantification and calibration with
a smaller number of monitored storms than are the quality aspects. Third,
while quantity lends itself to continuous monitoring, quality does not. More-
over, the parameters of concern in receiving water analysis may be
seasonal in impact, thus reducing the possible monitoring' period. Examples
of seasonal parameters are fecal coliforms, when contact recreation is a
goal, and dissolved oxygen, when fish propagation is desired.
One of the objectives of monitoring overflow quality is to develop
relationships between each drainage area's rainfall and runoff and pollutant
loading rate. Since it is not feasible to collect data at all of the CSO loca-
tions for an unlimited number of storms, representative drainage areas are
monitored for a few storms in the hope of capturing an adequate variety of
rainfall-runoff conditions for projection purposes. The data must be
analyzed statistically to determine if there is any correlation between runoff
quantity and quality. If a good correlation exists, the resulting relation-
ships can be used for predictive purposes with a relatively high degree of
confidence.
On the other, hand, adequate data may not exist to develop any meaning-
ful correlation between runoff quantity and quality. Literature quality data
should only be used as a last resort, and then only for a very rough assess-
ment, since conditions may vary considerably from one local area to
another. It is desirable, at least, to obtain runoff quality data for a small
number of rain events since this local data may reveal atypical runoff
quality. If the data is within the normal or average values for a particular
land use, then the use of literature data can be used with a higher degree of
certainty for preliminary assessments.
Pollutant Characteristics
The actual water quality parameters which should be measured in
stormwater overflow studies will vary and will depend upon the overall
objectives of the analysis, the uses of the receiving water, and the quality
52
-------
aspects of the receiving water, both existing and projected. The next
section of this report discusses modeling of the receiving water in detail.
The modeling of pollutants in sewer systems has not progressed to the
same state of development as hydrologic and hydraulic simulations. Pollu-
tant parameters commonly modeled include BOD, suspended solids,
settleable solids, and coliform bacteria. Nutrients, such as nitrogen and
phosphorus, are also included in some models along with a number of other
conservative substances; when the receiving water is a lake, nutrients can
be particularly important because of their effect on algal growth. A com-
mon mechanism used to model pollutant washoff and subsequent routing
through the sewer system consists of applying certain;fractions, by weight,
of a particular pollutant to the solids characteristic .of the overflow. The
simpler models employ statistical methods based upon land use, population
densities, curb length, average rainfall parameters, and other factors.
Characterizations of the overflow pollutants, along with total and peak
flow rates, are used to design storage and treatment facilities. Solids and
organic loadings and disinfection requirements are all important considera-
tions in the design of such facilities. An important consideration of over-
flow quality is the "first flush". The "first flush" usually contains the
highest concentrations of a number of contaminants and its characterization
is an important factor in abatement planning. It may be feasible to capture
the "first flush" in storage and/or treatment facilities and substantially
reduce the overall pollutant load on the receiving water. The location of
the "first flush" relative to the total overflow will assist in the sizing of
the abatement facility in a cost-effective manner.
CSO MODELING FOR ONONDAGA COUNTY
Figure 2 shows the steps employed in the sewer system modeling
phase of the CSO abatement program for Ononda
-------
I2-T-
10--
8--
3
6--
4-
2 -
OVERFLOW 073
SYRACUSE, NEW YORK
OVERFLOW = 3.50 TOTAL RAINFALL-0.09
0
0.0
NOTES: I. CIRCLED DATA POINTS
REPRESENTS ACTUAL DATA
RECORDED AT MONITORING
SITE.
2. IMG = 3785 m3
3.1 in = 25.4 mm
1.0
TOTAL RAINFALL (IN)
AS MEASURED AT HANCOCK AIRPORT
—I
2.0
Figure 12. Typical correlation between total rainfall and total overflow
for an individual drainage area (6).
54
-------
80-
70-
60-
o
£ 50-
3
_
cc
ID
UJ
£
CO
t
40--
30--
20--
10--
OVERFLOW =48.12 TOTAL
RAINFALL 2i57
0.0
OVERFLOW= 4789 TOTAL RAINFALL - 5.16
NOTES: |. IMG=3785m3
2.1 in =25.4 mm
— LEAST SQUARE FIT OF
MONITORING DATA.
— SIMPLIFIED SWMM
PROJECTIONS
O SWMM (WITH EXTRAN)
PROJECTIONS
1.0 2.0
TOTAL STORM RAINFALL (IN)
Figure 13. Comparison of SWMM and SSWMM output to measured data
for Onondaga County, New York (6).
55
-------
SSWMM was used to predict total system overflow volumes and to
develop and screen preliminary storage and treatment alternatives. The
model was calibrated by a mass balance of the total system overflow
volumes (Figure 13). SSWMM is not capable of predicting peak overflow
rates; hence, only total volume could be used for calibration purposes.
Also, SSWMM could only be operated in the daily mode since a linked
system was employed because of the large size of the Syracuse CSO
network.
The entire CSO system, including the interceptor system, was later
modeled, using detailed SWMM with the EXTRAN option. The predicted
output of SWMM for a small number of independent storms was compared
to both quantitative outputs of SSWMM and the monitored data, as shown
in Figure 13. Although good correlation was found to exist, the detailed
SWMM results showed that surcharging was a significant factor in the over-
flow volumes, thus precluding further SSWMM analyses.
The long-term rainfall records collected by the National Weather
Service at Hancock Field were analyzed in addition to the data gathered
from several rain gages placed in the study area. The long-term data
were compared to the data collected from the study area. It was found that
the local data correlated closely with the 20 years of National Weather
Service data. For use with SSWMM, the 20-year record was reduced to
two years of random rainfall. This random two-year data set was found to
be equivalent to the 20 years of record at Hancock Field after applying a
random selection technique. This process also reduced the operational
costs of SSWMM.
In addition to the development of the random two-year data set, a
synthetic hyetograph was developed for the area, as shown in Figure 11.
This hyetograph was developed for use in detailed SWMM.
To arrive at pollutant loadings, the Onondaga County CSO plan
employed statistical analysis in conjunction with the calibrated quantitative
models of SWMM and SSWMM. Pollutant loadings were expressed as
functions of rainfall through the following equation:
Quality = Cblm
where Quality = loading of pollutant in Ibs (kg)
q
G = geometric mean of pollutant, in Ib/MG (kg/m )
Q
b = intercept of total system overflow in MG (m ) (see
Figure 14)
m = slope of overflow curve (see Figure 14)
I = total rainfall in inches (mm)
56
-------
This relationship was developed for BOD, TSS, NHgN, TIP, OrgN and
UOD. Table 6 shows the geometric mean pollutant concentration for five of
these parameters in relation to two of the area's receiving waters.
TABLE 6. GEOMETRIC MEAN CONCENTRATIONS OF
POLLUTANTS ENTERING ONONDAGA
CREEK AND LEY CREEK FROM COMBINED
SEWER OVERFLOWS
BOD, Ib/MG (kg/m )
TSS, Ib/MG (kg/m3)
NELN, Ib/MG (kg/m3)
3
TIP, Ib/MG (kg/m )
OrgN, Ib/MG (kg/m3)
Onondaga Creek
505 (0.29)
2085 (1.21)
20.7 (0.012)
2.6 (0.0015)
11.5 (0.0067)
Ley Creek
230 (0. 13)
1530 (0.89)
2. 5 (0.0015)
1.2 (0.00070)
6.5 (0.0038)
Loading curves were then developed for varying return period storms,
using the preceding equation and the values contained in Table 6. Figure
14 illustrates a loading curve for BODg discharged into Onondaga Creek.
Once the two models were calibrated, the rainfall and quality data
analyzed, and the loading curves developed, the prediction phase was
initiated. Quantity and quality responses were simulated for a number of
storm conditions. The results of these simulations were entered in the
receiving water quality model to assess water quality impact. From these
analyses, the required levels of abatement were determined. SSWMM was
initially employed to screen the various pollutant removals. Best manage-
ment practices, in addition to structural alternatives, were considered in
the overall abatement planning. Detailed SWMM was used to better define
the predicted volumes, determine the output hydrographs, including peak
flow rates, and to further screen and select abatement alternatives. These
procedures are considered further in the following sections of this report.
57
-------
10'
10"
3
(9
.10
8
-*~
10
RAINFALL (INCHES)
10.0
Figure 14. Total BODs load discharged from the combined sewer system
to Onondaga Creek for storms of various return periods (6).
58
-------
SECTION 8
MATHEMATICAL MODELING OF RECEIVING WATER
It is impractical to design a sampling program with the expectation of
monitoring a storm which borders on contravention of a particular water
quality standard. The probability of monitoring such a storm under critical
environmental conditions is even more remote. Therefore, the determina-
tion of the allowable pollutant discharge under these conditions requires a
predictive tool. Mathematical water-quality models provide such a tool. A
mathematical model is defined as an expression or series of expressions
which describe the cause-and-effect relationships of a real event. Specifi-
cally, water quality models describe the influence of pollutant discharges o
receiving waters. Many receiving water models exist; they may be selecte
or adapted to accommodate wide variations in level of effort.
The type of model employed in predicting pollutant impact on a
receiving water is dependent on several factors: parameters of concern,
time frame, desired accuracy of prediction, and project scope. The first
two factors are interrelated, since the selection of parameter to be modele
generally sets the time frame of the model, an example being the modeling
of dissolved reactive phosphorus. If the desired end product is merely to
predict phosphorus concentrations within a lake, a steady state model can
be employed using annual loads; if the response of algal biomass to
dissolved reactive phosphorus concentrations is desired, however, then a
dynamic model which incorporates algal nutrient uptake on a continuous
basis is required. Figure 15 illustrates parameters as they relate to time.
Figure 16 shows interrelationships between sampling and modeling.
PRELIMINARY APPROACH
The seven parameters shown in Figure 15 are commonly related to
combined sewer overflows and should be considered in all CSO programs. O
these seven, three are commonly modeled for water quality impact; namel;
bacteria, dissolved oxygen, and nutrients. As previously discussed,
nutrients generally can be appraised on a long-term basis (e. g., average
annual load) and a steady state model may be appropriate. For bacteria ar
dissolved oxygen, however, the time frame of concern is much shorter an<
may require a dynamic or time-variable approach. In any event, a simple
59
-------
FLOATABLES
BACTERIA
r
r
T
DISSOLVED OXYGEN
run
SUSPENDED SOLIDS
1
NUTRIENTS
DISSOLVED
ACUTE TOXIC EFFECTS
LONG TERM
TOXIC EFFECT
DAY
MONTH
YEAR
HOUR
WEEK
SEASON
DECADE
Figure 15. Time frame of concern for various water-quality
para meters. (7).
60
-------
0>
T3
§
o
cr
Q.
«4-
O
C
Q.
-------
preliminary approach should be taken initially to determine the severity of
the problem.
Since the worst case of water quality impact involves transport of an
undispersed, potentially critical, pollutant load directly to a recreational
area, this case should be evaluated first. If no adverse impact is found
under these conditions, one will not occur under less severe conditions.
The following steps can be taken to make such a preliminary estimate:
1. Since transport is one of the most important factors affecting
water quality, the selection of a realistic current speed is
essential to water quality predictions. Long-term current
measurements generally are not available for most lakes;
therefore, a current speed for predictive purposes must be
derived from another source. Since the primary force
affecting currents is wind speed and direction, historical
wind records can be analyzed to determine under some pre-
defined criteria a critical wind speed. If a correlation
between wind vectors and current vectors can be made by
monitoring both vectors over a short time period, then the
critical wind speed can be converted to a critical current.
2. Segment the lake so that any actual multi-dimensional flow
can be viewed as a one-dimensional water body (i. e.,
river). As shown in Figure 6, rather than having Cell 1
flow to both Cells 2 and 3, treat the flow as if it hugs the
shore and does not interact with the mid-lake segments.
3. Compute a mass balance for a critical cell (Cell 1, as shown
in Figure 6) using the maximum concentration anticipated
from the combined sewer overflows and the mass transfer
of water throughout the lake associated with the maximum
current velocity determined in Step 1. Since the continuity
equation (Q = AV) applies, this flow can be computed as the
product of the average velocity times the cross-sectional
area of the interface between adjacent cells (e. g., Cells 4
and 1). The concentration of pollutant in a given volume of
water passing through Cell 1 can then be computed as follows:
i- %oad( Cload)
62
-------
where CL
Q
, .
lake
= concentration of pollutant in Cell 1
+ Qload
= flow across the interface between Cells
.. , ,
4 and 1
QI = flow leaving Cell 1 =
CL , = in-lake concentration of pollutant entering
iafce Cell 1
Q = flow associated with combined sewer discharges
C, , = concentration of pollutant associated with
load 1.1 i.
combined sewer discharges
4. In the case of a non- conservative pollutant, an estimate of
reaction rate must be obtained either from the literature or
from special ecosystem studies. For example, if the para-
meter under consideration is bacteria, the literature contains
a wide range of data on bacterial die-off (31). The most
conservative case should be utilized; for bacteria it would be
the minimum die- off rate. Critical conditions are discussed
in greater detail in Section 9.
5. Knowing the area proposed for recreational uses and the
approximate time of travel from the loaded cell (Cell 1) to
this area, the natural reduction in non- conservative pollutant
loads can be computed. In the case of bacterial die-off and
BOD-DO relationships, the first order Chick's Law and
Streeter-Phelps equations can be used, respectively.
Onondaga County
Using Onondaga Lake as an example (see Figure 7), .the current associ-
ated with maximum wind velocity is 0. 26 mph (0. 12 m/sec), the area of the
interface between Cells 1 and 2 is 25, 400 ft2 (2360 m2), and the load asso-
ciated with combined sewer discharges is 2.3 x lO*6 cells per day. The
time of travel between the loaded cell (Cell 1) and Cell 11 is approximately
0. 4 days, and the bacterial die- off rate is 1.16 per day. This die- off rate
was determined through laboratory measurements of Onondaga Lake water.
Using the above step-by-step approach, the concentration expected in the
recreational zone (Cell 11) was calculated to be 5. 4 x 104 cells/100 ml.
Since this preliminary approach yielded concentrations in the recreational
zone exceeding the allowable standards for its classification (200 cells/
100 ml), a more detailed evaluation was required. However, had this initial
procedure yielded concentrations which met standards for the intended best
63
-------
usage of the water, it would have sufficed and resulted in significant cost
savings.
A similar approach generally can be employed for dissolved oxygen
using an expanded form of the Streeter-Phelps equation; in the case of
Onondaga Lake, however, this method was not practical because of the
influence of fluctuating algal biomass which masked the BOD-DO relation-
ships.
DETAILED MODEL
The model (29) developed in this section describes the three-
dimensional transient distribution of fecal coliform, total phosphorus, BOD
and dissolved oxygen in Onondaga Lake following stormwater overflow
events from the City of Syracuse. The model includes advective and dis-
persive transport and mechanisms for settling, scour, and chemical and
biochemical reaction.
Advective transport is the transfer of mass between adjacent volumes
of water as a result of the unidirectional flow, whereas dispersive transport
actually mixes adjacent volumes of water as a result of the mass gradient of
a particular constituent (30). Figure 17 presents a visual distinction
between an advective and an advective-dispersive transport system. For
simplicity, the figure depicts a slug discharge of a non-conservative sub-
stance to a one-dimensional body of water and shows both the slug traveling
intact with the flow (advective transport) and the spreading of the slug as it
proceeds downstream with the flow (advective-dispersive transport). The
simplified "preliminary approach" previously presented treats the system
as if advective were the only transport mechanism, whereas in actuality
the maximum concentration will be further diminished by dispersion.
Figure 17 shows that pollutant concentration is the maximum under a purely
advective condition. The potential sources and sinks of pollutant concentra-
tion are presented in Figure 18.
Most of the material presented under "Detailed Model" has been
quoted or paraphrased directly from Limno-Tech, Inc. , (29).
The model, is based on the conservation of mass as expressed by the
following equation:
Accumulation = Net Input of Mass + Net Input of Mass
of Mass from Advective from Dispersive
Transport Transport
— Sources and Sinks of Mass
64
-------
o
I
z
tu
o
I
- MASS OF ADVECTIVE
ONE DIMENSIONAL SYSTEM
W/FIRST ORDER DECAY
-MASS OF ADVECTIVE-DISPERSIVE
ONE DIMENSIONAL SYSTEM
W/FIRST ORDER DECAY
Mc = CENTROID OF MASS
Xs = SPREAD OF MASS DUE TO DISPERSION
DISTANCE DOWNSTREAM (MILES)
Figure 17. Comparison of the effects on concentration of advective
and advective- dispersive transport.
65
-------
CD
o
O)
i
o>
o
CD
CO
CO
(O
CO
I* to*
"o *
° "55
CO >-
o ex
iSiEi
00
66
-------
or mathematically *£ = v.EvC_v.UC+r
at
where C = concentration of pollutant (mg/1)
t = time (days)
o
E = dispersion coefficient (m /day)
U = velocity (m/day)
r = reaction sources or sinks (mg/1/day)
Since a direct solution of this equation is usually not possible, the
solution requires the use of approximate solutions. The finite difference
solution, which is equivalent to representing the continuous lake system as
a collection of completely mixed cells that are interconnected by advective
and dispersive transport, was used in this model. This approach does
sacrifice some spacial accxiracy; however, the careful segmentation of the
water body to closely represent the overall circulation and mixing minimizes
these inaccuracies. Consistent with the above approach, the model treats
the lake as two vertical layers with 21 interconnected horizontal segments
in the epilimnion and six interconnected horizontal segments in the hypo-
limnion (Figure 7). Measurements from U. S. Lake Survey Chart No. 180
were analyzed and used to define the morphology of Onondaga Lake and to
establish the dimensions of each model cell.
The model utilizes steady-state circulation patterns and hydraulic
loads and time-variable pollutant loads to simulate the lake's water quality
response to stormwater discharges. The model linearly interpolates the
transient pollutant loadings from tabular input data to formulate the polluto-
graphs discharging to each model cell. Time-varying changes in flow
associated with stormwater input are normally small in comparison with
dispersive and advective flow of the lake circulation pattern; therefore, the
use of steady-state hydraulic loads was considered applicable. The assump-
tion of steady-state in-lake hydraulics may, however, be invalid for major
changes in lake circulation patterns caused by fluctuations in wind during
the period of simulation. These variations in lake hydraulics can be roughly
simulated by using stepwise changes in steady-state hydraulic patterns
whenever a marked change in current is required.
Fecal Coliform Model Kinetics
In addition to the simulation of the transport of bacteria (fecal
coliform) throughout the lake, the model accounts for the loss of organisms
due to die-off and settling. The die-off term encompasses the reduction of
bacterial numbers as a result of protozoan predation, sunlight disinfection,
natural death and respiration. Regrowth of fecal coliform in Onondaga
67
-------
Lake was assumed negligible. The overall die-away rate is represented in
the model by Chick's Law:
rf = kf(N)
where rf = rate of fecal coliform die-away (cells/1/day)
k = die-away coefficient (day"1)
N = fecal coliform concentration (cells/1)
The rate of bacterial die-off is dependent on temperature, with the
organisms being more persistent at lower temperatures. The die-off rate
for coliform bacteria is considered linearly dependent on temperature
according to the following equation (32):
k = kf (0.31 + 0. 0345T)
T 20
where kf = coliform die-off rate at temperature T (day"1)
T
kf = coliform die-off rate at 20°C (day"1)
20
T = temperature ( °C)
The loss of bacteria from the water column due to settling is des-
cribed in the following expression:
VSF „
where rf = rate of loss due to settling (cells/1/day)
s
D = depth (m)
VSP = settling velocity of fecal bacteria (m/day)
N - bacteria concentration (cells/I)
Total Phosphorus Model Kinetics
Total phosphorus was considered to be a relatively conservative sub-
stance in which net changes in concentration were limited to settling,
sediment release or bottom scour. While only particulate forms of phos-
phorus settle, the non-settleable soluble organic and soluble reactive forms
are converted to settleable material through biochemical processes such as
zooplankton grazing and fish predation, and chemical conversion (appatites).
68
-------
A model which individually describes these complex interactions was well
beyond the scope of the Onondaga project; therefore, the model uses a net
effective settling velocity for total phosphorus which incorporates the
overall removal rate in one term. The following simple kinetic expression
is used to describe the rate of phosphorus loss to the sediments and
sediment release of phosphorus:
r = -(VSP/D)P + SBP/D
P
where r = overall kinetic rate for total phosphorus (mg/l/day)
P
VSP •= average settling velocity (m/day)
D = depth (m)
P = total phosphorus concentration (mg/1)
SBP = sediment release of phosphorus caused by anaerobic
conditions or scour (g/m^/day)
Dissolved Oxygen Model Kinetics
The dissolved oxygen model includes kinetic expressions describing
reaeration, oxidation and settling of carbonaceous and nitrogenous oxygen
demand material (CBOD and NBOD), algal photosynthesis and respiration,
and sediment oxygen demand (SOD). The following paragraphs describe the
model kinetics for each of these processes.
Removal Rates for Organic Components —
CBOD occurs in both the dissolved- and particulate forms and the
appearance of this material is governed by mechanisms such as settling,
sediment scour, stripping, adsorption and biological oxidation. The rate of
bacterial oxidation of organic carbon has been traditionally described by
first-order kinetic expressions. The model uses such expressions,
modified by a non-linear decay coefficient, to account for inhibition of bio-
logical oxidation at low dissolved oxygen concentrations (33). In addition to
biochemical decay, a physical removal mechanism for settling is used in
the model. The overall kinetic term for CBOD is:
= (kc*+VSC/D)CBOD
69
-------
where
kc
VSC
D
CBOD
= overall removal rate for CBOD (mg/l/day)
= non-linear biological oxidation decay coefficient (day-1)
= settling velocity of CBOD (m/day)
= depth (m)
= carbonaceous biological oxidation demand (mg/1)
NBOD is exerted as a consequence of the bacterial oxidation of
ammonia to nitrite to nitrate. Municipal and many industrial wastes contain
nitrogenous material in both the organic and ammonia forms, the organic
compound of which must be hydrolyzed to ammonia before biochemical oxi-
dation can occur . Autotr ophic oxidation of ammonia to nitrite and nitrate has
been described by first-order kinetic rate laws. The Onondaga Lake model
uses a first-order expression for nitrification with a non-linear modification
to account for inhibition of nitrification at low levels of dissolved oxygen
(34). The model also uses a first-order expression to describe the settling
of nitrogenous material. Both the organic and ammonia forms of nitrogen
are present as dissolved and particulate matter. The particulate material,
especially organic nitrogen, can settle before it has a direct impact on the
lake's dissolved oxygen resources, although ultimately the material will
exert a benthal demand. The overall kinetic expression for NBOD is given
by the following equation:
NBOD
VSN/D) NBOD
where
= overall removal rate for NBOD (mg/l/day)
= non-linear nitrification decay coefficient (day"1)
= settling velocity of NBOD (m/day)
NBOD = nitrogenous biochemical oxygen demand (mg/1)
^
VSN
Biological Deoxygenation —
While the overall removal rate for the organic constituents (CBOD
and NBOD) results from a combination of physical (e. g. , settling) and bio-
chemical (e.g., bacterial utilization) phenomena, only the biochemical
changes have a direct effect on the dissolved oxygen. The settling
mechanism indirectly affects the DO by removing oxidizable material from
the water column. The oxidizable organic material deposited on the lake
bottom ultimately exerts a benthal oxygen demand.
The total deoxygenation rate (r ,) due to the decay of organic matter
remaining in the water column is represented by the following equation:
70
-------
_ r, = kp (CBOD) + k (NBOD)
^•* *i- •*-'o>
"I* *(-
Benthal Oxygen Demand--
The oxygen depletion rate by organic sediments is considered constant
(zero order) with inhibition at low dissolved oxygen concentration. The rate
of oxygen demand due to benthal deposits, B, is defined by the following
equation:
where r, = oxygen uptake due to benthal demand (mg/l/day)
B* = non-linear benthal oxygen demand (g/m2/day)
D = depth (m)
Reaeration —
Dissolved oxygen has limited solubility in water and its presence in
adequate concentrations is critical to the preservation of healthy aerobic
aquatic communities. Oxygen solubility in water is dependent mainly upon
the partial pressure of atmospheric oxygen, temperature and the concen-
tration of dissolved and suspended impurities. The dissolved oxygen satu-
ration concentration is dependent on the temperature and salinity of the
water (35).
Any gradient between the oxygen saturation concentration and the
actual in-lake oxygen concentration causes the natural forces of gas
transfer to act to restore the lake concentration to the equilibrium level.
This rate of restoration is called reaeration,
matically as:
r , and is described mathe-
r = k (DOQ-DO)
a a o
where r
'a
DO
= rate of reaeration (mg/l/day)
= reaer-ation rate coefficient (day" )
= dissolved oxygen concentration at saturation (mg/1)
= in-lake dissolved oxygen concentration (mg/1)
Wind action on the water surface is primarily responsible for reaera-
tion in lake systems. Figure 19 describes the atmospheric reaeration
coefficient as a function of wind speed and mixed-layer depth (36).
71
-------
CD
ro
TD"
o>
o>
o.
U)
•o
c
c
o
(O
o
o
0>
o
0>
a:
22
0)
o>
iZ
o
72
-------
Temperature Corrections —
The oxidation rates for the various mechanisms are dependent on tem-
perature and correctable using the modified Arrenhius equation. The basic
equation is as follows:
where kT = corrected rate coefficient at temperature T (day"1)
k2Q = rate coefficient at 20° C (day'1)
e = rate correction factor dependent on parameter
T = temperature (°C)
The values of e used in this model are presented in Table 7.
TABLE 7. TEMPERATURE CORRECTION FACTORS (31)
_ __ Rate Coefficient^ _________ 9
Carbonaceous Oxygen Demand 1.047
Nitrogenous Oxygen Demand 1.09
Sediment Oxygen Demand 1. 065
Reaeration 1. 024
Net Productivity —
Planktonic algae and rooted macrophytes have both a beneficial and
detrimental effect on the oxygen resources of natural water bodies. During
periods of sunlight, under favorable nutrient and temperature conditions,
macrophytes and algae fix carbon dioxide and produce oxygen through the
process of photosynthesis. However, these plants must also constantly
oxidize organic matter for energy through the process of respiration. Con-
sequently, during prolonged periods of darkness, algal or macrophyte
respiration can seriously deplete the oxygen resources of a lake. While a
sophisticated phytoplankton model with nutrient cycles would be very useful
for understanding water quality dynamics, the development of this type of
model was well beyond the scope of the Onondaga Lake study.
The model does, however, compute the effect of algae on the lake's
dissolved oxygen budget. Algal photosynthesis oxygen production is calcu-
lated using chlorophyll a, nutrient and incident light measurements
according to the following equation (30, 37):
73
-------
photo
= Y(Gp, L> (Gp, N> (G
-------
MODEL CALIBRATION AND VERIFICATION
The interaction of data collection and mathematical modeling phases of
any project occur during the model calibration and verification. While
establishment of the general, framework of the various sub-models is based
on widely-accepted relationships, the particular forcing functions, kinetics
and system responses applicable to each body of water must be formulated
from the data collected during real storms. The model calibration for
Onondaga Lake, for example, was completed initially in the steady-state
mode using average yearly pollutant concentrations. Calibration was
further refined in the dynamic state using a single storm with minimal
interferences. Once the variables were determined from the calibration
phase, the model was then verified through data collected during three
separate phases.
The lake model requires detailed data input for hydrology, influent
loads, circulation, dispersive transport, reaction coefficients, and environ-
mental conditions. Each of these inputs must be specified for every model
simulation and projection. The following paragraphs describe each model
input required and their source in the case of 0nondaga Lake.
Hydrology, Influent Loads, Circulation, Dispersive Transport
Basic to the model is quantification of influent flow and waste loads
and measurement of in-lake transport. In the case of Onondaga Lake, all
major tributaries were gaged by the U. S. Geological Survey and records
were available for all storms. Since the model can only accommodate
steady-state hydraulics, an average flow for the entire storm is all that is
required. If no stream gages are available, a temporary gage can be
established in all major tributaries at a convenient cross section. Periodic
measurements with a velocity meter over a range of flows will establish a
stage-discharge curve for the particular cross section. Field personnel
are thus enabled to obtain flow during the storm by periodically measuring
depth; the frequency of measurement will depend on the response time of
the particular drainage basin. Pollutant loads can be computed using
measured flows and associated concentrations determined through storm
sampling.
The in-lake transport can be more critical than the hydraulic load,
depending on the morphology of the lake. The circulation of Onondaga Lake
during each storm sampling period was determined using velocity measure-
ments taken in both the epilimnion and hypolimnion at the various stations.
The individual velocity measurements were then time-averaged to provide
an overall velocity at each station; from these velocities the appropriate in-
lake circulation patterns (gyre) were formulated. Figure 20 shows a plot of
75
-------
CO
o
76
-------
individual time-weighted velocities (both epilimnion and hypolimnion) from
which the circulation patterns were determined. While this pattern, which
corresponds to the advective transport within the lake, can be measured
directly, the phenomenon of dispersion must be estimated. Initial estimates
of dispersion were based on literature values and then refined through model
calibration. The dispersion coefficients used in the model are summarized
in Table 8.
TABLE 8. DISPERSION COEFFICIENTS (29)
Location
Epilimnetic near shore
Epilimnetic mid-lake
Epilimnetic basin interface
Hypolimnetic horizontal
Hypolimnetic-epilimnetic
vertical
Range of
Dispersion Coefficients
(cm2/sec)
2.9 - 4. 6 x 103
3.5 - 6.0 x 10
4.1 - 8.1 x 104
8.1 - 16.2 x 104
5.0 x 10
-1
Reaction Coefficients
Model reaction and settling coefficients were determined by a survey
of the literature (31, 38, 39), laboratory measurements (40), and model
calibration (29). The coefficients utilized in the Onondaga Lake study are
summarized in Tables 8, 9 and 10. As an initial approximation of the
deoxygenation coefficient for both carbonaceous and nitrogenous demanding
material, analysis of laboratory data was used (40). The reaction rate
determined from the laboratory studies proved to far exceed the reaction
rates finally determined through calibration of the model. It may have been
the result of low dissolved oxygen or the high chemical concentration in the
lake, which can limit the biological activity. As previously discussed, the
reaeration coefficients were determined as a function of wind speed, based
on literature values (36). The relationship between wind speed and reaera-
tion coefficient is presented in Figure 19. The mixed-layer depth in
Onondaga Lake during the summer months is 9 meters, and this value
was used throughout the study.
The benthal oxygen demand can be measured either through oxygen
uptake studies of undisturbed sediment cores or by constructing an oxygen
uptake chamber on the lake floor. Due to the difficulty of constructing a
chamber, the results of laboratory studies conducted previously (41) were
. 77
-------
TABLE 9. REACTION COEFFICIENTS AND SINKING RATES
VERIFIED THROUGH MODEL CALIBRATION (29)
Parameter
Fecal Coliform (F)
Phosphorus (P)
CBOD (C)
NBOD (N)
Reaction
Coefficient
kx(day-1f
1.16
--
0. 10
0. 0115
Near
0.
0.
0.
0.
Sinking
Shore
0
39
98
0
Rate - VSX
Mid-Lake
1. 97
2. 95
3. 28
1. 67
(ft /day)
Hypolimnion
0.0
0. 144
0. 98
0. 049
*base e @ 20°C
X - designated specific parameter
TABLE 10. DISSOLVED OXYGEN REACTION COEFFICIENTS (29)
Symbol
Y
K
•N
K-,
Coefficient Designation
Reaeration Coefficient
Chi a.
Algal Light Saturation Constant
Saturation Constant - Inorganic
Nitrogen
Saturation Constant - Soluble
Reactive Phosphorus
Reaction Coefficient
*
266 mgO2/mg Chi a
240 Langleys/day
0.015 mgN/1
0. 0025 mg P/l
* Varies with wind speed; see Figure 19.
78
-------
utilized. The benthal demand utilized was 0, 5 mgO2/m /day. This value
is extremely low when compared to other polluted systems (.31), but may be
associated with the high proportion of inorganics identified in the benthos
(41).
Subsequent to the sampling program of this project, a substantial
flocculent sediment layer was measured in Onondaga Lake which may exert
a greater oxygen demand than the compacted layer sampled with the corer.
The oxygen demand of this flocculent layer is currently being investigated.
Environmental Conditions
The environmental factors utilized in computing and correcting
reaction coefficients and computing photosynthetic oxygen production and
algal respiration for the various model runs are presented in Tables 7 and
10. Wind data, water temperature, secc'hi disc and algal nutrient concen-
trations were derived from the monitoring data, and incident light was
determined from long-term regional data. However, the incident light was
corrected for the percent cloud cover reported by the National Weather
Service at Hancock International Airport in Syracuse. Model input of
chlorophyll a was derived from limited data and had to be interpolated to
provide sufficient temporal data. The data utilized for pre-storm conditions
were derived from measurements made during the storm survey.
The calibration process for both the fecal coliform and dissolved
oxygen models is outlined in Figures 21 and 22. In order to demonstrate
impact, the Onondaga Lake verification plots of the most severe storm loads
are presented in Figures 23 through 27. In these figures, the curve pre-
dicted by the calibrated water quality model is superimposed on the actual
measured data points. Similar fits of model predictions to measured
results were obtained for the other three storms monitored, which varied
greatly in intensity.
Sensitivity Analysis
In order to determine the relative effects of the various parameters
on water quality, sensitivity analyses should be performed on several
variables. These analyses should be performed by varying several critical
parameters one at a time and observing the change in impact of the model
output. The parameters for which these analyses should be performed are:
all rate coefficients (i. e., carbonaceous and nitrogenous oxidation coeffi-
cients, reaeration coefficients, fecal coliform die-away rates), benthal
uptake and algal productivity in the case of DO, and sediment scour and
release in the case of nutrients. In general, a 25 to 50 percent plus or
79
-------
MEASURED
DRY WEATHER
COLIFORM
LOADS
MEASURED
ECOSYSTEM
RESPONSE
(MER)
DYNAMIC
FECAL
COLIFORM
MODEL
PREDICTED
MODEL
RESPONSE
(PMR)
MEASURED
WET WEATHER
COLIFORM
LOADS
TRANSPORT
SEDIMENTATION
COLIFORM
DIE-OFF
(LABORATORY)
ADJUST
MODEL
COEFFICIENTS
CALIBRATED
FECAL
COLIFORM
MODEL
Figure 21. Development of calibrated fecal coliform model.
80
-------
TIME VARIABLE
TRIBUTARY
FLOW
MONITORING
TIME VARIABLE
WATER QUALITY
ANALYSIS
DYNAMIC
PHOSPHORUS
MODEL
[TjONG TERM |
t PRODUCTIVITY '
' MODEL
ALLOCTHONOUS
(EXTERNAL)
ORGANIC INPUT
ALGAL
DIE-OFF
I
STANDING CROP
(CHLOROPHYLL a,
CARBON-14)
AUTOCTHONOUS
(INTERNAL)
ORGANIC INPUT
DYNAMIC
DISSOLVED
OXYGEN
MODEL
PREDICTED
MODEL
RESPONSE
(PMR)
PHOTOSYNTHESIS
ft RESPIRATION
BENTHAL
DEMANDS
TRANSPORT
SEDIMENTATION
ORGANIC
DECAY
RE AERATION
MEASURED
ECOSYSTEM
RESPONSE
(MER)
NO
ADJUST
MODEL
COEFFICIENTS
CALIBRATED
DISSOLVED
OXYGEN
MODEL
APPROACH TAKEN FOR THIS
PROJECT
DESIREABLE ADDITION TO
APPROACH (ESPECIALLY IN
HIGHLY PRODUCTIVE WATERS)
Figure 22. Development of calibrated dissolved-oxygen model.
81
-------
eot" <&t oo'f*
OH OOI/«) £101
' "7 ' i
/ i
* / i
j i
J i
7 i
••/ 1
^ • 1
oil1
8
8
C
8
<
'
B:
ti
8
9 " os-e. . co-f
CM
we ost OBT
(TM OOI/«)£Kn
eot
3
O
"o
E
c.
o
c:
8
E:
o
It
"o
o
§
o>
O
o>
1 E
i "s
LU Q
" tf)
0 o
i &
a:
00
OH MI/-) an
13
O>
E!
82
-------
•
1
-' i
1
1
V
c
8
§
O
B1
ci
OO'OZ
OO'OI
1/9H
'
i
•
•
•
nf
I
I
i
I
1
• \ 1
' \
I
. \ .
E
8
or
33
CD
I
8
OO'OZ OO'OI OO'rf"
1/3H
OO'tf
00'0°
<
a:
LU
O
1
a
1
a:
o
m
O
"S
0>
O
O
CO
O
TJ
tt>
(O
O
0)
O
.52
o
Q.
o
o
S
CP
83
-------
8
if- **
M
M
i
•
1
1
1"
1 »
t
1"
1
"•"
E
8
!
1
8
co-s,
i
co-si os't oo-(f oo-s
•
1
•
I
1 '
J
I
|
j-
1.
i"
i
1
H
co-.f
co-s
"
M
J
1 •
1
1
1 .
, CSV. 00'
M '
1
, .
'
• .
1
.
8
8
8
'
R
8
(P
8
8
C
8
i
i
R
8
8
, , . o
•
»
J
-p?
, •
1
I"
j,
4
1
1,
**
i
i
8:
B
8
: os-i ' 00-0°
1/3H
>l
H
••
_—4_^
1 •
1
1
'1
1
1
i-
1
a
1
1
J.-*
, OS't 00 '(
1/3W
OS't 00
I .?
OO'SJ
OS'i
n/sw
84
-------
OO'OI
w*i
00 'Ot
OO'OI
000°
!
\
•"
C
§
g
831
B
8
oo'o, OO;SH oo'«f
'
^> |
.Si
J 1
S |
• S '
^^ i
y 1
^X^ |
. i - .
8
8
S
8
5
«S
B
8
oo-m ^ oo-o-
8.
00'
\
1
i
I
1
i
.1
i'
__^^ Jt
S
B
§
|
K^"
a
8
51 ' «& M-
0>
u_
OO'OI
85
-------
1
.
•
1
i
1
.1
1
T
."I
i .
•1
I
os'e sz'e oo*
T/OH
OS'
H
0 B'»
•1
1
1
00
OS'O
1
"
"
"1
I
1
1
1
jf
•J
I
.
•
H
1
i , .
i
!•
ii
,1
i
i
'0 S2'« 00'
V3H
.
•
OS'
1
"
1
*
0 SZ'O
1/3W
1
1
1
1
. 1.
•1
[•
1
1'
1
00'
8
8
E
8
§
S'
„
B
8
0°' OS'
1 " II
S
0 SZ'O
1/OW
•
i
1.
1
00
S
8
G
§
R
R
8
0°'
OS'O
SZ'O
VOH
8
11
•
M
1
1
1-
•J
1
1
8
&
8
!
s'
8
0 SZ'O OO'O"
I/OH
•
•
1
1
1
•1.
•i
f\
,r
i.
i
0 SZ'O O0'«
1/OW
•1
"l
r
•1
[1
1
0 SZ'O 00"
0>
73
O
,o
I
o
Q.
10
O
OS'
" •' " 'T'"
•
0 SZ'O
1/01
,
f
,
J
.
•4
*
* '
V
00'
«
"
£
8
*"
§
O
sf
8
e
s
g
&
8
O
cc
1-
—p
LU
O
§
Q
0
ce
o
•
TD
e
3
to
0
O)
E
«^
o
§
to
^
O
Q.
E
o
O
•
^«
^~»
Sj
^^
•
"5
Q.
1
CVJ
v.
3
O>
L
86
-------
minus change in these rates or coefficients should be sufficient to determine
the sensitivity of the various parameters.
Tables 11 and 12 are presented as examples of the sensitivity analyses
performed on the parameters for Onondaga Lake. These analyses were
performed using the steady-state model since this mode of operation saved
considerably in computer time while still yielding usable results.
87
-------
^ int— CMCM
^ co CM in • • • • • •
T-l
OO ^t1^ COOO COCO OCM
o oo CMin . . . . ..
T-(T-| . . ^FCO CM'vt1 t— CM
CM
C!
0
a a a d a SP a
^a.° c! . 2 a. 2 a° ^>aO
«2 2 'a ra 2 'a 2 "^ 2 '^ O 2 'a
^ 'a g g 'a g 'a g 'a g -u 'a g
rfQ.>i racut^a f-\Q4i>^ i-^Dii>3 mG.>-5
OHM ^WM §WW gHW 3 H HI
fe CM PQ EH P
o
o
ra
9
o
II
ra
s
I
o
o
0
fc
-X-
ra
§
|
§
cd
0
o
ra
ra
a
o
oi
ra
0
.
§7
o a
0 Cti
0
"5
88
-------
TABLE 12. SENSITIVITY ANALYSIS OF PRODUCTIVITY*
ON IN-LAKE DISSOLVED OXYGEN
Dissolved Oxygen (mg/1)
Epillmnion Hypolimnion
Measured
(1975 Average Yearly)
Calibrated Model
(Including Average
Yearly Productivity)
Net Productivity
Increased by 25%
Net Productivity
Decreased by 25%
7.0
8.2
9.7
6.9
2.2
1.2
2.7
0.0
*Using steady state model (29)
89
-------
SECTION 9
MODEL PROJECTIONS
The ultimate goal of studies to determine the impact of waste dis-
charges on. a receiving water is to predict the waste load that it can
assimilate without violation of water quality standards. The impact of a
sufficient number of different loads should be determined so that a loading
curve such as shown in Figure 28 can be adequately defined. Formulation
of the loading curve is generally facilitated by use of the water quality
model to project receiving water response to approximately four different
levels of pollution. When determining allowable wet weather discharges,
the trial levels should include the dry-weather load alone and the dry-
weather load plus successively larger wet-weather loads until the receiving
water standard is violated. The allowable wet-weather load can then be
identified. Since the wet-weather pollutant load is comprised of combined
sewer overflows, urban stormwater and non-urban runoff, an estimate of
the latter two components is necessary before the degree of treatment
required for the combined sewer overflows can be determined.
Figure 29 presents the elements required to develop the loading
curve by using the calibrated water quality model. The next two sub-
sections deal with the selection of critical input conditions for the model,
including critical transport and environmental conditions, with the projec-
tion of both dry-and wet-weather pollutant loads for the model.
CRITICAL INPUT CONDITIONS
Prior to projecting the impact of combined sewer discharges on a
receiving water, the critical environmental conditions must be determined
for use in the water quality model. These conditions include background
water quality, wind speed, incident light, and temperature. Depending on
the parameter of concern, the environmental conditions that are critical
vary considerably. Table 13 outlines the critical imput conditions required
for several models. Even though the same input factors may be critical
for two or more water quality parameters, they can act on those param-
eters in a different manner. For example, fecal coliform impact on water
quality is most severe under conditions of low temperature causing mini-
mal die-off and high wind speed resulting in rapid transport to a sensitive
90
-------
6
Ul
z
Ul
§
a
LLJ
PRESENT DRY WEATHER
TOTAL OXYGEN DEMAND DISCHARGED TO LAKE (Ibs/day)
Figure 28. A typical loading curve relating pollutant load to water
quality response.
91
-------
FUTURE
DRY WEATHER
LOAD
FUTURE
WET WEATHER
LOAD
SELECTED
CRITICAL
TRANSPORT
CALIBRATED
DYNAMIC
MODEL
SELECTED
CRITICAL
ENVIRONMENTAL
CONDITIONS
PREDICTED
WATER
QUALITY
IMPACT
IS
INPUT LOAD
ICH EXACTLY
MEETS W.Q. STD.
DEFINED
WATER QUALITY (WQ)
STD. DICTATED
BY "BEST USAGE"
OF WATERS
Figure 29. General procedure for developing loading curve and
determining allowable load.
92
-------
TABLE 13. CRITICAL ENVIRONMENTAL CONDITIONS
FOR MODEL PROJECTIONS
Critical Input
Condition
Wind Speed
Temperature
Light Intensity
Tributary Flow
Tributary Load
Algal Standing Crop
Initial Concentration
in Each Cell of the
Receiving Water
Type of Water Quality Model
Fecal Dissolved Total
Coliform Oxygen Phosphorus
Yes
Yes
No*
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Yes
Yes
*Included only indirectly since light intensity affects die-off
rate.
**Phosphorus and productivity models are interrelated;
however, in this case, phosphorus was considered to be
conservative and any influence of algal activity is
reflected in the net settling rate.
93
-------
area of the lake, such as a bathing beach; the dissolved oxygen parameter,
on the other hand, is most adversely affected by high temperatures when
biochemical rates are maximum and at low wind speeds when reaeration is
minimal.
Critical water temperature depends on the locale as well as the para-
meter of concern. Information regarding critical water temperatures can
generally be obtained from the state's environmental quality section or from
the EPA regional office. It should be remembered that for some
parameters, low temperatures are critical, but for others, higher tempera-
tures will have a greater water quality impact. Since most regulatory
agencies stipulate a critical temperature for BOD-DO relationships but do
not have a critical temperature for bacterial die-off, some monitoring data
may be necessary for modeling bacterial parameters.
In predicting the dissolved oxygen content of productive lakes, light
conditions and the algal standing crop can have a dramatic effect and can
even dominate the DO picture. Historical light data can generally be
obtained from the nearest National Weather Service station.
Lake transport is dictated predominantly by wind speed (42). Since
historical records are generally not available for lake circulation patterns,
the flows to be used in the predictive water quality model must be deter-
mined through analysis of wind speed and direction. These data are
generally available from the local National Weather Service station or from
the NOAA office in Asheville, North Carolina. If the National Climatic
Center does not have a station in the community being studied, the national
office will provide a list of nearby stations which can provide such data for
purposes of approximation.
Onondaga County
Water temperatures used in the predictive model for Onondaga Lake
were selected on the basis of eight years of biweekly measurements (43, 44).
The contact recreation period was considered to extend from June through
September and a realistic low temperature of 20°;C(68°F) was chosen. For
the dissolved oxygen model, a critical temperature of 25° C (77°F) was
selected. Light conditions for the summer months were determined from
regional data (45) and corrected by the use of daily percent-cloud-cover
measurements from the municipal airport.
Current measurements taken during five storm periods in the summer
of 1976 indicated that in-lake circulation patterns (gyres) could be either
clockwise or counterclockwise. Based on this information, a circulation
pattern was chosen for the predictive runs which provided the most direct
transport to the lake shore area of concern. Since the proposed contact
94
-------
recreation area (Figure 6) is located along a northeastern shore, a pattern
of counterclockwise gyres in each basin was selected for transport in the
bacterial model. Since no one area of the lake is more critical than any
other with regard to dissolved oxygen, the gyre selected for the DO model
was immaterial. The flow pattern in both cases was directed through the
near-shore model cells (Figure 7), with only minor interaction with the mid-
lake area. This procedure provided minimum dilution and maintained the
wet-weather pollutant slug intact.
The in-lake velocity selected was based on data gathered during the
storm monitoring program (46). Ideally, a correlation between wind speed
and direction and current movement can be developed, and the critical
current velocity can be based on historical wind data. However, despite
measurements at ten stations twice daily during each storm, a predictable
relationship was not evident. The critical velocities utilized, therefore,
were taken directly from the measured data. The values used for the fecal
coliform and dissolved oxygen models were the maximum observed of 6. 3
mi/day (11. 8 cm/sec) and the minimum observed of 1. 13 mi/day (2.1 cm/
sec), respectively.
For the predictive bacterial model runs, the dispersion coefficients
were decreased so that the peak of the bacterial slug was minimally reduced
by dispersion. Thus, reduction of peak coliform concentrations in the pre-
dictive model runs were primarily the result of die-off. These transport
conditions produced maximum coliform impact in the epilimnion of the
contact recreation zone.
Wind speed is an integral part of the dissolved oxygen model since it
is the primary driving force of reaeration. The average monthly minimum
seven-consecutive-day wind speed was used in the dissolved oxygen model.
It was used instead of a more severe condition because the probability of
occurrence of alow frequency storm (e.g., one-year, two-hour storm)
coincidentally with this wind velocity was more realistic. For Onondaga
Lake, the value of 3. 9 mph (6. 4 km/hr) was determined from analysis of
six years of summer wind data.
PROJECTED POLLUTANT LOAD
The total projected pollutant load for storm-related water-quality
studies consists of a dry- and a wet-weather component. The dry-weather
component, comprised of both point and non-point sources, can generally be
expressed as a continuous steady-state function. The wet-weather component
is generally dynamic; however, it may be represented satisfactorily in
certain cases as a series of step functions. Representing the total load in
this approximate manner can save considerably in computer time and may
permit a simplified sewer system model to be used to estimate the load.
95
-------
The dry-weather component of the load is comprised of municipal and
industrial discharges and the contribution of tributaries that is not directly
associated with rainfall-runoff. The quantity and quality of discharges from
municipal and industrial wastewater treatment facilities can be determined
from information:
1. Stored in the data storage and retrieval system of the EPA
(STORE T);
2. Submitted by the discharger as part of the self-monitoring
program of the National (or State) Pollutant Discharge
Elimination System (NPDES or SPDES);
3. Collected during independent monitoring programs conducted
by regulatory or other governmental agencies or by educa-
tional institutions;
4. Published by local or state planning agencies concerned
with the area's growth over the design period.
The quantity of pollutants discharged to a lake by tributaries can be
determined from:
1. STORE T data;
2. Monitoring programs conducted under Section 303e of PL
92-500, which are available from the state's water quality
section or EPA;
3. Independent monitoring programs.
Sources 1 and 3 above could potentially provide enough long-term historical
data to determine a dry-weather loaji statistically. The 30 3 e Studies
generally do not provide long-term data but can provide usable quantity and
quality information under dry-weather conditions. Flow records for gaged
tributaries can be obtained from the local office of the United States
Geological Survey (USGS), the U. S. Army Corps of Engineers, or the
state's department of environmental protection. If quantity and quality
information is not available for a particular tributary, records from a
similar stream may be used if corrections are made for drainage area and
land use.
The wet-weather component of the pollutant load is comprised of con-
tributions from upland and non-urban runoff, urban stormwater, and dis-
charges from combined sewer overflows. The quantification of each of
these pollutant sources is required if the degree of treatment required for
96
-------
the CSO system is to be determined; however, complete projection of wet-
weather loads for these sources requires costly predictive modeling of each
component. Such modeling may be unnecessary and wasteful, depending on
both the relative magnitude of the discharges and whether the pollutographs
of the various elements superimpose on each other (7)- These two factors
are related to the size of the drainage areas, location of the urban center
within the drainage area, and overall response time of the various sources.
A pragmatic approach for determining the significance of the various
wet-weather components is to first evaluate the relative magnitudes of the
sources through an expanded monitoring program conducted simultaneously
with the monitoring of the CSO system. The sampling program requires
that quantity and quality measurements be made on all tributaries above and
below the urban area. The upstream sample identifies the upland runoff;
the difference between it and the downstream sample quantifies the urban
component. To distinguish between the CSO and stormwater portions of the
urban component, the load from the CSO system must be determined. For
each storm monitored, the data from the monitored areas can be applied to
the unmonitored areas so as to project an overall CSO load. Using this
method, the contributions of the various wet-weather sources can be com-
pared for relative impact.
If the contributions from either the urban stormwater or rural runoff
prove to be significant for a given parameter, projection of this component
for the design storm condition will be necessary. Since mathematical
modeling of these components is generally beyond the scope of CSO studies,
estimates must be made on the basis of measured data. The estimates are
generally based on data collected during the CSO study, since wet-weather
information is rarely available from other sources. Any historical data is
generally skewed toward dry-weather information. As a first trial, the
most severe upland load actually measured should be used atid'superimposed
on the projected CSO load.
Onondaga County
Onondaga Lake is unique in that the County of Onondaga has conducted
a yearly monitoring program (43, 44) in addition to conducting a compre-
hensive baseline study (41). The data collected during these two programs
made possible the selection of truly representative dry-weather tributary
waste loads. The number of years of record used in determining these
background loads depends on when and if major changes within the particular
drainage basin have occurred. Examples of such changes in Onondaga
County were abandonment of a landfill site, adoption of a more effective
sewer maintenance program, modification of treatment facilities, and the
state-wide banning of detergents containing phosphorus. Depending on the
97
-------
tributary, anywhere from two to six years of biweekly data were analyzed
to arrive at the average-summer dry-weather influent load. The discharge
load selected for the Metropolitan Wastewater Treatment Facility corres-
ponded to the design efficiency of the plant now under construction.
Table 14 presents the dry-weather discharges received by Onondaga
Lake. Since the times of concentration for the drainage basins tributary to
Onondaga Lake far exceeded the response time of the CSO system, the up-
land non-point component was ignored and pollutant loads generated during
storms were taken as the sum of the dry-weather .and CSO loads. Since only
a small percentage of the urban area has separate storm and sanitary
sewers, urban stormwater was also ignored.
Table 15 presents the wet-weather pollutant loads for the one-, two-
and ten-year storms; they were superimposed on the dry-weather loads
presented in Table 14.
LAKE MODEL PROJECTIONS
The useful end-products of water quality models, including lake
models, are primarily their cause-and-effect projections relating improve-
ments in receiving water quality to proposed or contemplated changes in
pollutant loads. The remainder of this section will be devoted to examining
the products of the Onondaga Lake modeling as examples of such projec-
tions.
Results - Onondaga Lake
In-lake fecal coliform concentrations were projected using the cali-
brated model for one dry-weather loading condition and three wet-weather
loads which bracketed contravention of the bacterial standard of 200 cells/
100ml. Table 16 shows the impact of the one-year (two-hour) storm on the
bacteriological quality in the contact recreation zone. Also shown is the
time of recovery (TOR), i. e., the time required for the quality to return to
an acceptable level. Dissolved oxygen deficit projections, also shown in
Table 16, will be discussed in later paragraphs.
Figure 30 shows the fecal coliform loading curve generated for the
lake. As may be seen from the figure and Table 16, the dry weather dis-
charges alone produce a peak fecal coliform concentration of 120 cells per
100ml in the recreation zone of the lake. However, the dry-weather effect
should not be as severe once the upstream pollutant sources of Onondaga
Creek are eliminated. This figure demonstrates the dramatic effect
that CSO's have on the bacteriological quality of Onondaga Lake. The
one-year, two-hour storm causes the fecal coliform concentration to exceed
cells/lOOml in the contact recreation zone and remain i.n violation of the
98
-------
GO
P
§
J
g
S
J
g
K
H
P
H
J
PQ
g
__^
J>j
K
0]
•a
ri
0
0
d
i — i
CM
•al
O r
0 £
"c
a
r™l £j P"
-J-5 £7 C
O °0 j:
^ o g
i_i C
-d
PH
0 c
CQ G1
3i
&
+j
3
.0
•1-1
JH
EH
CD
O O T-I O ^ O
co CM co CM in o
• * • • ' • B »
o in T-I ci o o o
CD CM
co
co
i-^
en
m co oo co T-I t- ^5
CO ^ 00 CO CO
CO CM T-I T-I
-
CD
CM
CO
-
* • • . • • • »
m o o o o o o
co o t— m CD o ^
CO O O CM C- CD
»-l CXI T-I T-I T-H
in
in
o
co"
o in o o o o in
t- co TH T-I in o
co t- T-I T-< m in
CO
o
00
t-
•V
CO
i o o m o o o
i en CM rt1 CM m T-I
in oo T-I co co
•V
T-I
o o in o o m in
o t- t- T-I en co co
o T-I t- m T-I j>
ura in* i-I" i-< -«j?
o o o o o o ^
O O CO O CO CD CD
co o CM T-I CM in
V"l f^
•
m
oo
CM
oo"
m
•V
CO
T-I
^
^
en
oo
ry.4 ]
^ « 5?
2 ^ 0 .2 -0
O o fa S 0 ^ en
,M S-iw 'E3r^
So J W g ^c§iS
to
TJ
ai
O'
O
d
• -v
03
1
JM
O
+->
'd
o
a
d
o
T3
0
00
ai
0
s
T!
03
.•s
a!
S
"S
0
I
0
£
I
*
d
rt
o
a
o
^
+->
0
0
a
ts
0
.
m
0
CQ CM
TO II
0 .t5 Cd
1> d O
T-I CM CO
99
-------
P
in
FQ
Cd
o
0
U
CXI
03
o a
d CQ
M 2
Ifi
ro
o
"
0
tu
bo
0
CQ
O
O txo
a
co
(M
CO
oo m
CO CO
CM
co
CO
co
CO
0
0
o m
i-H CO
co oo
CM co
CM CO
CO CO
CO
CM CO CO
CM
8 M
k ^
O o
..
0 cd
co
CO
Hi
P
CO CO
m
en
co
m co
co »H
co"
in
(M
co
co
0
0
in
co
co
(M
CD
(M CO
tO CM
LO
0
a
0
M
CO
co
CM CO
CO
o
co
CO
10
m
co
in
10 co
CO
co
CO
CO
oo
CM CO
in co co
I-H in ^
CD
•
co
in
0
cd
oo
CO CO
•^ oo
m co
0
OJW HOJW HOJ
0
CQ
0
Cd
a
a,
cd
tu
!n
(U
CQ
s
cd
0
(U
CD
0
0
02
•a
cd
0
cd
0
0
cd
0
8P
fc
0
Cd
CQ
T3
0
0
CQ
tin O O
0 iM
js p, pi
O CQ CQ
J^ 0 O
Q,
CQ ^M
TS
cd
CQ
cd
,S -6
0
0
o
cd
•3
-o
0
CQ
cd
cd
m
Scd"
0 T3
CQ
•O
Cd
0
cd
In
0
•H 0
cd s.
0 g
i M
-i, 0
0.8
CQ
5
o
CQ
0
CQ
0
I,6
a'
0
0 o 0
P EH
CO
T3
cd
CM
100
-------
200 cells/100ml standard for nearly 3. 3 days. To meet the water quality
standards, greater than a two-order magnitude of reduction in the fecal
coliform discharge from the CSO system is required.
TABLE 16. PROJECTED IMPACT OP WET-WEATHER
LOADS ON ONONDAGA LAKE
Storm Recurrence Interval
Pre-Storm 1 Year 2 Years
PC (cells/ 100ml)*
TOR (days)
DO Deficit (mg/1)
TOR (days)
120 104 **
3.3 **
0 0.39 0.48
Negl. 2/97
10 Years
**
**
0.90
4.37
* Occurring in epilimnion of contact recreation zone.
r#Not modeled since dramatic effects were noted with much more
frequent storms. Two additional model runs were performed
for storms occurring more frequently than once per year to
define the loading curve (Figure 30). ,
Onondaga Lake historically has had dissolved oxygen deficiency
problems (4) which were attributed to the large waste loads that the lake
receives. During this study it became evident that the dissolved oxygen
picture of the lake was controlled predominantly by fluctuations in the algal
biomass. Dramatic die-off of the algal standing crop occurs quite
frequently in Onondaga Lake and the dead cells contribute substantially to
the carbonaceous BOD of the lake. A die-off of biomass equivalent to
86 y g/1 chlorophyll a (the estimated peak concentration after tertiary treat-
ment is implemented (47)) over a two-week period produces an average
decrease of 9 mg/1 in dissolved oxygen. Biomass fluctuations of this
magnitude are not uncommon and, in fact, during the monitoring program
for the Storms Impact Study (46), a die-off of 130 y g/1 of chlorophyll a was
observed during a similar period. ~~
The DO deficit resulting from wet-weather discharges is negligible
when compared to the deficits resulting from algal die-off. Table 16 shows
the deficits produced by the one-, two- and ten-year storms together with
the time needed to return to within 5 percent of the initial condition (TOR).
The minimum dissolved oxygen concentration produced by the ten-year
storm load is 6. 9 mg/1 under the selected environmental conditions. Based
101
-------
I YEAR - 2 HOUR STORM
10
DRV WEATHER LOAD* I014 CELLS/DAY
' "ii
I0 10'
MAXIMUM LAKE CONCENTRATION (CELLS/100 ml)
10"
Figure 30. Fecal coliform loading curves for Onondaga Lake.
102
-------
on this information, the final abatement plan does not consider removal of
organics as a primary goal.
Due to the long-term effects of nutrients, the impact of phosphorus
contributed from the CSOfs was evaluated with the steady-state model. The
calibrated phosphorus model was run at the following two input conditions:
1.
2.
Average daily total inorganic phosphorus load (Table 14),
based on monitoring data, including projected phosphorus
reductions at the Metropolitan Sewage Treatment Plant.
Condition 1 above, plus the average daily CSO phosphorus
load; the average daily CSO load is approximately 13 Ibs/day
(28 kg/day) (6).
Table 17 compares the results of the runs and shows that the phosphorus
contributed from the combined sewer system produces only a 1 percent
change in concentration in the lake.
TABLE 17. AVERAGE ANNUAL PHOSPHORUS CONCENTRATIONS
IN ONONDAGA LAKE
Loading
Condition
Average Daily:
Without CSO
Component
Average Daily:
With CSO
Component
South Basin (mg/1) North Basin (mg/1)
Epilimnion Hypolimnion Epilimnion Hypolimnlon
0.221
0.223
0.508
0.513
0.204
0.206
0.495
0.500
The Wet-weather load produces such a minute change within the lake,
compared to other sources, that wet-weather treatment for phosphorus
removal was not considered in the abatement plan.
103
-------
SECTION 10
COST-EFFECTIVENESS ANALYSIS
The development of relationships between rainfall, overflow dis-
charges, and their receiving water impact is the essential basis for a cost-
effective abatement program. In the case of conventional dry-weather
treatment, receiving water quality is focused on established state standards
and prescribed low flow conditions. Such pre-set conditions do not exist
with respect to stormwater treatment because of the variable nature of
stormwater runoff and the premise that not all of the runoff can be economi-
cally captured and subsequently treated. The approach in stormwater treat-
ment has been to develop relationships between receiving water impact and
the cost of treatment for various storm conditions so that local governments
and regulatory agencies can be provided with the cost-benefit information
needed to make reasonable decisions. This section will summarize the
relationships unique to the CSO problem and will present a method for
selecting an optimum treatment level.
CSO RELATIONSHIPS
Rainfall Patterns
To determine the extent to which a particular area is subject to
stormwater pollution, the most fundamental information that must be ana-
lyzed is the seasonal and annual rainfall. As discussed in Section 7, a
number of statistical routines are available to analyze long-term rainfall
data. Rainfall can be classified according to monthly totals so that atypical
pattern can be developed and so that critical months can be determined, as
shown in Figure 31. Similar plots can be made depicting average rainfall,
intensity, duration, antecedent dry weather, or number of storms in the
respective months. These data, in conjunction with receiving water flows,
can be used to select the months or periods in which the most significant
pollution may occur. The simple planning models described in Section 7,
such as SSWMM and STORM, can be utilized to make gross estimates of the
number and quantity of overflows that occur over periods of days or years.
Given a long-term record of rainfall, the classic system of curves
relating rainfall intensity to duration can be developed for various storm
104
-------
3
* rt
g |
> 2">
5 _M
g S*
,fO J- .£
* o
'o .>•
I 1
V
i ?
a s
3 <
i I
i v.
l \
^^ s
iri
E "
E
to
IT
in
CM
II
> L
rO —
-
*+—
0 °
>
«i
c -C
C
= II
o ~a
C •£
O 0
>• j?
* «- *-
0 i o
wi •*j F^—
i &
S J =
J* *c
• •^™
o
~ - Cd
t >
fO
cf ..._,..,..
£^
— *?
fg
t
Gg
D T'
c- ^
CM o _
O-O
x^ ^^
u. O
< a>
II
o
r - 2!
"'11
sS
riil
03 ^55
S||
c
<
c g
.
4)
Q
O
O
O
S
• B
" a
_ ~o
3 «*-
»
•^ .c
"* c
o
D
3 O
~3
a
f |
— CO
.
1 ^>
a>
1 E
5
.a
u.
c
o
i i
OoP
(S3H3M) TttWNIVM
_
105
-------
frequencies or return intervals, as shown in Figure 32. Such curves are
used to classify storms in relation to pollution and to treatment costs, as
outlined later in this section.
Receiving Water Impact
In order to link pollutant quantities to their water-quality impact, a
relationship must be developed which accommodates dry-weather sources as
well as stormwater sources. Dry-weather inputs should reflect current
treatment requirements for both publicly owned and industrial treatment
works; if such facilities are not presently in operation, future dry-weather
inputs should be projected and included. Stormwater inputs should include
upstream, non-point sources, urban stormwater in areas separately sewered,
and combined sewer overflows. Since this last component represents the
greatest potential health hazard and is generally the most controllable entity,
it is the component for which cost-effective analyses will be described here-
in. It is recognized that at some point in the treatment of CSO's, urban
stormwater, non-point source pollutants, or additional dry-weather inputs
may be more cost effective to remove than the CSO's.
Figure 28 illustrates the foregoing pollutant sources as they might
relate to the dissolved oxygen requirement in a lake. Providing that the
receiving water can sustain all but a portion of the CSO sources, the pollu-
tant quantity requiring treatment can then be varied for different storm con-
ditions. While a pollutant quantity might represent a specific water-quality
goal, it is likely that regulatory agencies and the public would be interested
in the abatement costs associated with varying degrees of contravention and
compliance; thus, a range of treatment bracketing the identified storm should
be depicted. These relationships are explained in greater detail further in
this section.
CSO Discharges
The quantity of pollutants discharged by a combined system is depen-
dent on several factors. It can be expected that storms of greater intensity
will result in greater flows through the system and will generate higher
velocities which will enhance the scour of pollutants accumulated either on
the land surfaces or along pipe inverts. The duration over which an inten-
sity of rainfall occurs will determine the quantity of available pollutants that
are scoured from the system. The extent of pollutant accumulation prior to
a storm occurrence will depend on: (a) the residual of pollutants remaining
at the end of the previous storm; and (b) the amount accumulated during the
intervening period, commonly referred to as the AntecedenfDry Weather
Period (ADW).
Quantities of pollutants discharged under a particular set of storm
106
-------
10-,-
tr
x
CO
z
UJ
UJ
e>
ID
5 YR. STORM
0 YR. STORM
I YR. STORM
2YR. STORM^ x^\ v
XV"N
NOTE'
I IN/HR.= 25.4mm/hr.
H 1—
100 200
10
DURATION (MIN.)
Figure 32. Intensity - duration- frequency curves, Syracuse, New York.
107
-------
conditions can be calculated through the use of the SWMM, or statistically
derived relationships, or a combination of both. Generally, SWMM can be
employed to develop flow quantities or hydrographs, whereas SWMM or
statistically derived relationships can be used to arrive at qualitative yard-
sticks, such as the pollutographs and loadographs previously described and
illustrated.
If the residual pollutants and the ADW are kept constant, pollutant
load discharged can be related to various types of storms in terms of
recurrence interval. Appropriate residual pollutant and ADW quantities can
be derived from statistical analyses of data or from the literature. The
deposition is dealt with in greater detail elsewhere (48, 50).
ABATEMENT COSTS
Figure 33 illustrates the interrelationships between the various
phases of a CSO study and shows the development of a cost-effective treat-
ment solution.
Figure 33A illustrates the relationship between the pollutant loads
discharged from a CSO system and various storm conditions described in
terms of recurrence interval. Recurrence interval can be used in this
manner only if other significant factors such as pollutant residual and ADW
are held constant. Recurrence interval is an appropriate variable from the
standpoint of the general availability of data and relationship to abatement
costs.
• Generally several years of rainfall records are available whereby
frequency of overflow can be determined on the basis of sewer system capa-
city. Such information, in combination with everTlimited quality informa-
tion, can be used to determine the pollutants discharged from a system over
a period of years. This is accomplished either through a series of single-
event calculations or continuous simulation, as described in Section 7.
The significant factors in developing costs of abatement strategies
are flow rate and volume, which are derived from the characteristics of
recurrence interval, namely, rainfall intensity and duration.
Figure 33B illustrates the relationship between pollutant load dis-
charged from the sewer system and water quality response. This relation-
ship is developed and discussed in Section 8. Based on a selected water-
quality goal, the allowable pollutant load determined from Figure 33B can
be related to an allowable recurrence interval from 33A. Although the
relationships illustrated by Figures 33A and 33B can represent independent
study efforts, they can be interfaced through the development of the common
relationship to pollutant load.
108
-------
"Allowable
RECURRENCE INTERVAL
A. RAINFALL TYPE VS. POLLUTANT LOAD
WATER QUALITY RESPONSE
B. POLLUTANT LOAD VS. WQ RESPONSE
2
cc
R Allowable
RECURRENCE INTERVAL STORM (Rn) —».
C. RECURRENCE INTERVAL VS. REQUIRED
TREATMENT
<
§
a
b
IU
to
o
o
RECURRENCE INTERVAL STORM (Rn) »-
or ^.
NO. OF WATER QUALITY VIOLATIONS (WQV) PER YEAR
D. RECURRENCE INTERVAL VS. COST
Figure 33. Procedure for establishing the most cost-effective treatment
level to meet a given water quality goal.
109
-------
Figure 33C illustrates the relationship between treatment required
and recurrence interval. This relationship is based on a series of points
selected from Figure 33B, and can be expressed in terms of the pollutant
load in excess of the allowable (Wn-WauOWable)- The treatment require-
ments set the basis upon which the cost of a treatment facilities can be
determined. The various methods of source control and the various treat-
ment options are reported elsewhere (17). The specific unit processes
available and the associated unit costs for both construction and operation
and maintenance are well documented (18, 49).
Figure 33D illustrates the cost of treatment as it relates to both
recurrence interval and the number of water-quality violations. It should be
noted that the costs represent the least expensive mix of control and treat-
ment options in each case. These costs are developed on the basis of engi-
neering judgment and relationships developed through previous studies (50,
51).
COST-EFFECTIVE SOLUTION
Treatment costs can be related directly to either recurrence interval
or a measureable benefit, which in this case has been illustrated as the num-
ber of water-quality violations per year. Since recurrence interval (Rn)
serves as the basis for determining the frequency of overflow in the system,
the number of water-quality violations can be directly equated to recurrence
interval. Another method to determine the number of water-quality viola-
tions is through the use of continuous simulation, whereby the number of
violations per year for a particular treatment scheme are calculated. Costs
can then be plotted against the average annual benefit as shown in Figure 33D.
For the example selected, the knee of both curves shown occurs at
the same cross point. The "knee" of the curves represents the point of
diminishing returns. Below the knee, treatment may be as cost effective
but will result in less removal and subsequently more frequent violations of
the water-quality goal. At all points beyond the knee, the additional benefits
derived from incremental increases in expenditure decrease rapidly and are
commonly referred to as being beyond the point of diminishing returns.
While it may be desirable to meet water-quality standards beyond the "knee",
justification should be based on factors other than economics (e. g. , legisla-
tive decisions or "best usage" of the receiving water).
A more comprehensive method of presenting abatement costs and
associated benefits involves performing the above procedure for various
water-quality goals (see Figure 33B). Figure 34 shows a plot of costs
versus water-quality violations for various goals in order to illustrate the
economic impact of more or less stringent goals. Such an approach can be
used to determine the most cost-effective water-quality goal or "general
110
-------
I
o
cc.
I
tu
111
CO
o
o
C|
>Cn- CONCENTRATION SELECTED FOR WATER
QUALITY GOAL. GOAL IS INCREASING IN
STRICTNESS FROM Cj TO Cn.
On- OPTIMUM COST-BENEFIT FOR VARIOUS
WATER QUALITY GOALS.
GENERAL OPTIMUM SOLUTION
FREQUENCY OF WATER QUALITY VIOLATIONS (NO./YR.)
Figure 34. Procedure for determining the cost effective water quality goal.
111
-------
optimum solution" (GOS). The term GOS is used herein to denote a solution
which encompasses the essential factors necessary to arrive at a reasonable
solution. Figure 34 presents a summary of these factors from which the
decision-makers can select a reasonable water-quality objective.
The GOS can be determined by drawing a curve (A-B in Figure 34)
through the knees of a series of cost curves similar to the one presented in
Figure 33D. The general optimum solution to the cost-benefit analysis
occurs at the knee of this curve. This point not only defines the water-
quality standard and associated cost, but through the use of Figure 33D,
indicates the probable frequency with which that standard will be contra-
vened.
It should be noted that in some cases data will not yield a noticeable
break or knee at which the GOS occurs. However, in any event, presenta-
tion of the data as shown in Figure 34 will be desirable since it provides the
complete picture of the water-quality benefits which can be expected for any
given expenditure. Even though a GOS exists, it may be desirable for a
variety of reasons (e. g. , limited funds or inability to live with the cost-
effective water-quality goal) to select another water-quality goal. The cost
data presented in Figure 34 enables the decision-makers to balance the
water-quality goals against the total cost of achieving them.
Onondaga County
As discussed in Sections 7 and 9, curves relating various recurrence-
interval storms, CSO discharges and water quality were developed for the
Onondaga Lake study. In that case, the analyses showed fecal coliform to
be the only parameter to show significant water-quality impact from the CSO
discharges. Therefore, only examples of abatement solutions related to
meeting the contact recreation standard of 200 cells/100 ml established by
the New York State Department of Environmental Conservation will be dis-
cussed in this section.
Extensive statistical analyses of rainfall patterns were performed
on the historical records for Syracuse, New York, and the results com-
pared to the findings of both the CSO and the water-quality programs. This
procedure revealed that there are 170 rainfall events per year; based on
peak intensity, 65 of these produce overflows. On the average, the com-
bined sewer system is capable of accepting the runoff from a storm with
peak intensity of less than 0. 05 inches/hour without causing any overflow.
Of these 65 overflow-producing events, only 38 result in discharges of
sufficient magnitude to cause a violation of the fecal coliform standard in
the contact recreation zone. Figure 35 presents the number of violations
that occur annually as a result of Creating storms of various intensities or
recurrence intervals for a fixed two-hour duration. Figure 36 shows the
112
-------
CO
a:
13
O
_CD
m
. 00
.0
\ *
[
.
-CM
K)
.-*
CVJ
-
^
±:
o
3
0
o>
o
^
«*-
o
o"
2!
CO
o
0>
5
o
o>
o
1
C
o
_c
>,
"a
^^
^
c
o
^CD
li
O o
O -*T
CO
-I'S
•*- c
o • w
±: o
?!
o .E
0 o
5 £
C\J —
OOI^ CD IT) -sj- rO CM —
(SLjjUOLU) |DAJ8|U|
in
ro
o>
3
o>
113
-------
CD
t ID
0>
o
.3
T3
V
V)
O
o> ">
-I
c °
11
o>
o>
-J= o
o
M-
O
SS"
V) *t
1.2
CO
ro
(£86l)SJD||oa p suonniAl uj Sjsoo MIJOM
g>
UL
114
-------
present worth costs (1983) for constructing abatement facilities to treat
these storms.
Inspection of the curve presented in Figure 36 shows the "knee" of
the curve occurring in the vicinity of $12 million; even with that investment,
13 violations would occur per year. Of the 13 annual violations, over ten
will occur during the recreation season from June through September (Fig-
ure 37). This case is one where the knee-of-the-curve solution is unaccept-
able. Since ten violations would essentially prohibit use of the lake for con-
tact recreation, more stringent levels of treatment will ultimately be needed.
The selected goal, for which treatment will eventually be provided, is to
allow one violation of the contact recreation standard (200 cells/100 ml) per
year.
The final CSO abatement plan is presently being developed (6) and is
available through the Onondaga County Department of Drainage and Sani-
tation.
115
-------
(D
O>O
c-o
OT
E
o>
P^
'
f >
o o>
85
o o
•4= a?
|"5
c p
0) "
HINCWSNOIIVIOIA 39VM3AV
ro
i_
o>
k.
116
-------
.REFERENCES
1. Federal Water Pollution Control Act Amendments of 1972, Public Law
92-500. 92nd Cong. , 86 Stat. 816, October 18, 1978.
2. Clean Water Act of 1977, Public Law 95-217. 95th Cong. , 91 Stat.
1566, October 27, 1977.
3. Shelley, P. E. , and G. A. Kirkpatrick. Sewer Flow Measurement, A
State -Of- The-Art Assessment. EPA-600/2-75-027, U.S. Environ-
mental'Protection Agency, Cincinnati, Ohio, November 1975. 436pp.
4. Driscoll, E. D. Runnoff and Quality. In: Proceedings of Urban Storm-
Water Management Seminars. WPD 03-76-04, U.S. Environmental
Protection Agency, Washington, D. C. , January 1976.
5. Shelley, P. E. , and G. A. Kirkpatrick. An Assessment of Automatic
Sewer Flow Samplers. EPA-R2-73-261, U.S. Environmental Pro-
tection Agency, Washington, D, C. , June 1973. 233pp.
6. O'Brien & Gere, Inc. Combined Sewer Overflow Abatement Program.
Report Prepared for Onondaga County, New York (in preparation).
7. Areawide Assessment Procedures Manual, Volumes I, II and III.
EPA-600/9-76-014, U.S. Environmental Protection Agency,
Cincinnati, Ohio, July 1976.
8. Manning, M. J., et al. Nationwide Evaluation of Combined Sewer
Overflows and Urban Stormwater Discharges, Volume III: Characteri-
zation of Discharges. EPA-600/2-77-064C, U.S. Environmental
Protection Agency, Cincinnati, Ohio, August 1977. 290pp.
9. Report to the Congress by the National Commission on Water Quality.
U. S. Government Printing Office, Washington, D. C., Mar en 1976. 90 pp.
10. Report to Congress by the Office of the Comptroller General. Better
Data Collection and Planning is Needed to Justify Advanced Waste
Treatment Construction. U. S. Government Printing Office, Washing-
ton, D. C., December 1976. 70pp.
117
-------
11. Geldreich, E. E. Applying Bacteriological Parameters to Recreational
Water Quality. J. American Water Works Association, 56; 931, 1975.
12. Oliveri, V. P. , et al. Survival of Natural Populations of Salmonella
and Indicator Bacteria in Secondary Plant Effluents. Paper presented
at the Fiftieth Annual Conference of the Water Pollution Control
Federation, Philadelphia, Pennsylvania, October 1977.
13. Geldreich, E. E. , and B. A. Kenner. Concepts of Fecal Streptococci
in Stream Pollution. J. Water Pollution Control Federation,
4^:R336-R352, 1969.
14. Reid, G. K. Ecology of Inland Waters and Estuaries. Reinhold
Publishing Corporation, New York, New York, 1961. 375 pp.
15. Huber, W. C., et al. Interim Documentation, November 1977 Release
of EPA SWMM, Draft Report. U. S. Environmental Protection Agency,
Cincinnati, Ohio, November 1977.
16. Huber, W. C., et al. Storm Water Management Model User's Manual
Version n. EPA-670/2-75-017, U.S. Environmental Protection
Agency, Cincinnati, Ohio, March 1975. .367 pp.
17. Lager, J. A., et al. Urban Stormwater Management and Technology:
Update and Users' Guide,. EPA-600/8-77-014, U.S. Environmental
Protection Agency, Cincinnati, Ohio, September 1977. 331pp.
18. Heaney, J. P. , and S. J. Nix. Storm Water Management Model: Level
I - Comparative Evaluation of Storage-Treatment and Other Manage-
ment Practices. EPA-600/2-77-083, U.S. Environmental Protection
Agency, Cincinnati, Ohio, April 1977. 105pp.
19. Heaney, J. P. , et al. Storm Water Management Model: Level I -
Preliminary Screening Procedures. EPA-600/2-76-275, U.S.
Environmental Protection Agency, Cincinnati, Ohio, October 1976.
95 pp.
20. Amy, G., et al. Water Quality Management Planning for Urban
Runoff. EPA-440/9-75-004, U.S. Environmental Protection Agency,
Washington, D. C., December 1974.
21. U.S. Army Corps of Engineers. Storage-Treatment, Overflow,
Runoff Model "STORM" User's Manual. Hydrologic Engineering
Center, Davis, California, July 1976.
118
-------
22. Lager, J. A., et al. Development and Application of a Simplified
Stornawater Management Model. EPA-600/2-76-218, for U. S.
Environmental Protection Agency, Cincinnati, Ohio, August 1976.
153 pp.
23. Brandstetter, A. Assessment of Mathematical Models for Storm .and
Combined Sewer Management. EPA-600/2-76-175a, U.S. Environ-
mental Protection Agency, Cincinnati, Ohio, August 1976. 530pp.
24. Chow, V. T., and B. C. Yen. Urban Stormwater Runoff Determination
of Volumes and Flowrates. EPA-600/2-76-116, U.S. Environmental
Protection Agency, Cincinnati, Ohio, May 1976. 252pp.
25. U.S. Department of Commerce, Weather Bureau. Rainfall Frequency
Atlas of the United States for Durations from 30 Minutes to 24 Hours
and Return Periods from 1 to 100 Years. Washington, D. C. , May
1961.
26. Keifer, C. J., and H. H. Chu. Synthetic Storm Pattern for Drainage
Design. J. Hydraulic Division, Proceedings of the American Society
of Civil Engineers, 93 (HY4): 1332-1-1332-25. 1957.
27. Marsalek, J. Comparison of Runoff Simulations for Actual and
Synthetic Storms. Presented at Storm Water Management Model
Users Group Meeting, Milwaukee, Wisconsin, November 3-4, 1977.
Sponsored by U. S. Environmental Protection Agency, Washington,
D. C. 15 pp.
28. McPherson, M. B. The Design Storm Concept. Presented at Institute
of Stormwater Detention Design, University of Wisconsin, April 14,
1977, Madison, Wisconsin. 20pp.
29. Limno-Tech, Inc. Mathematical Modeling of the Impact of Storms on
Water Quality in Onondaga Lake. Prepared for Stearns & Wheler,
Civil and Sanitary Engineers, and Onondaga County, New York.
Ann'Arbor, Michigan, 1978. 133 pp.
30. DiToro, D. M., et al. A Dynamic Model of the Phytoplankton Popu-
lation in the Sacramento-San Joaquin Delta. In: Nonequilibrium
Systems in Natural Water Chemistry, R. F. Gould, ed. Advances in
Chemistry Series No. 106, American Chemical Society, 1971.
pp. 131-180.
31. Thomann, R. V. Systems Analysis and Water Quality Management,
Environmental Research and Applications, Inc. , New York, New York,
1972. 286 pp.
119
-------
32. Canale, R. P., R. L. Patterson, J.J. Gannon, and W. F. Powers.
Water Quality Models for Total Coliform Bacteria in Grand Traverse
Bay. J. Water Pollution Control Federation, 45:325-336, 1973.
33. Gaden, E. Aeration and Oxygen Transport in Biological Systems. In:
Biological Treatment of Sewage and Industrial Wastes, R. F. Gould
andW. W. Eckenfelder, eds. Reinhold Publishing Corporation,
New York, New York, 1956.
34. Hydroscience, Inc. Water Quality Analysis for the Markland Pool of
the Ohio River. Prepared for Malcolm-Pirnie Engineers, Metropoli-
tan Sewer District of Greater Cincinnati, Ohio, 1968.
35. American Public Health Association, et al. Standard Methods for the
Examination of Water and Wastewater, Thirteenth Edition, Washington,
D. C. , 1971. 774 pp.
36. Banks, R. B., and F. F. Herrera. Effect of Wind and Rain on Surface
Reaeration. J. Environmental Engineering Division, Proceedings of
the American Society of Civil Engineers, 103(EE3): 489-504.
37. Canale, R. P., et al. A Biological Production Model for Grand
Traverse. Technical Report No. 37, University of Michigan Sea
Grant, Ann Arbor, Michigan, 1974. 116pp.
38. Chapra, S. C. Total Phosphorus Model for the Great Lakes. J. Envi-
ronmental Engineering Division. Proceedings of the American Society
of Civil Engineers, 103(EE5): 1977. pp. 147-161.
39. Murphy, C. B., and G. J. Welter. Indices of Algal Biomass and
Primary Productivity in Onondaga Lake. Presented at National
Conference on Environmental Resources Development and Design,
ASCE Environmental Engineering Division, Seattle, Washington,
1976.
40. Stearns & Wheler, Civil and Sanitary Engineers. Progress Report:
Onondaga Lake Storms Impact Study. Department of Drainage and
Sanitation, Onondaga County, New York, 1978. 102 pp.
41. Onondaga County. Onondaga Lake Study. Project No. 11060,
FAE 4/71, U.S. Environmental Protection Agency, Cincinnati,
Ohio, 1971. 406 pp.
42. Wetzel, R. G. Limnology. W. B. Saunders Publishing Company.
Philadelphia, Pennsylvania, 1975.
120
-------
43. O'Brien & Gere, Inc. Onondaga Lake Monitoring Program.. Yearly
Report Prepared for Onondaga County, New York. Syracuse, New
York, 1970, 1971, 1972, 1973, 1974, 1975.
44. Stearns & Wheler, Civil and Sanitary Engineers. Onondaga Lake
Monitoring Program.. Yearly Report Prepared for Onondaga County,
New York, Cazenovia, New York, 1976, 1977.
45. Great Lakes Institute. Great Lakes Institute Data Record Surveys,
Parts I and H. University of Toronto, Canada, 1963.
46. Stearns & Wheler, Civil and Sanitary Engineers. Onondaga Lake
Storms Impact Study. Report Being Prepared for O'Brien & Gere,
Inc. , and Onondaga County. Cazenovia, New York, in preparation.
47. Walker, W. W. Some Analytical Methods Applied to Lake Water
Quality. Ph. D. Thesis, Harvard University, Cambridge, Massa-
chusetts, 1977.
48. Pisano, W. C. , and C. S. Queiroz. Procedures for Estimating Dry
Weather Pollutant Deposition in Sewerage Systems. EPA-600/2-77-
120, U. S. Environmental Protection Agency, Washington, D. C. ,
July 1977.
49. Turner, E.G., et al. 1976 Needs Survey. EPA-430/9-76-012, U.S.
Environmental Protection Agency, Washington, D. C. , February
1977. 400 pp.
50. Sonnen, Michael B. Abatement of Deposition and Scour in Sewers.
EPA-600/2-77-212, U.S. -Environmental Protection Agency,
Washington, D. C., November 1977. pp. 115.
121
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1, REPORT NO.
EPA-600/8-80-048
3. RECIPIENT'S ACCESSION-NO.
4, TITLE AND SUBTITLE
METHODOLOGY FOR EVALUATING THE IMPACT AND
ABATEMENT OF COMBINED SEWER OVERFLOWS
A Case Study of Onondago Lake, New York
5. REPORT DATE
November 1980 (Issuing Date)
6. PERFORMING ORGANIZATION CODE
7, AUTHOR(S)
Peter E. Moffa, John C. Byron, Steven D. Freedman,
John M. Karanik, and Randy Ott
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Stearns & Wheler
Civil and Sanitary Engineers
10 Albany Street
Cazenovia, New York 13035
10. PROGRAM ELEMENT NO.
A 35B1C
11. CONTRACT/GRANT NO.
R-805096
13, SPONSORING AGENCY NAME AND ADDRESS
Municipal Environmental Research Laboratory—Cin.,OH
Office of Research and Development
U. S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TYPE OF REPORT AND PERIOD COVERED
Final Report, 6/77 to 10/79
14. SPONSORING AGENCY CODE
EPA/600/14
15. SUPPLEMENTARY NOTES
Project Officer: Richard Field - Telephone (201) 321-6674
16. ABSTRACT
A general methodology is presented for the evaluation of the impact and abatement
of combined sewer overflows on receiving waters. It was developed from
experience with Onondaga Lake, an urban lake in central New York that receives
combined sewer overflows from the City of Syracuse via three tributary streams.
Field measurements were made of representative combined sewer overflows and the
receiving water for the purpose of developing individual mathematical models. These
models were employed to project the magnitude of pollutant load from a combined sewer
system for different storm conditions and the associated receiving water impact,
respectively. The results of these two models can be combined to express the
abatement cost to achieve different water quality standards.
A maximum DO deficit of 2.8 mg/1 in Onondaga Lake was predicted for a 10-year storm.
Bacterial violations can occur as many as 38 times during an average rainfall year.
Abatement beyond the "cost-effective" point will be necessary to assure the lake's
use for contact recreation.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSAT1 Field/Group
Water pollution, combined sewers, water
quality, wastewater, methodology,
mathematical models, cost-effectiveness
Lakes, receiving water
impacts, combined
sewer overflows, facilit.
planning
13B
18, DISTRIBUTION STATEMENT
Release to public
19. SECURITY CLASS (ThisReport)'
Unclassifipd
21. NO. OF PAGES
138
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
122
U.S. GOVERNMENT PRINTING OFFICE:1980—757-064/0177
------- |