Long-term preformance of stormwater detention facilities : a comparison of design methodologies

MISSING IMAGE

Material Information

Title:
Long-term preformance of stormwater detention facilities : a comparison of design methodologies
Series Title:
Florida Water Resources Research Center Publication Number 58
Physical Description:
Book
Creator:
Goforth, Gary F. E.
Heaney, James P. ( Thesis advisor )
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Notes

Abstract:
A general overview of empirical, analytical, statistical and simulation techniques for evaluating stormwater detention systems is presented. The benefits and limitations of these methods in designing a control device for water quality improvement are emphasized. A detailed analysis compares continuous simulation utilizing the Environmental Protection Agency's Storm Water Management Model with the statistical techniques advanced by Hydroscience, Inc. The general dynamics or storage and flow elements are discussed, emphasizing the importance of detention time in defining a time frame for evaluating systems.

Record Information

Source Institution:
University of Florida Institutional Repository
Holding Location:
University of Florida
Rights Management:
All rights reserved by the source institution and holding location.
System ID:
AA00001536:00001


This item is only available as the following downloads:


Full Text















Publication No. 58

LONG-TERM PERFORMANCE OF STORMWATER DETENTION FACILITIES:
A COMPARISON OF DESIGN METHODOLOGIES

by

Gay F.E. Goforth


University of Florida
Gainesville, FL 32611-2013











LONG-TERM PERFORMANCE OF STORMWATER DETENTION FACILITIES:
A COMPARISON OF DESIGN METHODOLOGIES



By



Gary F. E. Goforth


PUBLICATION No. 58




FLORIDA WATER RESOURCES RESEARCH CENTER



RESEARCH PROJECT TECHNICAL COMPLETION THESIS




ENGINEERING AND INDUSTRIAL EXPERIMENT STATION

PROJECT NUMBER 80-W31




THESIS SUBMITTED JUNE, 1981



The work upon which this thesis is based was supported in part
by funds provided by the Water Research Program,
Engineering and Industrial Experiment Station,
University of Florida, Gainesville















ACK.NOWLEDC.Ei- 1'-7TS


Interaction is free; as such, I owe thanks to many

individuals at Black Hall and the Center For Wetlands who

have contributed to the completion of this thesis and to my

career at the University of Florida. The diversity of

professionals in the department has provided a constant

challenge to maintain an awareness of the many, yet similar,

facets of environmental engineering sciences. A large debt

is acknowledged to Bob Dickinson who several times pulled me

up when I was close to drowning while SWMMing. Thanks also

go to Steve Nix for his help on S/T. The use of the

computer resources at the Center For Wetlands, Black Hall

and the Northeast Regional Data Center was invaluable.

Thanks to Anelia Crawford for the drafted figures. The

direction of this thesis is credited to Dr. J. P. Heaney;

the stability of the content is credited to Dr. W. C. Huber;

the influence of Dr. H. T. Odum is reflected in the holistic

approach to the problem assessment and solutions. Their

guidance and attention are greatly appreciated.

Certainly the greatest debt is owed to my wife, Karen.

Her patience, sacrifice, programming and typing ability and

overall good spirits in the face of adversity will be

forever appreciated.


















TABLE OF CONTENTS


ACKNOWLEDGEMENTS . .

LIST OF TABLES . . .

LIST OF FIGURES . .

LIST OF SYMBOLS . .

ABSTRACT . . .

INTRODUCTION . . .
GENERAL OVERVIEW . .
SYSTEM DESCRIPTION .
DEFINITION OF CONTROL UNIT
CLASSIFICATION OF METHODS .

CASE STUDY . . .
CATCHMENT CHARACTERIZATION


. . . ii



. . . v ii

. . . x

. . . .x i i


. 1
. 1
. 3
S 6
. 8


RAINFALL-RUNOFF CHARACTERIZATION .
Rainfall data. . . .
SYNOP. . . . .
Runoff Quantity. . .
BASIN CHARACTERIZATION . .
Flow Conditions. . .
Removal. . . . .
Theory of Settling. . .
Comparison of Flow Conditions


. 1 1
. 1 1
. 12
. 12
. 13
. 15
. 23
. 23
. 24
. 24
. 29


METHODOLOGIES . . . . .
ANALYTICAL. . . . .
EMPIRICAL . . . . .
SIMULATION. . . . .
STORM ... . . . .
SWMM . . . . .
Overview . . . .
Data Input for Runoff Block . .
Data Input for Storage/Treatment Block
Constant Discharge Simulation . .
Variable Discharge Simulation . .
Removal Mechanism.. . ... .
One-Year versus 25-Year Simulation. .
Runoff Block Results. . .
Storage/Treatment Block Results .


iii


. 32
. 32
. 33
. 36
. 36
. 36
. 36
. 38
39
. 45
. 46
... 147
S 50
. 50
. 52









STATISTICAL TECHNIQUES . . . 70
Event Definition . . . 70
Traditional Design of Flood Control Basins 72
Aggregate Statistical Methods. . .. 74
Data Input. . . . .. 75
Capture Performance Results . . 80
Removal . . . . 83
Analysis Using Simulated Data . 84
COMPARISONS . . . . 87

DISCUSSION . . . . . 91

GENERAL APPLICATION OF METHODOLOGIES . . 93

APPENDIX A Program Listing and Data Input . 97

APPENDIX B BASIN Development and Listing . .103

APPENDIX C Detention Time. . . . 105

REFERENCES . . . . . 112

BIOGRAPHICAL SKETCH . . . . .115
















LIST OF TABLES


Table

1. Case study catchment characteristics. 12

2. SYNOP results of 24.6 years of Atlanta,
Georgia rainfall. 16

3. Relationships between pollutant loads and
flow volume (FLOW). 35

4. Comparison( of Runoff Block results using
hourly, daily and weekly rainfall input. 38

5. SYNOP results of 24.6 years of simulated
runoff data. 41

6. Determination of normalized volume ratios
(Vb/Vro). 42

7. Calculation of constant discharge rates (Qc)
(Ti = 111. 17 hours; Vro = 18022 cubic feet). 46

8. Comparison of mean runoff event parameters
for 1953 with the 24.6-year record (minimum
interevent time = 4.0 hours). 52

9. Runoff Block 24.6-year simulation summary. 51

10. Examples of Storage/Treatment Block
summaries. 53

11. Estimates of flow capture efficiency (C) and
pollutant removal efficiency (R) as a
function of basin volume and constant
discharge rate: simulation results. 54

12. Estimates of flow capture efficiency (C) and
pollutant removal efficiency (R) as a
function of drawdown height and rate
(Vb/Vro=1.61): 1953 simulation results. 65

13. Estimates of flow capture efficiency (C) and
pollutant removal efficiency (R) as a
function of outlet height and outlet diameter
(Vb/Vro = 1.61): 1953 simulation results. 66










14. Computer costs of simulations.


15. Determination of normalized volume ratios. 76

16. Estimates of flow capture efficiency (C) as a
function of basin volume and constant
discharge rate: statistical results. 80

17. Comparison of simulated runoff mean event
statistics with rainfall conversion values. 85

18. Estimates of flow capture efficiency, (C) as a
function of basin volume and discharge rate:
statistical results with simulated runoff
means. 86

19. Estimates of hydraulic volume and detention
time of various control units associated with
a single rainfall event. 94
















LIST OF FIGURES


Figure

1. Representation of the hydrologic cycle. 4

2. Discretized subsystems of the hydrologic
cycle. 5

3. Processes defining the performance of a
control unit. 5

4. Schematic of Case Study catchment area. 7

5. Characterization of the various
methodologies. 9

6. Comparison of theoretical and observed
distributions of interevent times for
Minneapolis/St. Paul airport. 14

7. SYNOP values for mean event volume and
duration as a function of minimum interevent
time. 17

8. SYNOP values for mean event intensity and
interevent time as a function of minimum
interevent time. 18

9. A comparison of rainfall and runoff time
series depicting the reduction in number of
events and the reduction in the event volume. 20

10. Definition of interception and storage for
storm events. 21

11. Representation of a time series of runoff
flows. 22

12. Development of overflow rate in an ideal
settling basin. 25

13. Effluent concentrations for a first-order
removal process in n completely mixed plugs. 28

14. Removal efficiency for a first-order removal
process demonstrating the effect of increased


vii









turbulence; n=-l for quiescent and n=1 for
completely mixed. 28

15. Effluent responses to a step input. 30

16. Comparison of real and plug flow reactor
volumes for a first-order reaction. 30

17. Stage relationships as calculated by BASIN;
constant and variable' discharge. 40

18. Time series of constant influent pollutant
concentration. 44

19. Removal equation used in SWMM S/T Block. 48

20. Comparison of effluent concentrations under
ideal plug flow, ideal completely mixed and
as calculated with SWMM S/T completely mixed
routing; step input of pollutant. 49

21. Solution surface of flow capture efficiency
(C) as a function of basin volume and
constant discharge rate: 1953 simulation
results. 55

22. Solution surface of pollutant removal
efficiency (R) as a function of basin volume
and constant discharge rate: 1953 simulation
results. 56

23. Pollutant removal efficiency (R) as a
function of basin volume and constant
discharge rate: 1953 and 24.6-year
simul tion results. 57

24. Pollutant removal efficiency (R) as a
function of basin volume under variable
discharge conditions: 1953 simulation
results. 61

25. Solution surface of pollutant removal
efficiency (R) as a function of drawdown
height and drawdown rate (Vb/Vro = 1.61):
1953 simulation results. 63

26. Pollutant removal efficiency (R) as a
function of drawdown height and constant
discharge rate (Vb/Vro. = 1.61): 1953
simulation results. 64

27. Solution surface of pollutant removal
efficiency (R) as a function of outlet height


viii










and outlet diameter (Vb/Vro = 1.61): 1953
simulation results. 67

28. Pollutant removal efficiency (R) as a
function of outlet height and outlet diameter
(Vb/Vro = 1.61): 1953 simulation results. 68

29. Distribution of the largest sample value from
a sample of size n from an exponential
distribution. 73

30. Normal probability plot of Kentucky River
data. 73

31. Relationship of capture efficiency (C) with
normalized basin volume, normalized discharge
rate and mean volume coefficient of
variation. 79

32. Solution surface of flow capture efficiency
(C) as a function of basin volume and
constant discharge rate: 1953 statistical
results. 81

33. Flow capture efficiency as a function of
basin volume and constant discharge rate:
1953 statistical results. 82

34. Comparison of flow capture efficiency. 89

35. Comparison of pollutant removal efficiency. 90

B-1. Development of BASIN. 104

C-I Various control unit configurations:
steady-state conditions. 107

C-2 Various control unit configurations:
nonsteady-state conditions. 110
















LIST OF SYMBOLS USED


a,b coefficients of runoff quality power equation

ai percentage of flow passing through basin i

A cross-sectional area of flow

c effluent concentration

C flow volume capture efficiency

cO influent concentration

Cro runoff conversion factor

CV coefficient of variation

dr individual rainfall event duration

Dr mean rainfall event duration

Dro mean runoff event duration

e base of natural logarithm

H height of settling zone

Hd height at which discharge begins and ends

hp height of particle entering settling zone

k first-order reaction coefficient

I percent of catchment area that is impervious

ir individual rainfall event intensity

Ir mean rainfall event intensity

Iro mean runoff event intensity

kO initial first-order reaction coefficient

ki linear flow coeficient of basin i

L length of flow element

x











n

N



Qc

GcTi/Vro

R

SA

S/T

t

td

Ti

v

V

Vb

Vb/Vro

Ve

Vi

vo

vp

Vr

Vro

Vs

xo


turbulence coefficient

return period for design storm

volumetric flow rate

constant discharge rate

normalized discharge rate

pollutant removal efficiency

surface area of settling zone

storage/treatment

elapsed time

detention time of system

mean interevent time

velocity of flow

volume of flow element

empty volume of basin

normalized basin volume

effective volume

volume of basin i

overflow rate (surface loading rate)

particle settling velocity

mean rainfall event volume (per unit surface area)

mean runoff event volume (per unit surface area)

volume of settling zone

percentage of particles with vp less than vo










Abstract of Thesis Presented to the Graduate Council of the
University of Florida in Partial Fulfillment of the
Requirements
for the Degree of Master of Engineering








LONG-TERM PERFORMANCE OF STORMWATER DETENTION FACILITIES:
A COMPARISON OF DESIGN METHODOLOGIES










by
Gary F. E. Goforth
June 1981




Chairman: James P. Heaney
Major Department: Environmental Engineering Sciences


A general overview of empirical, analytical,

statistical and simulation techniques for evaluating

stormwater detention systems is presented. The benefits and

limitations of these methods in designing a control device

for water quality improvement are emphasized. A detailed

analysis compares continuous simulation utilizing the

Environmental Protection Agency's Storm Water Management

Model with the statistical techniques advanced by

Hydroscience, Inc. The general dynamics of storage and flow


xii










elements are discussed, emphasizing the importance of

detention time in defining a time frame for evaluating

systems.


Chairman


x i i i















INTRODUCTION


GENERAL OVERVIEW

In the urban environment, combined and separate storm

sewer overflows contribute the same amount of contaminants

to receiving waters as do secondary treatment effluents

(Heaney, et al. 1975, C.E.Q. 1978). Presently, simple

storage/treatment devices, i.e. one or two component systems

such as a stormwater detention basin, provide a

cost-effective tool for quantity as well as quality control

of these storm flows. While the design of these devices has

traditionally been based upon a single storm event, the

additional information provided by long-term analyses has

recently encouraged their adaptation. The engineer or

planner concerned with the design of a detention facility

for the quality control of stormwater runoff has a variety

of solution methodologies available: empirical approaches

utilizing average annual values; analytical methods based on

solutions to the flow governing equations; simulators

amenable to rigorous search techniques; and statistical

techniques involving rainfall-runoff parameter

distributions. No one method, or coordinated coupling of

methods, has been documented as the most cost-effective for









all applications. This is in part due to the lack of an

available long-term data base, but also reflects the lack of

comparative studies.

This thesis evaluates methodologies available for

analyzing the long-term performance of stormwater runoff

control devices. A detailed description of these methods is

not the intent. The manuals referenced for each provide

that service. Rather, the benefits and limitations of these

methods in designing a control device for water quality

improvement are emphasized. The comparison consists of

three criteria:

1. problem assessment, i.e. how does the particular

method define the system;

2. ease of application, e.g. data or computer

requirements, or cumbersome techniques, and

associated costs; and

3. accuracy of results, both absolute and relative to

data requirements.

Because the long-term performance of detention facilities

has not been well documented, there is no available data

base to definitively compare the accuracy of the methods.

Objectively, this study presents an opportunity to establish

relative estimates of the long-term performance of

storage/treatment devices designed for stormwater quality

control. The optimal design of a control device will depend

on problem specific constraints such as discharge quality

standards and economic considerations.









SYSTEM DESCRIPTION

The following definition of a system is presented as a

framework to maintain conceptual consistency.

A system is any structure, device, scheme, or
procedure, real or abstract, that interrelates in
a given time reference, an input, cause, or
stimulus, of matter, energy or information, and an
output, effect, or response, of information,
energy, or matter. (Dooge 1973, p. 4)

This functional interrelationship of inputs and outputs for

a given time reference provides a basis for addressing water

quality problems in a spectrum of hydrologic units, from

urban stormwater systems to lakes threatened with cultural

eutrophication.

Before evaluating the performance of a particular

system, the hierarchy of systems which influence that

performance must be recognized. A classical representation

of the hydrologic cycle is presented in Figure 1. Storages

and flows of water are the principal elements in the system,

although the influences of solar energies, land morphologies

and other factors are implicitly included. The system

depicted in Figure 1 can be partitioned into discrete

subsystems defined by characteristic storage and flows, as

shown in Figure 2. These subsystems can be further

subdivided into individual components, or control units,

whose boundaries similarly reflect the storage and flows

emphasized. The basic hydrologic characteristics which

define the performance of a control unit, as depicted in

Figure 3, are:















































S__- OCEAN
Ground Waler


Figure 1. Representation of the hydraulic cycle.


leep
percolation






































Figure 2.


Discretized subsystems of the hydraulic cycle.


Figure 3. Processes defining the performance of a
control unit.









1. The source of the mass, the unit's place in the

system and its relation to other units;

2. the dynamic storage and flow conditions; and

3. the removal mechanism.




DEFINITION OF CONTROL UNIT

The following analysis centers on the long-term

pollutant removal effectiveness of a hypothetical detention

facility. The control unit is a single basin which receives

the stormwater runoff from an urban catchment, and

discharges to an undescribed receiving water. A schematic

of the system is presented in Figure 4. The data source is

a 24.6-year record of hourly rainfall values obtained from

the National Weather Service. The system boundaries of the

control unit are drawn at the inlet and outlet. As such, it

is not just the rainfall which is the forcing function, but

the runoff, a result of the rainfall's interaction with the

catchment. The pollutant source is the constituent

contaminants of the runoff sand, debris, dust, etc. The

removal mechanism responsible for pollutant control is

sedimentation, and the removal characteristic is based on

the treatment time in the basin. The removal kinetics are

defined by the hydraulics within the basin, as determined by

basin geometry and the inflow and discharge characteristics.

These are the major influences on control efficiency and

become the design parameters.






































R-RAINFALL
Ro-RUNOFF
ET- EVAPOTRANSPIRATION
I- INFILTRATION
W- WIDTH OF CATCHMENT
L LENGTH OF CATCHMENT
DF- DETENTION FACILITY
D DISCHARGE TO RECEIVING WATER


Figure 4. Schematic of Case Study catchment area.









CLASSIFICATION OF METHODS

Several methodologies are available for estimating the

long-term performance of stormwater detention facilities.

The approaches are all models of the same complex process,

yet differ conceptually and mechanistically. Figure 5 is a

schematic depicting the relationship of the various

methodologies. For the purpose of this thesis, the

following classifications will be used:

1. Analytical approaches utilize some combination of the

general mass continuity equation and the

advective-dispersion equation to describe the flows,

storage, and pollutant removal characteristics of a

control unit.

2. Empirical approaches are derived from or guided by

experience. Although literally implying the lack of a

theoretical background, the expression is used to denote

methods which have been developed in scientific and

engineering practice.

3. Two types of statistical techniques are widely used

(Chow 1964). Frequency analysis methods are based on

approximating the value of a random variable with a

probability density function, from which frequencies of

occurrence may be assigned. Regression and correlation

analyses deal with the description of the relationship

between two or more variables.

















R9












EMPIRICAL
ANALYTICAL SIMULATION








STATISTICAL




Figure 5. Characterization of the various methodologies.







10

4. Digital simulation methods were developed to exploit the

ability of high-speed computers to manipulate

mathematical expressions. The main objective of these

methods is to deterministically model the dynamic

processes in a physical system.

Rarely does a definitive demarcation exist between solution

methodologies; there are overlaps and extensions from one to

the next. As a model, each method represents

simplifications, compromising between ease of application

and accuracy. The empirical and statistical approaches

provide first-cut approximations based on a small data

requirement. The more complicated simulations are generally

regarded as more accurate, although they may have extensive

data or computational requirements.















CASE STUDY


CATCHMENT CHARACTERIZATION

User-supplied catchment data are input for most models.

The extent of the data collection is dependent on the

requirements of the particular method employed. As

indicated in Figure 4, there are no streams, lakes or

groundwater flows. For simplicity, there was no initial

abstraction, areas of depression storage or other

consumption of water. The flow routing geometry was kept as

simple as possible. Conceptually, the catchment was a

sloping plane with no gutter or pipe networks. All the

runoff flowed directly to a dummy outlet on the downslope

side. The runoff from the entire catchment was routed to

the proposed basin, and was subsequently discharged to a

local receiving water. The data were based on observed

values for a drainage basin in Gainesville, Florida, and are

presented in Table 1 (Huber, et al. 1981). In an actual

catchment, waste characteristics would be obtained by

running column settling tests on runoff samples.









Table 1. Case study catchment characteristics.


Total area = 24.7 acres
Impervious area = 37 percent
No depression storage or initial abstraction
Average catchment slope = 0.040 ft/ft = 211 ft/mile
Maximum infiltration = 2.5 in/hr
Minimum infiltration = 0.52 in/hr
Evaporation = 0. 1 in/day
Population density = 500 people/square mile




RAINFALL-RUNOFF CHARACTERIZATION

Rainfall Data

As the storage and flows of water are the principal

elements in the catchment system, rainfall is the driving

force. Long-term rainfall data are available on several

time bases, e.g. continuous gages or discrete hourly, daily,

monthly or yearly records. Rainfall data are characterized

by volume (depth over the catchment area), average

intensity, duration and time between events. The rainfall

data source utilized for this study was the National Weather

Service (NWS) tape for 24.6 years (June 1948 December

1972) of hourly rainfall at Atlanta, Georgia. The standard

NWS format is to record hourly rainfall values in hundredths

of an inch on days when there is rain. Days without rain

are not recorded on the tape. Hourly data for the first day

of each month are recorded regardless of whether it rained

or not.









SYNOP

For methods requiring average event statistics, hourly

rainfall data may be analyzed with SYNOP, a computer package

developed by Hydroscience, Inc. (1979) to determine synoptic

statistics of data time series.

Rainfall volumes, intensities, durations and interevent

times are the principal parameters evaluated in SYNOP.

Available options include complete statistics on an event

basis and time basis, e.g. yearly averages. Cumulative

conditional probabilities (i.e., given that rain has

occurred) and return periods for hourly magnitudes are also

calculated, based on the California method of probability

plotting. The grouping of hourly data into storm events is

based on the minimum number of dry hours between rainfalls,

an input variable termed the minimum intervent time.

Assuming that the storm events occur as a Poisson process,

the time between events is exponentially distributed. The

exponential distribution is a special case of the gamma

distribution with the coefficient of variation equal to

unity. Figure 6 demonstrates the relationship between

theoretical and observed results for the cumulative

distribution of interevent times. The gamma distribution

has been widely applied in hydrology (Haan 1977). To define

events, the minimum interevent time is varied to obtain a

value close to unity for the coefficient of variation (cv)

associated with the interevent time. The SYNOP manual
































































Figure 6.


97
Li
> 96

0 95 =,25 -
94
< 93
2 92




0 9
9o

0



80

50 V= 1.25/

CL,

0 -T

0 2 a -5-6




MULTIPLES F THE MEAN
TIME BETWEEN STORMS


LECGND;
-THEONETICAL GAMMA DISTRIBUTION
0- OBSERVED DISTRIBUTION

NO TE
MINIMUM 6 DRY HOURS BETWEEN STORMS
( :;84HR,V g= 102).



Comparison of theoretical and observed
distributions of-interevent times for
Minneapolis/St. Paul airport.









recommends an initial trial of three hours for the minimum

interevent time. SYNOP was run on the entire 24.6-year

record of Atlanta rainfall to determine the storm

statistics. The results of these runs, presented in Table 2

and Figures 7 and 8, give some idea of the sensitivity of

the results to the choice of the minimum interevent time.

The computer costs averaged $6.25 per run. With eight hours

specified as the minimum number of dry hours defining an

event, the coefficient of variation for the mean interevent

time was 1.004. Values for the means of the parameters were

taken from this run, e.g. the mean volume (Vr) of a rainfall

event was 0.495 inches. Notice that Vr does not equal the

product of Ir and Dr. This is because Vr is the mean of the

products of the individual events' intensity (ir) and

duration (dr), which is not necessarily equal to the product

of the mean intensity (Ir) and the mean duration (Dr), i.e.

Vr = mean (irdr) which is not the same as (mean ir)(mean dr)

= IrDr.




Runoff Quantity

The interaction of rainfall and the catchment generates

runoff. The quantity of runoff is determined by the

influence of infiltration, evaporation, consumption patterns

and land use (Eagleson 1970). The watershed system response

to these interactions has been evaluated by hydrologists for

many years. A comparison of rainfall and runoff time

























Table 2. SYNOP results of 24.6 years of Atlanta, Georgia rainfall.


Minimum # Number Volume cv Duration
of dry hours Vr Dr
(in) (hr)

3 3215 0.367 1.540 4.642

5 2596 0.454 1.424 6.646

8 2331 0.495 1.384 7.824

12 2134 0.552 1.332 9.817


cv Intensity cv Interevent cv
Ir time Ti
(in/hr) (hr)

1.126 0.078 1.372 66.73 1.269

1.124 0.077 1.348 82.62 1.067

1.134 0.077 1.356 90.10 1.004

1.143 0.074 1.334 100.55 0.917










0.6





c 0.5


2
-J


0 0.4-

z
LU

0.3


3 6 9 12


1.5
Z
0
1.4 <


1.3 >
LL
1,2
1--



1.0 L
0

0.9


MINIMUM INTEREVENT TIME, hr


10.0 -


8.0 -


7.0 -


6.0-


5.0 -


H1.5


o0


1. 3


.2
0
1.2 0



L-
1.0 LLJ
0


0 3 6 9 12
MINIMUM INTEREVENT TIME, hr

Figure 7. SYNOP values for mean event volume and duration
as a function of minimum interevent time.


0-0 MEAN DURATION
A-A CV


-I


2











0.079


0. 0.078,

C
0.077


U5 0.076
Z
LJ
I--
Z 0.075


< 0.074
LI 0

0.073


MINIMUM


INTEREVENT


3 6
MINIMUM INTEREVENT


TIME, hr


9 12
TIME, hr


Figure 8.


SYNOP values for mean event intensity and
interevent time as a function of minimum
interevent time.


3 6 9 12


0

1.4
or

1.3
LLI
0
1.2


1.1 -U
LL

1.0 LJ
0
0
0.9






1.5
0
1.4


1.3
LL
1.2
F-
z


LL
U-
1.0 LJ
0
0
0.9


110


100


90


80


70


60


5.0 -
0









series, presented in Figure 9, depicts two phenomena

characteristic of the rainfall-runoff process:

1. a reduction in the number of events due to the capture

of low volume storms by the indigenous catchment storage

capacity, e.g. depression storage and soil moisture

capacity; and,

2. a reduction in the volume of the events due to the

catchment storage and flow interception; e.g.

infiltration rates.

Analyses that deal with single runoff events are not

sufficient to characterize these phenomena because the

catchment storage and interception capacities are functions

of antecedent soil moisture conditions, and are not

constant. It becomes necessary to retain as much.

information as possible on the time between successive

events. The event duration defines the reference time frame

for differentiating between storage and interceptor

elements. A storage element can detain up to a maximum

runoff volume per event, i.e. its detention time is greater

than the event duration. An interceptor, on the other hand,

can capture up to a maximum flow rate before bypassing

Stormwater runoff control devices can also be characterized

by storage and interception capacity, as presented in Figure

10.

A representation of a time series of runoff flows is

presented in Figure 11, a series of flow pulses separated by













MAR APR


Figure 9.


A comparison of rainfall and runoff time series
depicting the reduction in number of events and
the reduction in the event volume.


MAY


JUN


FEB


2.0


1.0


2.0


1.0


11, h 1


Maximum


Minimum
storage


S-or 0e


e -


I I


0,0













a) INTERCEPTION


b) STORAGE


q


nv


TIME
c) INTERCEPTION AND STORAGE





TIME


Figure 10.


Definition of interception and storage for
storm events.


M7T~


































I V.- ,


0 10 20 30


' 130 140 150 160 170


TIME, hr


Figure 11. Representation of a time series of runoff flows.


4.0 H


2.0-l


I I /\ I









relatively long periods of no flow. The determination of

runoff characteristics is a major step in the solution

process and is where the methods vary the most. The

approaches compared here offer a sharp contrast in the

representation of the rainfall-runoff process. The

statistical and empirical approaches summarize runoff

generation via linear conversion factors applied to rainfall

statistics. On the other hand, the simulation technique

utilizes some of the most refined concepts in deterministic

hydrology.




BASIN CHARACTERIZATION

Flow conditions

The input to the basin is the runoff from the catchment

area. The time series of runoff events depicted in Figure

11 suggests two realms of kinetics: rapid, relatively

well-mixed during the runoff event, followed by slower (less

dispersion, turbulence) kinetics and possibly quiescent

conditions during dry weather. Characterizing this time

series of discontinuous flows entering the basin is a major

obstacle in solution methods.

Basin discharge may be either variable, as in the case

of gravity drainage, or constant, via a pump or outlet

restriction. Negative feedback is inherent in gravity

systems, i.e. when the water level is high, the outflow is

high, and as such, tends stabilize the flow. It is







24

difficult to deal with this nonlinearity analytically; it is

far easier to analyze a constant discharge rate. There has

been no evidence to suggest that one is better than the

other for pollutant removal.




Removal

Theory of settling

Pollutant removal via settling is the most widely

useful operation in water and wastewater treatment (Fair,

Geyer and Okun 1968, Liptak 1974). In the design of

sedimentation basins, basic assumptions are incorporated:

1. inlet zone the influent is transformed to a

uniform vertical distribution of particles.

2. settling zone there is steady, uniform flow and

quiescent, discrete and unhindered settling.

3. bottom zone solids which enter the bottom zone are

not resuspended.

4. outlet zone solids that do not enter the bottom

zone leave in the effluent.

These four zones and particle settling paths are shown in

Figure 12. The major design parameter is the overflow rate

(vo), defined as


vo = H/td = (Vs/SA)/(Vs/Q) = Q/SA (1)


where H is the depth of the settling zone C L 3,

td is the detention time of the settling zone C T 1,

Vs is the volume of the settling zone E L3 1,










SURFACE AREA = SA
PERPENDICULAR TO FLOW
CROSS SECTIONAL AREA
= A


TRAVEL TIME L = L
r Q /A


L-A
Q


V
Q


OVERFLOW RATE = Vo = = =
t -/Q SA



Figure 12. Development of overflow rate in an ideal
settling basin.


Q









SA is the surface area of the settling zone C L2 3,

and

Q is the flow in the settling zone E L3/T J.

The settling process is relatively slow, such that it

is the removal rates that are important, rather than the

equilibrium state (Rich 1974). Under ideal conditions,

particles settle at a velocity (vp) governed by Stoke's Law,

and are removed if vp is greater than vo. Additional

particles are removed which enter the settling zone at a

height (hp) less than


hp = vp*td (2)


The total basin removal is given by


R = (1-xo) + (1/vo) 5 vp dx (3)


where R is the pollutant removal efficiency, and

xo is the proportion of particles with vp less than

vo.

Rarely do detention facilities perform under ideal

conditions. Most often, design efficiencies are reduced due

to violations of ideal assumptions caused by short

circuiting and turbulence, which alter the kinetics from

ideal quiescent conditions. Short circuiting is induced

through thermal currents, wind action, influent inertia,

etc. Resuspension of solids may occur as the flow rate

exceeds the scouring velocity. Thomas and McKee derived the







27

effect of longitudinal dispersion in a basin consisting of n

completely mixed plugs (Fair, Geyer and Okun, 1968). Figure

13 presents the relative effluent concentrations for an

instantaneous injection of dye undergoing a first order

decay as it passes through the basin. A completely mixed

basin is shown as n=1, while an ideal plug flow basin

(n=infinity) would be represented by a spike at t/td of

unity. The net effect of altering the flow regime from

quiescent to more turbulent conditions is the reduction of

the reaction coefficient k (Fair, Geyer and Okun 1968; Rich

1974). Although there is no way to predict before operation

the reduction for a particular basin, the phenomena can be

represented as in Figure 14, where the reduction of k is

given as


k/kO = (1-c/cO)**n (4)


where k is the reaction coefficient,

kO is the reaction coefficient under quiescent

conditions,

c is the effluent concentration,

cO is the initial pollutant concentration, and

n is the coefficient indicating the degree of

turbulence, which increases as the flow regime

diverges from ideal conditions.

Plug flow conditions are represented by n=0 where k=kO and

the pollutant decays according to


or, R = 1-c/cO = l-e**(-kt).


c/cO = e**(-kt)


(5)





























Relative time, t/ltd


Figure 13.


I.
o

0 0.
0
o0.



00.
0. -

0

0.

0. ^
_ 0C
[T (


Figure 14.


Effluent concentrations for a first-order
removal process in n completely mixed plugs.


0


8
7
6 -
5



02 . ..
0 _^ ,^___


0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 45 5.0
RELATIVE TIME, t/td


Removal efficiency for a first-order removal
process demonstrating the effect of increased
turbulence; n=-1 for quiescent and n=1 for
completely mixed.









Comparison of Flow Conditions

Plug flow and completely mixed removal regimes define

the extremes of pollutant removal performance. By

definition, the hydraulic regime in each plug is complete

mixing. Therefore, under static conditions, e.g. a column

settling test, there is no difference between the two. The

applicability of transferring column settling test results

to dynamic conditions was recently re-examined by White

(1976). In tank studies, he observed reasonable agreement

with column tests, although the results were highly

dependent on the waste characteristics. The effect of

longitudinal dispersion on plug flow performance has been

studied to correlate plug flow and complete mixing under

dynamic conditions (Weber 1972). Ideal plug flow is

represented as having zero dispersion while complete mixing

is assumed to have infinite dispersion. An example of the

effect of dispersion is reflected in Figure 15, showing

effluent responses to a step input. Weber (1972) presents

steady-state solutions that were derived from the general

continuity equation fitted to a dispersion model. For a

first-order reaction, the plug flow solution (zero

dispersion) is


c/cO = e**(-ktd) or, R = l-e**(--ktd) (6)


For a completely mixed basin, the steady-state solution is


or, R = 1-1/(1+ktd)


c/cO = 1/(1+ktd)












1.0
0.8
C. /C 0.6
o 0.4
0.2











1.0
0.8

C /C 0.6
0.4
0.2


INFLUENT


EFFLUENT
plug flow, 0 intermediate amount of dispersion
v x 0. 0025
large amour of dispersion d .
d complete mix, -
0,2 vxL
v L Dx
.,small amount of dispersion, -0. 002

0 0.5 1.0 1.5 2.0


B tvx/L




Figure 15. Effluent responses to a step input.








I c..- (complete mixing) First order

^64 kL/va 200 ....


16 100


50 (Lines of equal volume
l. or holding time)










0 (piug flow) -






Figure 16. Comparison of real and plug flow reactor
volumes for ia first-order reaction.









These two solutions represent the removal efficiency

extremes. The lower removal efficiency of complete mixing

is a result of the basin concentration being continuously

mixed with the influent, yielding a dilution of the influent

but a concurrent increase in the basin concentration. In

plug flow, the basin effluent more clearly reflects the

removal process occurring in the separate plugs. The

difference between the steady-state solutions for

first-order reactions is graphically presented in Figure 16,

which compares the volume of an ideal plug flow basin to the

volume of a basin with dispersion yielding the same removal

efficiency (Weber 1972). From Figure 16 it is possible to

predict the effect of implementing dispersion reducing

mechanisms such as baffles. For example, by reducing the

dispersion factor from infinity (completely mixed) to 1.0,

the volume necessary to provide 90 percent removal is

reduced by a factor of 2.

There has been no widely used method for sizing S/T

facilities with long-term stormwater quality control as the

main objective. Heaney, et al. (1979) presented the

mechanics for determining the optimal combination of S/T for

steady state conditions, subject to economic constraints.
















METHODOLOGIES


ANALYTICAL

Analytical models are developed from a combination of

the general mass continuity and advective-dispersion

equations. They range in complexity from simple

steady-state approximations to more complicated variable

flow solutions. The complexity of solution for detailed

problems intimidates users and prohibits widespread

application. Customarily, a simplification of the system's

dynamic processes is necessary to obtain a solution. The

analytical approach to stormwater detention facilities has

been advanced by Medina (1976). Constant and variable

(linear) flow and storage conditions were studied.

Conservative and nonconservative pollutant routing were

included. Effluent concentrations were derived for "simple"

forcing functions. Application of this method requires the

runoff flows and pollutants be converted to one of the input

functions for which solutions have been derived. The

solutions are limited in application due to their complexity

and sensitivity to storage and flow conditions, i.e. the

need to reduce the complexity of the dynamics to facilitate

the analytical solution. The complexity of these solutions










precludes their application to stormwater runoff events

without the use of detailed simulation.




EMPIRICAL

A widely used method for characterizing the

rainfall-runoff process is the application of a conversion

factor to mean rainfall values. Conversion factors are

catchment specific and have been correlated to

imperviousness, land use, population density and depression

storage (Hydroscience, Inc. 1979, Chow 1964). While this

method accounts for an average reduction in the volume of

storms, it does not correct for the decreased number of

events, i.e. it doesn't account for the small volume storms

that are retained in the catchment storage.

Empirical relationships for runoff water quality have

also been developed. Receiving water studies have

determined runoff coefficients for with nutrient loadings

based on land use (Reckhow 1980). In the urban arena,

Smolenyak (1979) determined coefficients of the power

equation


load = a(flow)**b (8)


where load is the pollutant load in the runoff E M 3,

flow is the runoff flow C L3/T 3, and

a,b are coefficients.










Values determined from data in the Urban Rainfall-Runoff

Quality Data Base are presented in Table 3.

The practice of water quality control via settling is

inherent in most applications of natural waters. Water and

wastewater treatment employ several variations of settling

basins. Natural hydrologic systems incorporate detention

for both the storage of kinetic energy and the deposition of

sediment load. As a result of the implementation of

reservoirs, siltation studies have contributed to the

identification of the relationships between sedimentation

and hydraulic parameters. The most popular sediment trap

studies are those of Brown, Brune, Churchill and Camp (Ward

and Haan 1977). Chen (1976) presents the historical

development of theoretical analyses of sediment retention.

These methods require the removal assessment of detention

facilities based on an annual time frame. As there was no

way to reconcile this with the stochastic nature of runoff

events, these empirical methods were not investigated

further. A recent review by Nix, et al (1980) summarizes

the benefits in utilizing detention facilities for

stormwater quality improvement. Several combinations of

empirical and statistical approaches were suggested for

design purposes.












TABLE 3.

Relationships between Pollutant Loads and Flow Volume (FLOW).


Dependent
Variable


BOD

COD

NH3N

NITN

NTOT

ORGN

TOTN

DOP

TOP

TOTOP

TOTP

TPHOS'

TOTS

TSS


Sig, Level
F-Test


.28

.76

.44

.80

.57

.88

.74

.83

.46

.27

.66

.91

.69

.56


.99

.99

.99

.99

.99

.99

.99

.99

.99

.90

.99

.99

.99

.99


No. of
Events



80

157

20

21

103

40

37

34

119

11

53

8

41

260


Model Load = a(FLOW)b
Reg. Coef.
a b


34.0

29.8

.215

.119

.0400

.856

.304

.0648

.0104

.0800

.426

.105

279

44.2


1.12

1.08

.72

.80

.71

1.04

1.07

.98

.78

.55

1.5

1.05

1.41

1.10


Source: smoltenyak 1,979









SIMULATION

Several models are available for the simulation of the

rainfall- runoff process and detention pond performance.

Simulation is amenable to a trial and error routine, leading

to search techniques for determining an acceptable design.




STORM

The Storage, Treatment and Overflow Model (STORM) of

the Corps of Engineers models wet-weather flow through

separate storage and treatment facilities (U.S.A.C.O.E.

1974). Runoff quantity is generated from an empirical

conversion factor applied to rainfall excess. Runoff

quality is generated as a function of land use. While flow

is routed through the treatment plant, there is no capacity

to model the quality improvement provided by detention.




SWMM Overview

The Storm Water Management Model (SWMM) is one of the

most comprehensive and well documented models available for

the analysis and design of urban stormwater systems (Huber

et al. 1980). SWMM was developed by Metcalf and Eddy, Inc.,

the University of Florida and Water Resources Engineers,

Inc. under a contract for the Environmental Protection

Agency. Version I was released in 1971 and has been

continually undergoing revision and updating. This study









utilized Version III, released in 1980, and also

incorporated further refinements. While the model allows

for extensive watershed simulation, this study restricted

the analysis to a simplified Runoff Block and concentrated

on the Storage/Treatment Block.

The Runoff Block deterministically models the

rainfall-runoff process in a catchment. Physical data such

as depression storage, infiltration rates, soil

characteristics, catchment area, ground slope, gutter

network, evapotranspiration rates and hourly rainfall values

are included as the model input. Pollutant build-up and

wash-off functions are available, as is pollutant generation

based on catchment land use. The Runoff Block generates

hydrographs and pollutographs as options. Several inherent

adjustment factors are available for calibration of the

model.

The Storage/Treatment (S/T) Block of SWMM models the

flow and pollutant routing through a storage and/or

treatment device which can be either a detention or

non-detention unit. Geometric and hydraulic relationships,

e.g. depth to surface area, and evaporation rates, and

incoming flows are included as data input for the S/T block.

Discharge can be modeled as either constant or variable

outflow. The package provides for a variety of removal

mechanisms and flow routing options, capable of tracking

both stormwater flows and constituent pollutants.









Data Input for Runoff Block

The Runoff Block accepts hourly rainfall data in the

National Weather Service (NWS) format. No gutter or pipe

networks were used in the example catchment; all runoff

exited the area via a dummy outlet. The impervious area was

modeled separately from the pervious area. Runoff generated

on either area went directly to the dummy outlet without

passing through the other area. Also, there were no areas

of depression storage in the modeled catchment. The data

input for the Runoff Block consisted of the catchment

characteristics presented in Table 1. A complete listing of

the data input is given in Appendix A.

Currently, continuous SWMM can only be run with hourly

rainfall input. As an aside, the hourly rainfall data were

transformed into daily and weekly records. SWMM Runoff

Block was run on the first 19 months of the Atlanta data and

the results are compared in Table 4.



Table 4. Comparison of Runoff Block results using
hourly, daily, and weekly rainfall input.


Rainfall Average Total Infilt. Evap. Runoff
Time Step Intensity Rainfall (in) (in) (in)

HOURLY HOURLY 81.68 51.72 6.39 28.68

DAILY DAYTOT/Dr 80.53 50.46 8.18 28.00

WEEKLY WEEKTOT/2Dr 81.62 51.73 6.11 28.62







39

Reasonable agreement was obtained, suggesting that if hourly

data were not available, continuous SWMM could be run using

daily or weekly rainfall data.




Data Input for Storage/Treatment Block

The basin geometry and hydraulic characteristics

required as input data for the S/T block were obtained from

a separate computer program, BASIN, written for that

purpose. BASIN calculates stage to surface area, stage to

volume and stage to discharge relationships for basins given

the dimensions, side slope and outlet configuration.

Examples are provided in Figure 17. The program development

and listing is provided in Appendix B.

As a reference, basin volumes (Vb) were normalized to

the mean runoff volume per event (Vro), yielding a

normalized volume ratio (Vb/Vro). Currently, SWMM does not

define storm event statistics, so the SYNOP program was

employed. This involved running the Runoff Block with the

complete 24.6-year (June 1948 December 1972) rainfall

record and generating 24.6 years of simulated runoff data.

These data were transformed to the format of the NWS

rainfall data, which is compatible with the SYNOP input

format. As with the rainfall data, the minimum interevent

time was varied to obtain the coefficient of variation (cv)

for the interevent time close to unity. The results of

these runs are presented in Table 5.










.i*-
C-
0a
0
0
0




LUI




C/)


I 2 3 4 5 6


DEPTH, ft


0 1 2 3 4 5 6

DEPTH, ft


30


25
4)
20
2n 20



0 15
W10 -


|._J 5
0


0 I 2 3 4 5 6


DEPTH, ft


UJ
D
_j
0
>

L)

-0-
0


S I I I I I
0 I 2 3 4 5 6

DEPTH, ft


Figure 17. Stage relationships as calculated by BASIN;
constant and variable discharge.


VOLUME











DISCHARGE





















Table 5. SYNOP results of 24.6 years of simulated runoff data.


Minimum # Number Volume CV Duration CV Intensity CV Interevent CV
of dry hrs Vro Dro fro Time Ti
(in) (hr) (in/hr) (hr)

3 2124 0.180 1.253 4.507 0.964 0.041 1.057 101.35 1.172

4 1998 0.192 1.221 4.980 0.978 0.041 1.058 101.75 1.114

5 1903 0.201 1.196 5.428 0.989 0.041 1.063 113.13 1.070

8 1760 0.217 1.194 6.438 1.030 0.040 1.073 122.32 0.999

12 1646 0.274 1.181 7.448 1.093 0.039 1.052 130.80 0.940









A minimum interevent time of four hours resulted in a

cv of 1.000. The mean runoff volume per event (Vro) was

determined to be 18022 cubic feet, based on the mean depth

of 0.201 inches over 24.7 acres. The number of events was

reduced from 2391 rainfall events (Table 2) to 1998 runoff

events. Continuity is checked by comparing the product of

the mean event volume and the total number of events with

the amount of runoff generated by the Runoff Block,

(0. 19Z inches/event)(."199 events) = 383,61 inches;

from Runoff, 388 inches.

Test basin volumes (Vb) were obtained from BASIN to closely

approximate volume ratios (Vb/Vro) of 0.5, 1.0, 2.0, 4.0,

10.0. For example, a Vb/Vro ratio of 0.50 implies that the

empty basin volume is 50 percent of the mean storm event

volume. The basin volumes used are presented in Table 6.



Table 6. Determination of normalized volume ratios (Vb/Vro).


Basin volume (Vb) Normalized volume
(cubic feet) ratio Vb/Vro

7744 0.430

15014 0.833

29064 1.613

74420 4.129

167835 9.313







43

The input data for the Runoff Block are provided in Appendix

A. The S/T Block utilized the runoff values generated from

the Runoff Block. To save execution time and money, the

Runoff Block was run once and the output stored on an

interface data disk. The interface data set served as the

input to S/T for the subsequent simulations. A constant

suspended solids concentration of 100 mg/1l was assigned to

the influent. This was chosen as opposed to generating

pollutants from the catchment area for four reasons:

1) it avoids concern over how the pollutants are

generated

2) it provides a base value (100 mg/1l) for future

comparisons;

3) it provides a blocked-off step input as shown in

Figure 18; and,

4) a constant influent concentration establishes that

the percent of flow bypassed is numerically equal to

the percent of the pollutant bypassed.

The actual flow condition in a basin is neither plug

flow nor completely mixed, but somewhere between, termed

"intermediate mixing". The complete mixing option of S/T

was chosen for the flow routing regime for its analytical

and computational simplicity, resulting in lower simulation

costs than the plug flow method.

Two options of basin discharge were explored: variable

outflow based on hydraulic head above an outlet, and a























I I


Figure 18.


Time series of constant influent pollutant
concentration.


LL
LL

:LUL


n7


H









drawdown scheme which emptied the basin at a constant rate

(see Figure 17). The pumping option was run for comparison

with the statistical technique, while the variable discharge

option was utilized for application to basins with gravity

drainage.




Constant Discharge Simulation

A variable volume, constant outflow unit was simulated

by using the pumping option of S/T. As a reference, the

constant drawdown rate was normalized as QcTi/Vro

where Qc is the drawdown rate in cubic feet per hour,

Ti is the mean interevent time in hours, and

Vro is the mean runoff volume in cubic feet

(Hydroscience, 1979).

As Ti and Vro are constants determined from SYNOP, variable

values of the ratio reflect different drawdown rates. These

rates were calculated to yield ratios of 1, 2, 4, 7, and 10

and are presented in Table 7.

It was assumed that drawdown occurs whenever there is

water in the basin. The effect of drawdown height (Hd) and

drawdown rate on capture and removal efficiencies was


analyzed.









Table 7. Calculation of constant discharge rates (Oc)
(Ti = 111. 17 hours; Vro = 18022 cubic feet).


Normalized discharge Drawdown rate (0c)
ratio OcTi/Vro (cubic feet/hr)

1 162

2 324

4 648

7 1135

10 1621




Variable Discharge Simulation

A variable volume, variable outflow control unit was

simulated with the S/T Block by utilizing a power equation

for basin discharge based on hydraulic head. This simulates

the hydraulics in a basin with gravity drainage. The outlet

characteristics were arbitrarily assigned as a six inch

circular opening placed one foot above the bottom. Instead

of a solution surface, as was provided in the constant

discharge simulations, a single removal curve was determined

for basin performance versus basin volume. In an analogous

manner to the drawdown height and rate combinations, outlet

elevation and cross-sectional area were recognized as design

parameters for basin performance and were analyzed.

Evaporation in the S/T unit was arbitrarily assigned a value

of 0. 1 inch per day.









Removal Mechanism

A removal equation was chosen of the form


R = Rmax(1-e**(-kt)) (9)


where R is the pollutant removal efficiency,

Rmax is the maximum removal efficiency,

k is the first-order rate coefficient E 1/T 3, and

t is the treatment time E T 3.

Fair, Geyer and Okun (1968) presents general removal

curves with k near 1.4 per hour for TSS and 0.50 per hour

for BOD, with Rmax of 0.75 and 0.45, respectively. For

design purposes, values for Rmax and k would be determined

from column settling tests with representative pollutants.

Values of 1.0 and 0.6 per hour, respectively, were

arbitrarily assigned for these parameters. In the S/T

Block, removal is accounted for once per time step, with the

length of the time step, one hour, as the treatment time.

The removal equation is presented in Figure 19. Because the

time step was held constant throughout the simulation there

was a constant percent removal (45 percent) of pollutant per

time step. A comparison of effluent concentrations for

removal governed by this equation in an ideal plug flow

basin, an ideal completely mixed basin and the S/T complete

mixing regime is presented in Figure 20. As shown, the S/T

results lie within the extremes of pollutant removal

efficiency provided by ideal plug flow and complete mixing.








1.00



0.9 -



0.8 -



0,7 Removal (

t

0.6 0
0
1
f 2

0.5 3
z 5
W 0.451

LL 0.4



O> 0.3 -
LUJ




0.2 -



0.1



0.0
0.0 ---- -------------
0 1 2. 3

TREATMENT TIME hr


Removal equation used in SWMM S/T Block.


Figure 19.













IDEAL PLUG FLOW


0.9


0.8


0.7


0.6


0.5


0.4


0.3


0.2


0.1


0.0


SWMM S/T COMPLETELY MIXED

IDEAL COMPLETELY MIXED


= 10 ft~hr


td = 3hr


3 4 5 6 7 8 9

TIME, hr


Figure 20.


Comparison of effluent concentrations under ideal
plug flow, ideal completely mixed and as
calculated with SWMM S/T completely mixed
routing; step input of pollutant.


0.83



0.72

0.64


0 I 2


SII I I.I...


I I I I ~









One-year Versus 24.6-year Simulation

The initial decision to use one year of data as opposed

to the full 24.6-year history was based on economic

considerations; mistakes and debugging were expensive enough

without extra data magnifying the costs. The first 12

months of the input record (June 1,1948 May 31,1949) were

utilized to get the simulator running. It was realized that

if a "typical" runoff year's simulation adequately

reproduced long-term basin performance, the costs of the

analysis would be reduced by as much as an order of

magnitude. A "typical" year, 1953, was chosen on the basis

of similar synoptic statistics, as determined by the SYNOP

run of the 24.6-year runoff data. A comparison of 1953

runoff parameters with those of the 24.6-year time series is

presented in Table 8.

The adequacy of one year's simulation for describing

the long-term basin performance was analyzed in a series of

simulations; the results are presented below.




Runoff Block Results

A variety of summary print options are available in the

Runoff Block, from detailed hourly results to the total

simulation summary, as presented in Table 9. On the

hypothetical catchment, 704 inches (60 percent) of the total

1179 inches of rain left via infiltration. A total of 388

inches (33 percent) accumulated as runoff, while 150 inches
























Table 9. Runoff Block 24.6-gear simulation summary.

MILLION
CUBIC FEET

TOTAL PRECIPITATION (RAIN PLUS SNOW) 105.512

TOTAL INFILTRATION 63. 106

TOTAL EVAPORATION 7.846

TOTAL GUTTER/PIPE/SUBCAT FLOW AT INLETS 34.830

TOTAL WATER REMAINING IN GUTTER/PIPES 0.000

TOTAL WATER REMAINING IN SURFACE STORAGE 0.000

$ ERROR IN CONTINUITY, % OF TOTAL PRECIP -0.256

* RUNOFF SIMULATION ENDED NORMALLY * *


INCHES OVER
TOTAL BASIN

1176.79

703.83

87. 51

388.47

0.00

0.00









Table 8. Comparison of mean runoff event parameters
for 1953 with the 24.6-year record
(minimum interevent time = 4.0 hours).


Volume (in)

cv

Duration (hr)

cv

Intensity (in/hr)

cv

Interevent time (hr)

cv


24.6-year

0. 201

1.222

5.066

0.972

0.043

1. 148

111.71

1.000


1953

0.217

0. 948

6. 190

1. 087

0.044

1. 164

104.83

0.986


(8 percent) were lost to evaporation. Mass continuity was

preserved within 0.3 percent over the total 24.6-year

simulation.




Storage/Treatment Block Results

The format for the S/T results are similar to the

runoff output with more emphasis on quality parameters.

Again, as shown in Table 10, the results were presented to

facilitate continuity checks.












Table 10. Examples of Storage/Treatment Block summaries.


DETENTION UNIT CHARACTERISTICS:
POLLUTANT ROUTING METHOD : COMPLETELY MIXED
RESIDUALS DRAW-OFF SCHEME: NEVER DRAWN OFF
DEPTH-AREA-STORAGE-FLOW RELATIONSHIPS :


DEPTH,FT.
0. 0
0. 50
1. 00
1. 25
1. 50
2. 00
2. 50
3. 00
3. 50
4.00
4. 50


SURFACE AREA,SG.FT.
5000. 0
5304. 0
5616. 0
5775. 0
5936. 0
6264. 0
6600. 0
6944. 0
7296. 0
7656. 0
8024.0


STORACE,CU. FT.
0. 0
2576.0
5306. 0
6729.9
8193. 7
11243. 7
14459.7
17845.7
21405. 7
25143.7
29063.7


* GOVERNED BY PUMPING

PUMPED OUTFLOW:
DEPTH AT WHICH FIRST PUMPING RATE DEGINS,FT.
DEPTH AT WHICH SECOND PUMPING RATE BEGINS,FT.
FIRST PUMPING RATE, CFS
SECOND PUMPING RATE,CFG
DEPTH AT WHICH ALL PUMPING STOPS,FT.


UNIT PARAMETER VOLUME
(CU. F. )
-- ---- --------- ----------


1 INFLOW,TOTAL
INFLOW,NET
BYPASS
TREATED OUTFLOW
RESIDUAL FLOW
REMOVED BY DECAY
REMAIN.TOT.VOL.
EVAPORATION


0.1456E+07
0.1386E+07
0.7027E+05
0. 1364E+07
0. 0
0.5285E+04
0. 1700E+05


CAT KAKA
LS.

0. 4547FA-04
0.4327E+04
0.0
0.1422E+04
0.0
0.2913E+04
0.3216E-01


0.0


0. 0
0. 0
0. 45
0. 45
0. 0









Constant Discharge Simulation

Performance results from the constant discharge

simulations are presented in Figures 21, 22 and 23, and in

Table 11.



Table 11. Estimates of flow capture efficiency (C) and
pollutant removal efficiency (R) as a function of
basin volume and constant discharge rate:
simulation results.


Normalized Volume Ratio

Normalized 0.43 0.83 1.61 4.13 9.31
Discharge
ratio 1953 24.6 yr 1953 1953 24.6 yr 1953 1953 25-yr

1 C 0.344 0.331 0.476 0.613 0.640 0.852 1.000 0.952
R 0.343 0.322 0.476 0.598 0.638 0.847 0.989 0.941

2 C 0.398 0.552 0.721 0.945 1.000 *
R 0.373 0.527 0.690 0.906 0.961 *

4 C 0.468 0.461 0.621 0.804 0.795 0.962 1.000 0.993
R 0.373 0.383 0.532 0.712 0.705 0.864 0.905 0.896

7 C 0.542 0.699 0.839 0.971 1.000 *
R 0.365 0.524 0.661 0.792 0.817 *

10 C 0.611 0.608 0.744 0.868 0.876 0.982 1.000 0.999
R 0.342 0.351 0.483 0.603 0.619 0.713 0.728 0.736


* indicates that simulation was not run.



The 1953 simulations duplicated the performance results

of the 24.6-year simulations within five percent over the

entire spectrum of basin volumes and discharge rates.

Figure 21 presents the solution surface for capture

efficiency as a function of drawdown rate and basin volume.









10.0



9.0


6.0



5.0



4.0



3.0


2.0

.70

1.0 0.60
50


0.0 I I I
0 1 2 3

NORMALIZED DI1

Figure 21. Solution surface
as a function of
discharge rate:


CHARGE, QcTi/Vro

of flow capture efficiency (C)
basin volume and constant
1953 simulation results.


















8.0



7.0


6.0



5,0



4.0



3.0


2.0 -



1.0




0


NbRMALIZED


Figure 22.,


DISCHARGE, QcTi/Vro


Solution surface of pollutant removal efficiency
(R) as a function of basin volume and constant
discharge rate; 1953 simulation results.






























0.5 -



04 -
0.43


0.3



0.2



1.0



0 .0 Ill1lll[
0 2 3 4 5 6 7 8 9 10

NORMALIZED DISCHARGE, QcTi/Vro


Figure 23.


Pollutant removal efficiency (R) as a function of
basin volume and constant discharge rate: 1953
and 24.6-year simulation results.









Isopleths of percent capture were drawn by linear

interpolation between calculated values. As expected, flow

capture was greater as the drawdown rate increased, due to

an increase in the effective volume. Also, as the basin

volume increased, the capture efficiency increased due to

less bypass. The vertical distance between the isoquants

represents the sensitivity of capture performance to basin

volume; the smaller the distance, the greater the

sensitivity. for drawdown rates greater than 4.0, there

appears to be uniform sensitivity to basin volume. The

lowest sensitivities occur at the lower drawdown rates

(QcTi/Vro less than 2.0). The isoquants converge slightly

toward the upper end of the abscissa. The horizontal

distance separating the isoquants represents the sensitivity

of capture performance to the drawdown rate. The isoquants

become parallel to the abscissa above QcTi/Vro of 4.0,

implying relative insensitivity to drawdown rate.

Sensitivity is increased as the drawdown rate is decreased.

Figure 22 presents the solution surface for pollutant

removal efficiency as a function of basin volume and

drawdown rate. Unlike the solution surface of capture

performance, the isoquants in Figure 22 slope upward after

an initial negative slope. The result is a solution surface

which allows more than one drawdown rate at a specific basin

volume to achieve the same removal performance. This

demonstrates the performance tradeoff of providing a larger









effective volume by emptying the basin quicker versus

providing a longer treatment time, although bypassing more

flow.

Combinations of basin volume and drawdown rate yielding

equivalent removal efficiencies are depicted along

isoquants. For example, the removal performance obtained by

a Vb/Vro of 4. 1 and a QcTi/Vro of 2.0 was the same as a

Vb/Vro of 9.3 and a normalized discharge rate of 4.0. The

greatest removal occurred in the region of large basin

volumes (Vb/Vro > 4.0) and low drawdown rates (QcTi/Vro <

4.0). Figure 23 presents the removal performance in a

different manner than in Figure 22. There is no increase in

information by presenting the results in this way, although

the communication of information is improved. For example,

in Figure 23, it is easier than in Figure 22 to see that the

sensitivity of removal to drawdown rate increases as the

volume ratio increases. Combinations of basin volume and

drawdown rates yielding equivalent removal as well as the

sensitivity of the removal performance to drawdown rate are

demonstrated. For example, the sensitivity of performance to

drawdown is represented as the slopes of the curves, and is

seen to increase as the volume ratio increases. As the

volume ratio increases, the maximum removal efficiency for

each volume occurs at decreasing drawdown rates. The 1953

performance curve for a Vb/Vro of 9.3 is depicted as a

straight line. This represents a divergence (5 percent at







60

QcTi/Vro = 1.0) from the 24.6-year results, possibly due to

the lack of a large storm during 1953. All of the curves

converge to 4.3 percent removal at QcTi/Vro of 0.0, that is,

in the case where there is no outlet. Long-term removal

efficiency would undoubtedly be smaller for this case,

tending to zero percent.




Variable Discharge Simulations

The effect of basin volume on capture and removal

performance in basins with variable outflow rates was

analyzed by running the SWMM S/T Block on five sets of basin

geometry and hydraulic characteristics. The results

presented in Figure 24 follow an intuitive removal

relationship with increased removal as the storage capacity

increases. The regions below Vb/Vro of 0.43 and above

Vb/Vro of 9.31 were not explored because of the unlikeliness

of such a small volume ratio. The resultant removal curve

is neither an exponential nor a power equation for the range

observed.

Only three 24.6-year simulations were run due to their

low marginal benefit, i.e. the one-year simulations gave

estimates close enough to the 24.6-year results to avoid

spending the extra money for the long-term simulations. The

costs of the runs averaged $0.60 for one-year and $6.25 for

24.6-year simulations.










1.00




0.9



0.8 -



0.7 -



0.6 -




0.5
z
0

00.4
Ld
-J
5 0.3 -

w
LUl

0.2 -



0. -









Figure 24.


1953 RESULTS
A 25-YEAR RESULT



25-year 1953

Vb/Vro C R C R

0.43 0.903 0.241 0.904 0.251
0.83 0.898 0.529
1.61 0.967 0.643 0.952 0.641
4.13 1.000 0.860
9.31 1.000 0.916 1.000 0.918



J.


1 2 3 4 5 6 7 8 9 10

NORMALIZED VOLUME, Vb/Vro

Pollutant removal efficiency (R) as a function of
basin volume under variable discharge conditions.:
1953 simulation results.









Optimal Basin Design

During preparation of the initial performance solution

surfaces, it was recognized that the heights at which

discharge began and ended would combine with the discharge

rate to affect basin performance. Intuitively, increasing

the height would decrease the capture efficiency, but due to

the completely-mixed flow routing regime, the remaining

volume would provide dilution of the influent.

Combinations of drawdown height (Hd) and rate (Gc) and

similarly outlet diameter and invert height, were simulated

in an attempt to develop guidelines for the optimal design

of detention facilities.

A basin with a Vb/Vro of 1.61 was utilized for these

simulations. The results from the constant drawdown

simulations indicated that a basin with this ratio had the

greatest performance sensitivity (26 percent capture and 11

percent removal) over the range of drawdown rates. In the

variable discharge runs, a basin with a Vb/Vro of 1.61

yielded results in the knee of the removal curve. It was

felt that a basin with this ratio was sensitive enough to

reflect the effect of height and discharge combinations on

basin performance.

The results of the constant drawdown rate simulations

are presented in Figures 25 and 26 and Table 12.

The solution surface in Figure 25 indicates that maximum

performance is achieved by a normalized discharge ratio



















3.0


.-o



z



Z:
0


L0
-J
Ld
z


0
C
c
cc
S


Figure 25.


2 4 6 8 10 12 14 16 18 20

NORMALIZED DISCHARGE, QcTi/Vro

Solution surface of pollutant removal efficiency
(R) as a function of drawdown height and
drawdown rate (Vb/Vro = 1.61): 1953 simulation
results.



















































0 2 4 6 8 10 12 14 16 18 20


NORMALIZED DISCHARGE,


Figure 26&


Q cT / Vro


Pollutant removal efficiency (R) as a function of
drawdown height and constant discharge rate
(Vb/Vro = 1.61).: 1953 simulation results.


0.5



0.4



0.3



0.2


1.0



0.0









Table 12.


Estimates of flow capture efficiency (C)
and pollutant removal efficiency (R) as a
function of drawdown height and rate
(Vb/Vro = 1.61): 1953 simulation results.


Drawdown Height (ft)


Normalized
Discharge
Ratio

1 C
R

2 C
R

4 C
R

7 C
R

10 C
R

12 C
R

15 C
R

18 C
R

20 C
R


* indicates


that simulation was not run.


of 12.0 with an Hd of one foot


Removal efficiencies for


basins with complete drawdown (Hd of 0.0) were lower than

limited drawdown (Hd greater than 0.0) for all but the

lowest rates. The general trend of the isoquants indicates

that similar performance can be achieved by a low drawdown


0


0.613
0. 598

0.721
0.690

0.804
0.712

0.839
0.661

0. 868
0.603

0.883
0. 570

0.900
0.520




*i
*t


0. 578
0. 578

0.679
0.679

0.756
0.734

0.804
0.752

0.837
0. 761

0.860
0.766

0. 874
0.749

0. 85
0.730

*


0. 517
0. 517

0. 610
0. 610

0. 687
0. 687

0. 749
0. 728

0. 778
0. 738

0. 795
0. 744

0. 831
0. 763

0. 845
0. 758

0. 851
0. 761


4


*
*

















0. 610
0. 597

0.648
0. 621

0.663
0.636


*f
*




*f
*f

*f





0. 717
0. 694

0. 745
0. 713

0. 765
0. 727

0. 776
0. 731









ratio and low height as well as a higher ratio and a

corresponding higher drawdown elevation.

Figure 26 more clearly presents the removal performance

associated with each height. As the height increased,

maximum removal occurred at higher drawdown rates. As the

drawdown rate increased, removal efficiency increased until

a maximum was reached, after which, further increase in

drawdown rate yielded decreased removal.

The results of the variable outflow simulations are

presented in Figures 27 and 28 and Table 13.


Table 13.


Estimates of flow capture efficiency (C) and
pollutant removal efficiency (R) as a function of
outlet height and outlet diameter (Vb/Vro=1.61):
1953 simulation results.


Outlet
Diameter
(ft)

0.0 C
R

0.25 C
R

0. 50 C
R

1.00 C
R

1. 50 C
R


0

0.043
0.043

0.887
0.708

0.976
0.293

1.000
0.018

*



Outlet Elevation (ft)

1 2 2. 5

0.043 0.043 0. 043
0.043 0.043 0.043

0.845 0.775 0.721
0.763 0.737 0. 696

0. 952 0.922 0.900
0.641 0.737 0. 753

1.000 1.000 1.000
0.460 0.650 0.705

1.000
*- 0.690


* indicates that simulation was not run.


3

0.043
0.043

0.635
0.619

0.876
0.757

0.995
0.747

1.000
0.734


4

0.043
0.043

*


0. 678
0.625

0.951
0.791

1.000
0.798







67
4.0
+63



0.70
0.60 0.75




3.0 75
0.75




4- i^

Z 0.70 ,-
-0
2.0 +74 -





0.60

0

1.0 64









+29
0 .0 1.. I 1 I I i
0,0' 0,25 0.5 .0 1.5

OUTLET DIAMETER, ft

Figure 27. Solution surface of pollutant removal efficiency
(R) as a function of outlet height and outlet
diameter (Vb/Vro = 1.61): 1953 simulation results.






















































0,25 0.50 1.00

OUTLET DIAMETER, ft


Figure 28.


Pollutant removal efficiency (R) as a function of
outlet height and outlet diameter (Vb/Vro = 1.61):
1953 simulation results.


1.00



0.9



0.8



0.7



0.6


0.5



0.4



0.3


0.2



0.I1



0.0 -
0.0


1.50







69

The capture performance describes intuitive results. As the

outlet height increases, the flow captured decreases. As

the outlet diameter increases, the capture efficiency

increases. The preliminary increase in removal efficiency

as the outlet height increases (holding diameter constant)

demonstrates the effect of dilution in a completely mixed

basin. Capture efficiency decreases, but the remaining

volume undergoes continuous pollutant decay, creating a

sizeable dilution potential for the influent. For each

outlet diameter, a unimodal removal curve was observed as

the outlet height increased. This reflected the effect on

removal efficiency of decreased capture performance combined

with increased dilution.

As the outlet diameter increased for a constant outlet

elevation, the capture efficiency increased. As observed

before, the removal performance peaks and then decreases as

the diameter increases, again reflecting the tradeoff of

capture and treatment.

The maximum removal was observed at an outlet height of

4.0 feet and an outlet diameter of 1.5 feet. Even though

the effective volume is small, the rate of discharge was so

great that 100 percent of the year's runoff flows were

captured.

This suggests that a basin with a large outlet,

possibly a weir, offers the best capture and removal

performance. This is a different case than a basin with no









outlet. With no outlet the maximum depth is maintained

(minus evaporation etc. losses) and any runoff is

immediately bypassed with no treatment. With a large outlet

below the maximum depth, the runoff will pass through the

basin and receive some degree of treatment by dilution.

Extrapolating these results to areas outside the solution

surface, the limiting maximum removal configuration would be

a drawdown rate as high as the maximum inflow rate, with the

discharge height at the top of the basin, i.e. no bypass,

but all the runoff would be diluted to some degree as it

passed through the basin. These trends suggest that optimal

quality control would be provided by a basin that is

partially full of water to provide dilution. This

conclusion is based on the supposition that pollutant

removal follows the exponential removal curve and that there

is complete mixing in the basin. The latter assumption is

hardly appropriate for sedimentation.

The computer costs associated with the simulations are

presented in Table 14.




STATISTICAL TECHNIQUES

Event Definition

A preliminary task in statistical analysis is the

grouping of raw data into independent events. A commonly

used method is the separation of hourly rainfall values by a

minimum number of hours with no rain. Serial









Table 14. Computer costs of simulations.


Procedure Cost *
Dollars
Runoff Block
24.6-year 3.25
One-year 0.75

Storage/Treatment Block
24.6-year 6.25
One-year 0.60

Synop
Interface 1.25
24.6-year runoff 1.75




* Average costs for low priority execution on
the University of Florida system. Normal
priority is approximately four times as much.



autocorrelation has been used to define this minimum

interevent time (Medina 1976). Howard (1976), Hydroscience

(1979) and others have suggested that rainfall events occur

as a Poisson process, and the time between events is

exponentially distributed. The exponential distribution is

a special case of the gamma distribution with the

coefficient of variation (standard deviation divided by the

mean) equal to unity. A cumulative distribution of time

between events was presented in Figure 6. The gamma

function has been widely applied in hydrology (Haan 1977).









Traditional Design of Flood Control Basins

The traditional sizing of stormwater detention

facilities has been based on the control of a single

"design" storm event. These basins are designed to capture

the runoff resulting from a storm expected to occur for a

given duration on the average once every N years (Haan

1977). Typical values of N range from 1-50, although

incorporation of downstream risk assessment governs the

value of N. The typical design storm analysis employs

ranking particular storms (e.g. annual maximum 60 minute and

24 hour rainfalls) over a time period and assigning an

extreme value probability distribution to the resultant

sequence. From this distribution, return periods are

assigned to storms of given magnitudes and durations.

Figure 29 depicts the relationship between the parent

distribution and resulting extreme value distributions.

Figure 30 presents a cumulative extreme probability

distribution (normal) used to determine the return period of

river flood flows. Because this method does not retain

information on the time between events, there is a loss of

information and the method is said to be inefficient in

terms of the data (Haan 1977). One unobtainable parameter

whose value is necessary for accurate flow routing is the

effective volume, defined as the actual storage volume

available at the beginning of a storm, which is a function

of antecedent conditions. Some municipalities require that















































Figure 29.


-n-2


.n,9


y (days)


Distribution of the largest sample value from

a sample size n from an exponential distribution.


110 -.



0 -o

ioi



0 70

so


Z 0-

*n Bo ~


RETURN PERIOD ( yrs )
1.01 1.It 2 5
PERCENT GREATER THAN
II ai I6 D0 0 70 ,O 60 40 10 20


I10 50 so oo zoo

10 A I I 0 0 0


iJThiTh~ ~


1 1' 11, ,, I I 1 7 1 7-1II I I T


Figure 30. Normal probability plot of Kentucky River data.


CA Oi 0 I I 1 0o 0 0 0 Do 70 Ao T AU 0 1 tA I t
PERCENT LESS THAN


H-HIIIII I1'11 H -H+HHttH


V M., IM, 1-0 i--,IlllilliTi-11-l-",IIF,-l-li-i,- I-L-4-


11I .It









ponds be built to contain the runoff from urbanized areas

(subdivisions, parking lots, etc. ) resulting from a design

storm. The Rational Method, an extension of the design

storm concept, is widely used for this purpose. The design

storm method is relatively straightforward and does not

require computer solution, hence it receives high ratings on

ease of application, though as Haan (1977) points out, there

is no direct theoretical connection between the solution and

the underlying mechanisms governing the storm events. While

this method is successfully used for flood control design,

no criteria have been established for determining a design

storm for quality control.




Aggregate Statistical Methods

Independently, two groups have developed statistical

approaches for designing urban stormwater detention

facilities. Howard (1976) presented the theoretical

framework for analyzing the interevent times and volumes of

combined sewer overflows resulting from various

storage/treatment configurations. His derivation was based

on approximating intensity, duration and interevent time as

independent and exponentially distributed random variables.

As part of a study evaluating the long-term performance

of stormwater control devices, Di Toro, et al. (1979)

derived an analytical expression for the effective volume of

a stormwater detention facility available at the beginning









of a storm event. Like Howard's method, this technique

retains information on the time between events an important

parameter for flow routing analyses. The method is based on

the SYNOP results for rainfall event statistics: intensity,

depth, duration and interevent time; converting them to

runoff values; setting up the respective equations for

effective volume and solving them. The result is a set of

graphs which enable the user to plot the solution surface

for percent of flow captured (C) and effective volume (Ve),

as a function of the constant drawdown rate (Qc) and the

ratio of the empty basin volume (Vb) to the mean runoff

volume (Vro). The algorithm requires repetitive application

to obtain an optimal basin size for maximum capture

efficiency. The method presents a straightforward procedure

for a simple catchment, single basin system (Dever, 1980).




Data Input

The Atlanta rainfall event statistics were obtained

from the previous runs of SYNOP. Runoff event parameters

were obtained from the rainfall values via a linear

conversion factor. The STORM equation was used to determine

the volume conversion factor, based on the percent

imperviousness for the catchment,


Cro = 0.15 + 0.751 (10)


where Cro is the runoff conversion factor, and










I is the percent of the catchment area that is

impervious.

With 1=0.37 for the catchment data, Cro=0.4275. The

conversions produced a mean runoff volume of 19000 cubic

feet based on the mean event depth of 0.212 inches over the

catchment area of 24.7 acres. The mean event volume is

similar to the value obtained by the SWMM simulation (18022

cubic feet). As mentioned earlier, the conversion method

does not account for the reduction in the number of events

resulting from catchment capture. This results in a total

of 504 inches of runoff for the entire 24.6-year record, as

compared with the 388 inches obtained by the Runoff Block.

Vb/Vro ratios were calculated to correspond to the

volumes used in the simulations. The resulting ratios are

presented in Table 15.



Table 15. Determination of normalized volume ratios.


Basin volume Normalized volume ratio
(cubic feet) Vb/Vro

7744 0.41

15014 0.79

29064 1.53

92420 3.74

167835 8.83










There was a misrepresentation of the normalized discharge

ratio in Di Toro's work.

1. The product QcTi was expressed as the average

drawdown between storms. The concept of average

drawdown between storms implies that mass

continuity be preserved. With no bypass, the

average volume entering the basin is equal to the

average runoff volume, and represents a limiting

value of the average long-term drawdown between

storms. The average drawdown would be even less if

the average volume entering the basin is less than

the average runoff volume, due to bypass. If the

product of Qc and Ti were truly the average

drawdown between storms, then the ratio (QcTi/Vro)

would always be less than or equal to unity; yet

relationships are developed for values up to

infinity.

2. The product of Oc and Ti would only equal the

average drawdown between storms if the time series

was complete homogeneous, with mean event volumes

occurring at mean interevent intervals. For all

other time series:

a. the product of Oc and Ti has no relationship to

average values;

b. the product of Oc and Ti is always greater than

the true average drawdown between storms; and,









c. the ratio (QcTi/Vro) is not bounded by

continuity constraints. This is why values

greater than unity are realized.

The average drawdown concept developed by Hydroscience,

Inc. is a plausible yet erroneous expression. The ratio

QcTi/Vro is used in this study merely as a normalized

discharge rate.

A solution surface of capture performance was prepared

as a function of basin volume and pumping rate. The grid

was created by using Figure 31 (see arrows) as follows:

1. enter the lower graph at the respective volume ratio

(Vb/Vro);

2. move horizontally until intersecting with the

normalized discharge curve (QcTi/Vro);

3. move to the upper graph at the effective volume

ratio (Ve/Vro), the common side between the graphs;

4. continue up until intersecting the runoff volume

coefficient of variation curve (cv); and

5. finally move horizontally and exit at the estimate

of capture efficiency (C).

This process was repeated for 25 combinations of volume

sizes and drawdown rates.









NORMALIZED EFFECTIVE VOLUME Ve/Vro
1.0 2.0 3.0 4.0 50-


0.4




0.2



0.0



1.0




2.0



3.0




4.0


LU

0.6-



CV = 1.38 0.8


CV= 1.22
I I I 1.0























0.5 0.75 1.0 2 510
NORMALIZED DISCHARGE, QcTi/Vro

Relationship of capture efficiency (C) with
normalized basin volume, normalized discharge
rate and mean volume coefficient of variation.


5.0 1 1
0,1



Figure 31 .









Capture Performance Results

The results are presented in Figures 32 and 33 and

Table 16.



Table 16. Estimates of flow capture efficiency (C) as a
function of basin volume and constant discharge
rate: statistical results.


Normalized Discharge Ratio
Normalized
Volume 1 2 4 7 10
Ratio

0.41 0.22 0.23 0.24 0.30 0.30

0.79 0.40 0.47 0.48 0.49 0.49

1.53 0.61 0.68 0.70 0.70 0.71

3.92 0.86 0.90 0.91 0.92 0.92

4.63 0.90 0.93 0.93 0.94 0.94




Isopleths of capture efficiency were drawn which emphasized

the apparent insensitivitiy of the performance to varying

pumping rates. This relationship exists due to the combined

shapes of the upper and lower curves. The capture

efficiency is most sensitive to the coefficient of variation

and effective volume ratio (Ve/Vro) at the lower end of the

Ve/Vro axis. However, at the lower end of the Ve/Vro axis,

the effective volume ratio is relatively insensitive to the

pumping rate, for the lines converge near a QcTi/Vro of 1.0.

Conversely, where the effective volume is most sensitive to

the pumping rate, at the upper end of the Vb/Vro axis, the

percent capture is least sensitive to effective volume, for








81
10.0



9.0



8.0



7,0

o

% 6.0


LU
n 5.0
0 492

S4.0 -
N 4.0 +91 0.90
-J


nz 3.0
00.80


2.0
+70 0.70


1.0 0.50


+29
0.0o I IIII
0 1 2 3 4 5 6 7 8 9

NORMALIZED DISCHARGE, QcTi/Vro


Figure 32. Solution surface of flow capture efficiency (C)
as a function of basin volume and constant
discharge rate: 1953 statistical results.












0.9 3. .3




0.8



0.7 1.50




0.6
0


Z 0.5 0.77


LL.
L 0.4
U-


-- 0,40
O_ 0.3



0.2



0.1




0.0 I II!---
0 I 2 3 4 5 6 7 8 9 10

NORMALIZED DISCHARGE, QcTi/Vro


Figure 33. Flow capture efficiency as a function of basin
volume and constant discharge rate: 1953
statistical results.









the slopes are the flattest. The greatest overall

sensitivity lies in the "middle" region of the graphs. The

maximum variation for any given volume was an 11 percent

increase from 64 percent to 75 percent for a ratio of 1.6.

The largest Vb/Vro ratio available in Figure 31 is 5.0.

This limitation precluded a complete comparison with

simulation results, where Vb/Vro ratios up to 9.3 were

analy zed.




Removal

A major weakness of using the statistical method for

estimating quality control is the inherent assumption of

absolute pollutant removal efficiency. In the design of a

detention facility for quality control, the determination of

a removal efficiency is the primary objective. It was

difficult, therefore, to assign a removal expression. It

was tempting to use the constant removal value incorporated

during each time step in the S/T simulation, however, the

two terms are not conceptually equivalent. The statistical

method essentially treats the removal mechanism in the basin

as a black box in which the constant removal refers to the

total captured flow during the complete time history. The

simulation employs time steps to route the flow through the

basin, and the constant removal term applies only to the

volume within the basin during that time step. This implies

that for any runoff volume which remains in the basin longer









than one time step, the total removal efficiency will be

greater than the constant removal term.

To make an estimate of removal efficiency would be

presupposing the solution in this study. However, an

estimate of removal efficiency can be obtained by taking the

product of the percent capture and an assigned constant

removal percentage. The resulting solution surface will

have the same shape as the capture performance presented in

Figure 32, but the value of the isoquants will be altered by

the removal factor.




Analysis Using Simulated Runoff Data

For further study, the statistical analysis was

repeated on runoff data obtained from the SWMM simulation.

The values determined by SYNOP for the data generated from

the Runoff Block (see Table 5) of SWMM are compared to the

converted values in Table 17.

Several differences were noted. The most obvious is that

there are 24 percent more events modeled in the statistical

method than the simulation. This difference affects the

total amount of runoff predicted by the two methods. The

simulation yields 308 inches of runoff while the use of a

conversion factor yields 504 inches of runoff, or 30 percent

more runoff over the 24.6-year period. This difference is

not apparent from comparison of the estimates of mean runoff

volume, where there is only a 5.47 percent difference in the









Table 17. Comparison of simulated runoff mean event
statistics with rainfall conversion values.


Number of events

Volume (in)

cv

Duration (hr)

cv

Intensity (in/hr)

cv

Interevent time (hr)

cv

Minimum interevent
time (hr) to yield
cv near 1.0


Simulated

1920

0.201

1.222

5.066

0.972

0.043

1. 148

111.71

1.000

4


Conversion

2391

0.212

1.384

7.824

1.134

0.032

1.356

90.10

1.004

8


two methods. This is an interesting point. Because the

statistical method deals with basin volumes normalized to

the mean event volume, the 30 percent continuity difference

between the methods is not reflected in the flow capture

solution surface. This implies that satisfying the

continuity equation may not be a requisite for a good

solution methodology. An important difference is that the

minimum interevent time was reduced from 8 hours to 4 hours

to obtain a coefficient of variation for interevent time

close to unity. At first, this appears to be due to the

presence of an effective detention time inherent in the









catchment storage, characterized by the attenuation of

magnitude and lengthening of duration. The values for the

mean duration contradicts this thought, being two hours less

than the rainfall mean., This decreased duration may be the

result of round-off error in transferring runoff data into

the NWS format for subsequent SYNOP runs. Flows less than

0. 125 cubic feet per second are rounded down to 0.0 inches

per hour, based on a catchment area of 24.7 acres. Volume

ratios and pumping rates calculated from these new values

are identical to those used in SWMM, as the same SYNOP run

was the basis for those values. The performance

determination was repeated on these new volume ratios and

pumping rates. The results are summarized in Table 18.


Table 18.


Estimates of flow capture efficiency (C)
as a function of basin volume
and discharge rate: statistical
results with simulated runoff means.


Normalized
Volume
Ratio

0.43

0.83

1.61

4.13

5.00


Normalized

1 2

0.28 0.28

0.44 0.49

0.64 0.70

0.89 0.93

0.92 0.95


Discharge Ratio

4 7 10

0.29 0.29 0.30

0. 50 0. 51 0. 51

0.73 0.74 0.75

0.93 0.94 0.94

0.95 0.96 0.96




Full Text

PAGE 1

Publication No. 58 LONG-TERM PERFORMANCE OF STORMWATER DETENTION FACILITIES: A COMPARISION OF DESIGN METHODOLOGIES by Gay F.E. Goforth University of Florida Gainesville, FL 32611-2013

PAGE 2

LONG-TERM PERFORMANCE OF STORMWATER DETENTION FACILITIES: A COMPARISION OF DESIGN METHODOLOGIES By Gary F. E. Goforth PUBLICATION No. 58 FLORIDA WATER RESOURCES RESEARCH CENTER RESEARCH PROJECT TECHNICAL COMPLETION THESIS ENGINEERING AND INDUSTRIAL EXPERIMENT STATION PROJECT NUMBER 80-W31 THESIS SUBMITTED JUNE, The work upon which this thesis is based was supported in part by funds provided by the Water Research Program, Engineering and Industrial Experiment Station, University of Florida, Gainesville

PAGE 3

ACKNOWLEDGEMENTS In tel' act ion i s fl l' e e ; ass u c h lowe t han k s to ma n y individuals at Black Hall and the Center For Wetlands who have contributed to the completion of this thesis and to my career at the University ofl Florida. The diversity ofl professionals in the department has provided a constant challenge to maintain an awareness of the many, yet similar, flacets ofl environmental engineering sciences. A large debt is acknowledged to Bob Dickinson who several times pulled me up .when I was close to drowning while SWMMing. Thanks also go to Steve Nix flor his help on SIT. The use ofl the computer resources at the Center For Wetlands, Black Hall and the Northeast Regional Data Center was invaluable. Thanks to Anelia Crawford flor the drafted fligures. The direction ofl this thesis is credited to Dr. J. P. Heaney; the stability ofl the content is credited to Dr. W. C. Huber; the inflluence of Dr. H. T. Odum is reflected in the holistic approach to the problem assessment and solutions. Theil' guidance and attention are greatly appreciated. Certainly the greatest debt is owed to my wifle, Karen. Her patience, sacrifice, programming and typing ability and overall good spirits in the flace of adversity will be florever appreciated. ii

PAGE 4

I \. TABLE OF CONTENTS ACKNOWLEDGEMENTS . . . i i LIST OF TABLES LIST OF FIGURES LIST OF SYMBOLS . . . . v vii x ABSTRACT .X i i I NTRODUCT I ON . . 1 1 3 6 CASE GENERAL OVERVIEW .. ..... SYSTEM DESCRIPTION . . . DEFINITION OF CONTROL UNIT .. ... CLASSIFICATION OF METHODS STUDY . CATCHMENT CHARACTERIZATION RAINFALL-RUNOFF CHARACTERIZATION data .... SYNOP. . Quantity .... BASIN CHARACTERIZATION Flow Conditions .. Removal ...... Theory Settling. Comparison Flow Conditions .8 1 1 1 1 1 2 1 2 1 3 1 5 23 23 24 24 29 METHODOLOGIES. .. ... 32 ANALYTICAL. . 32 EMPIRICAL . . 33 SIMULATION. . 36 STORM. .. .... ....... 36 SWI'1M . .. ... 36 Overview. . . 36 Data Input Block 38 Data Input Storage/Treatment Block 39 Constant Simulation 45 Variable Di$charge Simulation ... 46 Removal Mechanism .................. 47 One-Year versus.25-Year Simulation. 50 Block Results. . 50 Storage/Treatment Block Results. 52 iii

PAGE 5

STATISTICAL TECHNIQUES Event Definition ............ Traditional Design of Flood Control Basins. Aggregate Statistical Methods .. Data Input. . . Capture Performance Results Removal.,. . . Analysis Using Simulated Data COMPARISONS. .. DISCUSSION . GENERAL APPLICATION OF METHODOLOGIES APPENDIX A Program Listing and Data Input APPENDIX B BASIN Development and Listing APPENDIX C Detention Time. REFERENCES BIOGRAPHICAL SKETCH iv 70 70 72 74 75 80 83 84 87 9 1 93 97 1 03 1 05 1 1 2 1 1 5

PAGE 6

Table 1. 2. 3. 4. 5. 6. 7. 8. 9. LIST OF TABLES Case study catchment characteristics. SYNOP results of 24.6 years of Atlantal Georgia rainfall. Relationships between pollutant loads and flow volume (FLOW). Comparison! of Runoilil Block results using hourlYI daily and weekly rainfall input. SYNOP results of 24.6 years of simulated runoflf data. Determination of normalized volume ratios (Vb/Vro). Calculation oil constant discharge rates (Qc) (Ti = 111.17 hours; Vro = 18022 cubic feet>. Comparison of mean runoff event parameters for 1953 with the 24.6-year record (minimum interevent time = 4.0 hours). Runoff Block 24.6-year simulation summary. 10. Examples -of Storage/Treatment Block summaries. 11. 12. 13. Estimates of flow pollutant removal function of basin discharge rate: capture efficiency efficiency (R) as a volume and constant simulation"results. (C) and Estimates of fllow capture efficiency (C) and pollutant removal" efficiency (R) as a function of drawdown height and rate (Vb/Vro=1.61): 1953 simulation results. Estimates of flow capture efficiency (C) and pollutant removal efficiency (R) as a function of outlet height and outlet diameter (Vb/Vro = 1.61): 1953 simulation results. v 1 2 1 6 35 38 41 42 46 52 51 53 54 65 66

PAGE 7

14. Computer costs simulations. 71 15. Determination normalized volume ratios. 76 16. Estimates capture (C) as a function of basin volume and constant discharge rate: statistical results. 80 17. Comparison simulated mean event statistics with rainfall conversion values. 85 18. Estimates of flow capture ef.piciency, (C) as a function of basin volume and discharge rate: statistical results with simulated runoff means. 19. Estimates of hydraulic volume and detention time of control units associated with a single rainfall event. vi 86 94

PAGE 8

LIST OF FIGURES Figure 1. 2. 3. 4. 5. 6. 7. 8. 9. Representation of the hydrologic cycle. Discretized subsystems of the hydrologic c y c 1 e. Pro c e sse s de fin i n g ( the per for ma n ceo r a control unit. Schematic of Case Study catchment area. Characterization Or the various methodologies. Comparison of theoretical and observed distributions or interevent times ror Minneapolis/St. Paul airport. SYNOP values for mean event volume and duration as a function of minimum interevent time. SYNOP values for mean event intensity and interevent time as a function of minimum i ntereven t time. A comparison Or rainfall and runoff time series depicting the reduction in number of events and the reduction in the event volume. 10. Definition of interception and storage for storm events. 11. 12. 13. 14. Representation of a time series of runoff rlows. Development of overflow rate in an ideal iettling basin. Effluent concentrations for a first-order removal process in n completely mixed plugs. Removal efficiency for a first-order removal process demonstrating the effect of increased vii 4 5 5 7 9 1 4 1 7 1 8 20 21 22 25 28

PAGE 9

turbulence; n=-1 for and n=1 for completely mixed. 28 15. Effluent responses to a step input. 30 16. Comparison or real and plug flow reactor volumes for a first-order reaction. 30 17. stage relationships as calculated by BASINi constant and variable' discharge. 40 18. Time series of constant influent pollutant concentration. 44 19. Removal eQ.uation used in SIT Block. 48 20. Comparison of effluent concentrations under ideal plug flowl ideal completely mixed and as calculated with SWMM SIT completely mixed routingi step input of pollutant. 49 21. Solution surface of flow capture efficiency (C) as a function of basin volume and constant discharge rate: 1953 simulation results. 55 22. Solution surface of pollutant removal efficiency (R) as a function of basin volume and constant discharge rate: 1953 simulation results. 56 23. Pollutant removal efficiency (R) as a function of basin volume and constant discharge rate: 1953 and 24.6-year simul tion results. 57 24. Pollutant removal efficiency (R) as a function of basin volume under variable di'scharge conditions: 1953 simulation results. 61 25. Solution surface of pollutant removal efficiency (R) as a function of drawdown height and drawdown rate (Vb/Vro = 1.61): 1953 simulation results. 63 26. Pollutant removal efficiency (R) as a function of drawdown height and constant discharge rate (Vb/Vro. = 1.61): 1953 simulation results. 64 27. Solution surface of pollutant removal efficiency (R) as a function of outlet height viii

PAGE 10

and outlet diameter (Vb/Vro = 1.61): simulation results. 1953 28. Pollutant removal (R) as a function of outlet height and outlet diameter 67 (Vb/Vro = 1.61): 1953 simulation results. 68 29. Distribution of the largest sample value from a sample size n from an exponential distribution. 73 30. Normal probability plot Kentucky River data. 73 31. Relationship of capture efficiency (C) with normalized basin .volume, normalized discharge rate and mean volume coefficient of variation. 79 32. Solution surface of flow capture (C) as a of basin volume and constant discharge rate: 1953 statistical results. 33. Flow capture as a function basin volume and constant discharge rate: 1953 statistical results. 81 82 34. Comparison 89 35. Comparison of pollutant removal efficiency. 90 B-1. Development of BASIN. C-1 Various control unit configurations: steady-state conditions. C-2 Various control unit configurations: nonsteady-state conditions. ix 1 04 1 07 11 0

PAGE 11

a, b ai A c C cO Cro CV dr Dr Dro e H Hd hp k I ir Ir Iro kO ki L LIST OF SYMBOLS USED coerricients or runorr power percentage Or flow passing through basin i cross-sectional area Or rlow errluent concentration rlow volume capture erriciency inrluent concentration runorr conversion ractorcoerricient Or variation individual rainrall event duration mean rainrall event duration mean runorr event duration base or natural logarithm height or settling zone height at which discharge begins and ends height or particle entering settling zone rirst-order reaction coerricient percent Or catchment area that is impervious individual rainrall event intensity mean rainrall event intensity mean runorr event intensity initial rirst-order reaction coerricient linear rlow coericient or basin i length or rlow element x

PAGE 12

n N G Qc GcTi/Vro R SA SIT t td Ti v V Vb Vb/Vro Ve Vi vo vp Vr Vro Vs xo turbulence coefficient return period for design storm volumetric flow rate constant discharge rate normalized discharge rate pollutant removal efficiency surface area of settling zone storage/treatment elapsed time detention time of system mean interevent time velocity of flow volume of flow element empty volume of basin normalized basin volume effective volume volume of basin i overflow rate (surface loading rate) particle settling velocity mean rainfall event volume (per unit surface area) mean runoff event volume (per unit surface area) volume of settling zone percentage of particles with vp less than vo xi

PAGE 13

Abstract Thesis Presented to the Graduate Council the University Florida in Partial the Req,uirements the Degree Master or Engineering LONG-TERM PERFORMANCE OF STORMWATER DETENTION FACILITIES: A COMPARISON OF DESIGN METHODOLOGIES by Gary F. E. Gororth June 1981 Chairman: James P. Heaney MaJor Department: Environmental Engineering Sciences A general overview empirical, analytical, statistical and simulation techniq,ues ror evaluating stormwater detention systems is presented. The benerits and limitations of these methods in designing a control device water quality improvement are emphasized. A detailed analysis compares continuous simulation utilizing the Protection Agency's Storm Water Management Model with the statistical techniques advanced by Hydroscience, Inc. The general dynamics or storage and rlow xii

PAGE 14

elements are discussed, emphasizing the importance of detention time in defining a time frame for evaluating systems. Chairman u xiii

PAGE 15

'-'I INTRODUCTION GENERAL OVERVIEW In the urban environment, combined and separate storm sewer overflows contribute the same amount of contaminants to receiving waters as do secondary treatment efrluents (Heaney, et al. 1975, C. E. Q. 1978>' Presently, simple storage/treatment devices, i. e. one or two component systems such as a stormwater detention basin, provide a cost-errective tool for quantity as well as quality control Or these storm flows. While the design Or these devices has traditionally been based upon a single storm event, the additional information provided by long-term analyses has recently encouraged their adaptation. The engineer or planner concerned with the design of a detention facility for the quality control of stormwater runoff has a variety of solution methodologies available: empirical approaches utilizing average annual values; analytical methods based on solutions to the flow governing equations; simulators amenable to rigorous search techniques; and statistical techniques involving rainfall-runoff parameter distributions. No one method, or coordinated coupling of methods, has been documented as the most cost-efrective for

PAGE 16

2 all applications. This is in part due to the lack of an available long-term data base, but also reflects the lack of comparative studies. This thesis evaluates methodologies available for analyzing the long-term performance of stormwater runoff control devices. A detailed description of these methods is not the intent. The manuals referenced for each provide that service. Rather, the benefits and limitations of these methods in designing a control device for water quality improvement are emphasized. three criteria: The comparison consists of 1. problem assessment, i. e. how does the particular method define the system; 2. ease of application, e. g. data or computer r equ i rements, or c umb ersome tec hn i ques, and associated costs; and 3. accuracy of results, both absolute and relative to data requirements. Because the long-term performance of detention facilities has not been well documented, there is no available data base to definitively compare the accuracy of the methods. ObJectively, this study presents an opportunity to establish relative estimates of the long-term performance of storage/treatment devices for stormwater quality control. The optimal design of a control device will depend on problem specific constraints such as discharge quality standards and economic considerations.

PAGE 17

SYSTEM DESCRIPTION The following definition of a system is presented as a framework to maintain conceptual consistency. A system is any structure, device, scheme, or procedure, real or abstract, that interrelates in a given time reference, an input, cause, or stimulus, of matter, energy or information, and an output, effect, or response, of information, energy, or matter. (Dooge 1973, p. 4) This functional interrelationship of inputs and outputs for 3 a given time reference provides a basis for addressing water quality problems in a spectrum of hydrologic units, from urban stormwater systems to lakes threatened with cultural eutroph ication. Before evaluating the performance of a particular system, the hierarchy of systems which influence that performance must be recognized. A classical representation of the hydrologic cycle is presented in Figure 1. Storages and flows of water are the principal elements in the system, although the influences of solar energies, land morphologies and other factors are implicitly included. The system depicted in Figure 1 can be partitioned into discrete subsystems defined by characteristic storages and flows, as shown in Figure 2. These subsystems can be further subdivided into individual components, or contT'ol units, whose boundaT'ies similarly reflect the storages and flows emphasized. The basic hydrologic characteristics which define the performance of a control unit, as depicted in Fi g UT'e 3,are:

PAGE 18

4 I -, I ",_-.;., == --OCEAN ..-, : f 't' --, _____ ---.,... j -Ground Water ,', .. ,',M".. ..... ':1 :,"-,: .... ",.': .. ..:. ..... ,.:, ........... ... Figure 1. Representation of the hydraulic cycle.

PAGE 19

Figure 2. Figure 3. Throughfo II -l---f--!--.l--L Aquifers Streams, lakes, and rivers Precipito lion Oceans Discretized subsystems of the hydraulic cycle. Variables Variables characterizing characterizing E:lement af physical system Inflow outllow processes processes Processes defining the performance of a control unit. 5

PAGE 20

1. The source o-P the mass, the unit's place in the system and its relation to other units; 2. 3. the dynamic storage and -Plow conditions; and the removal mechanism. DEFINITION OF CONTROL UNIT 6 The -Pollowing analysis centers on the long-term pollutant removal effectiveness of a hypothetical detention facility. The control unit is a single basin which receives the stormwater runoff -Prom an urban catchment, and discharges to an undescribed receiving water. A schematic o-P the system is presented in Figure 4. The data source is a 24.6-year record o-P hourly rainfall values obtained from the National Weather Service. The system boundaries 6f the control unit are drawn at the inlet and outlet. As such, it is not Just the rainfall which is the forcing function, but the runoff, a result of the rainfall's interaction with the catc hment. The pollutant source is the constituent contaminants of the runoff -sand, debris, dust, etc. The removal mechanism responsible for pollutant control is sedimentation, and the removal characteristic is based on the treatment time in the basin. The removal kinetics are defined by the hydraulics within the basin, as determined by basin geometry and the inflow and discharge characteristics. These are the major influences on control ef-Piciency and become the design parameters.

PAGE 21

R-RAINFALL Ro-RUNOFF ETEVAPOTRANSPIRATION I INFILTRATION W WIDTH OF CATCHMENT' L LENGTH OF CATCHMENT OF -:-DETENTION FACILITY' 0-DISCHARGE TO RE;CEIVING WATER Figure 4. Schematic of Case Study area. 7

PAGE 22

; CLASSIFICATION OF METHODS Several methodologies are available ror estimating the long-term performance of stormwater detention facilities. The approaches are all models of the same complex process, yet differ conceptually and mechanistically. Figure 5 is a schematic depicting the relationship of the various methodologies. For the purpose of this thesis, the following classifications will be used: 1. Analytical approaches utilize some combination of the general mass continuity equation and the advective-dispersion equation to describe the rlows, storage, and pollutant removal characteristics Or a control unit. 2. Empirical approaches are derived from or guided by experience. Although literally implying the lack of a 8 theoretical background, the expression is used to denote methods which have been developed in scientific and engineering practice. 3. Two types of statistical techniques are widely used (Chow 1964). Frequency analysis methods are based on approximating the value of a random variable with a probability density function, from which frequencies of occurrence may be assigned. Regression and correlation analyses deal with the description of the relationship between two or more variables.

PAGE 23

9 f (v ,q) EMPIRICAL ANALYTICAL SIMULATION t STATISTICAL Figure 5. Characterization of the various methodologies.

PAGE 24

10 4. Digital simulation methods were developed to exploit the ability of high-speed computers to manipulate mathematical expressions. The main objective of these methods is to deterministically model the dynamic processes in a physical system. Rarely does a definitive demarcation exist between solution methodologies; there are overlaps and extensions from one to the next. As a modell each method represents simplificationsl compromising between ease of application and accuracy. The empirical and statistical approaches provide first-cut approximations based on a small data The more complicated simulations are generally regarded as more accuratel although they may have extensive data or computational

PAGE 25

CASE STUDY CATCHMENT CHARACTERIZATION User-supplied catchment data are input for most models. The extent of the data collection is dependent on the requirements of the particular method employed. As indicated in Figure 4, there are no streams, lakes or groundwater flows. For simplicity, there was no initial abstraction, areas of depression storage or other consumption of water. The flow routing geometry was kept as simple as possible. Conceptually, the catchment was a sloping plane with no gutter or pipe networks. All the runoff flowed directly to a dummy outlet on the downslope side. The runoff from the entire catchment was routed to the proposed basin, and was subsequently discharged to a local receiving water. The data were based on observed values for a drainage basin in Gainesville, Florida, and are presented in Table 1 (Huber, et al. 1981>' In an actual catchment, waste characteristics would be obtained by running column settling tests on runoff samples.

PAGE 26

Table 1. Case study catchment characteristics. Total area = 24.7 acres Impervious area :::: 37 percent No depression storage or initial abstraction Average catchment slope:::: 0.040 ft/ft :::: 211 ft/mile Maximum infiltration = 2.5 in/hr Minimum infiltration = 0.52 in/hr Evaporation:::: 0.1 in/day Population density = 500 people/square mile RAINFALL-RUNOFF CHARACTERIZATION Rainfall Data ,As the storages and flows of water are the principal elements in the catchment system, rainfall is the driving force. Long-term rainfall data are available on several 12 time bases, e. g. continuous gages or discrete hourly, daily, monthly or yearly records. Rainfall data are characterized by volume (depth over the catchment area), average intensity, duration and time between events. The rainfall data source utilized for this study was the National Weather Service (NWS) tape for 24.6 years (June 1948 -December 1972) OT hourly rainfall at Atlanta, Georgia. The standard NWS format is to record hourly rainTal1 values in hundredths oT an inch on days when there is rain. Days without rain are not recorded on the tape. Hourly data Tor the first day oT each month are recorded regardless of whether it rained or not.

PAGE 27

\ ..... j 13 SVNOP For methods requiring average event statistics, hourly rainfall data may be analyzed with SVNOP, a computer package developed by Hydroscience, Inc. (1979) to determine synoptic statistics of data time series. Rainfall volumes, intensities, durations and interevent times are the principal parameters evaluated in SVNOP. Available options include complete statistics on an event basis and time basis, e. g. yearly averages. Cumulative conditional probabilities (i. e. I given that rain has occurred) and return periods for hourly magnitudes are also calculated, based on the California method of probability plotting. The grouping of hourly data into storm events is based on the minimum number Or dry hours between rainfalls, an input variable termed the minimum intervent time. Assuming that the storm events occur as a Poisson process, the time between events is exponentially distributed. The exponential distribution is a special case Or the gamma distribution with the coefficient of variation equal to unity. Figure 6 demonstrates the relationship between theoretical and observed results for the cumulative distribution Or interevent times. The gamma distribution has been widely applied in hydrology (Haan 1977). To define events, the minimum interevent time is varied to obtain a value close to unity for the coefficient of variation (cv) associated with the interevent time. The SVNOP manual

PAGE 28

F gure 6. 99 w/ V w 98 :J -1 g 97 Z w > 96 I / V"00i; / .j / -j (.!) 0 95 r-/; {/ i'-l/'1.25 -1
PAGE 29

15 recommends an initial trial of three hours for the minimum interevent time. SYNOP was run on the entire 24.6-year record Or Atlanta rainfall to determine the storm statistics. The results of these runs, presented in Table 2 and Figures 7 and 8, give some idea of the sensitivity of the, results to the choice of the minimum interevent time. The computer costs averaged $6.25 per run. eight hours speciried as the minimum number of dry hours defining an event, the coerficient of variation for the mean interevent time was 1.004. Values for the means of the parameters were taken from this run, e. g. the mean volume (Vr) of a rainfall event was 0.495 inches. Notice that Vr does not the product of 11' and Dr. This is because Vr is the mean of the products of the individual events' intensity (ir) and duration (dr), which is not necessarily to the product of the mean intensity (Ir) and the mean duration (Dr), i. e. Vr = mean (irdr) which is not the same as (mean ir)(mean dr) = IrDr. Runoff Quantity The interaction Or rainfall and the catchment generates runoff. The of runoff is determined by the influence of infiltration, evaporation, consumption patterns and land use (Eagleson 1970). The watershed system response to these interactions has been evaluated by hydrologists for many years. A comparison of rainfall and runoff time

PAGE 30

" ( cTable 2. SYNOP results or 24.6 years of Atlanta, Georgia rainfall. Minimum # Number Volume cv Duration cv Intensity cv Interevent cv of dry hours Vr Dr Ir time Ti (in) (hr) (in/hr) (hr) 3 3215 0.367 1.540 4.642 1. 126 0.078 1.372 66. 73 1.269 5 2596 O. 454 1.424 6.646 1.124 0.077 1.348 82. 62 1.067 8 2381 0.495 1.384 7.824 1.134 0.077 1.356 90. 10 1.004 12 2134 O. 552 1.332 9.817 1. 143 0.074 1.334 100. 55 0.917 0'

PAGE 31

17 0.6 1.5 Z 0 1.4 tr CC c 0.5 1.3 .. lI-W a :E 1.2 ::> I--1 Z 0 0.4 W > 00 MEAN VOLUME 1.1 U Z A A CV LL
PAGE 32

18 1.5 Z 0 -.... 0.078 1.4 t( .c. "-0::: c: .0.077 1.3 ,. >-tJ... I-0 en 0.076 1.2 Z IW Z I-W Z 0.075 1.1 () z 0-0 MEAN INTENSITY I..l.. lJ... 0.074 CV 1.0 W W 0 ::E () 0.073 0.9 0 3 6 9. 12 MINIMUM INTEREVENT TIME, hr U t... 110 0-0 MEAN INTER EVENT 1.5 .r::. Z .. &--A CV 0 W 100 ti 1.4 I-0:: !-=. 90 1.3 f Z W lJ... > 80 1.2 0 W 0:: I-W Z I-70 1.1 W Z () lJ... Z 60 lJ... 1.0 W W 0 ::?: u 5 0.9 3 6 9 12 MINIMUM INTEREVENT TIME, hr Fi gu re 8 SYNOP values fo r mean event intensity and interevent time as a function of minimum L.interevent time.

PAGE 33

series. presented in Figure 9. depicts two phenomena characteristic of the rainfall-runoff process: 19 1. a reduction in the number of events due to the capture of low volume storms by the indigenous catchment storage capacity. e. g. depression storage and soil moisture capac i ty i and. 2. a reduction in the volume of the events due to the catchment storage and flow interception, e. g. infiltration rates. Analyses that deal with single runoff events are not sufficient to characterize these phenomena because the catchment storage and interception capacities are functions of antecedent soil moisture conditions. and are not constant. It becomes to retain as much, information as possible on the time between successive events. The event duration defines the reference time frame for differentiating between storage and inter.ceptor elements. A storage element can detain up to a maximum runoff volume per event. i. e. its detention time is greater than the event duration. An interceptor. on the other hand. can capture up to a maximum flow rate before bypassing Stormwater runoff control devices can also be characterized by storage and interception capacity. as presented in Figure 10. A representation of a time series of runoff flows is presented in Figure 11. a series of flow pulses separated by

PAGE 34

<-; FEB MAR I I II 0.0 II 2.0 ---I:l 'M JZ.i 1.0 JZ.i 0.0 APR MAY .1 I I JI J11 II. JUN 1.1 20 Maximum storage Minimum storage Figure 9. A comparison of rainfal I and runoff time series depicting the reduction in number of events and the reduction in the event volume.

PAGE 35

a) INTERCEPTION b) STORAGE D EI"] TIME c) INTERCEPTION AND STORAGE 0: W0Z1F gure 10. Definition of interception and storage for storm events. TIME 21

PAGE 36

22 U') '+-0 ... 0 ...J lL. 4.0 .lL.. lL. 0 2.0 Z ::> a:: 0 0 20 30 r 130 140 150 160 170 TIME, hr Figure 11. Representation of a time ser.ies of runoff flows.

PAGE 37

23 relatively long peroiods oT no Tlow. The determination of runoff characteristics is a maJor step in the solution process and is where the methods vary the most .. The approaches compared here offer a sharp contrast in the representation of the rainfall-runoff process. The statistical and empirical approaches summarize runoff generation via linear conversion factors applied to rainfall statistics. On the other hand, the simulation technique utilizes some of the most refined concepts in deterministic hydrology. BASIN CHARACTERIZATION Flow conditions U The input to the basin is the runoff from the catchment area. The time series of runoff events depicted in Figure 11 suggests two realms of kinetics: rapid, relatively well-mixed during the runoTf event, followed by slower (less dispersion, turbulence> kinetics and possibly quiescent conditions during dry weather. Characterizing this time series of discontinuous flows entering the basin is a maJor obstacle in solution methods. Basin discharge may be either variable, as in the case of gravity drainage, or constant, via a pump or outlet restriction. systems, 1. e. Negative feedback is inherent in gravity when the water level is high, the outflow is high, and as such, tends stabilize the flow. It is

PAGE 38

24 difficult to deal with this nonlinearity analyticallYi it is Tar easier to analyze a constant discharge rate. There has been no evidence to suggest that one is better than the other Tor pollutant removal. Removal Theory of settling Pollutant removal via settling is the most widely useful operation in water and wastewater treatment (Fair, Geyer and Okun 1968, Liptak 1974). In the design oT sedimentation basins, basic assumptions are incorporated: 1. inlet zone -the influent is transformed to a uniform vertical distribution of particles. 2. settling zone there is steady, uniform flow and quiescent, discrete and unhindered settling. 3. bottom zone -solids which enter the bottom zone are not resusp end ed. 4. outlet zone -solids that do not enter the bottom zone leave in the effluent. These four zones and particle settling paths are shown in Figure 12. The maJor design parameter is the overflow rate (vo), defined as vo = H/td = (Vs/SA)/(Vs/Q)' = O/SA (1) where H is the depth of the settling zone ( L J, td is the detention time of the settling zone ( T J, Vs is the volume of the settling zone [ L3 J,

PAGE 39

Q _---.INLET ZONE H SURFACE AREA = SA PERPENDICULAR TO FLOW CROSS SECTIONAL AREA = A Vo BOTTOM ZONE OUTLET ZONE TRAVEL TIME = L 1/ = L Q/A LA V = -= Q Q H H Q OVERFLOW RATE = Vo = = = t -if/Q SA Figure 12. Development of overflow rate in an ideal 5 e ttl i n g ba 5 in. 25

PAGE 40

26 SA is the the settling zone ( L2 J, and Q is the in the_settling zone [ L3/T J. The settling is slow, such that it is the removal rates that important, than the state (Rich 1974). ideal conditions, settle at a velocity (vp) by Stoke's Law, and are removed iT vp is greater than vo. Additional particles are removed which enter the settling zone at a height (hp) less than The total basin removal is given by 'to R = (1-xo) + (l/vo) vp dx (3 ) where R is the pollutant removal eflTiciency, and xo is the proportion oT particles with vp less than vo. Rarely do detention perTorm ideal conditions. Most OTten, design eTfliciencies are due to violations ideal assumptions caused by short circuiting and turbulence, which alter the kinetics flrom ideal quiescent conditions. Short circuiting is induced through thermal currents, wind action, influent inertia, etc. Resuspension of solids may occur as the .plow rate exceeds the scouring velocity. Thomas and McKee derived the

PAGE 41

27 effect of longitudinal dispersion in a basin consisting of n completely mixed plugs (Fair, Geyer and Okun, 1968>' Figure 13 presents the relative effluent concentrations for an instantaneous injection of dye undergoing a first order decay as it passes through the basin. A completely mixed basin is shown as n=1, while an ideal plug 'Plow basin (n=infinity) would be represented by a spike at t/td ofun i t y. The net effect of altering the flow regime from quiescent to more turbulent conditions is the reduction of the reaction coefficient k (Fair, Geyer and Okun 1968; Rich 1974) Although there is no way to predict before operation the reduction for a particular basin, the phenomena can be represented as in Figure 14, where the reduction of k is given as k/kO = (1-clcO)**n where k is the reaction coefficient, kO is the reaction coefficient under quiescent conditions, c is the effluent concentration, cO is the initial pollutant concentration, and n is the coefficient indicating the degree of turbulence, which increases as the flow regime diverges from ideal conditions. (4 ) Plug flow conditions are represented by n=O where k=kO and the pollutant decays according to cleO = e**(-kt) or, R = 1-c/eO "'" l-e**(-kt). ( 5)

PAGE 42

:t Relative time, t/td Figure 13. Effluent concentrations for a first-order removal process in n completely mixed plugs. 1.0 o (.) 0.9 "-(.) 0.8 .. 0.7 cj Z 0.6 8.5 W 0.4 '\ \ 1\ \1 \ \ r-... l'---.: r--f-!!-:' I---" In: > 0.3 ti 0.2 iiJ 0.1 0:: 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 FIgure 14. RELATIVE TIME, t/td Removal effIcIency for a first-order removal process the effect of increased turbulence; for quiescent and n=l for completely mixed. 28

PAGE 43

Comparison Flow Conditions Plug flow and completely mixed removal regimes deTine the extremes of pollutant removal performance. By definition, the hydraulic regime in each plug is complete mixing. Therefore, under static conditions, e. g. a column 29 settling test, there is no difference between the two. The applicability oT transferring column settling test results to dynamic conditions was recently re-examined by White ( 1976). In tank studies, he observed reasonable agreement with column tests, although the results were highly dependent on the waste characteristics. The effect of longitudinal dispersion on plug flow performance has been studied to correlate plug flow and complete mixing under dynamic conditions
PAGE 44

1.0 0.8 0.6 0.4 0.2 1.0 0.8 0.6 0.4 0.2 INFLUENT EFFLUENT Dd plug flow, Yr' 0 x large dispersion VT"' 0.2 x a 0,5 a. b. intermediate amount of dispersion Dd "'VL'O 025 x Dd complete mix, -L (l) Vx Figure 15. Effluent responses toa step input. 100 '* I "-/ v. (I) lcomplete mixing) first Qrder "'-I I .......... 64 i',.. Ill'." 200 K" ''', -" ....... ylOO r-... Ii 10 >. K -----r--50 (lines of equal volume ----i. or holdlnll time) II _-: l--I-1---...... b-( 20 .......... r-. -L-t'><' ...... !---i"'--r--, 10 -----------r.: -r----f-----r0,25 -" r-....... -,,2 .... -><... -----0,0625 -.... --.......... .. 0 (plug lIow) ./ ..... t:-". Q,QI Q,l Figure 16. Comparison of real and plug flow volumes foria first-order reaction. 30

PAGE 45

31 These two solutions represent the removal erficiency extremes. The lower removal efficiency of complete mixing is a result or the basin concentration being continuously mixed with the inrluent, yielding a dilution or the influent but a concurrent increase in the basin concentration. In plug fiowl the basin effluent more clearly reflects the removal process occurring in the separate plugs. The difference between the steady-state solutions for first-order reactions is graphically presented in Figure 16, which compares the volume of an ideal plug flow basin to the volume of a basin with dispersion yielding the same removal efficiency (Weber 1972>' From Figure 16 it is possible to predict the effect of implementing dispersion reducing mechanisms such as baffles. For example, by reducing the U dispersion factor from infinity (completely mixed) to 1. 0, the volume necessary to provide 90 percent removal is reduced by a factor of 2. There has been no widely used method for sizing SIT facilities with long-term stormwater quality control as the rna in 0 b J e c t i ve. Heaney, et al. (1979) presented the mechanics for determining the optimal combination of SIT for steady state conditionsl subJect to economic constraints.

PAGE 46

t'lETHODOLOG I ES ANALYTICAL Analytical models are developed from a combination Of the general mass continuity and advective-dispersion equations. They range in complexity f-rom simple steady-state approximations to more complicated variable flow solutions. The complexity of solution for detailed problems intimidates users and prohibits widespread application. Customarily, a simplification of the system's dynamic processes is necessary to obtain a solution. The analytical approach to stormwater detention facilities has been advanced by Medina (1976). Constant and variable (linear) flow and storage conditions were stUdied. Conservative and nonconservative pollutant routing were i"ncluded. Effluent concentrations were derived for "simple" forcing functions. Application of this method requires the runoff flows and pollutants be converted to one of the input functions for which solutions have been derived. The solutions are limited in application due to their complexity and sensitivity to storage and flow conditions, i. e. the need to reduce the complexity of the dynamics to facilitate the analytical solution. The complexity of these solutions

PAGE 47

precludes their application to stormwater runoff events without the use of detailed simulation. EMPIRICAL A widely used method for characterizing the rainfall-runoff process is the application of a conversion ractor to mean rainfall values. Conversion factors are catchment specific and have been correlated to 33 imp ervi ousness, land use, pop u lat i on dens i ty and depress i on storage (Hydroscience, Inc. 1979, Chow 1964). L..Jhile this method accounts for an average reduction in the volume of storms, it does not correct for the decreased number of events, 1. e. it doesn't account for the small volume storms that are retained in the catchment storage. Empirical relationships for runoff water quality have also been developed. Receiving water studies have determined runoff coefficients for with nutrient loadings based on land use (Reckhow 1980). In the urban arena, Smolenyak (1979) determined coefficients of the power equation load = a(flow)**b where load is the pollutant load in the runoff [ M J, flow is the runoff flow [ L3/T 3, and a, b are coefficients.

PAGE 48

Values determined from data in the Urban Rainfall-Runoff Quality Data Base are presented in Table 3. The practice of water quality control via settling is 34 inherent in most applications of natural waters. Water and wastewater treatment employ several variations of settling basins. Natural hydrologic systems incorporate detention for' both the storage of kinetic energy and the deposition of sediment load. As a result of the implementation of reservoirs, siltation studies have contributed to the identification oT the relationships between sedimentation and hydraulic parameters. The most popular sediment trap studies are those of Brown, Brune, Churchill and Camp (Ward and Haan 1977). Chen (1976) presents the historical development of theoretical analyses of sediment retention. These methods require the removal assessment of detention facilities based on an annual time frame. As there was no way to reconcile this with the stochastic nature of runoff eventsl these empirical methods were not investigated further. A recent review by Nixl et al (1980) summarizes the benefits in utilizing detention facilities for stormwater quality improvement. Several combinations of empirical and statistical approaches were suggested for design purposes.

PAGE 49

35 ( TABLE 3 Relationships beb/een Pollutant Loads and Flo\'! Volume (FLOW). Dependent R2 S;g. Level No. of Model Load = a(FLOW)b Variable F-Test Events Reg. Coef. a b BOO .2S .99 SO 34.0 1.12 COO .76 .99 157 29.S LOS NH3N .99 20 .215 .72 NITN .SO .99 21 .119 .80 NTOT .57 .99 103 .0400 .71 \ .. ./ ORGN .S8 .99 40 .S56 1.04 TOTN .74 .99 37 .304 1. 07 OOP .S3 .99 34 .064S .98 TOP .46 .99 119 .0104 .78 TOTOP .27 .90 11 .OSOO .. 55 TOTP .66 .99 53 .426 1.5 TPHOS' .91 .99 S .105 1. 05 "TOTS .69 .99 41 279 1. 41 TSS .56 .99 260 44.2 1.10 Source: Smolenyak 1.979

PAGE 50

36 SIMULATION Several models are available ror the simulation or the rainrall-runofr process and detention pond performance. Simulation is amenable to a trial and error routine. leading to search techniques for determining an acceptable design. STORM The Storage. Treatment and Overflow Model (STORM) of the Corps of Engineers models wet-weather flow through separate storage and treatment facilities (U.S.A.C.O.E. 1974>' Runoff quantity is generated from an empirical conversion factor applied to rainfall excess. Runoff quality is generated as a function of land use. While flow is routed through the treatment plant, there is no capacity to model the quality improvement provided by detention. SWMM Overview The Storm Water Management Model (SWMM) is one of the most comprehensive and well documented models available for the analysis and design of urban stormwater systems (Huber et al. 1980). SWMM was developed by Metcalf and Eddy, Inc., the University of Florida and Water Resources Engineers, Inc. under a contract for the Environmental Protection Agency. Version I was released in 1971 and has been continually undergoing revision and updating. This study

PAGE 51

L utilized Version III, released in 1980, and also incorporated rurther rerinements. While the model allows ror extensive watershed simulation, this study restricted the analysis to a simpliried Runorr Block and concentrated on the Storage/Treatment Block. The Runorr Block deterministically models the 37 rainrall-runorT process in a catchment. Physical data such as depression storage, inriltration rates, soil characteristics, catchment area, ground slope, gutter network, evapotranspiration rates and hourly rainrall values are included as the model input. Pollutant build-up and wash-orr runctions are available, as is pollutant generation based on catchment land use. The RunoTf Block generates hydrographs and pollutographs as options. Several inherent adJustment ractors are available ror calibration oT the mod e 1. The Storage/Treatment (S/T) Block Or SWMM models the rlow and pollutant routing through a storage and/or treatment device which can be either a detention or non-detention unit. Geometric and hydraulic relationships, e. g. depth to surface area, and evaporation rates, and incoming flows are included as data input ror the SIT block. Discharge can be modeled as either constant or variable outrlow. The package provides ror a variety Or removal mechanisms and rlow routing options, capable Or tracking both stormwater rlows and constituent pollutants.

PAGE 52

38 Data Input Tor Runoff Block The Runoff Block accepts hourly rainfall data in the National Weather Service (NWS) format. No gutter or pipe networks were used in the example catchmenti all runoff exited the area via a dummy outlet. The impervious area was modeled separately from the pervious area. Runoff generated on either area went directly to the dummy outlet without passing through the other area. Also, theT'e were no areas of depression storage in the modeled catchment. The data input for the Runoff Block consisted of the catchment characteristics presented in Table 1. A complete listing of the data input is given in Appendix A. Currently, continuous SWMM can only be run with hourly rainfall input. As an aside, the hourly rainfall data were transformed into daily and weekly records. StmM Runoff Block was run on the first 19 months of the Atlanta data and the results are compared in Table 4. Table 4. Comparison of Runoff Block results using hourly. daily, and weekly rainfall input. Rainfall Average Total Infilt. Evap. Runoff Time step Intensity Rainfall ( in) (in) ( in) HOURLY HOURLY 81.68 51.72 6. 39 28. 68 DAILY DAYTOT/Dr 80. 53 50. 46 8. 18 28. 00 WEEKLY WEEKTOT/2Dr 81.62 51.73 6. 11 28. 62

PAGE 53

39 Reasonable agreement was obtained, suggesting that if hourly data were not available, continuous SWMM could be run using daily or weekly rainfall data. Data Input for Storage/Treatment Block The basin geometry and hydraulic characteristics required as input data for the SIT block were obtained from a separate computer program, BASIN, written for that purpose. BASIN calculates stage to surface area, stage to volume and stage to discharge relationships for basins given the dimensions, side slope and outlet configuration. Examples are provided in Figure 17. The program development and listing is provided in Appendix B. As a reference, basin volumes (Vb) were normalized to the mean runoTf volume per event (Vro), yielding a normalized volume ratio (Vb/Vro). Currently, SWMM does not define storm event statistics, so the SYNOP program was employed. This involved running the Runoff Block with the complete 24.6-year (June 1948 -December 1972) rainfall record and generating 24. b years of simulated runoff data. These data were transformed to the Tormat of the NWS rainfall data, which is compatible with the SYNOP input format. As with the rainfall data, the minimum interevent time was varied to obtain the coefficient of variation (cv) for the interevent time close to unity. these runs are presented in Table 5. The results of

PAGE 54

40 -9 30 (, OJ OJ 0-00 -25 (1) 0 8 ill 4-0 () 0 :.0 20 :::I .. () 0:: -1 5 :::> 0 C/) > 5 0 0 2 3 4 5 6 0 2 3 4 5 6 DEPTH, ft DEPTH, ft. 18 15 (f) W .... 2 () 12 => .. -1 W 0 (!) > 0.:: 9
PAGE 55

t r Table 5. SYNOP results of 24.6 years of simulated runoff data. Minimum' Number Volume CV Duration Oro (hr) CV Intensity I ro CV Interevent Time Ti (h r) CV of dry hrs Vro 3 4 5 8 12 2124 1998 1903 1760 1646 ( j n) ( in/hr) 0.180 1.253 4.507 0.964 0.041 1.057 101.35 1.172 0.192 1.221 4.980 0.978 0.041 1.058 101.75 1.114 0.201 1.196 5.428 0.989 0.041 1.063 113.13 1.070 0.217 1.194 6.438 1.030 0.040 1.073 122.32 0.999 0.274 1.181 7.448 1.093 0.039 1.052 130.80 0.940

PAGE 56

L 42 A minimum interevent time of four hours resulted in a cv of 1.000. The mean runoff volume per event (Vro) was determined to be 18022 cubic feet, based on the mean depth of 0.201 inches over 24.7 acres. The number of events was reduced from 2381 rainfall events (Table 2) to 1998 runoff events. Continuity is checked by comparing the product of the mean event volume and the total number of events with the amount of runorf generated by the Runoff Block,' (0. 194 events) =: 383.61 inchesi from Runoff, 388 inches. Test volumes (Vb) were obtained from BASIN to closely approximate ratios (Vb/Vro) Or 0.5, 1.0, 2.0, 4.0, 10. O. For example, a Vb/Vro ratio or 0.50 implies that the empty basin volume is 50 percent of the mean storm event volume. The basin volumes used are presented in Table 6. Table 6. Determination of normalized volume ratios (Vb/Vro). Basin volume (Vb) feet) 7744 15014 29064 74420 167835 Normalized volume ratio Vb/Vro 0.430 O. 833 1. 613 4. 129 9.313

PAGE 57

43 The input data for the Runoff Block are provided in Appendix A. The SIT Block utilized the runoffvalues generated from the Runoff Block. To save execution time and money, the Runoff Block was run once and the output stored on an interface data disk. The interface data set served as the input to SIT for the subsequent simulations. A constant suspended solids concentration aT 100 mgll was assigned to the influent. This was chosen as opposed to generating pollutants rrom the catchment area for four reasons: 1) it avoids concern over how the pollutants are generated; 2) it provides a base value (100 mg/l) ror ruture comparisons; 3) it provides a blocked-orf step input as shown in Figure 18; and, 4) a constant inrluent concentration establishes that the percent of flow bypassed is numerically equal to the percent Or the pollutant bypassed. The actual flow condition in a basin is neither plug flow nor completely mixed, but somewhere between, termed "intermediate mixing". The complete mixing option of SIT was chosen for the rlow routing regime for its analytical and computational simplicity, resulting in lower simulation costs than the plug flow method. Two options of basin discharge were explored: variable outflow based on hydraulic head above an outlet, and a

PAGE 58

It.. g,?; 20 O:::!J... t Z O ti .-0::: Zl-W Z ::::>W -It) !J...Z zO -t) t Figure 18. Time series of constant influent pollutant concentration. 44

PAGE 59

45 drawdown scheme which emptied the basin at a constant rate (see Figure 17). The pumping option was run comparison with the statistical technique. while the variable discharge option was utilized for application to basins with gravity drainage. Constant Discharge Simulation A variable volume, constant unit was simulated by using the pumping option of SIT. As a reference, the constant drawdown rate was normalized as QcTi/Vro where Qc is the drawdown rate in cubic feet per hour, Ti is the mean interevent time in hours, and Vro is the mean volume in cubic (Hydroscience. 1979>' As Ti and Vro are constants determined from SYNOP, variable values the ratio drawdown rates. These rates were calculated to yield ratios of 1, 2, 4, 7, and 10 and are presented in Table 7. It was assumed that drawdown occurs whenever there is water in the basin. The of drawdown height (Hd) and drawdown rate on capture and removal efficiencies was analyzed.

PAGE 60

u Table 7. Calculation constant discharge rates (Oc) (Ti = 111. 17 hours; Vro = 18022 cubic reet>. Normalized discharge ratio OcTi/Vro 1 2 4 7 10 Drawdown rate (Oc) (cubic reet/hr> 162 324 648 1135 1621 Variable Discharge Simulation A variable volume. variable outrlow control unit was simulated with the SIT Block by utilizing a power equation 46 ror basin discharge based on hydraulic head. This simUlates the hydraulics in a basin with gravity drainage. The outlet characteristics were arbitrarily assigned as a six inch circular opening placed one root above the bottom. Instead or a solution surrace. as was provided in the constant discharge simUlations, a single removal curve was determined ror basin perrormance versus basin volume. In an analogous manner to the drawdown height and rate combinations. outlet elevation and cross-sectional area were recognized as design parameters Tor basin and were analyzed. Evaporation in the SIT unit was arbitrarily assigned a value OT O. 1 inch per day.

PAGE 61

Removal Mechanism A removal equation was chosen oT the Torm R = Rmax(l-e**(-kt where R is the pollutant removal eTTiciency. Rmax is the maximum removal eTTiciency. k is the Tirst-order rate coeTTicient [ lIT J. and t is the treatment time [ T J. 47 (9) Fair. Geyer and Okun (1968) presents general removal curves with k near 1.4 per hour Tor TSS and 0.50 per hour Tor BOD. with Rmax oT 0.75 and 0.45. respectively. For design purposes. values Tor Rmax and k would be determined Trom column settling tests with representative pollutants. Values oT 1.0 and 0.6 per hour. respectively. were arbitrarily assigned Tor these parameters. In the SIT Block. removal is accounted Tor once per time step, the length oT the time step, one hour. as the treatment time. The removal equation is presented in Figure 19. Because the time step was held constant throughout the simulation there was a constant percent removal (45 percent) oT pollutant per time step. A comparison oT eTfluent concentrations for removal governed by this equation in an ideal plug flow basin, an ideal completely mixed basin and the SIT'complete mixing regime is presented iri Figure 20. As shown. the SIT results lie within the extremes of pollutant removal efTiciency provided by ideal plug flow and complete mixing.

PAGE 62

48 1.00 ) 0.9 0.8 0.7 Removal (R) = l.O(1_e-O 6t) i t(hr) R 0.6 0.25 0.143 0.50 0.259 1.00 0.451 0:: 2.00 0.699 --3.00 0.835 >0.5 4.00 0.909 () 'S.OO 0.952 Z W 0.451 () lL.. 0.4 lL.. W ..J 0.3 0 W 0:: 0.2 0.1 0.0 0 2 3 4 5 TRE.ATMENT TIME hr .'0 Figure 19. Removal equation used in SWMM SIT Block.

PAGE 63

Figure 20. Comparison of effluent concentrations under .deal pl"ug flow, ideal comp1etely mixed and as calculated with SWMM SIT completely mixed routing; step input of pollutant.

PAGE 64

50 One-year Versus 24.6-year Simulation The initial decision to use one year of data as opposed to the full 24.6-year history was based on economic considerations; mistakes and debugging were expensive enough without extra data the costs. The 12 months of the input record (June 1,1948 -May 31, 1949) were utilized to get the simulator running. It was realized that if a IItypical" year's simUlation adeq,uately reproduced long-term basin the costs the analysis would be reduced by as much as an order magni tude. A IItypicalli year, 1953, was chosen on the basis of similar synoptic statistics, as determined by the SYNOP run of the 24.6-year data. A comparison 1953 runoff parameters with those the 24.6-year time series is presented in Table 8. The adeq,uacy of one year's simulation for describing the long-term basin was analyzed in a series simulations; the results are presented b.elow. Block Results A variety of summary print are available in the Runoff Block, from detailed hourly results to the total simulation summary, as presented in Table 9. On the hypothetical catchment, 704 inches (60 percent) of the total 1179 inches of rain left via infiltration. A total of 388 inc hes (33 percent) ac cumu lated as runoff, wh i 1 e 150 inc h es

PAGE 65

r C Table 9. Runoff Block 24.6-year simulation summary. TOTAL PRECIPITATION (RAIN PLUS SNOW) TOTAL INFILTRATION TOTAL EVAPORATION TOTAL GUTTER/PIPE/SUBCAT FLOW AT INLETS TOTAL WATER REMAINING IN GUTTER/PIPES TOTAL WATER REMAINING IN SURFACE STORAGE $ ERROR IN CONTINUITY. /. OF TOTAL PRECIP MILLION INCHES OVER CUBIC FEET TOTAL BASIN 105. 512 63. 106 7.846 34.830 0.000 0.000 -0.256 1176.79 703.83 87. 51 388.47 o. 00 0.00 * RUNOFF SIMULATION ENDED NORMALLY * \J1

PAGE 66

Table 8. Comparison of mean runoff event parameters Tor 1953 with the 24.6-year record (minimum inter event time = 4.0 hours). 24.6-year 1953 Volume (in) 0.201 0.217 cv 1.222 0.948 Duration (hr) 5. 066 6. 190 cv 0.972 1.087 Intensity (in/hr) 0.043 0.044 cv 1.148 1.164 Interevent time (hr) 111. 71 104.83 cv 1.000 0.986 (8 percent) were lost to evaporation. Mass continuity was preserved within 0.3 percent over the total 24.6-year simulation. Storage/Treatment Block Results The format Tor the SIT results are similar to the runoff output with more emphasis on quality parameters. Again, as shown in Table 10, the results were presented to facilitate continuity checks. 52

PAGE 67

Table 10. Examples of Storage/Treatment Block summaries. DETEtolTION UNIT CHARACTERISTICS: POLLUTANT ROUTING METHOD : COMPLETELY MIXED RESIDUALS DRAW-OFF SCHEME: NEVER DRAWN OFF 0.0 DEPTH-AREA-STORAGE-FLDW RELATIONSHJPS DEPTH, FT. SURFACE AREA, SQ. FT. 0.0 O. 50 1. 00 1. 25 1. 50 2.00 2.50 3. 00 3. 50 4.00 4. 50 GOVERNED BY PUMPING PUMPED 5000.0 5304. 0 5t.J1b.O 5775. 0 5']36.0 6264. 0 6600. 0 6944. () 7296.0 7656.0 8024. 0 STORAGE, CU. FT. O. 0 2576. 0 5306. 0 6729. 9 8193.7 11243.7 14459. 7 17845.7 21405.7 25143. 7 29063. 7 DEPTH AT WHICH FIRST PUMPING RATE DEGINS, FT. : 0.0 O. 0 O. 45 O. 45 O. 0 DEPTH AT WHICH SECOND PUMPING RATE BEGINS, FT. : FIRST PUMPING : SECOND PUMPING RATE,CFS : DEPTH AT WHICH ALL PUMPING STOPS, FT. : UNrr PARAMETER VOLUME CAT K(,K/-\ (CU. Ff. ) LE}!-1. -------------------'-----------------1 INFLOW, TOTAL O. 11).56E+07 O. INFLOW,NET O. 1386E+07 O BYPASS O. 7027E+05 0.0 TREATED OUTFLOW O. 136f1.E+07 O. 1422E+04 RESIDUAL FLOW 0.0 0.0 REMOVED BY DECAY 0.:29131:=:+04 REMAIN. TOT; VOL. O. 5283E+04 O. 3216E-'01 EVAPORATION 0.1700E+05 53

PAGE 68

54 Constant Discharge Simulation Performance results from the constant discharge simulations are presented in Figures 21, 22 and 23, and in Table 11. Table 11. Estimates capture (C) and pollutant removal (R) as a of basin volume and constant discharge rate: simulation results. Normalized Volume Ratio Normalized 0.43 0.83 1.61 4.13 9.31 Discharge ratio 1953 24.6 yr 1953 1953 24.6 yr 1953 1953 25-yr 1 C 0.344 0.331 0.476 0.613 0.640 0.852 1.000 0.952 R 0.343 0.322 0.476 0.598 0.638 0.847 0.989 0.941 2 C 0.398 R 0.373 O. 552 O. 721 0.527 0.690 0.945 1.000 0.906 0.961 4 C 0.468 0.461 0.621 0.804 0.795 0.962 1.000 0.993 R 0.373 0.383 0.532 0.712 0.705 0.864 0.905 0.896 7 C 0.542 0.699 0.839 0.971 1.000 R 0.365 0.524 0.661 0.792 0.817 10 'C 0.611 0.608 0.744 0.868 0.876 0.982 1.000 0.999 R 0.342 0.351 0.483 0.603 0.619 0.713 0.728 0.736 indicates that simulation was not run. The 1953 simulations duplicated the results of the 24.6-year simulations within five percent over the entire spectrum basin volumes and discharge rates. Figure 21 presents the solution surface capture efficiency as a of drawdown rate and basin volume.

PAGE 69

U 55 10.0 r---------------------------, 9.0 8.0 7.0 o ::> 6.0 .0 > W 5,0 ::::> -1 0 > C\ 4.0 W N -1
PAGE 70

( 56 10,0 ,....----------------_____ ---, 9.0 8.0 7.0 0 $ "-6.0 ..c > .. W 5,0 :::::> ..J 0 > 0 4.0 W N. -.J 0::: 3.0 0 Z 2.0 1.0 +53 0.40 +37 0, I 2' NbRMALIZED DISCHARGE, QcTi IVro Figure 22 Solution surface of pollutant removal efficiency (R) as a function of basin volume.and constant discharge rate: 1953 simulation results. 10

PAGE 71

U 57 1.0 r-----------------------------., 1953 RESULTS A 25-YEAR RESULT 0.9. 0.8 0.7 0.6 0:: .. 0.5 >-() Z W () 0.4 lJ.. 0.43 lJ.. W --I 0.3 0 W 0:: 0.2 1.0 0.0 o 2 3 4 5 6 7 8 9 10 NORMALIZED DISCHARGE. Qcl1 IVro Figure 23. Pollutant removal efficiency (R) as a function of basin volume and constant discharge rate: 1953 and 24.6-year. simulati6n results.

PAGE 72

Isopleths of percent capture were drawn by linear interpolation between calculated values. As expected, flow capture was greater as the drawdown rate increased, due to an increase in the effective volume. Also, as the basin volume increased, the capture efficiency increased due to less bypass. The vertical distance between the isoquants represents the sensitivity of capture performance to basin volume; the smaller the distance, the greater the sensitivity. for drawdown rates greater than 4.0, there appears to be uniform sensitivity to basin volume. The lowest sensitivities occur at the lower drawdown rates (QcTi/Vro less than 2.0). The isoquants converge slightly toward the upper end of the abscissa. The horizontal 58 distance separating the isoquants represents the sensitivity of capture performance to the drawdown rate. The isoquants become parallel to the abscissa above QcTi/Vro of 4.0, implying relative insensitivity to drawdown rate. Sensitivity is increased as the drawdown rate is decreased. Figure 22 presents the solution surface for pollutant removal efficiency as a function of basin volume and drawdown rate. Unlike the solution surface of capture performance, the isoquants in Figure 22 slope upward after an initial negative slope. The result is a solution surface which allows more than one drawdown rate at a specific basin volume to achieve the same removal performance. This demonstrates the performance tradeoff of providing a larger

PAGE 73

59 errective volume by emptying the basin providing a longer treatment time. although bypassing more r low. Combinations of basin volume and drawdown rate yielding equivalent removal erriciencies are depicted along i soquants. For e x amp Ie, the rem 0 val per for ma n ceo b t a in e d b Y a Vb/Vro of 4.1 and a QcTi/Vro of 2.0 was the same as a Vb/Vro of 9.3 and a normalized discharge rate of 4.0. The. greatest removal occurred in the region of large basin volumes (Vb/Vro ) 4.0) and low drawdown rates (GcTi/Vro < 4.0), Figure 23 presents the removal performance in a different manner than in Figure 22. There is no increase in information by presenting the results in this way, although ( ) "-" the communication or information is improved. For example, in Figure 23. it is than in Figure 22 to see that the sensitivity of removal to drawdown rate increases as the volume ratio increases. Combinations of basin volume and drawdown rates yielding equivalent removal as well as the sensitivity of the removal performance to drawdown rate are demonstrated. For example, the sensitivity of performance to drawdown is represented as the slopes of the and is seen to increase as the volume ratio increases. As the volume ratio increases, the maximum removal erficiency for each volume occurs at decreasing drawdown rates. The 1953 performance curve for a Vb/Vro of 9.3 is depicted as a straight line. This represents a divergence (5 percent at L;

PAGE 74

60 GcTi/Vro = 1.0) from the 24.6-year results, possibly due to the lack of a large storm during 1953. All of the curves converge to 4.3 percent removal at GcTi/Vro of 0.0, that is, in the case where there is no outlet. Long-term removal efficiency would undoubtedly be smaller for this case, tending to zero percent. Variable Discharge Simulations The effect of basin volume on capture and removal performance in basins with variable outflow rates was analyzed by running the SWMM SIT Block on five sets of basin geometry and hydraulic characteristics. The results presented in Figure 24 follow an intuitive removal relationship with increased removal as the storage capacity I increases. The regions below Vb/Vro of 0.43 and above Vb/Vro of 9.31 were not explored because of the unlikeliness of such a small volume ratio. The resultant removal curve is neither an exponential nor a power equation for the range observed. Only three 24.6-yearsimulations were run due to their low marginal benefit, i. e. the one-year simulations gave estimates close enough to the 24.6-year results to avoid spending the extra money for the long-term simulations. The costs of the runs averaged $0.60 for one-year and $6.25 for 24.6-year simulations.

PAGE 75

L" 61 1.00 r--------------------------, 0.9 0.8 0.7 1953 RESULTS 0.6 ... 25 -YEAR RESULT 0:: .. )0-0.5 U 25-year 1953 2 Vb/Vro C R C R W U 0.43 0.903 0.241 0.904 0.251 I.!-004 I.!-' W 0.83 0.898 0.529 1.61 0.967 0.643 0.952 0.641 4.13 1. 000 0.860 .J 9.31 1.000 0.916 1.000 0.918 0.3 0 :?: w .... n:: 0.2 0.1 :3 4 5. 6 7 8 9 10 2 NORMALIZED VOLUME, Vb/Vro Figure 24. Pollutant removal efficiency (R) as a function of basin volume under variable discharge conditions.: 1953 simulation results.

PAGE 76

62 Optimal Basin Design During preparation of the initial performance solution surfaces, it was recognized that the heights at which discharge began and ended would combine with the discharge rate to affect basin performance. Intuitively, increasing the height would decrease the capture efficiency, but due to the completely-mixed flow routing regime, the remaining volume would provide dilution of the influent. Combinations of drawdown height (Hd) and rate (Qc) and similarly outlet diameter and invert height, were simulated in an attempt to develop guidelines for the optimal design of detention facilities. A basin with a Vb/Vro of 1.61 was utilized for these simulations. The results from the constant drawdown simulations indicated that a basin with this ratio had the greatest performance sensitivity (26 percent capture and 11 percent removal) over the range of drawdown rates. In the variable discharge runs, a basin with a Vb/Vro of 1.61 yielded results in the knee of the removal curve. It was felt that a basin with this ratio was sensitive enough to reflect the effect of height and discharge combinations on basin performance. The results of the constant drawdown rate simulations are presented in Figures 25 and 26 and Table 12. The solution surface in Figure 25 indicates that maximum performance is achieved by a normalized discharge ratio

PAGE 77

U 63 4.0 '" +60 ,/ '" '" ,/ '" ,/ / ,/ ,/ / ./ / / / / 3.0 / / 71 + / / .. / / / -/ I, -0 :c j' j .. Z 2.0 76 0 + ...-:> ,/ ,/ W ..J W Z ;: 0 Cl 1.0 0.75 ,/ 0:: ,/ ,/ Cl .......0.70 0.0 .L.--I...-L..--'-_..L.-I...--I... __ '-_-t:... __ L-_-L---::;;..-J'--_-L-_--' o "Figure 25. 2 4 6 8 10 12 14 16 18 20 NORMALIZED DISCHARGE, QcTi IVro Solution surfa6e of pollutant removal efficiency (R) as a function of drawdown height and drawdown rate (Vb/Vro = 1.61): 1953 s.imulation results.

PAGE 78

0. 1.0 64 0.9 0.8 0.7 0.6 a:: .. 0.5 >() Hd (fi) SYMBOL z W 0 0 () I.J.. I A I.J.. W 2 0 ...J 3 0.3 4 "-0 W 0:: 0.2 1.0 0.00 NORMALIZED DI,SCHARGE, QcTi IYro Fi g u re Pollutant efficiency (R) as a function of drawdown height and constant di;charge rat'e (Vb/Vro = .1.61): simulation results.

PAGE 79

Table. 12. Estimates of flow capture efficiency (C) and pollutant removal efficiency (R) as a function of drawdown height and rate (Vb/Vro = 1.61): 1953 simulation results. Normalized Discharge Ratio 1 C R 2 C R 4 C R 7 C R 10 C R 12 C R 15 C R 18 C R 20 C R o 0.613 O. 598 O. 721 0.690 0.804 O. 712 O. 839 O. 661 O. 868 0.603 O. 883 O. 570 0.900 O. 520 * Drawdown Height (ft) 1 O. 578 O. 578 0.679 O. 679 O. 756 O. 734 0.804 O. 752 0.837 O. 761 0.860 O. 766 O. 874 O. 749 O. 885 O. 730 2 0.517 0.517 0.610 O. 610 0.687 0.687 O. 749 O. 728 O. 778 O. 738 O. 795 O. 744 O. 831 O. 763 0.845 O. 758 O. 851 O. 761 3 * * 0.717 O. 694 O. 745 O. 713 O. 765 O. 727 O. 776 O. 731 indicates that simulation was not run. 4 * * O. 610 O. 597 0.648 O. 621 0.663 O. 636 or 12.0 with an Hd of one foot. Removal efficiencies for basins with complete drawdown (Hd of 0.0) were lower than limited drawdown (Hd greater than 0.0) for all but the 65 lowest rates. The general trend of the isoquants indicates that similar performance can be achieved by a low drawdown

PAGE 80

66 ratio and low height as well as a higher ratio and a corresponding higher drawdown elevation. Figure 26 more clearly presents the removal performance associated with each height. As the height increased, maximum removal occured at drawdown rates. As the drawdown rate increased, removal efficiency increased until a maximum was reached, after which, further increase in drawdown rate yielded decreased removal. The results of the variable outflow simulations are presented in Figures 27 and 28 and Table 13. Table 13. Estimates of f-Iow capture efficiency (C) and pollutant removal efficiency (R) as a function of outlet height and outlet diameter (Vb/Vro=1. 61): 1953 simulation results. Outlet Outlet Elevation (ft) Diameter (ft) 0 1 2 2. 5 3 4 0.0 C 0.043 0.043 0.043 0.043 0.043 0.043 R 0.043 0.043 O. 043 0.043 0.043 0.043 0.25 C 0.887 0.845 O. 775 O. 721 0.635 R O. 708 O. 763 O. 737 O. 696 0.619 O. 50 C 0.976 0.'952 0.922 0.900 0.876 0.678 R 0.293 0.641 O. 737 O. 753 O. 757 0.625 1. 00 C 1.000 1.000 1.000 1.000 0.995' O. 951 R 0.018 0.460 0.650 O. 705 O. 747 O. 791 1. 50 C 1.000 1.000 1.000 R 0.690 O. 734 O. 798 indicates that simulation was not run.

PAGE 81

G 67 4.0 3.0 ---.,. ,.;" / ,; .... 'to.. Z / .0 ........ /" 2.0 ./ /' ::> ,/ ./ W ,/' ....J ,/ w ./ Iw ....J I-:::) 0 I.'" 1.0 0,0 __ ___________ ______ ____ -L __ 0.0' 0.25 0,5 1.0 1.5 OUTLET DIAMETER) ft Figure 27. Solution surface of pollutant removal efficiency (R) as a function of outlet height and outlet diameier (Vb!Vro =' 1.61): simulation results.

PAGE 82

,0 68 1.00 ------------------------, 0.9 0.6 0: .. >-0.5 U Z W U Hd ( f t) SYMBOL l.L.. 0.4 l.L.. 0 0 W -I I IJ. 2 0 0.3 0 3 :;E W 4 0: 0.0 L.. __ ...J__ 0.0 0.25 0.50 1.00 1.50 OUTLET DIAMETER) ft FIgure 28. Pollutant removal efficiency (R) as a function of outlet height and outlet diameter (Vb/Vro = 1.61): 1953 simu1ation results.

PAGE 83

69 The capture performance describes intuitive results. As the outlet height increases, the flow captured decreases. As the outlet diameter increases, the capture efficiency increases. The preliminary increase in removal efficiency as the outlet height increases (holding diameter constant) demonstrates the effect of dilution in a completely mixed basin. Capture efficiency decreases, but the remaining volume undergoes continuous pollutant decay, creating a sizeable dilution potential for the influent. For each outlet diameter, a unimodal removal curve was observed as the outlet height increased. This reflected the effect ori removal efficiency of decreased capture performance combined with increased dilution. As the outlet diameter increased for a constant outlet elevation, the capture efficiency increased. As observed before, the removal performance peaks and then decreases as the diameter increases, again reflecting the tradeoff of capture and treatment. The maximum removal was observed at an outlet height of 4.0 feet and an outlet diameter of 1.5 feet. Even though the effective volume is small, the rate of discharge was so great that 100 percent of the year's runoff flows were captured. This suggests that a basin with a large outlet, possibly a weir, offers the best capture and removal performance. This is a different case than a basin with no

PAGE 84

) 70 outlet. With no outlet the maximum depth is maintained (minus evaporation, etc. losses) and any runoff is immediately bypassed with no treatment. With a large outlet below the maximum depth, the runoff will pass through the basin and receive some degree of treatment by dilution. Extrapolating these results to areas outside the solution surface, the limiting maximum removal configuration would be a drawdown rate as high as the maximum inflow rate, with the discharge height at the top of the basin, i. e. no bypass, but all the runoff would be diluted to some degree as it passed through the basin. These trends suggest that optimal quality control would be provided by a basin that is partially full of water to provide dilution. This conclusion is based on the supposition that pollutant removal follows the exponential removal curve and that there is complete mixing in the basin. The latter assumption is hardly appropriate for sedimentation. The computer costs associated with the simulations are presented in Table 14. STATISTICAL TECHNIGUES Event Definition A preliminary task in statistical analysis is the grouping of raw data into independent events. A common1y used method is the separation of hourly rainfall values by a minimum number of hours with no rain. Serial

PAGE 85

Table 14. Computer costs or simulations. Procedure Cost Dollars Runorf Block 24.6-year One-year Storage/Treatment Block 24.6-year Synop One-year Interface 24.6-year runo;; 3.25 O. 75 6.25 0.60 1. 25 1. 75 Average costs for low priority execution on the University or Florida system. Normal priority is approximately four times as much. autocorrelation has been used to deTine this minimum interevent time (Medina 1976). Howard (1976), Hydroscience 71 (1979) and others have suggested that rainfall events occur as a Poisson process, and the time between events is exponentially distributed. The exponential distribution is a special case oT the gamma distribution with the coeT;icient 0; variation (standard deviation divided by the mean) equal to unity. A cumulative distribution of time between events was presented in Figure 6. The gamma function has been widely applied in (Haan 1977).

PAGE 86

72 Traditional Design or Flood Control Basins L The traditional sizing or stormwater detention has been based on the control a single "design" storm event. These basins are designed to capture the runorr resulting a storm expected to occur for a given duration on the average once every N years (Haan 1977 >. Typical values of N range from 1-50, although incorporation of downstream risk assessment governs the value of N. The typical design storm analysis employs ranking particular storms (e. g. annual maximum 60 minute and 24 hour rainralls) over a time period and assigning an extreme value probability distribution to the resultant sequence. From this distribution, return periods are assigned to storms of given magnitudes and dUTations. Figure 29 depicts the relationship between the paTent distribution and resulting extreme value distTibutions. Figure 30 presents a cumulative extreme probability distribution (normal) used to determine the return peTiod of Tiver flood flows. Because this method does not retain infoTmation on the time between events, theTe is a loss of information and the method is said to be inefficient in terms of the data (Haan 1977). One unobtainable parameteT whose value is necessary for accurate flow routing is the effective volume, defined as the actual stoTage volume available at the beginning of a storm, which is a function of antecedent conditions. Bome municipalities require that

PAGE 87

.25 .15 p (y) 16 Y (days) 73 Figure 29. Distribution of a sample size n the largest sample value from from an exponential distribution. 1.01 1.11 ...... II .. II H-I-+-i 1'1-'''''' :j 110 "0 II. 100 1/1 .... 0 u. 0 0 VI -0 70 0 .Q 00 !: a '0 I. .0 }---t---rh '0 '0 l.ri" 1-'-. 0,\ 0.. 0,," '0 RETURN PERIOO ( Y") 2 5 10 20 50 100 200 PERCENT GREATER THAN 'rO ..., .. r t -+. .;,..;..;., : t )0 20 10 o,m 0.1 0,1 .;..t .'1=FR= +---t--" ASSUMED HISTOR!CAL ,J.+ --t=:q::r:;::,. +H-'--.. '-=: HF --In+, .--1-+ I-+., 10 )0 40 ,0 .0 )b IU u .. ,. lOt. PERCENT LESS THAN Figure 30. Normal probabi I ty plot of Kentucky River data.

PAGE 88

u 74 ponds be built to contain the runoff from urbanized areas (subdivisions, parking lots, etc.) resulting .prom a design storm. The Rational Method, an extension of design storm concept, is widely used for this purpose. The design storm method is relatively straightforward and does not require computer solution, hence it receives high ratings on ease of application, though as Haan (1977) points out, there is no direct theoretical connection between the solution and the underlying mechanisms governing the storm events. While this method is successTully used for flood control design, no criteria have been established Tor determining a design storm for quality control. Aggregate Statistical Methods Independently, two groups have developed statistiial approaches for designing urban stormwater detention facilities. Howard (1976) presented the theoretical framework for analyzing the interevent times and volumes of combined sewer overflows resulting from various storage/treatment His derivation was based on approximating intensity, duration and interevent time as independent and exponentially distributed random variables. As part of a study evaluating the long-term performance of stormwater control devices, Di Toro, et al. (1979) derived an analytical expression for the effective volume of a stormwater detention facility available at the beginning

PAGE 89

75 of a storm event. Like Howard's method, this technique retains information on the time between events, an important parameter for flow routing analyses. The method is based on the SYNOP results for rainfall event statistics: intensity, depth, duration and interevent time; converting them to runoff values; setting up the respective equations for effective volume and solving them. The result is a set of graphs which enable the user to plot the solution surface for percent of flow captured (e) and effective volume (Ve), as a function of the constant drawdown rate (Oc) and the ratio of the empty basin volume (Vb) to the mean runoff volume (Vro). The algorithm requires repetitive application to obtain an optimal basin size for maximum capture efficiency. The method presents a straightforward procedure for a simple catchment, single basin system (Dever, 1980), Data Input The Atlanta rainfall event statistics were obtained from the previous runs of SYNOP. Runoff event parameters were obtained from the rainfall values via a linear conversion factor. The STORM equation was used to determine the volume conversion factor, based on the percent imperviousness for the catchment, ero = O. 15 + 0.75I (10) where ero is the runoff conversion factor, and

PAGE 90

76 I is the perc.nt of the catchment area that is impervious. With 1=0.37 for the catchment data, Cro=0.4275. The conversions produced a mean runoff volume of 19000 cubic feet based on the mean event depth of 0.212 inches over the catchment area of 24. 7 acres. The mean event volume is similar to the value obtained by the SWMM simulation (18022 cubic feet). A$mentioned earlier, the conversion method does not account for the reduction in the number of events resulting from catchment capture. This results in a total of 504 inches of runoff for the entire 24.6-year record, as compared with the 388 inches obtained by the Runoff Block. Vb/Vro ratios were calculated to correspond to the volumes used in the simulations. The resulting ratios are presented in Table 15. Table 15. Determination of normalized volume ratios. Basin volume Normalized volume ratio (cubic feet) Vb/Vro 7744 O. 41 15014 O. 79 29064 1. 53 92420 3. 74 167835 8. 83

PAGE 91

77 There was a misrepresentation oT the normalized discharge ratio in Di Toro's work. 1. The product QcTi was expressed as the average drawdown between storms. The concept oT average drawdown between storms implies that mass continuity be preserved. With no bypass, the average volume entering the basin is to the average runoTT volume, and represents a limiting value oT the average long-term drawdown between storms. The average drawdown would be even less iT the average volume entering the basin is less than the average runoTT volume, due to bypass. IT the product oT Qc and Ti were truly the average drawdown between storms, then the ratio (QcTi/Vro) would always be less than or to unity; yet relationships are developed Tor values up to in-Pinity. 2. The product oT ac and Ti would only the average drawdown between storms iT the time series was completey homogeneous, with mean event volumes occurring at mean interevent intervals. For all other time series: a. the product oT Gc and Ti has no relationship to averag e va I ueSi b. the product oT Qc and Ti is always greater than the true average drawdown between stormsi and, r U

PAGE 92

l G 78 c. the ratio (QcTi/Vro) is not bounded by continuity constraints. This is why values greater than unity are realized. The average drawdown concept developed by Hydroscience, Inc. is a plausible yet erroneous expression. The ratio QcTi/Vro is used in this study merely as a normalized discharge rate. A solution surface of capture performance was prepared as a function of basin volume and pumping rate. The grid was created by using Figure 31 (see arrows) as follows: 1. enter the lower graph at the respective volume ratio (Vb/Vro)i 2. move horizontally until intersecting with the normalized discharge curve (QcTi/Vro)j 3. move to the upper graph at the effective volume ratio (Ve/Vro), the common side between the graphs; 4. continue up until intersecting the runoff volume f coefTicient oT variation curve (cv); and 5. Tinally move horizontally and exit at the estimate of capture efficiency (C). This process was repeated for 25 combinations OT volume sizes and drawdown rates.

PAGE 93

(.) NORMALIZED E'FFECTIVE VOLUME J Ve/Vro 1.0 r__ -,-1.0 ___ 2::; ..:...0 __ --=.3:r:-.O ____ 4:.:,::.0=___ 5:.::;.0 0.8 0.2 >a () w z C/) w C/) () >0.6 0.4 tt co w :z w 0 0:: I-0:4 0.6 :::> u l-e:{ a.. a:: ........ ..c 1.0 > .. w ,:::> -l 2.0 0 W N 3.0 ..J 0:: 0 Z 4.0 5.0 I--L.. ___ ..I..___ ---l ____ __ 0,1 0.5 0.i5 1.0 2 510 NORMALIZED DISCHARGE, QcTi/Vro Figure 31. Relationship of capture efficiency (C) with normalized basin volume, normalized discharge rat e and m'e a n vol u m e C 0 e f fie 1 e n t 0 f v a ria t ion 79

PAGE 94

lJ L Capture Results The results are presented in Figures 32 and 33 and Tab Ie 16. Table 16. Estimates flow capture efficiency (C) as a function of basin volume and constant discharge rate: statistical results. Normalized Discharge Ratio Normalized Volume 1 2 4 7 10 Ratio O. 41 0.22 0.23 0.24 0.30 0.30 O. 79 0.40 0.47 O. 48 0.49 0.49 1.53 0.61 0.68 O. 70 O. 70 O. 71 3.92 0.86 0.90 O. 91 0.92 0.92 4.63 0.90 0.93 0.93 0.94 0.94 80 Isopleths of capture efficiency were drawn which emphasized the apparent insensitivitiy the performance to varying pump i ng rates. This relationship exists due to the combined shapes of the upper and lower curves. The capture efficiency is most sensitive to the coefficient of variation and volume ratio (Ve/Vro) at the lower end of the Ve/Vro axis. However, at the lower end of the Ve/Vroaxis, the effective volume ratio is relatively insensitive to the pumping rate, -For the lines converge near a QcTi/Vro of 1. O. Conversely, where the volume is most sensitive to the pumping rate, at the upper end of the Vb/Vro axis, the percent capture is least sensitive to effective volume, for

PAGE 95

'L; 81 10.0 .9.0 8.0 I-7.0 0 $ ........ 6.0 .c 'r> ... W 5.0 ::::J I--l 0 +92 > 0 W 4.0 N -. +91 0.90 -l
PAGE 96

L 82 1.0 Vb/Vro=5.00 0.9 3.83 0.8 0.7 L50 0.6 C,) .. >C,) Z 0.5 0.77 W U lJ.. lJ.. W 0.4 W 0:: ::::> I-0.40 a.. 0.3 C,) 0.2 0.1 0.0 "--_-L-_--I.. __ .l...-_...l-_-L __ L..-_-l-_-L __ l-_..J o 2 3 4 5 6 7 8 9 10 NORMALIZEOOISCHARGE QcT! IVro Figure 33. FJowcapture efficienct as a function of basin volume and constant discharge rate: 1953 statistical results.

PAGE 97

83 the slopes are the The greatest overall sensitivity lies in the "middle" region the graphs. The maximum variation any given volume was an 11 percent increase 64 percent to 75 percent a ratio of 1.6. The largest Vb/Vro ratio available in Figure 31 is 5.0. This limitation precluded a complete comparison with simulation results, where Vb/Vro ratios up to 9.3 were analyzed. Removal A maJor weakness of using the statistical method for estimating control is the inherent assumption of absolute pollutant removal efficiency. In the design of a detention facility for control, the determination of a removal efficiency is the primary obJective. It was to assign a removal expression. It was tempting to use the constant removal value incorporated during each time step in the SIT simulation, however, the two terms are not conceptually The statistical method essentially treats the removal mechanism in the basin as a black box in which the constant removal to the total captured flow during the complete time history. The simulation employs time steps to route the flow the basin, and the constant removal term applies only to the volume within the basin during that time step. This implies that for any runoff volume which remains in the basin longer

PAGE 98

than one time step, the total efficiency will be than the constant removal 84 To make an estimate of efficiency would be the solution in this study. an estimate of efficiency can be obtained by taking the product of the percent and an assigned constant The solution will have the same shape as the capture in 32, but the value of the will be by the factor. Analysis Using Simulated Runoff Data For further study, the statistical analysis was on obtained the SWMM simulation. The values by SYNO? for the data the Rtinoff Block (see Table 5) of SWMM to the converted values in Table 17. differentes were noted. The most obvious is that 24 percent events modeled in the statistical method than the simulation. This affects the total amount of by the two methods .. The simulation yields 308 inches of runoff while the use of a conversion yields 504 inches of 30 runoff the period. This is not apparent of the estimates of mean volume, is only a 5. 47 in the

PAGE 99

Table 17. Comparison of simulated runoff mean event statistics with rainfall conversion values. Simulated Number of events 1920 Volume (in) 0.201 cv 1.222 Duration (hr) 5.066 cv 0.972 Intensity (in/hr) 0.043 cv 1. 148 Interevent time (hr) 111.71 cv 1.000 Minimum interevent 4 time (hr) to yield cv near 1. 0 Conversion 2381 0.212 1.384 7.824 1.134 0.032 1.356 90. 10 1.004 8 two methods. This is an interesting point. Because the statistical method deals with basin volumes normalized to 85 the mean event volume, the 30 percent continuity difference between the methods is not reflected in the flow capture solution surface. This implies that satisfying the continuity equation may not be a requisite for a good solution methodology. An important difference is that the minimum interevent time was reduced from 8 hours to 4 hours to obtain a coefficient of vari.tion for interevent time close to unity. At first, this appears to be due to the presence of an effective detention time inherent in the

PAGE 100

86 catchment storage, characterized by the attenuation of magnitude and lengthening of duration. The values for the mean duration contradicts this thought, being two hours less than the rainfall mean.. This decreased duration may be the result of round-off error in transferring runoff data into the NWS format for subsequent SYNOP runs. Flows less than 0.125 cubic feet per second are y'ounded down to 0.0 inches per hour, baSed on a catchment area of 24.7 acres. Volume ratios and pumping rates calculated from these new values are identical to those used in SWMM, as the same SYNO? run was the basis for those values. The performance determination was repeated on these new volume ratios and pump ing rates. The results are summarized in Table 18. Table 18. Estimates of flow capture efficiency (C) as a function of basin volume and discharge rate: statistical results with simulated runoff means. Normalized Volume Ratio O. 43 O. 83 1. 61 4.13 5. 00 Normalized Discharge Ratio 1 2 4 7 10 0.28 0.28 O. 29 0.29 O. 30 O. 44 0.49 O. 50 0.51 O. 51 0.64 O. 70 O. 73 O. 74 O. 75 O. 89 O. 93 O. 93 0.94 0.94 0.92 0.95 0.95 0.96 O. 96

PAGE 101

) 87 Because of the low sensitivity of flow capture efficiency to small changes in either volume or drawdown ratio, the solution surface obtained is almost identical to Fi g ure 32. The only computer costs associated with the statistical technique were for the SYNOP runs, which averaged $6.50 for rainfall and $1.75 for runoff data. COMPARISON In comparing the methods, it is necessary to recognize that the statistical, empirical and analytical techniques are first-cut approaches, while the simulation yields a greater design and analysis flexibility, although requiring a larger data input. As emphasized earlier, there was no established data base to definitively compare the results obtained. However, certain aspects of the statistical and simulation methodologies can be compared. Both techniques utilize the same rainfall data. The simulation generates its own runoff data, while the statistical technique relies on a conversion factor. With no established data base, there was no way to say whether the simulated runoff values are more accurate than the statistical conversions. However, previous implementations of SWMM which tested \ predicted runoff with documented values reinforce the assumption that the simulation results are accurate. Both methods allow calibration of results with observed data when

PAGE 102

88 present. Mean event runorr values obtained rrom the two methods dirrered by only 5 percent, but the dirrerent'number or events (2381 ror the statistical method versus 1920 ror the simulation> resulted in the statistical method predicting 30 percent more total runorf over the 24.6 year record. Figure 34 is a comparison of the rlow capture solution surfaces obtained from the simulation and statistical methods. For dischage ratios less than 1.0 the two methods are in reasonable agreement. However, at discharge rates above this ratio the estimates diverge, with the statistical estimates predicting up to 20 percent less capture than the simulation results. Figure 35 is a comparison Or the estimates of removal efficiencies obtained from the simulation and statistical methods, where r is the constant removal term employed by the statistical technique. The most striking dirference is the shape of the isoquants. The simulation estimates reflect the combination of rlow bypass and treatment time in the slopes or the isoquants. The ability to depict this fundamental property Or storage/treatment devices is relinquished by the statistical method with the direct inclusion of a removal expression, r. Estimates of r yielding comparable removal as the simulation results ranged rrom near unity at discharge ratios than 1.0 to less than 0.8 at higher drawdown rates.

PAGE 103

( L 89 10.0 r--------------------------, o :> .c > W ::E ::> -.J o > 9.0 8.0 7.0 6.0 5,0 o W 4.0 N -.J
PAGE 104

90 10.0 r----------------------------, 9.0 8.0 7.0 0 ;;; "-..c 6.0 > ... w ::::> 5,0 :.J 0 G > 0 W 4.0 N -I 0:: 3.0 .0 Z ------2.0 ___ 1.0 -.----------------. 0.50 40 0.0 0 NORMALIZED DISCHARGE, QcTl /Vro Figure 35. Comparison of pollutant removal efficiency.

PAGE 105

L/ DISCUSSION This analysis was concerned with comparing solution methodologies. Because there are no actual data to obtain removal characteristics or calibrate the results, the results are not meant to stand alone, but are a runction Or the assumptions involved. It was relt that empirical and analytical methods were applicable for determining long-term perrormance of stormwater detention facilities. Application of analytical methods to stormwater runoff events may be obtained with computer simulation (Medina 1980). Continuous simUlation was found to provide the most detailed performance analysis. SWMM can handle complicated rlow networks and SIT combinations, and give detailed summaries for any time period. The simulation provided more interesting (i. e. no explicit relationship between capture and removal performance) relationships between the overall removal efficiency and the basin parameters. There was a noticeable removal performance tradeoff between increasing the effective volume at the beginning of a storm and increasing the pollutant treatment time. These relationships were closer to observed results than the constant removal results of the statistical method (Heaney J

PAGE 106

92 1975 >. The simulation permits a sensitivity analysis of basin parameters and pumping strategies, while there is limited mechanism for this in the statistical technique. This feature allows the simulation to be used to search for the optimal basin characteristics for pollutant removal. The relative effectiveness of the statistical technique in estimating flow capture performance for the hypothetical catchment was demonstrated by comparison with the results from the simulation. Conversion of rainfall volume was within 5 percent of the simulated results, although total volume of runoff over the 24.6 year record was 30 percent higher in the statistical method. The statistical method in the form utilized precludes determination of pollutant removal efficiency. Modifications are needed to account for variable removal efficiency within the detention facility.

PAGE 107

GENERAL APPLICATIONS OF METHODOLOGIES A spectrum Or hydrologic control units is depicted in Figures 1 and 2. An example Or these units and estimates or their volumes and detention times associated with a single rainfall event are enumerated in Table 19. Both hydrologic storages and transport elements are classified as control units. While flows of water are not generally recognized as control elements, they possess the same characteristics necessary for pollutant removal, e. g. even though a detention f-acility is stationary, the volume of-water passing through it is not. The hydrologic and f-unctional characteristics alluded to in Figure 3 are 'inherent in all water bodies, whether it is a river reach which has a detention time of less than a day or a water supply reservoir with a detention time on the order ofmonths. The classificati6n of-a hydrologic unit as a storage or rlow is based on a relative time scale; a flow element turns over "f-aster" than a storage device. This time dependence dictates the reaction kinetics and is the basis for recognizing a spectrum of water bodies in Table 19. In all cases, the actual mechanisms or removal are not altered, but rather, the inherent kinetics which determine

PAGE 108

94 Table 19. Estimates OT hydraulic volume and detention time of various control units associated with a single rainTall event. Percent of Even!; Va 1 ume 1. interception by above ground structures I") "'-. the initial abstraction volume 3. the transport through the soil via inTi1tration 4. the overTlow runoTT 5. the Tlow in gutters 6. areas OT depression storage in the basin 7. evapotranspiration 8. the -Plow in stream channels and Tlood plains 9. detention Tacilities 10. ponds and lakes 11. the ultimate receiving water; here we recognize an even larger system, either subsurTace, oceanic or atmospheric. Source: modiTied Trom Medina 1976. 0-5 0-5 30-60 20-35 10-30 0-10 0-10 10-30 0-15 10-35 5-20 Decention Time 0-7 days 0-3 hours 5-12 hours 1-18 hours 1-18 hours 0-1 month 0-12 hours 0-9 days 0-10 days 0.1-1000 years 1-10000 years the extent Or perrormance are changed. The hydrologlc kinetics dictate the predominance or dirTusive ,transport or turbulent advective transport (Rich 1974). The omission OT the time scale criteria in evaluating system perTormance presented diTriculties in the early works

PAGE 109

on lake eutrophication (Vollenwieder 1968, Dillon 1975, Reckhow 1978). The derivation of the first models began with the mass continuity equation but quickly veered to empirical concepts. Vollenwieder (1968) originally developed a general expression for the trophic state of lakes based on aerial loading and mean depth. The same 95 empirical model was used for all water bodies, ranging from urban ponds with detention times on the order of days, to mammoth lakes with detention times of hundreds of years. The early predictive models were based on data gathered from a number of large, public lakes threatened with water quality problems. Later studies revealed a set of lakes whose trophic conditions were considerably different than predicted. In one study, Dillon (1975) compared the phosphorus concentrations in two lakes of similar depths. Their areal phosphorus loadings and, hence, their predicted phosphorus levels, differed by a ractor of twenty. Due to the difference in detention times, 26 days versus 1738 days, the observed concentrations in the lakes were similar. It was recognized that detention time, which had been neglected in the early models, was a fundamental factor in the lake concentration of phosphorus. The influence of detention time was obscured in the earlier studies as a result of using a one year time frame on lakes with detention times greater than one year. In 1973, Vollenwieder incorporated the detention time concept in the trophic state model with

PAGE 110

96 an rate, as the ratio mean depth to detention time (Vollenwieder 1976). Other intuitively detention time the continuity and similar to the ones which Vollenwieder and others ultimately arrived at (Reckhow 1978, Schnoor 1975). It is possible to apply the solution methodologies analyized in this thesis to problems of a nature. The solution process should follow the .same as was presented: 1. identify the system; a. unit's place in system; b. influencing processes; 2. ide n t i f y ma J 0 flo w san d s tor age s ; and, 3. recognize the time dependence on removal mechanisms i. e. detention time, which affects the performance of the reaction kinetics.

PAGE 111

0000 0001 OOUl 0003 0004 0005 00U6 0007 0008 0009 0010 0011 0012 0013 0014 0015 0016 Oll17 0018 0019 0020 0021 I '022 \",:/023 0024 0025 0026 0027 0028 0029 Ou30 0031 0032 Ll033 0034 0035 0036 0037 0038 0039 0040 0041 0042 Q 0 3 0044 0045 0046 APPENDIX A. PROGRAM LISTING AND DATA INPUT \ Table A-l Runoff Block Simulation IISWMM JOB (2006,3400,100,20,0), 'GARY F. GOFORTH',CLASS=1,REGION=512K, II MSGLEVEL=(I,l) I*PASSWORD 2,RAIN I*ROUTE PRINT REMOTE6 II EXEC FORTXCLE,LPARM='LIST,NOMAP,OVLY',OPTIONS='NOSOURCE,NOMApi IIFORT.SYSIN DO SUBROUTINE COMBIN RETURN END SUBROUTINE TRANS RETURN END SUBROUTINE RECEIV RETURN END SUBROUTINE EXTRAN RETURN END SUBROUTINE STRT RETURN END /*INCLUDE /*INCLUDE /*INCLUDE /*INCLUDE /*INCLUDE RHYDfWl RHYDR02 RHYDR03 GUTTERl GUTTER2 1* /ILKED.SYSLMOD DO SPACE=(CYL,(5,1,1 //LKED.SYSUTI DD SPACE=(CYL,(5,Z /ILKED.LlB DO DSN=UF.A0063473.S\JMM,DISP=SHR I/LKED.SYSIN DO INCLUDE LIB(MAIN) INCLUDE LIB(GRAPH,CURVE,PINE,PPLOT,SCALE,HYSTAT) OVERLAY ALPHA INCLUDE LIB(RUNOFF,RBDATA,CTRAIN) INSERT OVERLAY BETA INCLUDE LlB(HYDR2,GAt.1P,QSHED,\JSHED) INCLUDE LIB(MELT,AREAL,FINDSC,GQUAL,HCURVE) INSERT \'JSHED OVERLAY BETA INCLUDE L I BC PRIHR3) //GO.FTOIFOOl DO UNIT=SYSDA,SPACE=(TRK,(SO,10, II VOL G S ERg \J 0 R K a 1 DC [3 G ( R E C F M:: V [3 S 0 L 'K S I Z E ::I 4 2 40 B U F NO = 1 ) //GO.FT02FOOI DO UNIT=SYSDA,SPACE=(TRK,(SO,lO, / / VOL=SER=\JORKOl,DC8=( RECF1,1=VBS, BLKS I ZE=4240, BUFNO=1) //GO.FTU3F001 DO UNIT=SYSOA,SPACE=(TRK,(SO,lO, 17 II VOL=SER=WORKOl,DCB=(RECFM=V8S,BLKSIZE=4240,BUFNO=1) f -.. 1.,-_/ 97

PAGE 112

Table A-I (cont.) //GO.FT04F001 DD SYSOUT=A,DCD=(RECFM=FSA,LRECL=133,BLKSIZE=133) IIGO.FT08FOU1 DD UfJIT=SYSDA, II DSN=UF.AOOG3473.RAlfJA,SPACE=(TRK,(100,50),RLSE), I I 0 I S P = ( N HJ C AT L G ) I 0 C B = ( R E C F Iv! = VB S,B L KS I Z E = 4 2 4 0, BJ F N 0 = 1 ) IIGO.FT09FOOI DO UtJIT=SYSDA,SPACE=(TRK,(50,lO, II /IGO,FT10F001 DD UNIT=SYSDA, // DCB=(RECFM=FB,LRECL=80,BLKSIZE=GIGO), /1 DISP=SHR,DSt'J=UF.B0063400.SG.RAIN IIGO.SYSIN DO 10 8 8 9 123 4 RUNOFF 8 9 9 8 o 8 o o TH t SIS A S t MULAT I OfJ RUN OF T\JENTY-F I VE YEARS OF HOURLY RAINFALL (ATLANTA AIRPORT) ON 10 HA THROUGH A BASIN. 1 0 0 60.0 1 25.0 0 0 0.5 7 31 12 72 090451ATLANTA, GEORGIA 9 9 G 98 o 48 1 0 OU40 OU49 o l) 50 OU51 0052 OU53 0054 0055 0056 0057 005H 0059 0060 0061 000L 00G3 0064 0065 0066 0067 1 1 1 103724.7 37.0 0.04 0.0130.25 0.0520.184 2.5 0.52 0.00115 0 00G9 0 1 0070 EtJDPROGRAM 0071 1* END OF WORK FILE

PAGE 113

Ouoo OU01 0002 0003 0004 0005 0006 0007 0008 0009 0010 Oll11 OU12 OU13 OU14 0015 OU16 0017 0018 0019 0020 OU21 00L2 '----.J 024 ." '0025 0026 0027 0028 0029 0030 0031 0032 0033 0034 0035 0036 0037 0038 0039 OU40 0041 0042 0043 0044 0045 0046 0047 )48 '-.J049 0050 0051 005:l Table A-2 StoragelTreatment Block Simulation 99 I/SWMM JOB (2006,3400,200,50,0),'GARY F. GOFORTt1',CLASS=1,REGION=512K, II I*PASSWORD 2,RAIN I*ROUTE PRINT REMOTE6 II EXEC FORTXCLE,LPARM='LIST,MAP,OVLY',OPTIONS='NOSOURCE,NOMAP' IIFORT.SYSIN DD SUBROUT I tJE COtB IN RETURN END SUBROUTINE TRANS RETURN END SUBROUTINE RECEIV RETURN END SUBROUTINE EXTRAN RETURN END SUBROUTINE RUNOFF RETURN END 1* IIJC LUD E urn TDNT* /*INCLUDE 5TRT&&& /*INCLUDE STRDAT* I*INCLUDE PLUGS* 1* I/LKED.SYSLMOD DD SPACE=(CYL,(S,l,l IILKED.SYSUTI DO SPACE=(CYL,(S,2 /ILKED.LIB DO DSN=UF.A0063473.SWMM,DISP=SHR IILKED.SYSIN DO INC L U 0 ELI B 01 A I r J) I NCLUOE LI B (GRAPH, CURVE, PI fJE, PPLOT, SCALE, HYSTAT) OVERLAY ALPHA INCLUDE LIB(CONTRX,STCOSX) INSERT Sl,S2. OVERLAY BETA INCLUDE LIB{EQUATX,INTERX) IIGO.FTOIFOOI DO UNIT=SYSDA,SPACE=(TRK,(SO,lO, I I VO L=S ER=\IORKO 1, DCB=( R ECFM=VBS, B LKS I Z E=4 2 40, BUFNO=I) IIGO.FT02FOOI DO II IIGO.FT03FOUI DO UNIT=SYSDA,SPACE=(TRK,(SO,lO, II VOL=SER=WORKOl,DCB=(RECFM=VBS,BLKSIZE=4240,BUFNO=1) IIGO.FT04FOOl'DD SYSOUT=A,DCB=(RECFM=FSA,LRECL=133,BLKSIZE=133) IIGO.FT08FOOl DD UNITaSYSDA,SPACE=(TRK,(SO,50, II VOL-SER=WORK01,DCB=(RECFM=VBS,BLKSIZE=4240,BUFNO=1) IIGO.FT09FOOI DO UrJIT=SYSDA,SPACE=(TRK,(50,10, II VOL=SER=WORKOl,DCB=(RECFM=VBS,BLKSIZE=4240,BUFNO=1) IIGO.FTIOFOOI DD UNIT=SYSDA, II DCB=(RECFM=VBS,BLKSIZE=4240,LRECL=80,BUFNO=1), II DISP=SHR,DSN=UF.A0063473.RAINAA IIGO.SYSIN DO 10 8 10 B 10 8 10 8 10 8 10 8 10 8 10 8 1 ? h

PAGE 114

Table A-2 (cont.) 100 l OU54 STORAGE 0055 BASIN ONE 0056 0057 B1. 0 1 1 1 0 OU58 Cl 530101 0.0 0 0 0 0059 01 0.1 0.1 0.1 0.1 0.1 0.1 0.1 ,0060 01 0.1 0.1 0.1 0.1 00G1 Ell OU62 F1 BASIN 0063 F2 10000.0 1 100 100 100 0064 G1 1.0 0065 G2 0 0 1 0066 G2 0067 G3 -1.666E-04 OOG8 G3 1.0 -1.0 0069 HI OU70 11 1 2 -3 0071 13 0.0 5000 a 0072 13 0.5 5304 a 0073 13 1.0 5616 0 0074 13 1.25 5775 3.151 0075 13 1.5 5936 4.457 'I U76 13 2.0 6264 6.303 13 2.5 6600 7.719 0078 1 3 3.0 G944 8.913 0079 13 3.5 7296 9.965 0080 13 4.0 7656 10.912 0081 13 4.5 8024 11.791 OU82 I 3 0033 15 0.0 0.0 0.450 0.450 0.0 0084 16 0.0 0085 /*INCLUOE Gl 0086 ENOPROGRAM 0087 / END OF \WRK FILE.

PAGE 115

Table A-3 Program Listing of Interface Program 1 01 l 0000 IIFLY JOB (2006,3400,lOO,5,0),FLY,CLASS=1,REGION=512K, -0001 II MSGLEVEL=(l,l) 0002 I*PASS\JORD 2,RAI N 0003 I*ROUTE PRINT rcp 0004 II EXEC WATFIV 0005 $JOB 0006 DIMENSION OUTFlW(24l,FLOH(24),POLl(24) 0007 DIMENSION IFLmJ(24) 0008 DIMENSION IDATE(24) 0009 Dlt .. 1ENSI01J ISAVECIO),tJDIM(lO) 0010 CHARACTER SOURCE*4(S),PNAME*4(2,2),PUNIT*4(2,2) 0011 CHARACTER TITLEl*4(40),TITLE2*4(40) 0012 LEND=721231 0013 LOCAT=090451 0014 NQSDUM=l 0015 READ(lO) TITLEl 0016 READ(lO)' IDATEZ,ZZERO 0017 READ(lO) TITLE2 0018 REAO(lO) (SOURCE(K),K=l,S),NSTEP,OELT,INLETS,NQSDUM,TAREA 0019 READCIO) (ISAVE(K),K=l,INLETS) o 0 2 0 REA D ( 1 0) P N AM E ( K, K W ) K = 1 2 ) K \ J = 1 N Q S DUM) 0021 RfAD(10) PUNIT(K,KW),K=1,2),KW=1,NQSDUM) 0022 READ(10) (NDIM(K),K=l,NQSDUM),QCONV 0023 10 ITFLOW=O 0024 DO 250 J=l,24 a a 2 5 REA D ( 1 0 ) AT I ME 1, I D ATE ( J ) T I tv! E I, 0 U T F UH J ) PO L L( J ) C********************* CONVERT CFS TO HUIWREDTHS OF AN INCH **** 0027 C FLOvJ(CFS) 3GOOSEC 100% 121N PER FOOT / ( 43S60FT2 PER ACRE 0028 C 24.7 ACRES) = FLOW 4.015124 o 0 2 9 2 0 0 F L mJ( J ) = 0 U T F UH J ) 4 0 15 1 2 4 o U3 0 C OU31 C********** COfJVERTING TO INTEGER AND ROUNDING UP ( >0.5) *********** 0032 I F LmJ( J ) = I F I X ( F LmJ( J ) ) OU33 DIFF=(FLOW(J)-FLOAT(IFLOW(J) 0034 IFCDIFF.GE.0.5) IFLOW(J)=IFLOW(J)+l 0035 C 0036 ITFLOW=ITFLOW+IFLOW(J) 0037 250 CONTINUE 0038 JDATE=IDATE(l) 0039 INT=JDATE/IOO 0040 .INT1=lrn*100 0041 INT2=JDATE-INTI IF(INT2.EQ.1)GO TO 280 0043 IF(ITFLOH.LT.l)GO TO 10. o U 4 4 2 8 0 R I T E ( 8 3 U 0) L 0 CAT J D ATE, ( I F L 0\ H J ) J = 1 1 2 ) L 0 CAT J 0 ATE, ( I F L mJ( J ) 0045 IJ=13,24), 0046 300 FORMAT(2IG,'1',1213,/,216,'Z',1213) 0047 IF CJDATE.LT.LEtlD) GO TO 10 0048 350 CONTINUE 0049 STOP 0050 END ,,0051 $ENTRY 0052 //GO.FT08FOOI DD UNIT=SYSDA, 0053 II DSN=UF.BOU63400.S6.RAINC,SPACE=(TRK,(125,10),RLSE), o 0 5 4 / I DIS P = (t J E \I CAT L G ) ,DC B = (f( E C F M = F B, lJ L K S I Z E ::: 6 1 GO, L R E C L = [: 0 ) 0055 IIGO.FTI0FOOI DO UNIT=SYSDA, 0056 II DCB=CRECFM=VBS,LRECL=BO,BLKSIZE=4240,BUFNO=1), o U 5 7 / / DIS P = S H R D S '" U F A 0 0 G 3 4 7 3 R A I r J A 1*r-n,1

PAGE 116

Table A-4 SYNOP Execution 0000 IISYNOP JOB (200G,3400,SO,5,O),'GRIZZLY',CLASS=1 0001 I*PASSI-JORD 0002/*kOUTE PRINT OUU3 II EXEC FORTXCLE,SUBLIB1='CIRCA.FORTLIB',LPARM='LIST,NOMAP' 0004 IIFORT.SYSIN DD 0005 1* 0006 IILKED.LIB DO DSN=UF.A0063473.SYNOP,DISP=SHR 0007 IILKED.SYSIN DO 0008 INCLUDE LIB(MAIN) OU09 INCLUDE LIB(COAD2,SETIA,NUMER,IDATE,SHELL,SHELR,DAIDA) 0010 IIGO.SYSIN DO 0011 90451 004110 0012 IIGO.FT10F001 DO UNIT=SYSOA,OISP=SHR,OStJ=UF.B0063400.S6.RAINC, 0013 II OCB=(RECFM=FB,LRECL=80,BLKSIZE=61GO) 0014 IIGO.FT11F001 DO UNIT=SYSOA,SPACE=(TRK,(20,10, 0015 II VOL=SER=WORK01,DCB=(RECFM=FB,LRECL=80,DEN=3,BLKSIZE=800) I 1016 1* OF WORK FILE 1 02

PAGE 117

L, G APPENDIX B BASIN Development and Listing BASIN is a small BASIC program written to generate hydraulic relationships variety Or basin geometries. This is necessary as input to the Storage/Treatment Block Or SWMM. The development BASIN is depicted in Figure B1. The wetted area of a basin is a function the basin side slope and the depth. The resulting stage-surface area relationship is integrated over the basin depth to yield the stage-volume relationship. I BASIN also calculates the stage-discharge relationship based on gravity drainage. 10 20 30 40 50 60 70 80 90 100 110 120 130 140 BASIN INPUT "WHAT ARE THE DEPTH, BASE, LENGTH, AND SLOPES?"iD,B,L, S1,82,S3,S4 INPUT "WHAT ARE THE DIMENSIONS OF THE OUTLET DIAMETER AND HEIGHT?";PI,HT FOR D1=0 TO D VOL=B*D1*L+Dl*D1*D1/3*(S1*S2+S2*S3+S3*S4+S4*S1)+ D1*D1/2*(B*
PAGE 118

L B \ y __ 1. \ = dt?s' en .. w 0:: C/) 0 dmax DEPTH 0 .. W ,(9 0:: :c () en Cl 0 dmax DEPTH Fi.gure B-1. Development of BASIN. 104 SA = BL + kId + k 2 d 2 kl = B(sl + 83 ) + L(82 + 84 ) k2 = 8182 + 8283 + 8384 + 8481 dV = SAdd V = S SAdd V = BLD + k3d2 + k 4 d3 k3 = kl/2 k4 = k2/3 > .. w ::> -.J 0 > 0 DEPTH Q = k (d -d )1/2 6 0 dmax k6 = (CE)SA 1 (2g)I/2 out et

PAGE 119

APPENDIX C Detention Time The concept hydraulic detention time is widely applied in many areas science and engineering. Typically, storage/treatment are designed under steady state and uniform conditions. In practice, detention present in estimating a hydraulic detention time due to characteristics which violate ideal assumptions. A simplirication rrequently employed in reservoir design and lake studies is averaging the rlow ovei a relatively long time interval, usually a year, and assuming that the change in volume over the time interval is zero, implying a steady state condition. The and storages in urban stormwater detention are characterized by a smaller (less than one year) time scale, due to quicker turnovers. As a result, there is a need to examine the rlow routing process on a smaller time scale in order to analyze the treatment characteristics Or the control unit.

PAGE 120

Behind td=V/Q: Steady-State Conditions Under unirorm flow conditions, the velocity Or water in a control unit is equal to 106 v==Q/A (C-l) where v is the velocity r LIT J, Q is the flow [ L3 IT J, and A is the cross-sectional area Or .plow [ L2 J. The length or time (t) required to travel a distance (L) is t==L/v (C-2) Assuming that the same steady and unirorm rlow conditions prevail in a detention basin, the travel time (detention time) is td == L/v == L/(Q/A) == (LA)/Q == vIa (C-3) where V is the basin volume in [ L3 J Figure C-1 shows some common configurations of control units at steady state conditions. For a system composed Or bas ins ins e r i e s ( C -1 b ) the tot a 1 d e ten t i on t i me i s the sum of the individual basin detention times. For parallel basins (C-lc), the average detention time ror the system is the sum of the product Or the percentage or total flow traversing each circuit and the detention time or that circuit. In Figure C-lc, the percentage Or flow passing through basin 1 is (alQ/Q), and the detention time is

PAGE 121

-1 v IQ I =td = VIQ a) VI I -/ Va td = VI/Q + V2/Q b) VI Q Q td = VI/Q+V2/Q V2 a2Q a2.Q c) KQ Q (I+K )Q Q td = (-'-.)(:!! ... E) VI V2 I+K. Q Q d) Fi 9 u r e C -1. V a rio usc 0 n t r 0 1 u nit con fig u rat ion s : steady-state conditions. 1 07

PAGE 122

\ 108 (Vl/alG). The 1 to the system detention time is (alG/G) (Vl/alG) = V1/O (C-4) calculations basin 2 yields the system detention time td = Vl/G + V2/Q (C-S) Expanding this concept to multiple tank in td = summation i=l to n (Vi/G) (C-6) a system containing a loop, as in C-ld, the detention time is to the weighted sum of the tanks' detention times and the detention time. The passing the system without is Q/(l+k)G = 1/(1+k) (C-7) k is the while the is kG/(l+k)G = k/(l+k) (C-8) The system detention time is td = (1/(1+k(Vl/Q+V2/Q) (C-9)

PAGE 123

Notice that the rlow in the recycle system receives less treatment time than the rlow through basins in series. Hence, settling basins are rarely designed with recycle. Analysis or Nonsteady-State Conditions By employing rinite dirrerence techniques, equations ror detention times or the in Figure C-l may be 109 developed ror nonsteady-state conditions. Using the average volume during the time interval dt, the expression dv/dt by derinition is zero. The rlow passing through a tank is described by both an inflow and a outflow term. By derining the instantaneous rlow as the mean or the inflow and outrlow magnitudes, the average rlow over a time interval can be determined as the mean instantaneous rlow. Q = (in + out)/2 Q(ave) = (0 at t + Q at t+dt)/2 (C-I0) Expanding this concept to average detention time, td(ave) = V(ave)/O(ave) (C-11) Control unit configurations under nonsteady but linear rlow conditions are presented in Figure C-2, where the errluent is proportional to the amount Or storage. The detention time may be represented as td = (2V) I (O+kV) (C-12)

PAGE 124

,,-... 1 1 0 v I K,V. (:-. td = V/OtKV) a) I .E.1 VI V td = 2V, Q + KIVI + 2V2 KIV, + K2V2 b) -VI Q K,VI+K2V2 2V, + 2V2 td = oIQI'+-KIVt -o2Q+K2V2 I..-e-V2 ....:::-0 a2Q K2V2 K2V2 KJVt K2 V2 td = R+St (A) VI V2 S :.' -Fjgure C-2. Various control configurations: nonsteady-state conditions.

PAGE 125

111 For a series or linear units, the system detention time is equal to the sum or the components# detention times. td = summation Trom i=l to n [(2V(i I (k(i-l)V(i-l)+k(i)V(i] For parallel tanks the detention time is td = summation rrom i=l to n C(2a(i)V(i I (a(i)Q + k(i>VCi] (C-13) (C-14) For tanks with recycle, the rlow recycled is proportional to the volume in basin 2. The amount flowing in the system (8) is 8 = (Q+2klVl+2k2V2+k3V2)/4 (C-15) The amount passing straight through the system (St) is (k3V2), and the recycle Tlow (R) is (k2V2>. The detention time is td = St+R)/S)(A) (C-16) where A = Vl/(Q+k2V2+klVl + (2V2)/(klVl+k2V2+k3V2 It is emphasized that these equations will describe steady state systems and will reduce to the equations in Figure C-l when in.plows equal 'out.plows. Also, these equations are equally valid .por any system which can be described by similar (linear) kinetics.

PAGE 126

L REFERENCES 1. Council on Environmental Quality. Environmental Quality. The Ninth Annual Report OT the Council on Environmental Quality, Non Point Sources. December 1978. 2. Chow, V. T Ed., Handbook oT Applied Hydrology A Compendium oT Water-resources Technology. McGraw-Hill Book Company, New York, N. Y., 1964. 3. Dever, R. Discussion on Stormwater Interception and Storage by Dominic Di Toro. oT the Environmental Engineering Division. ASCE. Vol. 106, No. EE4. April 1980. 4. Dillon, P. The Phosphorus Budget OT Cameron Lake, Ontario: The Importance oT Flushing Rate to the Degree oT Eutrophy oT Lakes. Limnology and Oceanography, Vol. 20, 1975, pp. 28-39. 5. Di Toro, D. M., and M. Sma 11. Stormwater Interception and Storage, oT the Environmental Eng i neer i ng Di vi s ion. ASCE, vo 1. 105, No. EEL February 1979, pp. 43-54. 6. C.!., Linear Theory or Hydrologic Systems, Technical BUlletin No. 1468, Agricultural Research Service, USDA, D. C., 1973. 7. Eagleson. P. S Dynamic Hydrology. McGraw-Hill Book Company, New York, New York, 1970. 8. Fair, G. M C. Geyer, and D. A. Okun, Water and Wastewater Engineering, Vol. 2, Water Purification and Wastewater Treatment and Disposal, Wiley and Sons, Inc., New York, New York, 1968. 9. Haan, C. T., Statistical Methods in Hydrology, The Iowa State University Press, Amos Iowa, 1977. 10. P., W. C. Huber, H. Sheikh, M. A. Medina, 11. R. Doyle, W. A. Peltz, and E. Darling. Urban Stormwater Management Modeling and Decision-Making, EPA Report No. 670/2-75-022, National Environmental 'Research Center" Cincinnati, Ohio, May 1975. Heaney, P., Economic/Financial Analysis OT Urban Water Quality Management Problems, EPA Grant No. R-802911-02-4, Municipal Environmental Research Laboratory, Cincinnati, Ohio. March 1979.

PAGE 127

L/ 113 12. Howard, C. D. D., of Storage and Treatment Plant Overflows, Journal of the Environmental Engineering Division, ASeE, Vol. 102, No. EE4, August 1976, pp. 709-722. 13. Huber, W. C., J. P. Heaney, and S. J. Nix, Stormwater Management Model User's Manual-Version III, EPA Draft Report, National Environmental Research Center, Cincinnati, Ohio, 1980. 14. Huber, W. C., P. L. Brezonik, J. P. Heaney, G. F. Goforth, M. C. Cullem and D. J. Pollmann, An Environmental Study of Hogtown Creek in Gainesvil.le, Fla., Final Report to Florida Dept. of Environmental Regulation, Gainesville, Fla., March 1981. 15. Hydroscience, Inc., A Statistical Method for the Assessment of Urban Stormwater, EPA-440/3-79-023, Office of Water Planning and Standards, Washington, D. C., May 1979. 16. Liptak, B. G., Environmental Engineers' Handbook, VOl.1 1: Water Pollution, Chilton Book Company, Radner, Pennsylvania, 1974. 17. Med ina, M. A., Jr., Interaction of Urban Stormwater Runoff, Control Measures and Receiving Water Response, Dissertation presented to the Graduate Council of the University of Florida-in partial fulfillment of the reguirements for the degree of Doctor of Philosphy, Gainesville, Florida, 1976. 18. Medina, M.A., Discussion on Water Quality Trap Efficiency of Storm Water Management Basins, by R. H. McCuen, Water Resources Bulletin, Vol. 17, No. 1, February 1981. 19. Nix, S. J., J. P. Heaney, and W. C. Huber, Water Quality Benefits of Detention, Chapter 12 of APWA Manual, January 1981. 20. Reckhow, K. H., Lake Quality Discriminant Analysis, Water Resources Bulletin, Vol. 14, No.4, August 1978. 21. Reckhow, K. H., M. N. Beaulac and T. Simpson, Modeling Phosphorus Loading and L.ke Response Under Uncertainty: A Manual and Compilation of Export Coefficients, EPA Clean Lakes Section, Washington, D. C., June 1980. 22. Rich, L. G. I Environmental Systems Engineering, McGraw-Hill Book Company, New York, New York, 1974.

PAGE 128

114 23. Sc hnoor, J. L., and D. J. 0 'Conner, A Simp liT i ed Approach Tor Eutrophication Modeling of Lakes, A Papaer Presented at the 51st. Annual ConTerence OT WPCF, Anaheim, CaliTornia, October 1978. 24. Small, M. and D. M. Di Toro, Stormwater Treatment Systems, Journal OT the Environmental Engineering Division, ASCE, Vol. 105, No. EE3, June 1979, pp. 557-569. 25. Smolenyak, K., Urban Wet-Weather Pollutant Loadings, A thesis presented to the Graduate Council OT the University OT Florida in partial fulfillment of the requirements for the Degree or Master Or Engineering, University of Florida, Gainesville, Florida, 1979. 26. Sonzogni. W. C., P. C. Uttormark and G. F. Lee, A Phosphorus Residence Time Model: Theory and Application, Water Resources Research, Vol. 10, March 1975, pp. 429-435. 27. U. S. Army Corps Or Engineers, Urban Stormwater Runoff STORM, Generalized Computer Program, 723-S8-L2520, 28. H. E.C., May 1974. Vollenweider, R. A., Water Management Research, DECO-Report 68-72, Paris, 1968. 29. Vollenweider, R. A., and P. J. Dillon, The Application of the Phosphorus Loading Concept to Eutrophication Research, Canada Centre for Inland Waters, National Research Council of Canada, 1974. 30. Vollenweider, R. A., Advances in Defining Critical Loading Levels for Phosphorus in Lake Eutrophication Men. Ita!., Idrobio1., 1976. 31. Ward, A. J., C. T. Haan and B. J. Barfield, Simulation of the Sedimentology OT Sediment Detention Basins, OWRT B-046-KY, Water Research Institute, Lexington, Kentucky, June 1977. 32. Weber, W. J., Physicochemical Processes ror Water Guality Control, New York, New York, 1972. 33. White, J. B., and M. R. AlIos, Experiments on Wastewater Sedimentation, Journal WPCF, Vol. 48, No.7, July 1976, pp. 1741-1751.

PAGE 129

BIOGRAPHICAL SKETCH Gary Frank Edmon Goforth was born on December 12, 1956. Just outside the prison walls in Huntsville, Texas. He grew up in Texas and Indiana and has lived in Florida since 1966. He enrolled at the University Florida in 1975, and received a Bacheior of Science in Engineering (Environmental Engineering) in 1979. He subsequently entered the Graduate School at the University of Florida, working on a Master of Engineering Degree in Environmental Resources Management. He was a member of Tau Beta Pi and Kappa Phi Kappa honorary societies, the Water Pollution Control Federation, the American Water Works Association, and the Aircraft Owners and Pilots Association. To support his academic habit, he worked throughout his undergraduate and graduate career. Beginning as a haberdasher, he has monitored the effects thermal effluent from a nuclear power plant on a saltwater marsh, researched tertiary treatment of domestic waste in a freshwater marsh, and assessed the impact of urbanization on the water quality of an urban stream. He has coauthored several annual reports associated with these proJects. In 1980, he married a wonderful little Polish girl, Karen. 115