States
Environmental Protection
Agency
Office of Air
Land and Water Use
Washington, DC 20460
EPA 600/9-79-003
November 1978
Research and Development
v>EPA
Proceedings
Stormwater Management
Model (SWMM)
Users Group Meeting
November 13-14, 1978
Miscellaneous Reports Series
-------
EPA-600/9-79-003
November 1978
PROCEEDINGS
STORMWATER MANAGEMENT MODEL (SWM)
USERS GROUP MEETING
13-14 NOVEMBER 1978
Project Officer
Harry C. Torno
Office of Air, Land, and Water Use
Office of Research and Development (RD-682)
U.S. Environmental Protection Agency
Washington, D.C. 20460
OFFICE OF AIR, LAND, AND WATER USE
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
-------
DISCLAIMER
This report has been reviewed by the Office of Air, Land and Water
Use, Office of Research and Development, U.S. Environmental Protection
Agency, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the U.S. Environ-
mental Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
IX
-------
FOREWORD
A major function of the Research and Development program of the
Environmental Protection Agency is to effectively and expeditiously trans-
fer, to the user corenunity, technology developed by those programs. A
corollary function is to provide for the continuing exchange of information
and ideas between EPA and users, and between the users themselves. The
Stormwater Management Model (SWMM) users group, sponsored jointly with
Environment Canada/Ontario Ministry of the Environment, was established to
provide such a forum.
This report, a compendium of papers presented at the last Users Group
meeting, is published in the interest of disseminating to a wide audience
the work of Group members.
Courtney Riordan
Acting Deputy Assistant Administrator
Office of Air, Land and Water Use
111
-------
ABSTRACT
This report includes ten papers, on various model - related topics,
presented at the semi-annual joint U.S. - Canadian Stornwater Manage-
ment Model (SVMM) Users Group Meeting, held November 13-14, 1978 in
Annapolis, Maryland.
Topics covered include a description of the new SWMM Storage/Treat-
ment program, several papers on storrnwater retention, a methodology for
evaluating agricultural BMP's, two papers on model applications in facility
planning, and other papers on model applications.
IV
-------
CONTENTS
Page
Number
Foreword iii
Abstract iv
REVISED SWMM STORAGE/TREATMENT
BLOCK, S.J. Nix, J.P. Heaney and W. C. Huber 1
ON-SITE CONTROL OF NONPOINT SOURCE
POLLUTION, R. H. McCuen 28
STORMWATER MODELING APPLICATIONS,
A. R. Perks 39
MODEL FOR SELECTION OF STORMWATER
CONTROL ALTERNATIVES, R. Bedrosyan
and J. Ganczarcyk 50
METHODOLOGY FOR EVALUATING AGRICULTURAL
BEST MANAGEMENT PRACTICES (BMP's)
J. Kuhner, W. W. Walker and J. J. Wineman 88
COMPUTER SIMULATION OF FLOOD RELIEF WORKS
UTILIZING INLET CONTROL AND DETENTION
STORAGE, P. E. Theil and A. M. Candaras 105
ANALYSIS OF DETENTION BASIN SYSTEMS,
R. G. Mein 131
CSO FACILITIES PLANNING IN CINCINNATI
USING SWMM (A CASE STUDY) ,
J. D. Sharon 147
SWIM USAGE IN FACILITIES PLANNING,
G. D. Cole, L. W. Varner, J. W. Shutt 190
A PRE- AND POST-PROCESSING PROGRAM
PACKAGE FOR THE STORMWATER
MANAGEMENT MODEL, W. James 218
List of Attendees 235
v
-------
REVISED SWMM STORAGE/TREATMENT BLOCK
by
Stephan J. Nix, James P. Heaney, and Wayne C. Huber
Department of Environmental Engineering Sciences
University of Florida
Gainesville, Florida 32611
Presented at the
Storm Water Management Model Users
Group Meeting
Annapolis, Maryland
November 13-14, 1978
PREFACE
The following paper is Section 5 of a forthcoming documentation report
of SWMM procedures for the Environmental Protection Agency. The research has
been supported by EPA Grants R-802411 and R-805664.
-------
SECTION 5
STORAGE/TREATMENT SIMULATION
OBJECTIVES
The primary objectives of the Storage/Treatment Block are to:
1. provide the capability of modeling a larger number
of processes in both the single event and continuous
modes;
2. simulate the quality improvement provided by each
process;
3. simulate the handling of sludges; and
4. provide estimates of capital, operation and
maintenance costs.
Although the objectives of the Storage/Treatment Block have not changed
appreciably from earlier versions (1), the model has been virtually re-
written. The earlier versions were too limited in use and scope. This
version is much more flexible in terms of the control units available, pol-
lutant routing and cost estimating. However, the user is advised that in-
creased flexibility implies increased user input and knowledge of the pro-
cesses to be modeled. In other words, the model does not provide sevei^j.
dozen specialized designs, but provides the tools necessary to simulate the
desired processes. Naturally, flexibility precludes ultrasophistication.
Several precautions should be noted before setting up the S/T Block.
1. The need for local waste characterization data cannot be
overemphazised. Treatment unit performance is a function
of the nature of the waste.
2. Lab or pilot plant performance data should be used whenever
possible to derive performance functions.
3. Dry-weather treatment performance functions should be
applied cautiously to wet-weather units.
\
PROGRAM DEVELOPMENT AND OVERVIEW
Development
Past versions of the Storage/Treatment Block simulated various pro-
cesses on the basis of limited empirical data and operating experience.
-------
Often the data were localized and/or specialized. Thus, they were of ques-
tionable applicability to a wide variety of situations. Additionally, the
model did not account for the physical characteristics of the incoming waste
stream or the handling of sludges.
To improve the storage/treatment modeling capabilities of SWMM the
following considerations were instrumental in creating a new model.
1. There should be a high degree of flexibility in the simulation
of individual units and the interaction among units.
2. In addition to simulating the mass of pollutants, it is
important to account for the physical characteristics
(i.e., particle size and specific gravity distribution)
of each pollutant.
3. Sludge handling is an important part of any wastewater
treatment scheme and should be simulated.
4. All costing routines should be as flexible as the
performance algorithms.
5. The model should be capable of modeling wet- and
dry-weather facilities.
Overview
The present Storage/Treatment Block is approximately 1500 Fortran
statements in length and consists of eight subroutines. The routing of flow
and pollutants through the entire block is controlled by subroutine STRT
which is called from the Executive Block. STRT also provides the main
driving loop for the model and generally acts as the central coordinating
subroutine. Subroutine STRDAT is called in STRT and is responsible for
reading the input data provided by the user. Subroutine CONTRL is called
each time-step from the main driving loop in STRT. CONTRL directs flow and
pollutants from one unit to another as prescribed by the desired scheme and
coordinates the majority of the printed output. Subroutine UNIT is called
from CONTRL for each unit modeled and is the heart of the Storage/Treatment
Block. It contains the necessary flexibility and capability to model most
storage/treatment processes (units). Subroutine EQUATE is used by UNIT to
provide several forms of pollutant removal equations. Subroutine INTERP is
employed by UNIT for linear interpolation. Subroutine PLUGS is used by UNIT
to model perfect plug flow through a detention unit. Subroutine STCOST is
called from STRT to determine capital, and operation and maintenance costs.
The model has become user-intensive rather than program-intensive. The
user is responsible for providing the program with the desired storage/
treatment scheme and operating characteristics of each unit (along with
other information). However, input guidelines are provided in the User's
Manual for several types of units. Again, the strength of this approach is
to maximize flexibility and applicability to local conditions and design
criteria.
-------
SIMULATION TECHNIQUES
Introduction
Flow and pollutants are routed through one or more storage/treatment
units by several techniques. The units may be arranged in any fashion,
restricted only by the requirements that inflow to the plant enters at only
one unit and that the products (treated outflow, residuals, and bypass flow)
from each unit not be directed to more than three units. The flows into,
through and out of a unit are shown in Figure 5-1. Treatment and sludge
handling units are modeled by the same subroutine (UNIT). Additionally,
both wet- and dry-weather facilities may be simulated by the proper
selection of unit arrangement and characteristics. Units may be modeled as
having a detention capability or instantaneous throughflow. Pollutants or
sludges may be represented as a simple mass or further characterized by a
particle size distribution. A unit may remove pollutants (or concentrate
sludges) as a function of particle size and specific gravity, detention
time, incoming concentration, the removal rate of another pollutant, or a
constant percentage. The S/T Block can receive the flow and any three
pollutants from any one outlet in any other block of SWMM. Also, flows and
pollutants may be provided by the user and fed directly to the S/T Block.
If both sources are present they are combined and treated as one input. For
example, the user may enter directly dry-weather flows and enter wet-weather
flows from the Runoff Block. All flows and pollutants in the S/T Block are
assumed to be' averages over a time step. This includes the input data and
internal calculations. '
The following sections describe the techniques available for flow and
pollutant routing which allow the user to model several types of storage/
treatment units.
Flow Routing
Detention vs. Instantaneous Throughflow --
A unit may be modeled to handle flow in one of two ways; as a detention
basin (reservoir) or a unit instantaneously passing all flow. The idea of a
detention basin is not limited to storage basins and sedimentation tanks but
also includes such processes as dissolved air flotation, activated sludge,
and chlorination. Processes that may be modeled as having instantaneous
through-flow include microscreens, fine screens and other forms of screening.
-------
'tot
BYPASS EXCESS
(YES)
Q
by
LEGEND
Qtot
T6S
Q
by
in
Qout
Qres
= TOTAL INFLOW, ft3/sec
= MAXIMUM ALLOWABLE INFLOW,
ft^sec
= BYPASSED FLOW, ft3/sec
= DIRECT INFLOW TO UNIT, ft3/sec
= TREATED OUTFLOW,
= RESIDUAL STREAM, ft°/s©c
Figure 5-1. Flows Into, Through, and Out of a Storage/Treatment Unit,
-------
Detention Basins --
The rate of change of storage in a detention basin is found by writing
a mass balance equation for the system shown in Figure 5-2.
V
Figure 5-2. Time Varying Inflow and Outflow Rates for a Reservoir.
The rate of change of storage equals inflow minus outflow, or
AV/At =1-0 (5-1)
_ 2
where I = average inflow rate during At, ft /sec.
_ 3
0 = average outflow rate«during At, ft /sec.
V = reservoir volume, ft , and
At = time step, seconds.
Let subscripts 1 and 2 denote the beginning and end of the time period,
respectively. Then, the average inflow rate I, is
I =
+ I2)/2
The average outflow rate, 0, is
0 = (0, + 00)/2
i z
Also, the change in reservoir volume is
AV = V0 - V,
(5-2)
(5-3)
(5-4)
Substituting equations 5-2, 5-3, and 5-4 into equation 5-1 and multiplying
through by At yields the desired expression for the change in volume, i.e.,
V - V, =
At -
At .
(5-5)
-------
For a given time step, I,, I~, 0.., and V, are known and 0? and V~ need to
be determined. Grouping the unknowns on the left hand siae of tne equation
and rearranging yields one of the two required equations:
(0.5)02 At + V2 = Q.5(Il + I2)At - [(0.5) OjAt - Vj] (5-6)
The second required equation is found by relating CL and ¥„, each of which
is a function of reservoir depth. The procedure is illustrated in the
following example.
Table 5-1 presents geometric data for a hypothetical reservoir with a
base elevation of 343.0 ft and a maximum pool elevation of 353.0 ft. The
corresponding depths are shown in column 3. Surface area, as a function of
depth, is presented in column 4. If the reservoir has an irregular geometry,
the surface area is measured from a topographic map. The eleven depth-area
data pairs shown in columns (3) and (4) of Table 5-1 are required input data.
The model calculates the depth versus storage relationship by linear inter-
polation of the area between any two adjacent depth-area data pairs.
Recalling equation 5-6, the objective is to find
0.5 02 At = f(0.5 02At + V2) (5-7)
The data in Table 5-2 give CL and ¥„ as functions of depth over the range
of depths for which outflow occurs. In this case, outflow occurs only if
the reservoir depth exceeds 8.0 ft. Thus, eleven data triplets may be
selected which cover the range of interest. Then, as an option, the
user inputs these eleven triplets of columns (3), Depth, (6), Outflow
Volume (02DT2), and (7), Outflow & Reservoir Volume (SATERM). During
the simulation the model uses this relationship to estimate end-of-period
storage and outflow.
Alternatively, the model will continue to generate parameters 02DT2
and SATERM from a specified weir, orifice or pumped outflow configuration.
If a weir or orifice is at an elevation above the reservoir bottom (e.g.,
at elevation 351 ft in the example), the program will generate seven pairs
of 02DT2 and SATERM at and above the weir/orifice elevation and four below.
In both cases, eleven depth-area pairs are also input. In the latter
situation discussed above, these are used to calculate the depth-volume
relationship for the reservoir. In the former situation in which the data
triplets are input, the volume computation is not required. In both cases,
the surface area is needed to estimate the volume of evaporation lost from
the reservoir.
-------
Table 5-1. Geometric Data For Hypothetical Reservoir
n
(1)
1
2
3
4
5
6
7
8
9
10
11
Elevation
h
ft
(2)
343.
344.
345.
346.
347.
348.
349.
350.
351.
352.
353.
Depth
y
ft
(3)
0
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
Surface
Area
A 2
1000 ft Remarks
(4) (5)
0 Base of reservoir
3.
15.
45.
121.
225.
365.
550.
790. Weir elevation
1080.
1440. Maximum pool
Column
(1) Counter
(2) Elevations from topographic map
(3) Depth = h - 343.
(4) Measured from topographic map or may be calculated (by user)
if geometry is regular.
-------
Table 5-2. Routing Data For Hypothetical Reservoir
Elevation Depth
n h y
ft ft
(1) (2) (3)
1
2
3
4
5
6
7
8
9
10
11
Column
(1)
(2)
(3)
(4)
(5)
(6)
(7)
351-0 8.0
351.2 8.2
351.4 8.4
351.6 8.6
351.8 8.8
352.0 9.0
352.2 9.2
352.4 9.4
352.6 9.6
352.8 9.8
353.0 10.0
Volume Discharge
v2 o2
1000 ft3 ft3/sec
(4) (5)
1720.
1850.
2000.
2220.
2400.
2650.
2900.
3100.
3400.
3700.
3900.
0
10.
20.
35.
50.
65.
80.
105.
130.
165.
200.
Counter
Elevation from topographic map
Depth = h - 343.0
Volume measured or calculated volume
Measured data or calculated from discharge
Calculated using 02 (col. 5), At = 21,600
Calculated using col. 4 and col. 6
02DT2
.5 02 At
1000 ft3
(6)
0
108.
216.
378.
540.
702.
804.
1134.
1404.
1782.
2160.
formulas
sec
SATERM
.5 02At +
1000 ft3
(7)
1720.
1958.
2216.
2598.
2940.
3352.
3764.
4234.
4804.
5482.
6060.
V,
-------
The computational procedure is summarized as follows:
1. Known values of I.. , !„, 0 , At, and V.. are substituted into the right
hand side of equation 5-6. The result is the first value of 0.5 CLAt + V~.
2. Knowing (0.5 CLAt + V2) the value of 0.5 CLAt is obtained by interpolating
between adjacent values of 02DT2 and SATERM.
3. Twice the value of 0.5 CLAt is subtracted from V~ to give the new
0.5 OjAt - Vr
4. Add 0.5(1, + !„) At to the new value of 0.5 0,At - V. to get the new
value of 0.5 C>2At + V2-
5. Continue this process until all inflows have been routed.
To summarize the input alternatives, the earlier version of the
storage model permitted the user to read in depth-area data and an
outflow condition of a weir, orifice, or pumping. It could not handle
the case of a natural reservoir with an irregular stage-discharge relation-
ship. The updated model allows the user to input the required relationship
between storage volume and outflow rate in the form of the eleven data
triplets discussed previously. This approach permits the user to select
the eleven data points which best approximate the desired functional
relationship. This approach is felt to be preferable to adding more
complexity to the model to analyze automatically the wide variety of
reservoir geometries and operating policies encountered in practice.
An excellent description of this level-surface routing procedure
(the Puls Method) is presented in Viessman et al. (2). Sound engineering
judgement is essential in setting up this routing procedure. The input
data and associated assumptions should be checked carefully.
Evaporation --
Evaporation losses are also accounted for in detention units. The loss
is computed, at the end of each time step, by
ey = A • ed • At (5-8)
3
where e = evaporation loss, ft /time step, ~
A = surface area at the water level in the unit, ft ,
e, = evaporation rate, ft/sec., and
At = time step, sec.
The user must supply the values of e, for each month of the simulation
period. Losses are taken from the total storage volume present at the
end of each time step. In the case where routing by plug flow is specified
(see below) the loss is proportionally taken from each plug, depending on
its proportion of the total storage volume.
10
-------
Instantaneous Throughflow --
If the unit is specified to have no detention capability, then the
model assumes that what arrives during a time step leaves as treated out-
flow that same time step less removed volumes. In this case, the volume of
removed water is taken as a simple percentage of the inflow.
Pollutant Routing
Complete Mixing --
Pollutants are routed through a detention unit by one of two modes:
Complete mixing or plug flow. For complete mixing, the concentration of
the pollutant in the unit is assumed to be equal to the effluent concentration.
The mass balance equation for the assumed well mixed, variable volume
reservoir shown in Figure 5-3 is (3):
= 0(t) C(t) - I(t) CT(t) - K C(t) V(t) (5-9)
dt
3
where V = reservoir volume, ft
C = influent pollutant concentration, mg/1,
C = effluent and reservoir pollutant concentration, mg/1,
I = inflow rate, ft /sec,
0 = outflow rate, ft /sec,
t = time, sec and ,
K = decay coefficient, sec.
Equation 5-9 is very difficult to work with directly. It may be approximated
by writing the mass balance equation for the pollutant over the interval, At:
Change in Mass entering Mass leaving Decay during
mass in tank = during At - during At - At
during At
c i + c1 i co + co c v + c v
At - At - K At (5-10)
where subscripts 1 and 2 refer to the beginning and end of the time step,
respectively.
From the reservoir routing, 1^ I2> QI , Q^ V^ and V£ are known. The
concentration in the reservoir at the beginning of the time step, C^} and
the influent concentrations, C and C are also known as are the decay rate,
K and the time step, At. Thus, the only unknown, the end of period con-
centration, C~, can be found directly by rearranging equation 5-10 to yield
11
-------
Kt),
-------
(C? I + cJ IJ CO K C V
C V. + L 1 0 l Z At - -f-^ At ^i-i At
C2 = 1 * * 2 ^ (5-11)
V0(l + V^) - ^ At
/ z 2
Equation 5-11 is the basis for the complete mixing model of pollutant
routing through a detention unit.
Plug Flow --
If the user selects the plug flow option, the inflow during each time
step, herein called a plug, is labeled and queued through the detention
unit. Transfer of pollutants between plugs is not permitted. The outflow
for any time step is comprised of the oldest plugs, and/or fractions thereof,
present in the unit. This is accomplished by satisfying continuity for the
present outflow volume (which was calculated earlier):
LP
I V. • f. = V (5-12)
j=JP ' J °
3
where V = volume leaving unit during the present time step, ft ,
V. = volume entering unit during j time step (plug j), ft ,
f. = fraction of plug j that must be removed to satisfy
J continuity with V , 0 ^ f. S 1,
JP = time step number of the oldest plug in the unit, and
LP = time step number of the youngest plug required to
satisfy continuity with V .
The detention time (sec) for each plug j is calculated as
(t ) = (KKDT - j) At (5-13)
J
where KKDT = present time step number.
For a plug j leaving the unit, the amount of pollutant leaving is
(P ) = (P ) f (1.0 - R ) (5-14)
J Jo «J
where (P ). = amount of pollutant leaving unit in plug j, Ibs,
tj
(P.). = amount of pollutant entering unit with plug j, Ibs, and
R. = removal fraction for plug j
The manner in which R. is calculated is decided by the user; however, as
with the complete mixing option, R. should be function of (t ,) . . The
technique for developing removal equations is discussed later.-' The remaining
pollutants in each plug leaving the unit are totaled to give the total amount
discharged during the present time step.
13
-------
In either mode, removed pollutants are accumulated in the unit and exit
with the removed water to form the residuals stream. The fraction of accumu-
lated pollutants removed from the unit in any one time step is equal to the
fraction of stored water removed from the unit. The water removal rate is speci-
fied by the user as a fraction of the remaining storage or a constant rate.
Instantaneous Throughflow --
Pollutants are routed instantaneously through units modeled as having no deten-
tion capability. In other words, the pollutants arriving during a time step
leave the same time step less the removed portion. The removed pollutants are
combined with the removed water to form the residuals stream. The rate at which
water is removed is a user specified fraction of the inflow.
Sludge Handling --
Sludge (or sludge volume) is assumed to consist of the removed suspended
solids and water from any storage/treatment unit. Sludge is not routed differ-
ently from any other flow in the system - it is simply one of the flows leaving
a unit. In other words, the residuals stream from a unit can only be termed
"sludge" if suspended solids are routed. The removal techniques discussed below
are applicable to sludge handling units as well as mainline units.
Pollutant Characterization
Pollutants are characterized by their magnitude (i.e., mass flow and
concentration) and, if the user desires, by particle size and specific
gravity distributions. Describing pollutants by their particle size dis-
tribution is especially appropriate where small or large particles dominate
or where several storage/treatment units are operated in series. For ex-
ample, if the influent is primarily sand and grit, then a sedimentation unit
would be very effective; if clay and silt predominate, sedimentation may be
of little use. Also, if several units are operated in series, the first
units will remove a certain range of particle sizes thus affecting the per-
formance of downstream units. Therefore, the need for describing pollutants
in more detail is obvious for modeling purposes. The pollutant removal
mechanism peculiar to each characterization is discussed below.
Pollutant Removal
Characterized by Magnitude --
If pollutants are characterized only by their magnitude then the model
improves the quality of the waste stream by removal equations. Removal of any
pollutant may be simulated as a function of detention time (in minutes,
detention units only), incoming concentration, inflow rate, the removal
fraction of another pollutant, the incoming concentration of another pollutant
or any combination of the above "removal factors". This selection is left to
the user. Two functional forms are provided by the program to construct the
desired removal equation:
aixi ao a-3X-a a/ at;xc; a£ 37X7 ae
R . a,e ' ' */ * a10e 3 3 X, * * .„. 5 5 K/ * .„. 7 7 */ (5-15,
14
-------
or
(alxl + a2X2 + a3X3 + 34X4) a5 a6 a7 38
R = age x5 x6 x? xg (5-16)
where x. = removal equation variables,
a. = coefficients, and
RJ = removal fraction, 0 £ R £ 1.0.
Within the prograifi, the user may assign removal equation variables, x., to
be parameters such as detention time, flow rate, inflow concentration of
the parameter being removed, inflow concentration of another parameter, etc.
These are seen to be parameters which are computed within the program at each
time step. (If they are not assigned to be specific parameters, the remaining
x. are set equal to 1.0 for the duration of the simulation.) The coefficients,
a., are directly specified by the user. There is considerable flexibility
contained in these two forms, and with a judicious selection of coefficients and
factors, the user can probably create the equation he desires. Example appli-
cations of equations 5-15 and 5-16 are given below to illustrate the procedure.
Earlier versions of the Storage/Treatment Block employed the following
removal equation for suspended solids in the sedimentation unit (5):
—K>
R = R ( 1 - e d) (5-17)
max
where R = suspended solids removal fraction, 0 S R S R ,
T. . T ,. max
R = maximum removal fraction,
t, = detention time, min., and
K = first order decay coefficient, 1/min.
This same equation could be built from equation 5-15 by setting aq = R ,
a = -R , a_= -K and letting x = detention time, t,. All other
coefficients, a., would equal zero.
Another example is taken from a study by Lager et al. (6). Several
curves for suspended solids removal from microstrainers were derived with
a variety of aperture sizes. Fitting a power function to the curve
representing a 35-micron microstrainer yields
R = 0.0963 SS°'286 (5-18)
where R = suspended solids removal fraction, and 0 S R S 1.0, and,
SS = influent suspended solids concentration, mg/1.
Equation 5-16 can be used to duplicate this removal equation by setting
aQ = 0.0963, a = 0.286 and x = influent suspended solids concentration, SS.
_7 J -~)
All other a. are zero.
J
Sludge handling may also be modeled with equations 5-15 and 5-16.
Figure 5-4 shows the reduction in volatile solids in raw sludge (suspended
solids - see earlier discussion) by a digester as a function of percent
volatile solids and detention time (4). These curves can be approximated by
15
-------
O
z>
o
UJ
(T
O
CO
UJ
-J
O
90
80
70
60
50
40
30
20
10 20 30 40 50 60 70
RETENTION BASED ON RAW SLUDGE
FEED, DAYS
Figure 5-4. Reduction in Volatile Solids in Raw Sludge (4).
16
-------
, t 0.33 ,.,
R = 1.31 x 10'4 () pl'67 (5-19)
where R = volatile solids reduction, 0 S R S 1.0
t = detention time, min,
P = percent volatile solids in raw sludge,
T7C
Pv = 100 § ' (5-20)
where VS = influent volatile solids concentration, mg/1, and
SS = influent suspended solids (raw sludge) concentration, mg/1.
Equation 5-16 can be used to construct equation 5-19 by setting
a = (1.31 x 10~4)(1440)~°'33, a = 0.33, a^ = 1.67, a = -1.67, x = detention
time, t , , x, = influent volatile solids concentration, VS, and x? = influent
suspended solids (raw sludge) concentration, SS. A current description of
sludge handling can be found in references 7-9.
Characterized by Particle Size and Specific Gravity Distribution --
Distribution - If pollutants are characterized by their particle size
and specific gravity distribution, then they are removed from the waste
stream by particle settling or obstruction. Many storage/treatment
processes use these physical methods to treat wastewater; sedimentation
and screening are among the most obvious examples.
In this mode, pollutants are apportioned over several (up to 10)
particle size/specific gravity ranges (e.g., ten percent of the BOD
is found in the range from 10 to 50 microns). Each of the ranges is
preset' by the user and assigned an upper and lower bound on the particle
diameter and a value for specific gravity. The user also specifies the
apportionment of pollutants over the various ranges as they enter the first
unit. This distribution is modified as it passes through the storage/
treatment plant. Unfortunately, the distribution entered at the first unit
must remain constant over time since the other blocks of SWMM do not provide
a time-varying particle size distribution.
Each unit removes all or some portion of each range; the associated re-
moval of pollutants is easily determined. For example, if a sedimentation
unit removes 50 percent of the particles in the 50 to 100 micron range and
ten percent of the pollutant in question is found in this range, then five
percent of the total pollutant load is removed. The total removal is deter-
mined by summing the effects of several ranges passing through this unit.
Once certain particles are removed, the distribution of particle sizes for
the outflow can be determined and passed on to the next unit or receiving
water. The removed particles constitute the size distribution for the
residuals stream. The next several paragraphs describe the two mechanisms
available to the user for pollutant removal when pollutants are characterized
by particle size.
17
-------
Particle settling - There are several forms of settling: unhindered settling
by discrete particles, settling by flocculating particles, and hindered
settling by closely spaced particles (10). For simplicity, the unhindered
settling of discrete particles will be the removal mechanism simulated in
this model.
Discrete particles settling in a quiescent fluid accelerate to the
point where the drag force exerted by the suspending fluid reaches equil-
ibrium with the gravitational force exerted on the particle (10). At this
point, the particle settles at a constant velocity known as the terminal
velocity. By equating the forces acting on such a particle, the equation
for the terminal or settling velocity of the particle is derived and approx-
imated by (10)
'
=/4 _g (S - 1) d (5-21)
where v = terminal velocity of particle, ft/sec,
g = gravitational constant, 32 ft/sec ,
Cj, = drag coefficient,
S = specific gravity of particle, and
d" = diameter of particle, ft.
Additionally,
C= I4- , if N < 0.5, or (5-22)
u JNR
24 _3 4
CD = N + 0j + 0.34, if 0.5 ^ NR < 10 , or, (5-23)
CD = 0.4, if NR 2 104. (5-24)
where NR = Reyonlds number, dimensionless,
NR = vg d/v (5-25)
2
and v = kinematic viscosity, ft /sec.
The procedure for finding v under any of the above conditions is demonstrated
by Sonnen (11) and is computed at the beginning of the simulation and stored
for later use. The average of the high and low ends of each particle size
range is used as the representative particle size for use in the above
calculations.
A range of conditions may exist in an actual detention unit, from very
quiescent, to highly turbulent and nonquiescent. Camp's (12) ideal removal
efficiency, £„, will be used for quiescent conditions, and an adaptation of
his sedimentation trap efficiency curves (12, 13, 14) as described by Chen (15)
will be used to make the extension to nonquiescent conditions, as described
below.
18
-------
For quiescent conditions,
EQ = rain J 1 (5-26)
^ Vs'Vu
where EQ = particle removal efficiency as a fraction, 0 £ E~ £ 1,
v = terminal velocity of particle, ft/sec, and
v = overflow velocity, ft/sec.
u
Additionally,
Ay/td = y/t . (5-27)
3
where Q = flow rate, ft /sec, „
A = surface area of detention unit, ft ,
y = depth of water in unit, ft, and
t, = detention time, sec.
Equation 5-27 assumes a rectangular detention unit with vertical sides.
However, a circular unit (with vertical sides) may also be modeled when
characterizing pollutants by particle size. In other words, equation 5-27
is restricted to units that allow the surface area to remain constant at
any depth. Applying this equation (and, thus, the entire particle size
methodology) to other unit types should only be done when the surface area is
independent of depth.
Equation 5-26 represents an ideal quiescent basin in which all particles
with settling velocities greater than v will be removed. Deviations from
quiescent conditions can be handled explicitly based on Camp's (12) sedimentation
trap efficiency curves, which were developed as a complex function of particle
settling velocity and several basin parameters.
v y v A v £ v
E = f < 2!- » i- = ^ =-^r > <5-28>
H tj u
where E = particle removal efficiency, 0 ^ E S 1, „
£ = vertical turbulent diffusivity or mixing coefficient, ft /sec,
v = flow through velocity of detention unit, ft/sec,
$, = travel length of detention unit, ft, and
other terms are defined previously.
Camp (12) solves for the functional form of equation 5-28 assuming a
uniform horizontal velocity distribution and constant diffusivity, £. A
form of the advective-diffusion equation then results in which local changes
in concentration at any vertical elevation are equal to the net effect of
settling from above and diffusion from below. The diffusivity will be
constant if the horizontal velocity is assumed to have a parabolic distribution,
(although this assumption is clearly at variance with the uniform velocity
distribution assumption above). For the parabolic distribution, & is then
found from
19
-------
e = 0.075 yVt /p (5-29)
o
where T = boundary shear stress, Ib/ft and
p = density of water = 1.94 slug/ft .
The term -Jt /p is known as the shear velocity, uy.., and can be evaluated
using Manning's equation for open channel flow (14),
(5-30)
u.,, = A/To/p /Q 1/6
where n = Manning's roughness coefficient.
The flow through ("horizontal") velocity, v , is also given by
v = £/td (5-31)
where £ = travel length of detention unit, ft, and
t = detention time, sec.
Equations 5-29 and 5-30 are then used to convert v y/2e to a more useable
form,
v y v v y176
a™ f\ 1 __~ — — g ,„ f C _ O O ^
— U.irt ~ -i i- — 7 \.3 3
-------
1.00
111
. .80
o
z
UJ
o
\L
u.
tu
0.
<
HE
ui
2
o
ki
CO
.60
0.01
0.10 l.O
TURBULENCE FACTOR
oc = v> y>/6 = _
10 2 €
Figure 5-5. Camp's Sediment Trap Efficiency Curves (12-15).
10.0
I I 1 I I I I I I
IDEAL QUIESCENT
CONDITIONS
TURBULENT FLOW
CONDITION (Ct'O.OI)
Et « l-txp(v§/v
I 1 I I I I I
.3 .4 .6 .8 1.0
C v. / vu )
2.0
4.0
Figure 5-6. Limiting Cases in Sediment Trap Efficiency (15).
21
-------
Quiescent conditions are assumed to exist for Of = 1.0 for which removal is
given by equation 5-26. Equations 5-26 and 5-33 are shown in Figure 5-6.
The parameter a may now be used as a weighting factor to obtain the overall
removal efficiency, E,
E = a EQ + (1-a) Et (5-35)
Thus, a linear interpolation is made between the curves of Figure 5-6
(equations 5-26 and 5-33). Within the program, values of the turbulence
factor are limited to 0.01 ^ a ^ 1-0. If a value computed from equation
5-32 is less than 0.01 and (more turbulent) it is set equal to 0.01 and
similarly for the quiescent boundary.
To summarize, the particle settling computations proceed as follows.
1. For each size and specific gravity range a settling velocity is
computed using equations 5-21 to 5-25. Then for each range all
steps below are performed.
2. The turbulence factor, a, is computed from equation 5-32.
3. EO is computed using equation 5-26.
4. E is computed using equation 5-33 or 5-34.
5. Finally, the removal efficiency for the particular particle size
and specific gravity range is computed from equation 5-35.
In a normal simulation, several plugs leave the detention unit in any
given time step. The effluent is all or part of^a number of plugs depending
on the required outflow as determined by the storage routing techniques dis-
cussed earlier. Thus, the effluent particle size distribution is a composite
of several plugs. This composite distribution is determined by taking a
weighted average (by pollutant weight in each plug) over the effluent plugs.
This distribution is then routed downstream for release or further treatment.
The particles that were removed from each plug are also composited and are
combined with the sludge volume.
Particle obstruction - The second major removal mechanism used when pol-
lutants are characterized by particle size distributions is the obstruction
or screening of particles. Removal efficiencies are a result of two
actions: the straining of the screens, and the additional filtration pro-
vided by the mat produced by the initial screening (17). Screens vary widely
in the size of the aperture and the manner in which the waste water flows
through them. To simplify the calculations, the removal of particles is
assumed to be a function of screen size only and the filtration by the mat
is ignored. In other words, a particular screen size will remove only those
particles larger than that size. This is not entirely accurate, of course,
but the result is a conservative removal estimate that may be accurate in
cases where backwashing is at a relatively high rate. In fact, a study by
Maher indicates that this simplifying assumption is reasonable (17). In this
case. 3 microstrainer with a Hard "0" screen (aperture of 23 microns) was
22
-------
installed in a residential area of Philadelphia, Pennsylvania. The analysis
of the backwash material for two storms (one in which a coagulant was used)
revealed that, by particle count, 88 to 96 percent of the particles were
indeed smaller than 23 microns. However, by weight, over 99 percent of the
material was found in particles greater than 23 microns. Although Maher did
not report the distribution in terms of weight, it was a simple matter to
convert by assuming a specific gravity.
During the simulation, a screen alters the particle size distribution
for a particular time step without detention time. Again, only the par-
ticles larger than the screen aperture are removed. If the screen size
falls between the high and low ends of any range, the pollutants are removed
by simple linear interpolation. For example, if 20 percent of the suspended
solids are found in the range from 10 to 50 microns and the screen aperture
is 20 microns, then 75 percent of the suspended solids in that range will be
removed or 15 percent of the total suspended solids load. Of course, if the
entire range is larger than the size of the aperture, then all pollutants in
that range are removed.
Comment on characterization by particle size distribution - Pollutants char-
acterized by a particle size distribution are restricted by the model to the
two removal mechanisms discussed above. This limits the user somewhat if
this characterization is chosen. The types of units that could be con-
sidered in this case would include sedimentation tanks, regular storage
basins, bar racks, fine screens, microscreens and swirl concentrators. How-
ever, these units probably represent the bulk of the processes applied to
the problem of combined sewer overflow and stormwater runoff. Thus, limits
of the applicability of the model using this mode are probably not too
severe.
Cost Calculations
Initial capital and operation and maintenance costs are calculated at
the end of a simulation. These costs are computed using only the information
processed for the simulation period. In other words, no attempt is made to
derive costs for particular time intervals (e.g., annual). It is left for
the user to interpret the results produced by the subroutine STCOST.
The capital cost for each unit is computed as a function of a design
flow or volume specified by the user or is calculated by the model as a
function of the maximum value recorded during the simulation.
C = a Qb (5-36)
cap max
or C = a (Q. )b (5-37)
cap in max
or C = a Vb (5-38)
cap max
or C = a (V )b (5-39)
cap obs max
23
-------
where C = initial capital cost, dollars,
cap
3
0 = maximum allowable inflow, ft /sec,
max
3
(0 ) = maximum inflow encountered during the simulation, ft /sec,
xin max
3
V = maximum allowable storage (detention units only), ft ,
max
3
(V , ) = maximum storage encountered during the simulation, ft , and
obs max °
a, b = coefficients (specified by the user).
Power functions are frequently used in wastewater treatment cost estimations.
Therefore, the above equations should be widely applicable.
Operation and maintenance costs are calculated as functions of the
variables listed above and the total operating time (calculated as the number
of time steps with inflow to the unit).
C = d Q f + gD (5-40)
om max op
or C d(Q. ) f + gD (5-41)
om in max op
or C = d V f + gD (5-42)
om max op
or C = d(V , ) f + gD (5-43)
om obs max op
where D = total operating time during the simulation period, hours, and
d,f,g = coefficients (supplied by the user).
The user is cautioned not to misinterpret the cost calculated by the
model. For example, in a single-event simulation the calculated capital
cost could only be considered an estimator of the true capital cost when
the event simulated is a design event. Likewise, when operating time is
a factor in computing operation and maintenance costs, the calculated
costs can be a valid estimator of the true costs only when a long term
simulation is performed. Recent EPA publications provide useful information
for the proper selection of the coefficients required in equations 5-36
through 5-43 (18, 19).
24
-------
SUMMARY
A new Storage/Treatment Block has been developed that is somewhat
different from its predecessor. The model requires greater user input and
knowledge of the processes being modeled. Storage/Treatment units may be
modeled as detention or non-detention units. Pollutants may be characterized
by their magnitude alone or by magnitude and their particle size/specific
gravity distribution. Any three of the pollutants available from other
blocks may be routed through the S/T Block. A simple cost routine is also
included.
In summary, the Storage/Treatment Block offers the user a flexible
tool for modeling wet- and dry-weather facilities and evaluating their
performance and costs.
25
-------
REFERENCES
1. Metcalf and Eddy, Inc., University of Florida, and Water Resources
Engineers, Inc., "Storm Water Management Model, Volume I - Final
Report", EPA Report No. 11024DOC07/71, (NTIS PB 203 289) Environmental
Protection Agency, Washington, DC, July 1971.
2. Viessman, Jr., W., Knapp, J.W., Lewis, G.L., and Harbaugh, I.E.,
Introduction To Hydrology, IEP, New York, Second Edition, 1977.
3. Medina, Jr. M.A., "Interaction of Urban Stormwater Runoff, Control
Measures and Receiving Water Response", Ph.D. Dissertation, University
of Florida, Gainesville, 1976.
4. Rich, L.G., Environmental Systems Engineering, McGraw-Hill, New York,
1973.
5. Huber, W.C., Heaney, J.P., Medina, M.A., Peltz, W.A., Sheikh, H. and
Smith, G.F., "Storm Water Management Model User's Manual, Version II",
EPA-670/2-75-017, Environmental Protection Agency, Cincinnati, Ohio,
December, 1974.
6. Lager, J.A., Smith, W.G., Lynard, W.G-, Finn, R.F. and Finnemore, E.J.,
"Urban Stormwater Management and Technology: Update and Users' Guide",
EPA-600/8-77-014, (NTIS PB 275 264), Environmental Protection Agency,
Cincinnati, Ohio, September, 1977.
7. Gupta, M.R., et al., "Handling and Disposal of Sludges from Combined Sewer
Overflow Treatment: Phase I - Characterization," EPA-600/2-77-053 a,
Environmental Protection Agency, Cincinnati, Ohio, 1977.
8. Huibregtse, K.R., "Handling and Disposal of Sludges From Combined Sewer
Overflow Treatment: Phase II - Impact Assessment," EPA-600/2-77-053 b,
Environmental Protection Agency, Cincinnati, Ohio, 1977.
9. Osantowski, R., et al., "Handling and Disposal of Sludges From Combined
Sewer Overflow Treatment, Phase III - Treatability Studies,"
EPA-600/2-77-053 c, Environmental Protection Agency, Cincinnati, Ohio,
1977.
10. Fair, M.F., Geyer, J.C. and Okun, D.A., Water and Wastewater
Engineering; John Wiley and Sons, Inc., New York, 1968.
11. Sonnen, M.B., "Subroutine for Settling Velocities of Spheres", Journal
of the Hydraulics Division, ASCE, Vol. 103, No. HY9, Sept. 1977,
pp. 1097-1101.
12. Camp, T.L.. "Sedimentation and the Design of Settling Tanks", Transactions
ASCE, Vol. Ill, 1945, pp. 895-958.
13. Dobbins, W.E., "Effect of Turbulence on Sedimentation", Transactions
ASCE, Vol. 109, 1944, pp. 629-678.
26
-------
14. Brown, C.B., "Sedimentation Engineering", Chapter XII in Engineering
Hydraulics, ed. by H. Rouse, John Wiley and Sons, New York, 1950.
15. Chen, C.N., "Design of Sediment Retention Basins", Proceedings of
the National Symposium on Urban Hydrology and Sediment Control,
University of Kentucky, Lexington, Kentucky, July 1975.
16. Hazen, A., "On Sedimentation", Transactions ASCE, Vol. 53, 1904,
pp. 45-71.
17. Maher, M.B., "Microstraining and Disinfection of Combined Sewer
Overflows - Phase III", EPA-670/2-74-049, (NTIS-PB 235 771)
Environmental Protection Agency, Cincinnati, Ohio, August 1974.
18. U.S. Environmental Protection Agency, "Areawide Assessment Procedures
Manual: Volume III", EPA-600/9-76-014, Environmental Protection Agency,
Cincinnati, Ohio, July 1976.
19. Benjes, H.H., "Cost Estimating Manual -- Combined Sewer Overflow Storage
and Treatment", EPA-600/2-76-286, Environmental Protection Agency,
Cincinnati, Ohio, December 1976.
27
-------
ON-SITE CONTROL OF NONPOINT SOURCE POLLUTION
Richard H. McCuen
Associate Professor
Department of Civil Engineering
University of Maryland
College Park, Maryland 20742
INTRODUCTION
It is widely recognized that nonpoint source (NPS) pollution is a major
problem in both urban and rural areas. Table 1 provides some typical levels
for eight pollutants from 3 land uses. Considerable effort has been expended
in developing stormwater management programs, with a majority of the resources
being spent on evaluating regional effects and control. Much of this effort
has involved model development for the purpose of predicting regional pollu-
tion levels as a function of land use change. However, little effort has
been expended for analyzing the potential of on-site control as a means of
reducing NPS loadings. Two methods that have significant potential for con-
trolling NPS pollution in urban areas are street sweeping and detention stor-
age.
THE USE OF STREET SWEEPING FOR
CONTROL OF NONPOINT SOURCE POLLUTION
The recent concern about nonpoint source pollution in runoff from urban
and suburban areas should focus attention on the effectiveness of street
sweeping as a means of controlling nonpoint source pollution. Unfortunately,
past investigations on the efficiency of street sweeping are few and primar-
ily concerned about control of either radioactive fallout or the dust and dirt
fraction of street wastes. And the variation in the results is important.
Studies report removal fractions from 0.1 to 0.95, with the wide variation
being due to factors such as differences in particle size distribution, sweep-
er and broom type, sweeper speed, and pavement type and condition.
Because of the importance of street sweeping in controlling nonpoint
source pollution, the modeling approach appears to provide a convenient means
of examining the efficiency of street sweeping in controlling nonpoint source
pollution. A model was developed (8) that included three components:
28
-------
1) algorithms describing the accumulation of pollutants on residential, com-
mercial, and industrial land uses; 2) a hydrologic component that describes
the removal of pollutants due to storm water runoff; and 3) a component to
estimate the removal of pollutants from streets by street sweepers. Accumu-
lation of pollutants are defined in the model to be a function of time between
storm events or street sweepings, land use, pavement type and condition, popu-
lation density, and traffic density. The removal of pollutants by surface
runoff is a function of the initial pollutant loading estimated from accumu-
lation component, the total volume of rainfall, the percentage of impervious
area in the watershed, the slope of the gutter or surface, and the overland
flow length. The effectiveness of street sweeping in removing pollutants is
developed as a function of the characteristics of the initial pollutant load-
ing, the forward speed of the street sweeper, the type of street sweeper, and
the pavement type. The model can simulate the following pollutants: total
solids, volatile solids, 5-day biochemical oxygen demand, chemical oxygen
demand, Kjeldahl nitrogen, nitrates, phosphates, and total heavy metals.
The model was designed as a planning model to be used for a variety of
common planning problems, such as the evaluation of alternative corrective
measures, evaluation of alternative growth plans, the evaluation of the effec-
tiveness of public works scheduling, and the evaluation of the cost effective-
ness of policy constraints (4,9).
To measure the sweeping effectiveness of a sweeper a concept of relative
effort RE has been defined as:
RE = 12_00_ft/min
forward speed of sweeper in ft/min (1)
Using data from in situ tests reported in the literature (7) equations that
relate the mass remaining to relative effort were calibrated from selected
particle size ranges, pavement type (concrete and asphalt), and sweeper type
(motorized and vacuumized).
To demonstrate the flexibility and validity of the model, field surveys
were conducted at four specific locations within Washington, D.C. The test
sites included an industrial site, a commercial area, multi-family housing,
and single family land use. Also, sites were selected so that the model was
tested under a variety of important site characteristics.
29
-------
Table 2 shows the percentage removal of eight pollutants for the test
site in single-family residential land use, with the results for sweeping
frequencies of 2, 4 and 7 days, forward speeds of 3 mph and 4 pmh, and
both vacuumized and motorized sweepers. Table 2 indicates that the fre-
quency of sweeping was the most important variable, with sweeper type and
forward speed being considerably less important. For lower speeds the
relative effort is greater and thus the removal percentages are slightly
higher. The highest removal efficiency was for total solids while phos-
phates had the lowest removal efficiency. The results suggest that sweep-
ers are not highly efficient in removing pollutants from urban streets.
This stems from the inability of the sweepers to remove the smaller parti-
cles, representing the greater bulk of the pollutants.
While removal percentages may be low, the alternative is to have the
pollutants washed off during storm events. Table 1 shows the magnitude of
the problem. Values such as these indicate that nonpoint source pollution
is a serious problem and failure to sweep will lead to either serious de-
gradation of receiving streams or increased demand for treatment of storm
water runoff.
THE USE OF ON-SITE DETENTION FOR
CONTROL OF NONPOINT SOURCE POLLUTION
It should be evident that street sweeping cannot serve as an indepen-
dent means of NPS pollution control. First, with average removal fractions
less than 0.5, there is still a significant amount of pollutants available
for removal by stormwater runoff. Second, while sweepers can be used in
highly developed urban areas, there still remains a considerable contribu-
ting area where sweepers either are not an economical means of pollution
control or cannot effectively service the source. Third, because of the
importance of sweeping frequency (Table 2), the limitation of sweeper avail-
ability, and the random nature of runoff generating storms, it is not al-
ways possible to sweep an area before the pollutants are removed by runoff.
Thus, sweeping must be considered as just one component of a water pollu-
tion control program.
Recognizing that stormwater detention is effective in controlling both
flow rates and sediment, it should also be of interest in examining the
30
-------
impact of detention basins for water pollution control. The effectiveness
of basins for pollution control will help determine the most cost-effective
policies for stormwater management. By controlling pollution loadings at
the source, it will reduce both water treatment costs and the degradation
of water quality in streams, rivers, and estuaries.
The study site is a 148-acre urban subwatershed of Cabin John Creek
in southcentral Montgomery County. The subwatershed includes all of the
virtually total impervious Montgomery Mall shopping center, several moder-
ate and high-rise apartment complexes, portions of a major highway and two
secondary roadways, and two townhouse developments still under construction
during the study. Some intervening green space remains among the several
developments. Drainage from these areas outfalls into Montgomery Mall
Lake at six different points. The major inflow point is at the head (east
end) of the lake, where a 66-inch diameter RCP drains the shopping mall
and 64 percent of the total area (95 acres) tributary to the lake.
The lake is about 750 feet long and 350 feet wide and has a 5.9 acre
permanent pool, which at normal level varies from 3 to 13 feet in depth,
allowing 37 acre-feet of dead storage capacity. Primary outflow is con-
trolled by a 24-inch diameter CMP riser, topped with a 36-inch diameter
circumferential hood, and connected to an 18-inch diameter horizontal CMP
barrel. A 40-foot wide grassed overflow channel has a crest at an elevation
3 feet higher than the riser crest.
In addition to precipitation and flow rates both into and from Mont-
gomery Mall Lake, the eleven water quality parameters were sampled during
several storm events. Table 3 provides a summary of the characteristics
of the water quality parameters.
The efficiency of a stormwater management basin for the control of
pollutants can be measured using the trap efficiency TE:
TE = 1 - V /V.
where V and V- are the total volumes of a pollutant in the outflow and
inflow, respectively. Table 3 provides a summary of trap efficiencies
31
-------
computed from median values of pollution levels in the observed data from
the study site. While these values are based on median values, the values
suggest that trap efficiencies can be quite high if the basin is designed
for both flow and pollution control. Design parameters for pollution con-
trol are somewhat different than for flow control (3). The high values
shown in Table 3 result partially from the large storage volume provided
and the low return periods of the storms that produced the runoff.
McCuen (6) provided relationships between the inflow and outflow rates of
a detention structure and the levels of the pollutants in the inflow and
outflow. The trap efficiency is easily computed from these estimates.
CONCLUSIONS
In a recent statement concerning policy recommendations in water re-
sources, Ackermann, et al. (1), concluded,
"In order to mitigate the effects of pollution, the following
policies and procedures are recommended:
1. An intensive assessment of nonpoint sources of pol-
lution should be obtained and sound and implementable drain-
age and land use management practices should be developed
that will control them before they enter our surface water
and ground water."
Davis, et al. (3 ), discussed one such intensive effort into assessing
nonpoint source pollution, including a comprehensive data collection pro-
gram. The study by Davis, et al., was one of the few studies that con-
sidered NPS pollution at the source in the upland areas of a watershed.
While considerable effort has been expended on estimating NPS pollu-
tion at the regional level, comparative little effort has been expended
into NPS control on the local level. The results of this study show that
stormwater management can be an effective means for controlling NPS pollu-
tion. This is especially significant because many state and local govern-
ments require detention storage for the control of both peak dischange and
sediment in urbanizing areas. However, the design of detention basins
should be different for flow control than it would be for pollution control,
and thus, research on the multi-purpose design of detention facilities for
32
-------
both flow and quality control is needed. An advantage of controlling NFS
at the local level is that water quality in both rivers and small streams
should improve, whereas regional control measures may not result in the
improvement of the water quality of small streams.
33
-------
ACKNOWLEDGEMENT
The research was supported by funds provided by the U. S- Department of
the Interior, Office of Water Research and Technology, as authorized under the
Water Resources Act of 1964. The project was administered by the Water Resources
Research Institute, University of Maryland.
34
-------
REFERENCES
1. Ackermann, W.C., D.J. Allee, J. Amorocho, Y.Y. Haimes, W.A. Hall, R.A.
Meserve, R. Patrick, and P.M. Smith, Scientific and Technological Consid-
erations in Water Resources Policy, EOS, Transactions, American Geophysi-
cal Union, Vol. 59, No. 6, pp. 516-527, June 1978.
2. Curtis, B.C. and R.H. McCuen, Design Efficiency of Detention Basins, ASCE,
J. Water Resources Planning and Management, Vol. 103, No. WR1, pp. 125-140,
May, 1977.
3. Davis, W.J., R.H. McCuen, and G.E. Kameldulski, The Effect of Storm Water
Detention on Water Quality, International Symposium on Urban Hydrology,
Hydraulics, and Sediment Control, University of Kentucky, Lexington,
July, 1978.
4 Sutherland, R.C. and R.H. McCuen, Simulation of Urban Nonpoint Source Pol-
lution, Water Resources Bulletin. Vol. 14, No. 2, pp. 409-428, 1978.
5 McCuen, R.H., Stormwater Management Policy and Design, Journal of Civil
Engineering Design, Vol. 1, No. 1, 1979 (in press).
6. McCuen, R.H., Water Quality Trap Efficiency of Stormwater Management Ba-
sins, Technical Report, Department of Civil Engineering, University of
Maryland, 1978.
7. Sartor, J. D. and G. B, Boyd, Water Pollution Aspects of Street Surface
Contaminants, U.S. Environmental Protection Agency,, EPA-P2-72-081, Nov.,
1972.
8. Sutherland, R. C., A Mathematical Model for Estimating Pollution Loadings
and Removals from Urban Streets, M.S. Thesis, Department of Civil Engineering,
University of Maryland, January, 1975.
9. McCuen, R. H., R. L. Powell, and R. C. Sutherland, The Relative Importance of
Factors Influencing Loadings in Runoff from Urban Streets, Section 10 of
Utility of Urban Runoff Modeling, M.B. McPherson (ed.), ASCE Technical No. 31,
ASCE Urban Water Resources Research Program, July, 1976.
35
-------
Table 1. Mean Loadings of Selected Nonpoint Source Pollutants'
Pollutant (Ib/curb mile)
Total Solids
Volitile Solids
BOD5
COD
Kjeldahl Nitrogen
Nitrates
Phosphates
Total Heavy Metals
Residential
1200
86
11
25
2.0
0.06
1.1
0,58
Industrial
2800
150
21
100
3.9
0.18
3.4
0.76
Commercial
360
28
3
7
0.4
0.18
0.3
0.18
* Abstracted from Ref. 1
36
-------
Table 2. Percentage Removal of Pollutants: Single-Family Residential Land Use
Type of
Sweeper
Vacuumized
Pollutant
Total Solids
Volatile Solids
BOD
COD
Kjeldahl Nitrogen
Nitrates
Phosphates
Total Heavy Metals
Frequency
of Sweeping
Foward Speed of Street Sweeper
3 mph 4 mph
2
Days
70
55
53
38
55
42
25
62
4
Days
53
40
38
22
39
27
11
46
7
Days
35
26
25
13
25
16
6
31
2
Days
68
53
51
34
53
38
20
60
4
Days
52
39
37
20
38
25
9
45
7
Days
35
26
25
12
25
16
5
30
Motorized
U)
Total Solids
Volatile Solids
BOD
COD
Kjeldahl Nitrogen
Nitrates
Phosphates
Total Heavy Metals
69
55
53
39
55
42
27
62
52
39
37
22
59
27
12
46
34
26
25
13
25
16
6
30
67
E2
50
35
52
39
22
59
51
38
26
21
37
25
10
44
34
25
24
12
24
16
5
29
-------
TABLE 3. Water Quality Trap Efficiencies
Parameter
Units
Median
Inflow Rate
Median
Outflow Rate
Trap
Efficiency
UJ
00
5-day BOD
20-day BOD
Chemical Oxygen Demand
Zinc
Cadmium
Orthophosphat es
Total Organic Carbon
Total Phosphorus
Ammonia
Lead
Iron
10'hb/sec
10"hb/sec
lO'nb/sec
lO'nb/sec
10'hb/sec
10",lb/sec
10~::ib/sec
10~;:ib/sec
'
10"b/sec
10 Ib/sec
11.35
19.7
60.5
0.463
0.36
0.308
13.25
0.73
6.8
0.43
10.3
0.37
1.35
2.10
0.0026
0.0063
0.022
0.48
0.007
0.025
0.018
0.373
.88
.93
.97
.99
.98
.93
.96
.99
.99
.96
.96
-------
STORMWATER MODELLING APPLICATIONS1'
by Alan R. Perks2*
INTRODUCTION
In this brief paper, I would like to describe the background and framework
within which much of our stormwater modelling is carried out in Ontario,
and then review several different modelling applications recently carried
out by Proctor & Redfern Limited.
Under the Canada-Ontario agreement for Great Lakes Water Quality, the Urban
Drainage Subcommittee^ ' was constituted to promote and oversee research
into the problems of urban drainage. This committee adopted 3 basic
objectives:
1. to define the problems
2. to develop the capabilities for solution
3. to develop a strategy for implementing solutions
Examples of the projects carried out under each of these objectives were
(3)
monitoring studies; ; Storm Water Management Model development, ' and more
recently preparation of provincial policies for urban drainage. ' The
adoption of these policies in the near future will signify the end of the
subcommittee's program.
At present, however, the regulatory responsibility for urban drainage
remains somewhat divided between different levels of government and different
agencies. At the provincial level, the Ministry of the Environment is
primarily concerned about water quality, whereas the Ministry of Natural
Resources deals mainly with water quantity management. The design and
construction of storm sewer systems and small channels is controlled
primarily at the municipal level. There has fortunately been a vigorous
interest in stormwater management on the part of all 3 agencies. In fact,
several Ontario Municipalities have been among the first in Canada to adopt
1. A paper presented to the 3rd Canadian-USA SWMM User's Group Meeting,
Annapolis, Md., November, 1978.
2. Head of the Hydrotechnical Group, Proctor & Redfern Ltd. Toronto, Ontario,
39
-------
new drainage criteria incorporating stormwater management principles and
modelling concepts.
APPLICATIONS
Urban Stormwater models have been used extensively by Proctor & Redfern in
such cities as Toronto, Hamilton, Mississauga, St. Catharines, London and
Sault Ste. Marie. Generally, these applications related to the analysis
of existing system deficiencies, the design of expansions to existing
systems, the design of stormwater management systems for entirely new develop-
ments, or the study of entire small watersheds for drainage planning pur-
poses. Our approach has been to utilize a set, or hierarchy of models
ranging from simple to quite complex, and including STORM; ' HYMO, '
ILLUDAS;7' and SWMM^8'. Such a package of models affords the user the
capability to analyze most urban drainage problems. Fig. 1 shows the
typical uses to which these models may be put.
The following section briefly outlines several of our recent applications
which are indicative of the range of problems encountered.
West St. Catharines
West St. Catharines is served by several small combined sewer systems which
have a history of frequent flooding. An investigation of one of these
areas was launched to develop a flooding-relief program.
The SWMM was applied to this 175 acre area and was calibrated to observed
flooding instances and levels. An improved version of the EXTRAN routine
was applied to simulate surcharging. The results for both the recorded
storm and the 5 year design storm indicated a general inadequacy of all
pipes in the system and extensive flooding.
Because both trunks and lateral sewers were responsible for flooding,
alternatives for providing relief were limited to complete separation or
a program involving trunk replacement combined with several stormwater
management measures.
40
-------
This latter alternative appeared most promising and was investigated more
fully. Cost savings could be achieved by utilizing the existing lateral
sewers and replacing the trunk only. Detailed simulations of particular
blocks were carried out to determine surface hydrographs, the effects of
roof leader disconnection and inlet controls at catchbasins, and the
depth of gutter flows along the street. The results indicated that a
program of encouraging surface flows could reduce the need to replace
existing lateral sewers without causing excessive inconvenience. The
new trunk sewer would have adequate capacity to accept both lateral sewer
and surface gutter inflows. Fig. 2 shows the project area and some perti-
nent results.
The capability of the SWMM to model surface flows, pipe flows, and surcharge
conditions was indispensable to the analysis of this system.
The Idlewood Subdivision is a 185 acre development in Kitchener, Ontario.
Located in a small natural watershed, there was concern over the down-
stream impacts of this development. Stormwater management measures were
required to provide this watershed protection.
The subdivision featured an extensive recreational greenway system which
was also planned to serve as a stormwater facility. Street and lot runoff
would be directed to the greenways where it would be conducted to detention
areas via grassed swales. The detention areas would ultimately discharge
slowly to the creek.
The ILLUDAS model was used to develop design runoff hydrographs and to deter-
mine optimum sizes and locations for detention. It was demonstrated that
the 5 year post-development flow would be approximately the same as under
natural conditions. The ability of the model to handle pipes* channels,
and detention storage was the main reason for its use in this case. Fig. 3
shows the drainage features of this project.
41
-------
Kldd's Creek
Kidd's Creek Is a small partly-urban watershed located in Barrie, Ontario.
The downstream section is in an older residential area and is quite con-
gested for drainage. Flooding occurs here frequently. The upper two-
thirds of the watershed is undergoing rapid development, and concern has
been expressed about the effects of this development on downstream flooding.
Fig. 4 shows a schematic of the Kidd's Creek watershed.
HYMO was used to determine watershed flows at various points, and these
were then used to establish inadequacies in the downstream culverts and
channels. Floodlines were determined by water surface profile calcula-
tions. On this basis, several programs of remedial works were developed
for different design frequencies.
It appeared, however, that stormwater management practices in the upstream
developments would reduce the need for expensive downstream improvements.
Further simulations of drainage conditions and storages confirmed the
feasibility of this approach, and the final plan recommended rigorous
stormwater management coupled with minimum downstream improvements to
satisfy existing flows.
Further detailed SWMM and ILLUDAS modelling will be required at the
design stage to establish effective stormwater management schemes for
each area.
CONCLUSIONS
Modelling techniques aid the understanding and development of complex
stormwater management schemes, and provide the designer greater capacity
to evaluate alternatives. Our firm has found them useful in a great
variety of projects, typified by the examples cited above.
One of the more interesting areas of modelling applications is in the
implementation of stormwater management concepts in new developments.
In Southern Ontario, a great deal of development is planned over the
next decade which will significantly influence our urban watersheds.
42
-------
The competing demands for environmental protection and maximum land usage
and yields creates conditions where a great number of alternatives must
be thoroughly analyzed before a master plan can be adopted.
43
-------
REFERENCES
(1) Report of the Urban Drainage Subcommittee
Ontario Ministry of the Environment
Toronto 1976
(2) Stormwater Management Model Verification Study
M.M. Dillon Ltd.
Prepared for Environment Canada
Burlington 1976
(3) Stormwater Management Model Study, Volumes I - IV
Proctor & Redfern Limited
Prepared for Environment Canada
Ottawa 1976
(4) Urban Drainage Policy Document (3rd draft - unpublished)
Ontario Ministry of the Environment
Toronto 1978
(5) "STORM" - Storage, Treatment, Overflow, Runoff Model
The Hydro!ogic Engineering Centre
Corps of Engineers, U.S. Army
609 Second Street
Davis, California 95616
(6) "HYMO" - Problem Oriented Computer Language for Hydrologic Monitoring
Agricultural Research Service
U.S. Department of Agriculture
May 1973
(7) "ILLUDAS" - The Illinois Urban Drainage Area Simulator
Michael L. Terstriep and John B. Stall
Illinois State Water Survey, Urbana
Bulletin 58, 1974
(8) "SWMM" - Storm Water Management Model -Vol. I Final Report for the
Environmental Protection Agency
Water Pollution Control Research Series
by - Metcalf & Eddy, Inc., Palo Alto, California
University of Florida, Gainsville, Florida
Water Resources Engineers, Inc., Walnut Creek, California
(9) "The Development of Storm Modelling in Canada"
Alan R. Perks, P.Eng.
The Proctor & Redfern Group, Toronto
A paper presented to the International Conference on Urban Storm Drainage,
University of Southampton, England, April 1978
44
-------
FIGURES
Figure 1 - Use of Different Runoff Models
Figure 2 - West St. Catharines Project Area
Figure 3 - Idlewood Subdivision Drainage Features
Figure 4 - Schematic of Kidd's Creek Watershed
45
-------
METHOD
PLANNING
DESIGN
DEFINE
CRITICAL
EVENTS
ASSESS
HYOHOLOGIC
CHANGES
ASSESS
STORAGE
EFFECTIVENESS
ASSESS
REGIONAL
IMPACTS
SIMPLE
COMPLEX
RATIONAL
STORM
a\
HYMO
ILLUDAS
SWMM
(9)
FIGURE 1 - USE OF DIFFERENT RUNOFF METHODS
-------
Area 29
USE NO
HIIINIIIIINflH _ end STAGE
HZ") — DIAMETER FOR NEW TRUNK SEKER
— SURFACE FLOW
g,5 — RESTRICTED PIPE FLOW
(approKlmalily equal to capacity)
ALTERNATIVE RELIEF 3 - Temporary Street Storage
- Main Trunk Upgraded for 5 Year Level
- 60% Roof Disconnection
FIGURE 2
WEST ST. CATHERINES
PROJECT AREA
-------
Figure 3
, , •. i _iJ_J ! ! I I i f M i ! [ I ' | MM
V-fc.«—1 *-'-"—J fe.—.A.A.t^i:^ v^i.-iii^tJ^™. U,.Li
ed Storage Area
Figure 3 - fdlewood Subdivision Drainage Features
48
-------
SUBWATERSHED
BOUNDARY
? KEMPENFELDT
ff BAY
FIGURE 4
Schematic Of Kidd's Creek Watershed
49
-------
'MODEL FOR SELECTION OF STORMWATER CONTROL ALTERNATIVES" *
by
R. Bedrosyan and J. Ganczarcyk
Department of Civil Engineering, University of Toronto
Toronto, Canada
* Paper presented at SWMM Users Group Meeting,
Annapolis. Maryland, November, 1978
50
-------
"MODEL FOR SELECTION OF STORMMATER CONTROL ALTERNATIVES"*
by
I n
R. Bedrosyan and J. Ganczarczyk^
Department of Civil Engineering, University of Toronto
Toronto, Canada
ABSTRACT
In urban centres, the degree of pollution from combined sewer over-
flows and stormwater runoff is becoming ever more noticeable and critical.
Among the many stormwater management alternatives for combined sewer over-
flow abatement, remote monitoring/real-time control is emerging as a cost-
effective method in large urban centres. Its advantage is that it does
not require additional facilities, but uses the existing system more
effectively.
A general computer model is needed in the planning stage, which can
estimate the overall performance of real-time control and other stormwater
management alternatives, without employing expensive, single-event models
like WREM. STORM, the most well-established initial planning model is
unable to provide an overall system description. Therefore, a new pro-
gram, named RAFFI, was devised which can ba combined with STORM. RAFFI
is a simple model which provides hourly information on flows and pollut-
ants (BOD and SS) diverted into receiving waters from various overflow
1. Design Engineer, Department of Public Works, Borough of North York,
Willowdale, Ontario, Canada.
2. Professor of Civil Engineering, University of Toronto, Toronto, Canada.
51
-------
points, and treatment and by-pass operations at the plant.
The model can simulate the overall behaviour of existing and pro-
posed systems in multi-basin urban areas for single storms, or for many
storms over a long period, such as one year. In-system storage can.be
simulated by assuming that the interceptor acts as a common reservoir
which can store flows from all watersheds, while each trunk can store
flows only from the watershed that it serves. Dynamic regulators can be
simulated according to a built-in control strategy, which changes the flow
capacity of diversion pipes and outfalls, based on plant, trunk and inter-
ceptor flow and storage values at each hour. Capacities for primary,
secondary or by-pass operations at the plant can be varied during storms.
Treatment of overflows at each outfall can also be simulated. The model
is equally applicable to separate storm sewer systems.
The objective was not to produce yet another new model, but to inte-
grate program RAFFI with STORM, in order to increase its capabilities in
the initial planning stage. By using STORM-RAFFI, a decision-maker can
estimate the overall performance of an existing system during wet weather
(i.e. quantity and quality of discharges from multiple overflow points,
effluents from primary or secondary treatment, or by-passes at the plant),
to compare the mass pollution loading from overflows with that from the
plant, and to predict the long-term effect of alternatives such as real-
time control with in-system storage, or overflow treatment. Selection of
alternatives can be based on frequency, duration, volume and pollution of
overflows, and on the total pollution (from all plant effluents plus over-
flows) reaching receiving waters during storms, all of which can be out-
put from model RAFFI.
52
-------
After the model was built and tested successfully on a hypothetical
combined sewered area, it was applied to data from the City of Winnipeg,
Manitoba,which experiences combined sewer overflow and plant by-pass pro-
blems. Results for a six-month simulation period indicated that the ex-
pansion underway at the plant will not reduce overflow pollution, even
though it may eliminate by-passes.
Among the various abatement alternatives which were analyzed, real-
time control with in-system storage distributed in the interceptor and
trunk sewers produced the best results. Reductions were 20% for overflow
frequency, 30% for overflow duration, 22% for overflow volume and 40-50%
for overflow BOD and SS discharge, compared with the no-control system
behaviour. Microstraining treatment of overflows at each point produced
a 10-40% improvement in overflow quality but no change in overflow volume,
frequency and duration. In comparison, all of the real-time control
strategies, such as in-system storage in trunks only, or in line storage
in the interceptor with holding reservoirs along its length, gave 20-50%
improvements.
53
-------
INTRODUCTION
Combined sewer overflows, stormwater runoff and treatment plant by-
passes during storms are serious pollution sources in large urban centres.
Until recently, the effects of runoff quantity on communities downstream
(flooding due to increased flow), and of runoff quality on receiving water
bodies (degradation due to shock loads) were underestimated, ignored or
simply could not be determined. The prime objective of stormwater manage-
ment was to collect runoff from the streets as fast as possible and dis-
charge it to receiving waters. As a result of economical and environment-
al constraints, and increased monitoring of wet weather flow (WWF), the
situation has changed. On the other hand, advances in treatment plant
operations have improved the effluent quality significantly, thereby in-
creasing the percentage contribution of combined sewer overflow/stormwater
runoff in the overall pollution picture. Therefore, research has been
directed towards better stormwater management. This paper is an attempt
to provide a simple tool which may become quite useful in the planning
phase of a stormwater management project.
PROBLEM
The urban water resources system is illustrated in Figure 1. The
ultimate objective is to minimize the pollution discharge from paths 1 to
6 such that the downstream water quality will not be worse than the up-
stream quality. This paper deals with paths 3, 4, and 5 and provides a
tool to investigate pollution abatement alternatives for them.
Combined sewer systems belong to the past; however, their pollution
problems exist today and may even increase in the future. Increased dry
54
-------
weather flow (DWF), due to higher populations, will leave less capacity
for additional flow to be diverted into treatment during storms; in par-
allel, increased imperviousness, due to urbanization, will create higher
runoff, the net result being more frequent, longer and more polluted
overflows.
Combined sewer overflows are more polluted in flat cities because
low DWF velocities result in accumulation of pollutants in the sewer, only
to be flushed during wet weather flow(WWF). Combined sewer overflwos,
during the first hours in particular, create a shock load on receiving
waters because they carry most of the accumulated pollutants picked up by
precipitation from the air, by runoff from the ground surface, catchbasins
and from within the sewer system, in addition to pollutants from mixing
with municipal sewage. Usually, a runoff or overflow pollutograph reaches
the peak much earlier than a runoff hydrograph causing high pollutant con-
centrations to be discharged at the start of a storm. Unlike the treat-
ment plant effluent which is stable and has undergone biodegradation for
several hours, the combined sewer overflow contains freshly produced poll-
utants and thus, has a high deoxygenation rate and can be easily degraded
by all types of microorganisms present in receiving waters. Therefore, in
the short term, an overflow with 30 mg/1 BOD may have more detrimental
effects than a plant effluent with 30 mg/1 BOD, even though both will
exert 30 mg/1 BOD after 5 days. Various studies indicate that the annual
sanitary sewage overflowed in combined sewers ranges from 2 to 5% (Field
et al, 1973, Marsalek, 1972). Although this is a small volume, the
pollutant contribution of this percentage is high. In an area with a
secondary treatment plant, about 50% of SS and 30% of BOD reaching
55
-------
receiving waters annually come from overflows. Rosenkranz (1970) notes
that 5% of the annual raw sewage load will become 33% of the effluent
load, or 25% of the total load, if 85% reduction is achieved at the treat-
ment plant. As the treatment methods improve and removal efficiencies
increase, the pollution contribution of combined sewer overflows will be-
come more critical.
SOLUTIONS
Stormwater management alternatives can be divided into four catego-
ries as shown in Table 1. Since references (Field et al, 1972, Lager et
al, 1974, Marsalek, 1972) deal with all of them quite extensively, only
the remote monitoring/real-time control method will be briefly explained.
Remote monitoring/real-time control is a collection system control
that can be implemented in large cities with combined sewer systems. It
has the objective of assisting a dispatcher (a human supervisor or control
computer) in routing and storing combined sewer flows and making the
most effective use of interceptor, trunk sewer and treatment plant capa-
cities. This operation is controlled by predicting and monitoring pre-
cipitation, runoff, sewer flows and receiving water quality, and by accor-
dingly regulating and distributing the WWF in the system before allowing
any overflows. Those portions of the sewer system capacity not used for
flow transmission are used for flow storage. Flows are stored in the
pipes and distributed from high runoff to low runoff areas by means of
dynamic regulators, gates and cross-connections. Backwatering and sur-
charging are allowed. Treatment plant throughput is regulated to maxi-
mize the amount of WWF treated. Overflows are allowed when all of the
56
-------
in-system storage is utilized and only from the least critical points.
Information obtained from telemetered rain gauges, flow and depth sensors,
quality monitors at outfalls, treatment plant and receiving waters are
processed by a central computer, which calculate and implements the con-
trol operations, either automatically, based on pre-programmed strategies,
or by the decisions of a human supervisor. Figure 2 shows a typical real-
time control system.
Minneapolis-St. Pual, Minnesota (Anderson, 1970), Detroit, Michigan
(Detroit Metro Water Department, 1970), and Seattle, Washington (Leiser,
1974) have applied this extensive but cost-effective concept and achieved
substantial success in pollution reduction. Based on pre- and post-pro-
ject conditions, average reductions are 35-70% in overflow volume, 68% in
pollution, 68% in frequency and 88% in duration. Reported costs are $3-8
million capital and $200,000 - $270,000/year operational (Anderson, 1970,
Leiser, 1974, Lager et al, 1974). This alternative is feasible and read-
ily applicable in most large urban centres with combined sewer systems,
provided that static regulators be replaced with dynamic ones.
TOOLS FOR SELECTING SOLUTION
Since substantial amounts of time and money are involved in each
management alternative, thorough studies are required to evaluate each
one. Computer models are tools needed by the decision-maker, designer
and the operator of stormwater management facilities for selecting, an-
alyzing, designing and implementing the most cost-effective method.
According to their objectives, the models can be classified into
three groups:
57
-------
1. Initial planning, 2. Detailed analysis/design, 3. Operatic.].
Data requirements, computer costs, level of sophistication and De-
gree of accuracy increase, while output information, numU-r of alternat-
ives, length of study period and size of study area decrease from initLJ
planning to detailed analysis and design models. In general, critical
precipitation/runoff events can be identified and long-term effects cai,
be estimated by the initial planning models by continuous simulations
while the single-event analysis/design models evaluate the short-term
performance of existing or proposed control systems. Extensive inform-
ation about current models and application results is provided by Brand-
stetter (1975), MacLaren and Procter Redfern (1976), Water Resources
Engineers et al (1974).
NEED FOR A NEW TOOL
At present, there is no initial planning model to study stormwater
management alternatives such as real-time control or stormwater overflow
treatment, although models do exist to analyze and design them in the
latter part of a management project. At the start of a project, it is a
waste of time and money to use a highly sophisticated model which requires
much input data and outputs much detailed and unneeded information about
each alternative. In addition, models in the design/analysis class are
single-event simulation models and therefore, cannot estimate the overall,,
long-term behaviour of a control alternative (or of the existing system),
which is essential for decision-making and economic analysis.
Real-time control is becoming an attractive abatement alternative in
large combined sewered cities. After some success in a few American
58
-------
cities, some Canadian municipalities, Montreal, Toronto and Edmonton have
also begun to consider this concept. The Metro Toronto sewer system is
similar to that of Minneapolis-St. Paul, being equipped with dynamic gates
and water level sensory which are adaptable to real-time control operations. Al
though the hardware exists, the software to estimate and analyze effective
ways of real-time control do not exist yet. Decision-makers will require
some general answers to the following questions before they consider any
further action in this direction: What will happen with real-time con-
trol? Is it worth the effort, time and money? How much reduction in
overflow frequency, volume and pollution will be achieved with real-time
control? Will overflow treatment be any better? etc.
An initial planning study with the model proposed in this paper
could answer some of these questions. It would, at least, provide a
'feeling" of the system and possible solutions. This type of study would
not benefit from sophisticated models like Water Resources Engineers
Model (WREM), because of high cost, high degree of detail in input data,
unneeded and detailed information in output and shortness of test period
(single events).
Storm Treatment Overflow Runoff Model (STORM), the most established
initial planning model (U.S. Army Corps, 1975), is not suitable in this
.vrr ?i f-pcause it can only analyze quantity and quality of one outfall from
>,>:,!- v/ittTshed at a time, whereas a total system contains many watersheds
with many outfalls and, in case of a combined system, a treatment plant
whose performance changes during storms. Futhermore, STORM cannot si-
mulate variable treatment and variable storage during storms, which are
reqired operations for a real-time control alternative.
59
-------
OBJECTIVES OF RAFFI
In order to overcome the deficiencies of STORM, a new model named
RAFFI was built. Its objectives are:
1. To investigate the overall effects of a storm, or of many storms in
a given period, in a multi basin urban area; more precisely, to
provide hourly information on flows and pollutants (BOD and SS)
reaching receiving waters from multiple overflow points, treatment
plant secondary and primary effluents, or by-passes; to analyze per-
centage breakdown of all WWF quantity and quality, as they reach re-
ceiving waters from various components of the system.
2. To analyze any improvements due to real-time control in combined
sewers - by variable in-system storage in interceptor and trunks,
and variable treatment at the treatment plant.
3. To analyze any improvements due to combined sewer overflow or sep-
arate stormwater treatment at outfalls before discharging to receiv-
ing waters.
4. To compare pollution discharged from overflows/stormwater runoff,
with that from treatment plant effluents or by-passes.
5. Along with another program built at the University of Toronto, to
increase capabilities of STORM to analyze most of the available
stormwater management alternatives.
In addition to the above, the model computes information on the num-
ber of events contained in the system without overflowing, and the aver-
age duration of overflows at each point. The output can be (i) a summary
for all events in a continuous simulation; (ii) a summary for a single-
60
-------
event, and (iii) a detailed hourly analysis of an event for overflow hy-
drograph and pollutograph construction. It contains about 500 statements
and needs an average of 5 seconds to analyze a system with 12 watersheds
for a six-month continuous simulation. The model assumptions, flowchart,
program listing, applications and results are described in detail else-
where (Bedrosyan, 1976).
Figure 3 shows an overview of model RAFFI. It can analyze up to 20
watersheds simultaneously, which may have separate storm sewers, or may
be connected to a treatment plant with a combined sewer system. Input to
the model is in the form of hourly runoff hydrographs and BOD and SS
pollutographs for each watershed. STORM output can be used as RAFFI in-
put, although the input data may be supplied manually or from any other
model that can generate runoff from rainfall. Computations are in hourly
time-steps; units are acre-inch/hr for flow routing, Ib/hr for pollutant
routing and acre-inch for storage routing. The intention of this work is not
i
to produce yet another new model but to integrate it with STORM, in order
to increase its capabilities in the initial planning stage.
ASSUMPTIONS
The important assumptions used in formulating the model are briefly
discussed below:
1. Each watershed is served by one major trunk sewer. In combined
sewer systems, a diversion pipe connects the trunk to the interceptor,
which collects flows up to a low multiple of DWF from each watershed and
transports them to the treatment plant, Flows in excess of the diversion
pipe or regulator capacity are overflowed from each trunk, DWF from each
61
-------
watershed can be input as average flow in cfs and BOD and SS concentra-
tions in mg/1.
2. Each trunk-interceptor connection is equipped with a dynamic re-
gulator, which can change the diverted flow, according to plant treat-
ment rates or available storage space in the interceptor. The range of
regulation is from zero (regulator gate closed) to diversion pipe cap-
acity (gate fully open). Conventional combined systems are simulated by
assuming inactive regulators, set at a fixed diversion capacity above
which all excess flows in a trunk (i.e. flows above 2.5 x DWF) are over-
flowed.
3. In-system storage can be simulated in any trunk or interceptor
by assuming a pipe to act like a reservoir. The lower limit of storage
is zero (i.e. if the pipe is flowing full) while the upper limit is input
by the user. It can be computed as follows:
Maximum storage available = (free cross section) x (safe length),
where (free cross section) = cross section of pipe less area used by flow,
(safe length) = a distance upstream beyond which basement flooding
or high head may be critical.
4. The treatment plant is simulated by two variable treatment
operations with two outflows. The outflows may be chosen by the user,
such as (i) secondary effluent and (ii) primary effluent, or (i) treated
effluent, and (ii) untreated by-pass, etc. The treatment capacities of
the two operations are input in multiples of average total DWF and per-
cent BOD and SS removal efficiencies. It is assumed that an increase in
flow during storms will not affect the performance of a large plant, if
62
-------
this increase lies within certain hydraulic and/or organic limits and
is not prolonged. Variations in inflow are much more critical for a
small plant than for a large one; however, most of the combined sewered
areas are old and highly populated urban centres, served by large treat-
ment plants with average DWF's greater than 50 Imperial million gallons
per day (Imgd).
The secondary (biological) treatment process, designed for average
DWF, is more susceptible to inflow variations than the primary operations.
Based on plant performance data (Heinke, 1973, Lin, 1974, Salib, 1976),
it is generalized that, for a plant larger than 50 Imgd DWF, flow in-
creases up to 1.5 - 2.0 x DWF will not affect the biomass activity,if
this increase does not last longer than 6-15 hours. The momentum of the
process is stable enough to allow increases within these limits. BOD and
SS removal efficiencies may be in the range of 70 - 95%.
The primary (physical-chemical) treatment operations, designed for
maximum day DWF (usually 2-3 x average DWF),are readily available for
WWF treatment if DWF is at its average during the storm. Additionally,
most settling tanks have allowances for up to 1.5 - 2.0 x the design sur-
face loading rates (Heinke, 1973, Salib, 1976). Removal efficiencies will
decrease slowly with increasing surface loading rates until a certain
limit is reached, beyond which there will be a sharp decline in efficien-
cies and a wash out of pollutants (Heinke, 1973). It is generalized that
flow increases up to 2.0 - 4.0 x DWF will not affect primary operations
and that BOD and SS removal efficiencies may be 30 and 50%, respectively.
In the developed model, the capacity limits and removal efficiencies are
not defaulted but left to the user's choice, for greater flexibility.
63
-------
5. Overflow treatment at each outfall can be simulated and
different treatment methods for different overflow points can be studied.
Physical or physical-chemical operations such as microstraining, filtrat-
ion, screening/dissolved-air flotation can be simulated. Most models use
steady-state operations to simulate stormwater treatment, even though
stormwater is a dynamic andhighly variable phenomenon; these formulae give
results with a wide range of fluctuations (Lager, et al, 1974, MacLaren-
Proctor Redfern, 1976, Wisner, 1976) yet the mean values are calibrated
to match actual results. Since equations are not any more accurate, and
since the objective in initial planning is to estimate the long-term be-
haviour of a treatment alternative, the only input data needed is percent
BOD and SS average removal efficiencies, and constraints of minimum feas-
ible effluent concentrations, which can be directly obtained from some
pilot plant studies (Glover et al, 1973, Gupta et al, 1973, Harvey et al,
1973, Lager et al, 1974, Lee et al, 1974, White et al, 1973). There is
a need for specifying minimum effluent concentrations during simulation
because stormwater runoff/combined sewer overflows, unlike municipal
sewage, are highly variable, both in flowrate and pollutant concentration;
a treatment unit cannot be equally effective across the wide range of in-
flow quantity and quality. Effluent from a microstraining unit, for ex-
ample, cannot have less than 10 mg/1 SS, due to physical limitations such
as screen size, flux, etc. If the average SS removal efficiency is 80%
and the influent concentration 20 mg/1, the effluent cannot be 20 (1.0 -
0.80) = 4 mg/1, but 10 mg/1 at the lowest. Therefore, the program assigns
the minimum feasible effluent concentrations when the effluent or the
influent is below these limits.
64
-------
6. Real-time control of in-system variable storage and variable
treatment at the plant is simulated according to a built-in strategy.
This strategy is adopted in accordance with physical constraints and the
results of various strategies reported in literature (Anderson 1970,
Detroit Metro Water Department, 1970, Bell, 1974, Leiser, 1974, Wenzel et
al, 1974). It is assumed that the interceptor acts as a common reservoir
and can store flows from all watersheds, while a trunk can store flows
only from the watershed that it serves. The main advantage of intercept-
or storage over trunk storage is that the interceptor is long enough to
distribute flows during moving storms, from rainy areas of the city to
the dry areas, thereby preventing or at least retarding overflows.
When a storm begins,and before any in-system storage is utilized,the
treatment rates at the plant are increased to their maximum capacities,
in order to capture the polluted "first flush" at the beginning of the
storm. If the storm intensities generate runoff greater than the
treatment rates, the sewers are used for storage. Priority is given to
interceptor storage because the interceptor is long enough to distribute
flows from high runoff areas to low runoff areas of the city, during
localized storms. Another reason for interceptor storage to precede trunk
storage is that filling the individual trunks first may cause overflows
before there is a chance to fill the interceptor. This may happen if
the runoff intensity exceeds the reduced flow capacity of the trunk, or
the flow capacity of the regulator/diversion pipe. In this case, over-
flowing will be essential in order to prevent flooding, even though inter-
ceptor storage capacity still exists. Therefore, once the runoff intensity
exceeds the maximum treatment rates, interceptor storage is utilized; if
65
-------
the storm still continues after the interceptor is filled up, the trunks
are used for storage; and if these are also filled, only then are over-
flows allowed. The recession of the storm and emptying of the system
follows a similar pattern; the treatment rates are decreased to average
DWF only when all of the in-system storage is emptied and treated. The
control strategy is described in Figure 4,and the performance of each
system component shown in Figure 5. Routing is computed, based on mass
balance procedures, in hourly time-steps with continuous updating of
plant, trunk, interceptor flow and storage values. Pollutant movement
along the system is assumed to be flow-proportional.
7. The model is equally applicable for separate storm system simu-
lation. By inputting zero trunk-to-interceptor diversion pipe capacity,
the combined system is converted to a separate one. Various stormwater
treatment methods can be simulated continuously with multiple watersheds
and outfall points.
APPLICATION
The model was built and tested on a hypothetical combined sewered
city. Performance of a conventional combined system and various real-
time control alternatives were tested both for single events, and for all
events in a one-year period. Results similar to those reported in Seattle
(Leiser, 1974) and Detroit (Detroit Metro Water Dept., 1970) were obtained.
Next, the model was applied to one of the three interceptor/treat-
ment plant systems in the City of Winnipeg, Manitoba, which experiences
combined sewer overflow and plant by-pass problems. The study area
shown in Figure 6, is about 9,300 acres, generates about 15 Imgd DWF and
66
-------
is serviced by a 10-mile long interceptor which lies along the Assiniboine
and Red Rivers. There are twelve watersheds, each served by a combined
trunk sewer that can divert up to 2.75 x DWF into the interceptor/plant
system, all excesses being overflowed. The plant at the end of the inter-
ceptor is the North-End Sewage Treatment Plant, which is quite overloaded;
the average DWF that it treats is near its design capacity. This situation
has resulted in frequent by-passes during WWF, sometimes even during DWF.
Therefore, the City is enlarging the pi ant, from 55 to 100 Imgd for second-
ary and from 100 to 180 Imgd for primary treatment capacities (City of
Winnipeg, 1975, MacLaren, 1976).
Precipitation records from three different rain gauges were used to
illustrate the spatial distribution and movement of the storms. STORM was
applied to each of the twelve watershed to produce runoff hydrographs and
BOD, SS pollutographs for all events in a six-month period. Then, RAFFI
was used to combine the output of STORM and to simulate the overall be-
haviour of the existing system and the system with the proposed control
alternatives. DWF was simulated by inputting average flow and pollutant
data for each watershed.
Apart from the existing system simulations, three real-time control,
and two overflow treatment alternatives, and all their combinations were
analyzed as seen in Table 2.
The existing or no-control alternative was assumed to be the
Winnipeg system after the expansion underway at the plant was completed.
The simulations showed that this plant expansion will have virtually no
effect on the overflow pollution, even though it may eliminate the by-
passes.
67
-------
In the simulation of the existing system, it was found that total
runoff duration was 330 hours, 4% of the 6-month simulation period; yet,
the BOD and SS loads discharged to the receiving waters during WWF (4%
of the time) were equal to, or greater than the loads discharged during
DWF (96% of the time), based on the assumption of 80 - 90% removal
efficiency of the plant. This finding clearly indicated that the WWF
pollution was quite significant; moreover, it took place in a short
period of time and thus, its short-term effects were even more critical.
Results of the continuous simulation are shown in Figure 7. It can
be seen that in the expanded system, without any control.nearly 80% of
the total WWF volume and 30 - 40% of the total WWF pollution (or 60 - 80%
of the total pollution discharged out) reached receiving waters from
overflows, during all storms in a six-month period. Overflow frequency
was 100% (all events overflowed), and average overflow duration was 3.6
hours per event. Among the various control alternatives, 83, real-time
control with in-system storage distributed in trunks and interceptor, pro-
duced the best results. Improvements were 20% for overflow frequency,
30% for overflow duration, 22% for overflow volume and 40 - 50% for over-
flow BOD and SS discharge, compared with the no-control system behaviour.
By increasing the treatment rate and by storing the WWF in various parts
of the system, overflows during the first hours of an event were prevented,
thereby protecting the receiving waters from the polluted first flush.
Treatment of overflows was not as effective as real-time control be-
cause the minimum feasible effluent concentrations were usually higher
68
-------
than the effluent concentrations to be achieved by the removal efficien
cies; therefore, the facilities could not be used at optimum efficiency.
Improvements ranged in 10 - 40% in overflow quality, without any change in
overflow volume, duration and frequency. In comparison, all of the real-
time control strategies produced 20 - 50% improvements, based on results
for the no-control alternative.
Short-term performance of the alternatives was also analyzed.
Simulations with test storms and selected events showed that real-time
control alternatives could keep the WWF in the system for the first 1.0 -
3.0 hours of an average 6-hour storm, before allowing overflows. Results
were much more favourable during moving storms; this clearly showed the
advantage of "distributable" storage in the system for effective capacity
utilization. Figures 8 and 9 are examples of the results obtained from
the detailed analysis of a 6-hour event.
LIMITATIONS
„ Once the model is built and its potential use demonstrated, its cap-
abilities may be increased and limitations reduced, using this work as a
base or starting point. Some of the limitations and recommendations for
improvement are mentioned below:
1. The model uses only one real-time control strategy, i.e. one
method of sequencing the operations. Work should be done to provide more
than one control strategy and regulator operation rule.
2. Since every city has its own unique sewer network configuration,
the model should be able to handle more than one type of sewer system.
69
-------
Analysis of two or more interceptors (as in Toronto), two or more over-
flow and regulator points in a watershed should be incorporated; but the
program is flexible and most users can handle these changes.
3. Time lag during routing is not provided.
4. Quality change during system control is ignored,even though in
reality, there will be settling of pollutants in the sewers if storm-
water is retained in the pipes for 6-8 hours. Quality routing is assumed
to be flow-proportional.
5. Performance of treatment plant and overflow treatment facilities
are approximated as percent removal efficiencies, within some specified
limits and constraints.
6. Interfacing RAFFI with STORM is one of the challenging limitat-
ions. However, this limitation is not in model logic but only in computer
programming software and can be solved.
STORM simulates runoff events from a given precipitation record for
one watershed at a time. Only after the entire record is simulated for
one watershed, does it start on the next watershed and repeat the same
procedure. RAFFI, on the other hand, simulates one runoff event from all
watersheds and analyzes simultaneously the effects of this event on the
total system. When the event has ended, it proceeds to the next event and
covers all watersheds and the common system again. The two models can be
visualized to operate as the row and column of a matrix. Memory storage
or programmed punching of STORM output are not feasible solutions because
of the size of data output and card sequencing. A new subroutine should
be added to STORM, which will enable the model to proceed hourly and pro-
70
-------
vide "absolute" events rather than "relative" ones; in this fashion,
STORM should be able to identify the first hour of an event in absolute
terms. That is, as soon as runoff begins in one watershed, it should
register in all watersheds as a new event, regardless of whether there
is runoff in other wathersheds or not.
7. The incorporation of an economic analysis and/or optimization
model to the STORM-RAFFI package could be the next step in development
which would increase the capabilities of the decision-maker. Optimization
models for this purpose are discussed by Bell (1974), Lindholm (1974),
Wenzel et al (1975).
CONCLUSIONS
This work essentially demonstrates two things - first, that real-
time control is a cost-effective abatement alternative and second, that
model RAFFI can be used to evaluate this in the initial planning stage.
Despite the high initial cost, real-time control helps to utilize all
components of an existing system very effectively. Since it can be im-
plemented readily in most major combined sewered cities without requiring
any additional facilities, it should be considered an attractive solution.
Despite its limitations and simplistic assumptions, RAFFI can become a
useful tool when interfaced with STORM, in the initial planning stage of
a stormwater management study. It is suggested that at the beginning of
a project, the behaviour of an existing system and all alternative abate-
ment methods be estimated with a simplesinexpensive5continuous simulation
model like STORM-RAFFI over a long period. This will provide a starting
point and will help to screen a large number of alternatives at a low cost.
71
-------
Once a feeling of the problem and potential solutions is obtained, the
number of alternatives can be reduced and it may become feasible to con-
centrate on the more effective ones with more detailed and expensive mod-
els such as WREM (Water Resources Engineers Model) or DORSCH-HVM (Hydro-
graph Volume Method).
Finally, an example is cited which may help to indicate the potential
for STORM-RAFFI. Simulation with WREM on the Winnipeg study area would cost
$300 - $400/run, for a single 6-hour critical event, with surcharged con-
ditions (Clarke, 1976). The output would be in the form of very detailed
overflow hydrographs, water depths, height of surcharge, etc., at many
"nodes" for time intervals of less than a minute. The study with STORM-
RAFFI, would cost only $15 - $20/run, for a continuous six-month record,
with much more useful information to the decision-maker about real-time
control 0^ overflow treatment alternatives,
72
-------
REFERENCES
Anderson, J.J., "Real-Time Computer Control of Urban Runoff", ASCE,
Journal of Hydr. Div.. 96, HY-1, pp. 153-164, January, 1970.
Bedrosyan, R., "A Stormwater Management Planning Model for Real-Time
Control and Stormwater Treatment Alternatives", Master of Applied
Science Thesis, Dept. of Civil Engineering, University of Toronto,
Toronto, Canada, 1976.
Bell, P.W.W., "Optimal Control of Flow in Combined Sewer Systems", Metro
Water Ingelligence System, TR #12, Fort Collins, Colorado, May, 1974.
Brandstetter, A., "Assessment of Mathematical Models for Storm and Comb-
ined Sewer Management", U.S. EPA Office of Research and Development,
Cincinnati, Ohio, August, 1975.
Clarke, W., Personal communication, September, 1976.
City of Winnipeg, Department of Public Works, "Annual Report-1975".
Detroit Metro Water Department, "Detroit Sewer Monitoring and Remote
Control", in "Combined SewerOverflow Abatement Technology", pp. 219-
290, U.S. Dept. of Interior, Water Pollution Control Research Series
11024-06/70, July, 1970.
Field, R., and Struzeski, J., "Management and Control of Combined Sewer
Overflows", Journal of Water Pollution Contr.. Fed., 44, 7, pp.1393-
1415, July, 1972.
Field, R., and Tafuri, A.M., "Storm Flow Pollution Control in the U.S.",
in "Combined Sewer Overflow Seminar Papers", pp. 1-48, EPA-670/2-73-
077, U.S. EPA Office of Research and Development, Cincinnati, Ohio
November, 1973.
73
-------
Glover, M. K., and Herbert, G.R., "Microstraining and Disinfection of
Combined Sewer Overflows", EPA, Washington, D.C., EPA-670/2-73-071,
1973.
Gupta, M.K., and Agnew,R.W., "Screening/Dissolved-Air Flotation Tnv.tment
of Combined Sewer Overflows", in "Combined Sewer Overflow_Seminir
Papers", pp. 129-152, EPA-670/2-73-077, U.S. EPA Office of Research
and Development, Cincinnati, Ohio, November 1973.
Harvey, P., "High-Rate Multi-Media Filtration", in " Comb in ed_Se we r_0 ver -
flow Seminar Papers", pp. 115-128, EPA-670/2-73-077, U.S. EPA Office
of Research and Development, Cincinnati, Ohio, November, 1973.
Heinke, G.W., "Design and Performance Criteria for Settling Tanks for the
Removal of Physical-Chemical Floes", Research Report No.10, Environ-
ment Canada, 1973.
Lager, J.A., and Smith, W.G.,"Urban Stormwater Management and Technology:
An Assessment", EPA-670/2-74-040, U.S. EPA Office of Research and
Development, Cincinnati, Ohio, December, 1974.
Lee, J.A. et al., "Filtering Combined Sewer Overflows", Journal of the
Water Pollution Contr. Fed.. 44, 7, pp.1317-1333, July 1972.
Leiser, C.P., "Computer Management of a Combined Sewer Systetn", iMunicip-
ality of Metro Seattle, Washington, for EPA Storm and Continue! S. • /:T
Section, Cincinnati, Ohio, EPA-670/2-74-022, July, 1974.
Lin, K.C., "Significance of Temperature in the Activated Sludge Process",
Ph.D. Thesis, University of Toronto, 1974.
Lindholm, O.G., "Factors Affecting the Performance of Combined vs. Separate
74
-------
Sewer Systems", paper presented at the 7th International Conference
on Water Pollution Resources, Paris, September, 1974.
MacLaren, J.F., Limited, "Combined Sewer Overflows in the City of Winnipe-
Draft Report, February, 1976.
MacLaren, J.F., Limited, and Proctor-Redfern Limited, "Stormwater Manage-
ment Model Study", Report for Environment Canada and Ontario Ministry
of the Environment, presented at First Canadian Stormwater Management
Model Workshop, Toronto, Ontario, March, 1976.
Marsalek, J., "Abatement of Pollution Due to Combined Sewer Overflows",
Inland Water Branch, Dept. of Environment, Canada, Technical Bulletin
No. 66, 1972.
Rosenkranz, W.A., "Storm and Combined Sewer Demonstration Projects", U.S.
Dept. of Interior, Water Pollution Control Research Series, DAST-36,
January, 1970.
Salib, W., Plant Manager, Toronto Ashbridges Bay Sewage Treatment Plant,
Personal Communication, May, 1976.
U.S.Army Corps of Engineers, "Urban Storm Water Runoff-STORM", Hydrologic
Engineering Centre, Davis,California, January, 1975.
Water Resources Engineers and U.S.Army Corps of Engineers, "A Model for
Evaluating Runoff Quality in Metropolitan Master Planning", ASCE
Urban Water Resources Research Program, Technical Memo, No. 23,
April, 1974.
Wenzel, H.G., Labadie, J.W., and Grigg, N.S., "Detention Storage Control
Strategy Development", ASCE, Journal of Water Resources. Planning
75
-------
and Management Pivision, pp. 117-135, April, 1976.
White, R.L., and Cole, T.G., "Dissolved Air Flotation for Combined Sewer
Overflows", Public Works, No. 104, 2, 50^, 1973
Wisner, P.E., "Objectives of Modelling in Urban Drainage Studies", paper
presented at First Canadian Stormwater Management Workshop, Toronto,
March, 1976.
76
-------
Table 1
Stormwater Management Alternatives
1. Sewer Separation
2. Source Controls (i) on-site storage
(ii) improved street cleaning practice
(iii) wiser usage of chemicals
Collection system controls (i) infiltration/inflow control
(ii) sewer flushing
(iii) polymer injection
(iv) remote monitoring/real-time control
Storage/Treatment
~ (i) Upstream - in-line or off-line
Storage (ii) Downstream - in-line or off-line
(iii) In-system
(iv) Physical microstraining, filtration
(v) Physical-Chemical disinfection
Treatment
(vi) Biological - biological contactor
(vii) Combinations of (i) to (vi)
77
-------
TABLE 2
ALTERNATIVES ANALYZED
1. NO CONTROL
2. OVERFLOW TREATMENT
(i) Low-rate
(ii) High-rate
3. REAL-TIME CONTROL, IN-SYSTEM STORAGE
(i) In trunks
(ii) In trunks + interceptor
(iii) In interceptor + common reservoirs
4. COMBINATIONS OF 1, 2, 3.
78
-------
VD
WATER
PURIFICATION
Untreated Dis
Separate Storm
Combined Storm
£
u
d
•t
4
UJ
a:
H
RECEiVSNG WATER BODY
FIG. ! WATER RESOURCES SYSTEM OF AN URBAN AREA
-------
COMPUTER CONTROL CENTER
Prediction Model
(Computer Proc/am)
Control Model
(Computer Program)
Transmission
of Reid Data
Sensing
Elements
Interface
Tronsirs ssion
of Status
Transmission
of Instructions
Control Elements
(Regulators, Gates, Pumps,
etc.)
T
Interface
Water Resources
Sub System
FIG. 2 A COMBINED S€WER CONTROL SYSTEM
80
-------
PRECIPITATION
CO
Inttfcepfer
Regulator/
Diversion
Facility 2
Connection
Overflow
Trtatment
WWF
By-pcss
Primary/
By-pose
Effluent
RECEIVING WATER BODY
Secondary
Efflu«nt
FIG. 3 RAFFI- An Ovtrvi®*
-------
'////
V
OP O
\
c
2. Interceptor "*
r-/
•B
£
•ft
1
*
F'
rt
-y
1. Treolment Plant
. _J 1
y y y i — i
4. Overflow
STRATEGY
I. Increase treatment plant capacity to maximum, T
2. Fill tns common int«rc«^tor-subject ta canctraint of
connection pipe capacities ,and if R>T
3. Fill the individual trunks, if interceptor fled and R*T
4. Overflow, if R>Regulated inflow capacity
(a) SYSTEM FILLING STRATEGY
2, Interceptor
^"•*
•g
3
£
1
•^y
c
3
£
-™(
1
3
F
™
-f
3. Tre<
STRATEGY
I. Empty the trunks at T
Z Empty the interceptor at T
3. Reduce treatment rotes from maximum ,T, to nermal DWF
(b) SYSTEM EMPTYING STRATEGY
FIG. 4 CONTROL STRATEGY IN REAL-TIME CONTROL
82
-------
Trunk.
Inflow and
Outflow
Trunk
DWF
0
0
Watershed n
Overflow
Discharge
Watershed n
Overflow
Volume
Interceptor
Inflow
Interceptor
Storage
Volume
Interceptor
Outflow,
Treat merit plant
Inflow and
Outflow 0
« Ouff
Trunk to interceptor fW
~~ diversion capacity
23456 789
_lrunk storage capacity
5 S
5 6
Treo|ment pto«t ftow capacity
2 4
789
Jateratptor storage capacity
2 4
plant flow
capacity
8 9
FIG. 5 OF SYSTEM REAL-TIME CONTROL
83
-------
FIG. 6 CITY OF WINNIPEG
oo
SCALE
TREATMENT FACILITY
PRECIPITATION GAGE
INTERCEPTOR -*» OVERFLOW POINT
-------
CD
FIG. 7 PERCENTAGE POLLUTION LOADS REACHING RECEIVING WATERS
DURING WWF IN A SIX-MONTH PERIOD -WINNIPEG
BOD
Overflow frequency-100%
Overflow duration
3.6ht/event
BOD
SS
Overflow frequency-KX)%
Overflow duration
16hr/event
BOD
SS
Overflow frequency -80%
Overflow duration
25 hr/evsnt
BOD
SS
Overflow fr«qu«ney-8O%
Overflow duration
2.5 hc/tvent
BOD
SS
Overflow frequency-85%
Overflow duration
2.7 hr./event
BI; no control
Bias 81+ overflow treatment
B 3 *- controlled in-system storage in trunks and interceptor
8 3o 183 * overflow treatment
84 icontrolled in-system storage in interceptor and reservoir
along its length
LEGEND
combined sewer overflow load
*econdory effluent load
primary effluent load
pollution tood retained
-------
350-
81 * No control
83s Controlled in-system storage
in trunks + interceptor
B4=Controil«d storage in int«rc«ptor
* r«s«rvoirs along if
WWF« Runoff + 0WF
Bl
3 4
TIME (Mrs.)
FIG. 8 WINNIPEG -OVERFLOW SS POLLUTOGRAPHS SINGLE EVENTS
FOR SOME ALTERNATIVES
86
-------
B3x Can (relied in-vystan
in traito
3 4
TIME (hra.)
RG. 9 OVERFLOW §00 POLLUTOGRAPH
87
-------
Methodology for Evaluating
Agricultural Best Management Practices (BMP's)
by
J. Kiihner, W.W. Walker, and J.J. Wineman*
Meta Systems Inc
10 Holworthy Street, Cambridge, Massachusetts 02138
The goals of the Water Pollution Control Act Amendments of 1972
(FWPCA, PL 92-500) will be achieved only if nonpoint source, as well
as point source, pollution is controlled. Authority exists under the
FWPCA and the Clean Water Act of 1977 (PL 95-217) for the Environmental
Protection Agency (EPA), in conjunction with individual states, to
devise policies and initiate control programs to manage nonpoint source
pollution. Progress, however, has been slow due to various reasons.
Based on a feasibility study undertaken by Meta Systems Inc for
the EPA (September, 1978), a policy analysis methodology is presented
here that addresses the problem of developing and evaluating effective
strategies for the management of agricultural nonpoint source pollu-
tion. Figure 1 indicates that the proposed methodology is specifically
designed to assess both the instream water quality and socio-economic
impacts of agricultural practices and specific government policies
aimed at encouraging agricultural practices which decrease nonpoint
source pollution. Thus the methodology allows the simultaneous exami-
nation of 1) the water quality impacts of selected agricultural prac-
tices in the watershed and downstream waterbodies and 2) the economic
effects (upstream and downstream) that alternative practices and non-
point source pollution control policies have upon the farmer and down-
stream user. At this point the nonpoint source pollution control prob-
lems which the methodology addresses are limited to those that are
amenable to solutions by incremental on-farm adjustments for damage
reduction. Such problems would include, for example, the generation
and transport of sediment and nutrients.
In developing the methodology, it was felt that potential methods
for the assessment of environmental and socio-economic impacts of agri-
cultural practices would be effective only if they exhibit the following
characteristics,
• Compatibility between data • Ease of Understanding
availability and requirements
• Usefulness at the appropriate
• Robustness under a wide range planning level
of alternative agricultural . . .
futures * Applicability to the full
range of on-farm options
• Capability to evaluate major
policy options
*This paper was presented by Dr. Jochen Kuhner at EPA's SWMM Users Group
Meeting in Annapolis, Maryland on November 13, 1978.
88
-------
oo
i-D
FIGURE 1
HETHOUOL06V FOR ASSESSMENT OF WATER QUALITY
«MD SOCIO-ECONOMIC IMPACTS OF AGRICULTURAL PRACTICES
Soil T»ps
Slope/Lang!!;
Areo
Climotologlc
.
il
lurol
e
jtlves:
nology
1
1
I
1
1
FAR!
Farm
Model
I
MODEL
Economic
Evoluailon of
Agricultural
proclices
1
1
T~*
1
_J
Net Revenue
Practices
-------
In the present study it appeared that the best way to evaluate, tb.e
proposed methodology on the basis of the above cMractv-ri sties «v.-:i?-; ro
work through an example. Thus the Black Creek water "h^ol in northeastern
Indiana, a USEPA/USDA demonstration project (see Christiansen and
Wilson, 1976; Lake and Morrison, 1975), was selected and combined with
a. synthesized downstream impoundment with characteristics typicaj. of
those found in the Corn Belt (for details, see Meta Systems r^c, Septem-
ber, 1978; Wineman et al., 1978, and Walker and Ktihner, 1978a and 19'"'8b)
Methodology Development
Agricultural Practices and the Farm Model
While major market and regulatory pressures — such as prices,,
taxes, subsidies, government regulations — are exerted at a regional
or national level, it is the farmer who responds by choosing both his
crops and methods of farming. For this reason the methodology has to
include a farm budget.
It is assumed that the farmer desires to maximize net revenues,
Accordingly, he chooses a set of agricultural practices such as;
® crop rotation 9 structural erosion and drainage
control practices; and
© tillage practices
® levels of chemical application
These choices are represented as inputs to the farm model for the cal-
culation of a variety of costs associated with operating the tarrn, in
the specified manner-* This required developing a ..iat-i base for the
model. Each element of cost was updated for 1977 prices and modified
where necessary to adapt the model for the Black Creek area. Additional
inputs to the model specify expected yields and market prices for each
crop.
In the case study 11 practices (plus two modifications to include
custom hiring) are selected, and farm budgets are set up for uniform
farms of 250 acres on each of three soil types (upland, ridge, and low-
land) characteristic of the Black Creek watershed, Table 1 shows the
revenue, costs, arid net returns for three farms, each on one soil type/
assuming uniform adoption of each of the 11 farm practices and existing
government policies.
*The procedure described by Dr. Klaus Alt in Appendix C of Control of
Water Pollution from Cropland (USDA, U.S. EPA, 1976) was applied here.
90
-------
TABLE !>• SUMMARY OP FASWl *iOC'i?
SUNDER EXISTING Goviit..-
Farm
Practice
\
\
A
GROSS
REVENUE
A. Upland soil
E, Ridge soil
C. Lowland soil
COSTS
^. Upland soil
B. Ridge soil
C. Lowland soil
1
MET
RETURN
A, Upland soil
B. Rlrkie soil
C. Lowland soil
Tillage Practices
corn ,
conven -
ticnal
tillage
(cc-cv)
52.5
65,0
65.0
39,7
41.4
42.7
12.8
23.6
22.3
corn ,
chisel
plow
(CC-CH)
52.5
65.0
65.0
39.1
40.9
42.1
13.4
24.1
22.9
corn,
no-till
(CC-NT)
49.9
65.0
52.0
43.0
44.9
45.5
6.9
20.1
6.5
Rotations
corn
soybean ,=
conven-
tional
tillage
(CB-CV)
46.3
59.1
59.1
32.9
33.3
34.8
13.5
25.8
24.4
corn
soybean
chioei
pi.»
iCB-CH!
46.3
59.1
59.1
32.6
33.1
34.5
13.7
26.1
24.6
I
corn
soybean
no-till
(CE-NT)
44.4
57.9
50.7
32.3
32.7
34.1
12.2
25.1
16.6
corn.- soy-
bean, wheat.
hay ,
partial use
of herbicides
(CBWH:
43.0
51.8
49.9
34.4
34.7
35.4
8.5
17.4
14.5
* !CBWH)
!i i . 0
51.8
49.9
30.6
31.0
11.7
12.4
20.8
18.1
corn, soy-
bean, wheat
hay,
no- till
!CB'WH*-ST)
43.0
51.8
49.0
34.2
34.4
35.1
8.8
17.4
13.9
(CBWH-NT)
43.0
51.8
49.0
30.3
30.7
31.4
12.8
21.1
17.6
Terraces
corn
corr/en- i corn
tional chisal
tillage
(CC-CVT)
56.0
68.5
68.5
46.4
48.2
49.4
9.6
20.3
19.1
plow
CC-CHT
56,0
68.5
68.5
45.8
47.5
48.9
10.2
20.9
19.6
cors
soy-
bean
no-
till
(CB-i-i'iT,
47.4
60.9
53.7
38.9
3a . 3
40.7
8.6
J1.5
13.0
Coliimns may r.o'^ a<3d !u
-------
Water Quality Model
Examples of watershed (practice/soil type combinations) and water
quality analyses are based on the assumption of a homogeneous watershed
(as assumed in the farm model). At this preliminary stage of methodo-
logy development this approach was considered to be more appropriate
than one dealing with aggregate economic and environmental impacts in
a heterogeneous watershed (which would incorporate variations in soil,
slopes, farm sizes, and other characteristics).
The water quality analysis is separated into two major sections:
1) the watershed, which is characterized as generating
different loadings of pollutants depending on agri-
cultural activities and watershed characteristics.
2) the impoundment, where water quality is dependent
on the type and quantity of loadings from the
watershed and on impoundment characteristics.
In this scheme the river is represented as a medium for transporting
pollutant loadings from the watershed to the impoundment. Water
quality conditions in the river reflect these loadings, which enter
the river in surface runoff and groundwater base flow and are trans-
ported in dissolved and sediment-bound phases.
The methods developed for the watershed analysis are of an empiri-
cal nature and are concerned with long-term average emissions (though
modified, they are consistent with the mode of the Universal Soil Loss
Equation (Wischmeier and Smith, 1972)). Although the use of long-term
average time scales precludes direct assessment of responses under
extreme meteorologic conditions, effects of the timing of various agri-
cultural operations, seasonal variations in water quality, and analysis
of relatively short-lived compounds, we feel that given currently
available data and limited knowledge of the relevant physical processes,
a framework built from complex models would generally not be feasible
or useful at the regional planning level,
Pathways involved in the watershed analysis and impoundment water
quality analyses are summarized respectively in Figures 2 and 3. Average
annual export (delivery) rates of the following substances are evaluated
in the watershed analysis: 1) sediment (sand, silt, and clay fractions);
2) phosphorus (biologically available); 3) dissolved nitrogen? and
4) dissolved color. The computed concentrations of these components
are assumed to be representative of average water quality conditions
in rivers draining the agricultural watershed. These results of the water-
shed analysis can be linked with downstream empirical models for the pur-
pose of predicting quality impacts in impounded waters under steady-
state, seasonal, or long-term-average conditions. As Figure 3 indicates,
water quality components considered are suspended solids, phosphorus,
nitrogen, transparency, and epilimnetic chlorophyll-a concentration,
92
-------
00
WATERSHED
CHARACTERISTICS
Field Characteristics
Soil Characteristics
Climate
Morphometry
Agricultural
Practices
Crop Yields
TRANSPORT RATES
Sediment
Runoff
Percolation
TRANSPORT MEDIA
COMPOSITION
Sediment
Runoff
Percolation
Nitrogen Budget
AVERAGE RIVER
WATER QUALITY
AND COMPONENT
LOADINGS
Sediment
Phosphorus
Nitrogen
Color
Figure 2. Pathways in the Watershed Analysis
-------
LOADINGS
TRAPPING/
DECAY RATES
OUTFLOW
EPILIMNETIC
CONCENTRATIONS
Color
Color
-*• Color
DELIVERED
Sediment
*• Sediment
Phosphorus
Nitrogen
Suspended
Solids
IMPOUNDMENT MORPHOMETRIC
AND HYDROLOGIC CHARACTERISTICS
Transparency
Chlorophyll-o
Concentration
FIGURE 3 . PATHWAYS IN THE IMPOUNDMENT WATER QUALITY ANALYSIS
-------
The latter is used as an index of eutrophication and represents the
extent of the algal growth in the surface waters of impoundments dur-
ing the summer. The model developed for predicting chlorophyll-a
levels considers the possible limitation of algal growth by light,
phosphorus, and/or nitrogen. Transparency is computed as a function
of suspended solids, dissolved color, and chlorophyll-a concentrations.
Models are formulated for each of the above water quality components
based upon theoretical considerations and the results of previous
modeling efforts. They are calibrated and tested empirically using a
data base characterizing the behavior of these components in Corn Belt
impoundments and compiled from various sources (EPA/NES, 1975; USDA,
1969; ISBH, 1976; USAGE, 1977).
Important additional independent variables include total water-
shed area, impoundment surface area, and impoundment mean depth.
Assumed values for the examples discussed below are 200 km2, 5 km2,
and 4 meters respectively. It should be noted that the evaluations
of the relative impacts of the agricultural practices on impoundment
water quality may be somewhat sensitive to the choice of the watershed/
impoundment configuration.
Impacts on Downstream Users
To estimate the impacts of alternative farming practice/soil type
combinations on downstream users, changes in water quality must be
related to measurements of value to these users. Depending on the
use of the water and the land uses surrounding the watershed and im-
poundment, certain water quality components are of more or less inte-
est to different groups of people concerned with water quality (users),
Some benefit categories of interest in this case are human health,
municipal water supply, flood control, ecology, recreation, aesthetics,
and the local economy. Rigorous quantitative methods of benefit estima-
tion would vary depending upon the particular benefit category of inter-
est (see Appendix E of Meta Systems Inc, September, 1978), It is clear
that a comprehensive benefit estimation methodology cohering several such
measurement techniques would be a major undertaking, As an alternative appro-
priate for policy-making and agency resources, a simplified version is pre-
sented here that qualitatively assesses the direction of benefits (as per-
ceived by each user group) resulting from water quality changes induced by the
alternative practices. As the example (table 2) shows, a minus sign has
been chosen to indicate that an increase in a specific water quality com-
ponent has £=1 detrimental effect on the specified benefit group; for
example, an increase in nitrogen concentration in drinking water is poten-
tially harmful to human health. A zero indicates that an increase in the
parameter is of no importance to the benefit category. For instance, the
same increase in nitrogen concentration just mentioned would not impact
flood control operations in the impoundment. A water quality component in-
crease which has a positive impact on a benefit category is indicated
95
-------
by a plus sign. Increasing impoundment biomass, for example, might
improve sport fishing since more food might increase the available
fish population.
Use of the Farm and Water Quality Models
Comparison of Practices
The methodology described above can be used to illustrate: 1) how
agricultural practices can be evaluated in terms of water quality im-
pact to facilitate selection of BMP's (under existing government poli-
cies) ; 2) how government policies which encourage the implementation
of practices that are conducive to water quality improvements (i.e.,
BMP's) may be examined; and 3) how changes in economic conditions can
affect agricultural practices and water quality impacts.
The evaluation of agricultural practices under current policies
uses the 11 selected farm practices listed (and defined) in Table 1
(as if they constituted a comprehensive set of alternatives currently
available to farmers). In addition to gross soil loss (i.e., erosion),
six variables related to water quality were analyzed for the three
soil types (upland, ridge, lowland) and the 11 farm practices. For
illustrative purposes, the results (together with net revenues) for
one soil type, the lowland, are displayed as a set of bar graphs
(Figure 4). In the graph increasing pollutant loads or concentrations
are shown by higher vertical height of the bar; for net revenue ver-
tical height increases with higher returns.
In terms of net revenue ranking of the 11 farm practices, the corn-
soybean rotation is most profitable based on prices chosen for these
commodities in this example (i.e., corn, $2.00 per bushel; soybeans,
$5.00 per bushel; wheat, $2.50 per bushel; hay, $60 per ton). This
practice is also most profitable on the other two soil types.
Figure 4 can also be used to examine the differences in soil loss
(gross erosion) or any of the six water quality components for each
farm practice on each of the three soil types, For example, on the
lowland, the practice that maximizes net revenue ($24,600) is calcu-
lated to produce an annual gross soil loss of 0,49 kg per square meter*
(gross losses range from 0. 78 to 0.1 kg/m2 before application of sedi-
ment delivery ratios). This can be contrasted with results obtained for
the other two soil types. On the ridge the annual net revenue maximizing
($26,100) practice results in 1.2 kg/m2 annual gross soil loss (range:
2.1 to 0.2 kg/m2). On the upland the net revenue maximizing ($13,700)
practice shows an annual soil loss of 3.4 kg/m2 (range: 6.1 to 0.6 kg/m2).
On all soil types the no tillage corn-soybean-wheat-hay rotation (CBWH-NT)
minimizes gross soil loss.
*kg/m2 = 4.5 tons/acre.
96
-------
Figure 4. Comparison of Practices — Lowland
(see Table 1 for defini-
NET REVENUE tion of practices)
(K$> »nC...
J^
SOIL LOSS
(kg/rnz of
wafershed-yr) ,8
A
SEDIMENTATION
(kg/m* of
impoundment-yr)
0
RIVER NITROGEN
(g/ms) 20
10
RIVER PHOSPHORUS
(9/mS)
,10
RIVER LIGHT EXTINCTION COEFFICIENT
(m")
40
IMPOUNDMENT LIGHT EXTINCTION COEFFICIENT
(rn1) 2-0
IMPOUNDMENT BIOMASS
(g Chlorophyll- .03
A/m>)
97
-------
The water quality impacts of agricultural practices vary with field/
soil tvr/e - vater body (river versus impoundment) and specific pollutant.
Use of gross soil loss alone as the criterion for farm practice evalua-
tions can lead to erroneous conclusions because of the importance of
various dissolved components and the interactive effects of different
processes (e.g., decay, adsorption/desorption, sedimentation).
The types of gross soil loss (erosion) controls evaluated here do
not appear to have proportionate imapcts on phosphorus losses and im-
poundment eutrophication problems. This results from the influence of
the following factors which are considered in the predictive (modeling)
framework: 1) a small fraction of soil phosphorus (five to ten percent)
is biologically available; 2) minimum tillage methods tend to cause
phosphorus enrichment of the surface soil layer and create a potential
for leaching of phosphorus from crop residues during snowmelt periods;
3) impoundment phosphorus trapping efficiency is correlated with sedi-
mentation rate; 4) reduced erosion rates result in increased availability
of light to promote algal growth in downstream impoundments. Because of
the combined effects of these factors, the agricultural practices evalu-
ated do not appear to be effective in controlling eutrophication and in
some cases cause moderate increases in algal growth (see Figure 4). In
such cases corresponding increases in the potential for fish production
could be interpreted as water quality benefits, although additional
analysis would be needed in order to quantify such benefits. Impacts
of soil loss controls on suspended solids concentrations and transparency
levels are generally in the same direction as impacts on gross erosion
rates, but are usually less. The results are specific to the field/soil
types examined and to the assumed watershed/impoundment configuration.
The importance of one pollutant compared to another may also shift
from watershed to watershed and hence influence the selection of those
water quantity components of primary importance to the evaluation of the
BMP's. In assessing BMP's it seems reasonable to rank the different
pollutants on the basis of severity of local water quality issues.
Our report (Meta Systems Inc, September, 1978) shows that the method-
ology can be used equally well for the other purposes mentioned above. Policy
examples investigated in the study include prohibition of certain cultivation
practices, gross soil loss restrictions, and fertilizer control policies. Chang-
ing economic conditions are evaluated using an example of an energy future.
Downstream User Effects
In order to compare uhe various agricultural practices froui the
downstream users* point of view.- vc. i)9^c. to select a base case. Having
?. = sa;Tied that the farmer is a maxi^'.i.?i.c of net revenue,, we choo.T.-: the pr.37-
t.tce- producing the highest net revenue (the corn-soybean rotation using
chisel plowing (CE-CH)) .. To illustrate, Figure 5 depicts the relative
v.-ater quality, gross soil loss, and net revenue impacts (measured as p.vr
centage increases or decreases relative to the base case) of the other ten
practices on the lowland soil type. As described above, the downstream
98
-------
Figure 5. Percent Change of Highest Revenue Factor — Lowland
REVENUE
SOIL LOSS
(%)
Or
•40[
-80.1
80i
40
0
-40
^
SEDIMENTATION
(%) 80r
40H
0
-40-
-80
RIVER NITROGEN
(%) 120
80
40
0
-20
RIVER PHOSPHORUS
(%) 20r
J
RIVER LIGHT EXTINCTION COEFFICIENT
(%) 60
20-
0
-20
-60-
IMPOUNDMENT LIGHT EXTINCTION COEFFICIENT
(%) 20p ^^ p^^^, n^n,
_ ®l "^
IMPOUNDMENT SiiOMASS
'%} 20
0
99
-------
benefits of alternative fanning practices can be qualitatively compared
by mapping the quantitative practice/water quality relationships depic-
ted in Figure 5 onto each user or benefit group. The resulting relative
impacts of the 11 farm practices on the benefit categories of interest
are summarized in Table 2. The number of pluses, minuses, or zeros in
each cell indicates the beneficial, detrimental, or neutral impact res-
pectively of the change in each of the six water quality components,
displayed in Figure 5, when a switch is made from the base case (CB-CH)
to the compared practice. For example, a switch from the base case to
a corn-soybean-wheat-hay rotation (CBWH) would result in a decrease in
each of the six water quality components, ranging from a decrease in
impoundment sedimentation of 70 percent to a decrease in impoundment
biomass of three percent (see Figure 5). It can be seen from the six
pluses in the second row and seventh column of Table 2 that the benefit
category, municipal water supply, for example, has been impacted posi-
tively by the decreases in each of the six water quality components
caused by the switch in agricultural practices.
The analysis also shows that in general, with the possible excep-
tion of the beneficial impact of higher biomass levels on sport fishing,
no conflicts exist among the benefit categories} i.e., all categories
are either not influenced or negatively influenced by an increase in
any of the water quality components,
In contrast to many other case studies, no attempt has been made
here to weigh the water quality components or benefit categories, since
this would introduce complex aggregation problems and difficult value
judgments highly specific to individual cases.
Conclusion
A proposed methodology for assessing the water quality and socio-
economic impacts of agricultural practices has been described and tested
in a case study, and its potential uses have been illustrated. It
appears that the development of such a methodology for regional and
state level planning for nonpoint source pollution control is feasible
and would be of significant value for broad analyses of large numbers
of policy alternatives, including identification of best management
practices. However, as noted in the discussion, the methodology is cur-
rently at a preliminary stage of development- and further refinements, such
as those summarized below, are necessary to make it fully operational.
» While the f£.rm budget model as presented here captures
the major elements important for assessing the economic
impacts of alternative nonpoint source pollution con-
trol policies on the farmer, further modifications
vould be necessary before it could be used effectively
in a planning context. Most important, the model should
100
-------
Table 2. Summary of Relative Impacts of Farming Practices on Benefit Categories
Soil Type: Lowland
Fanning Practices
Benefit Categories
human health
(drinking water)
municipal water
supply
dredging
(flood control)
ecology
recreation
sport fishing
contact
non-contact
aesthetics
local economy
CC-CV
1(0)
1C+)
5(0)
1W
1(0)
S(-)
l(-)
1(0)
1(+)
2(0)
2(0)
3S-)
1(+)
CC-CH
1(0)
3<+)
5(0)
3( + )
1(0)
l(-)
2(4-)
1(0)
2( + )
2(0)
2(0)
3<->
CC-NT .
1(0)
><„
K+!
5(0)
3<+>
1(0)
K-)
1(0)
2(0)
2(0)
3( + )
CB-CV
2(0)
1(0)
5(0)
l(-)
1(0)
1(0)
2(0)
2(0)
K + 5
2(0)
1(0)
CB-CH
6(0)
6(0)
6(0)
6(0)
6(0)
6(0)
6(0)
6(0)
6(0)
CB-NT
1(0)
....
K+>
5(0)
3(+)
1(0)
K-)
1(0)
2(0)
2(0)
3<+)
CBWH
1(0)
"*>
5(0)
6(+)
1(0)
l(-)
1(0)
2(0)
2(0)
6(+)
CBWH-NT
2(0)
1(0)
5(0)
1(0)
2(0)
2(0)
3(0)
3(0)
1(0)
CC-CVT
2(0)
1(0)
5(0)
K-)
1(0)
2(0)
2(0)
11 + )
3(0)
3(0)
1(0)
CC-CHT
2(0)
1(0)
K->
5(0)
1(0)
2(0)
2( + )
2(0)
3(0)
K-)
3(0)
K-)
1(0)
CC-NTT
1(0)
««
5(0)
3(-J
1(0)
1(0)
2(0)
2(0)
3(+>
-------
be automated, perhaps employing a revenue-maximizing linear
programming model for policy analysis. In addition, a much
broader range of agricultural practices must be considered,
including variations in fertilizer applications, organic
farming, and livestock integration.
• While the model includes what we consider to be those
functional aspects of the watershed/impoundment system
which are required for evaluation of the impacts of farm
management practices on water quality, it requires addi<-
tional analysis, calibration, and testing before it can
be applied in a planning context. A preliminary sensi-
tivity analysis has helped to identify those aspects of
the model framework which are most critical to the pre-
dictions (See Meta Systems Inc, (Appendix D), September,
1978).
• Additional water quality components that must be included in
the framework are dissolved oxygen, biocides, and biocide
residues.
102
-------
References
Bruyne, G. M. "Trap Efficiency at Reservoirs." Transactions of the
American Geophysical Union, Vol. 34, June 1953, pp. 407-418.
Christensen, R. G. and C. D. Wilson, ed. Best Management Practices
for Non-Point Source Pollution Control. EPA-905/9-76-005, U.S.
EPA, Region 5, Chicago, November 1976.
Data Resources Inc. "Data Resources Outlook for the United States
Energy Sector: Control Case." Energy Review. Lexington, Mass.
Summer 1977.
Indiana State Board of Health. Reports__gn Limnologic Investigation
of Lakes Martin, Palestine, Sylvan, Waubee, Webster, _Cr_ooked_,
Long and Hamilton. Water Pollution Central Division, Biological
Studies and Standards Section, Indiana 1976.
Iowa State University, Cooperative Extension Service. Background
Information for use with Corp-Opt System. FM 1628, 8th revision,
Ames, Iowa, December 1976.
James, S. C. , ed, Midwest Farm Planning Manual. Ames, Iowa: Iowa
Lake, J. and J. Morrison, eds. Envi£oraii«ntal_Im]^^
Water Quality. Progress Preport - Black Creek Project, Allen
County, Indiana, Allen County Soil and Water Conservation District,
EPA-905/9-75-Q06, November 1975.
Meta Systems Inc.
Reducing
_^
_^
and Appendices . * U.S. Environmental Protection Agency; Athens,
Georgia, September 1978. (Requests to be directed to Thomas E. Waddc-11,
U.S. Environmental Protection Agency's Environmental Research Labora-
tory, Athens, Georgia 30605)
*Appendix A, Farm Model.
Appendix B, Methods for Predicting Watershed Loadings.
Appendix C, Methods for Predicting Impoundment Water Quality.
Appendix D, Water Quality Impact Results: Additional Interpre-
tations and Sensitivity Analysis.
Appendix E, A Discussion of Benefit Estimation.
Appendix F, Crop Response to Fertilizer,
103
-------
References Continued
Purdue University, Department of Agricultural Economics. Purdue Crop
Budget, Model B-94. July 15, 1977.
Taylor, C. R. and K. K. Frohberg. "The Welfare Effacts of Erosion
Controls, Banning Pesticides and Limiting Fertilizer Applications
in the Corn Belt," American Journal of Agr. Econ., February 1977.
U.S. Army Corps of Engineers. Miscellaneous water quality data from
Indiana and Ohio reservoirs, 1971-1977. Louisville, Kentucky, 1977.
U.S. Department of Agriculture. Soil Conservation Service. National
Engineering Handbook, Section 4, Hydrology. U.S. Government
Printing Office, Washington, D.C. 1971.
U.S. Department of Agriculture. Summary of Reservoir Sediment Deposition
Surveys made in the United States through 1965. Miscellaneous
Publication Number 1143, May 1969, 64 pp.
U.S. Department of Agriculture. U.S. Environmental Protection Agency.
Office of Research and Development, Control of Water Pollution from
Cropland, Vol. 1: A Manual for Guideline Development. EPA 600/
2-75-0266, June 1976.
U.S. Environmental Protection Agency, National Eutrophication Survey^
Series of Working Papers. Corvallis Environmental Research Labora-
tory. Las Vegas, Nevada, 1975-76.
Walker, W.W. and J. Ktihner. "Modeling the Land-Water Interaction: Needed
For Evaluation of Best Management Practice Effects on Water Quality."
Paper No. NA78-207, American Society of Agricultural Engineers, St.
Joseph, Michigan, 1978.
Walker, W. W. and J. Kiihner. "An Empircial Analysis of Factors Controllina
Eutrophication in Midwestern Impoundments," Proceedings of the Interna-
tional Symposium on the Environmental Effects of Hydraulic Engineering:r
University of Tennessee, Knoxville, 1978b.
Wineman, J. et al. "Evaluation of Controls for Agricultural Nonpoint Source
Pollution," Proceedings of the 10th Annual Cornell University Conference
"Best Management Practices for Agriculture and Silviculture", Rochester,
New York, 1978.
White, W. C. "Fertilizer — Food-Energy Relationships." Paper, The Fertilizer
Institute, Washington, D.C., August 28, 1974.
Wischmeier, W. H. and D. D. Smith. Predicting Rainfall ••— Erosion Losses from
Cropland East of the Rocky Mountains. ARS, U.S. Department of Agriculture,
Agriculture Handbook No. 282, 1972.
Woolhiser, D. A. "Hydrologic Aspects of Non-Point Pollution." USEPA/USDA, 1976.
104
-------
COMPUTER SIMULATION OF FLOOD RELIEF WORKS
UTILIZING INLET CONTROL AND DETENTION STORAGE
By: Paul E. Theil and A. M. Candaras
Paul Theil Associates Ltd.
Consulting Engineers
Bramalea, Ontario, Canada
Users Group Meeting
Annapolis, Maryland
November 13, 1978
105
-------
COMPUTER SIMULATION OF FLOOD RELIEF WORKS
UTILIZING INLET CONTROL AND DETENTION STORAGE
INTRODUCTION
Basement flooding as well as pollution from combined sewer overflows
have become a very serious problem for many urban areas throughout the
world.
The damage that results is staggering. For example, within Metropolitan
Chicago, the Corps of Engineers have estimated that about 400,000
buildings are subject to flooding, causing an average annual damage of
$280 million. In addition, pollution of streams, rivers and Lake Michigan
cause a further estimated damage of about $500 million annually. Chicago
may be the most effected area in North America, but the problem is wide-
spread and affects us all, either as a suffering homeowner, or as
citizens having, through taxes, to participate in the cost of providing
relief measures.
Nor is the problem related only to combined Beveisj i\io:.:?. 3-.J-. Icu 3 O^raaqe
has occurred in areas where foundation drains have been connected b/
gravity to separate storm sewers, a practice U3?^ hv T nuiriDt r cf
municipalities in Canada, In such cases, a hydrc«-t.ut,j..c :.\'. CB.:-^ v.j ll_
be created on basements during times when the storm sewer 5urch&ta-;s at.
a result of runoff exceeding the sewer capacity. Wher.L ti3><:> i-ir-el oi:
surcharge exceeds the floor level of any adjacent basement by cr,f toot
or even less, tht upti&t cAf&ttd iA Au&&4jCiA.
-------
INTRODUCTION - cont...
floor will enter the sanitary sewer system through the basement floor
drains, causing sanitary sewer back-up as well. Damage of this type
is well documented, but unfortunately it is rarely covered by insurance,
leaving the homeowners to pay the cost of repairs.
The cause of the problem is easy to understand; our sewers simply do not
have sufficient capacity to handle every runoff eventuality. To provide
that would not only be very costly indeed, but also result in moving
problems downstream often causing more serious damage. The. pauf>t
0(5 Amoving btomm wat&i cu> &at>t cu> po&& n.oj>uJU:, bat to the. wv
CONVENTIONAL SOLUTION
The conventional solution nevertheless has always been to provide more
sewer capacity by constructing relief sewers or tunnels. This method
has many shortcomings; in spite of the high costs, it generally is not
very effective in providing relief for areas not directly served by the
relief sewers. Due to the very high costs, we cannot afford to install
relief sewers on all the local streets and as a result areas upstream
of the relief sewers will generally not receive significant flood relief.
Sewer separation must, by necessity, commence from an adequate downstream
outlet. Quite often, however, the most frequent flooding problems occur
during high intensity-short duration storms, i.e. summer thunder storms.
Such rainfall events result in large volumes of runoff in a relatively
short period of time, causing overloading in upstream areas, whereas
downstream areas adjacent to trunk sewers with larger conduit capacity
are not as susceptible. To provide relief by conventional methods,
relief sewers therefore have to be extended into the upstream areas also.
107
-------
ALTERNATIVE SOLUTION
The solutions our firm has employed for solving drainage problems in
existing built-up areas were derived by approaching the problem ^fiom the,
cUAHC^tion, Since existing sewer systems usually will
handle all flows except during the most intensive storms, we concluded
that It wowtd be. mo/ie. appiopltate, Ji^ the. -Lnt^ : capacxLtt/ coutd be. c.on&wtte.d,
40 that the, tiata ofa /inflow too old not e.xeeed the, capac-ity ofi thu &xLf>£ing
We have termed this solution the INLET CONTROL METHOD.
Storm water detention along with inlet control does not only lend itself
as a means of solving drainage problems in existing areas, but it is also
suitable for new areas, particularly where zero increase in peak
runoff is required to prevent environmental damage or where treatment
of storm water runoff would otherwise be required to maintain pollution
at acceptable levels. Based on our research and experience to date, we
are confident that in many such cases a combined sewer system with inlet
control and source detention will give the highest value-effective solution.
The benefits of using inlet control and detention storage for either combined
or separate systems can be briefly summarized as follows:
1. In areas with combined sewers:
- eliminates the need for costly sewer separation on streets and
private property.
- can provide immediate relief in areas with most severe flooding
problems, leaving less prone areas until funds become available.
- provides a level of protection much higher than what usually is
available by conventional methods, and likely at much lower costs.
- permits polluted storm water ("first flush") to be treated by
existing treatment facilities.
- prevents pollution from combined sewer overflows.
- reduces peak flows in receiving waters, resulting in less flooding
damage and stream erosion downstream.
108
-------
ALTERNATIVE SOLUTION - cont...
2. In areas with separate storm sewers, but with foundation drains
connected by gravity, it further
- eliminates structural and flooding damage to basements during
times when the rate of runoff exceeds the sewer capacity.
The inlet control method has become more practical with the introduction
of the "Hydro-Brake" regulator. A Hydro-Brake is a patented flow
controller made of stainless steel. It is self regulating and has no
moving parts. It requires no power, but uses the static head of stored
water to create its own "energy" to retard the flow. The movement of
water through a Hydro-Brake involves a swirl action, dissipating energy
to control the rate of discharge.
Although the function of a Hydro-Brake is somewhat similar to an orifice,
it has certain important advantages:
1. It permits a much larger opening for passage of the same amount
of water. This is particularly important where clogging is a
possibility, such as for instance in catchbasins. It is also
important where sanitary or combined sewage flows are being regulated.
2. The flow rate of a Hydro-Brake is not significantly affected by a
variation in head. This is important where it is desirable to
maintain a relatively large passage for the water, yet also maintain
a fixed maximum rate of flow during peak conditions.
3. The outflow from a Hydro-Brake does not create a high velocity jet
stream as an orifice will, thus avoiding scouring inside sewer pipes.
Three case studies will be presented here, where the inlet control method
and detention storage have been designed with the aid of computer
109
-------
ALTERNATIVE SOLUTION - cont...
modelling, to alleviate flooding problems in existing urban areas in
three separate municipalities in the Toronto area.
BOROUGH OF YORK
The Borough of York, being an older municipality, is served almost
exclusively by combined sewers which were originally designed to
accommodate a 2 year design storm, but due to increased imperviousness
and physical weaknesses of the sewer system, now in many cases have much
less capacity. Further aggrevating the problem are the facts that
stormwater from roofs and ground water from foundation drains along with
the sanitary service are connected to the sewer. The result has been
more and more frequent flooding of basements. Studies carried out about
10 to 12 years ago recommended sewer separation, and for about eight years
the Borough, with a budget of $1 million annually, proceeded to do just
that. With the cost rapidly escalating, it became apparent that adequate
relief would not be in sight for many years without over-extending the
Borough's ability to pay.
In 1976 our firm was engaged by the Borough of York to carry out a study
with the objective of finding alternative solutions for solving the
flooding problems, which were most pronounced in four specific areas.
The study was to be based on providing relief to accommodate 2 year design
storms without causing surcharge above existing basement floors. This
criteria has been a standard for many years, not only in the Borough of
York, but also in the adjoining City of Toronto. This, at our suggestion,
was extended to include cost estimates for the 5 year and 10 year design
storms as well.
110
-------
BOROUGH OF YORK - cont...
Our study concluded that conventional methods of providing relief sewers
would be unfeasible due to the fact that three of the four areas were
located at a considerable distance from suitable storm water outlets.
This made us search for alternative methods, of which the inlet control
method and detention storage was found to be the most value effective by
far. The inlet control was achieved by placing Hydro-Brake regulators
in the system and by disconnecting downspouts or, where this is not
feasible, placing flow restrictors in them.
Catchbasins were sealed where positive drainage could be achieved thus
utilizing the major system to carry stormwater runoff. At low points
where runoff Would accumulate, storm sewers were provided to carry the
water to detention tanks (see Fig.l ). Stormwater would then be
discharged into the combined sewer at a predetermined release rate, such
that the combination of roof runoff, surface runoff plus dry weather flows
would not cause flooding damage.
In this particular case, the majority of storage was accommodated in
subsurface tanks due to the high urban density and lack of open space
which is characteristic of many older sections of the city.
The preliminary estimates prepared as part of the study for the remedial
works required for the four areas indicate a range of costs as follows:
2 year design storm $110,000
5 year design storm $285,000
10 year design storm $830,000
As a result of the findings of our study, the Borough decided to proceed
on the basis of providing protection against 10 year storms and to have
the work installed for three of the four areas as early as possible.
The Borough was particularly pleased with the fact that the flooding
111
-------
BOROUGH OF YORK - cont...
MODEL SELECTION - cont...
The program pays special attention to the unsteady character of all the
flow with respect to time, the retention capacity of the watershed
surface and sewer networks, the flow regulating effect of backwater and
the inter-connective effect of sewer branchings. Flow retardation and
diminishing processes in a watershed during a storm are simulated by the
surface runoff model which transforms the hyetograph into individual surface
runoff hydrographs. This model enabled us to simulate both the minor
and major systems simultaneously, along with the flow regulators and
detention facilities.
MODELLING
For each catchment area the existing sewer system carrying dry weather
flow plus a percentage of roof flow (the minor system) and the overland
drainage system (the major system) were modelled simultaneously.
Initial storage volumes were estimated by modelling a restrictor located
in the inlet from the road system to the sewer system with an imaginary
free outlet at the obvert of the sewer.
Hydrographs produced at the imaginary outlets were integrated to produce
the required storage volumes. After the preliminary volumes had been
produced, storage tanks were inserted into the model and the overall
system was simulated.
Downstream back water effects were also established by including the
downstream areas in the simulation.
It was found that allowing unrestricted roof water entry into the combined
112
-------
BOROUGH OF YORK - cont...
problems in the most affected areas could be rectified relatively quickly
and not have to wait for the sewer separation programme to be completed,
and at costs within their annual sewer separation budget.
MODEL SELECTION
For this particular application a model would have to possess the
capability of effectively simulating numerous flow dividers, overflow weirs
and many complex interconnections which are characteristic of modification
in older urban areas. Backwater effects from downstream areas where flood
relieve works were not to proceed would also have to be accounted for.
In short, a very sophisticated sewer hydraulic model was required. The
two models considered for this application were the SWMM model with the
Extran option, and the Dorsch HVM model.
Since the adjoining municipality, the City of Toronto, into which our
study areas outlet, was using the Dorsch HVM model,we decided to use the
same, permitting us to utilize the City of Toronto's input and output
results for the backwater effects.
The HVM model is a single event rainfall-runoff computer program package
for detailed hydraulic analysis of existing and planned sewer networks.
The program was developed by the German consulting firm Dorsch Consult
in Munich over a decade ago, but is constantly being modified to meet the
objectives of any particular project where it is to be applied.
The model incorporates the principles of continuity and energy (Saint
Venant's) equations. These equations are transformed into a system of
finite differential equations and modified by various mathematical
manipulations. The overall solving procedure is provided by a repetitive
downstream-upstream calculation scheme which involves further internal
iterations and loops.
113
-------
BOROUGH OF YORK - cont...
MODELLING - cont...
sewer while regulating only surface inflow would result in the system
surcharging. To reduce the surcharge to a level below the lowest
basements, it was found that to achieve a 10 year level of protection
about 50% of the flow from the downspouts would have to be discharged to
the surface.
IMPLEMENTATION
During the course of the detailed design, we found that the plans
showing existing utilities were rather inaccurate and in fact did not
include many of the utilities subsequently found. For this reason we
re-estimated the cost of the works prior to tender call, which resulted
in a final estimate of $987,633.00. Construction commenced in May 1978
and was essentially completed by October. The actual final construction
cost including contingencies will be very close to this amount.
In order to maximize the efficienty of the Hydro-Brake and minimize the
storage volume requirements, a special head-regulating chamber was designed.
This design, as shown on Fig.2 results in the maximum permitted release
rate to be obtained relatively early during the runoff event, with the
result that the required storage volume will be reduced. The chamber
should be located at low points on roadways where runoff accumulates,
or surface runoff should be directed into the chamber via pipes.
Should the depth of water in the tank exceed the water elevation in the
chamber i.e. when inflow into the chamber subsides, the flap gate will
automatically open and water from the tank will enter the chamber until
the static heads equalize.
As simulated in the modelling analysis, 50% of the roof area was assumed
114
-------
BOROUGH OF YORK - cont...
IMPLEMENTATION - cont...
to be disconnected. Legal and physical constraints did not allow us
to proceed with the disconnection, thus regulators are to be placed in
the downspouts which will have the same effect in dampening the peak
flows from the roofs.
PERFORMANCE TO DATE
The first installation in one of the study areas was completed late in
September of 1978. Although rainfall events this year have not been as
intense as those experienced last year, intensities which had previously
caused problems have not produced any flooding complaints.
In order to evaluate the performance of the system as installed, as well
as calibrating the computer model used in the analysis, we have now taken
steps to monitor the system. These results will enable us to verify
or improve the design techniques for future installations.
BRIDLEWOOD COMMUNITY
This area which is about 150 acres, consisting primarily of single family
homes, is located within the Bridlewood community of the Borough of
Scarborough and has been plagued by severe floodings since it was built.
Development of this area commenced about 1962 with storm drainage designed
solely on the basis of a piped system (minor system) with a theoretical
capacity to handle storms expected to occur once every five years.
No provisions were made to prevent flooding when runoff in excess of
sewer capacity would occur. Neither did the design include for an over-
115
-------
BRIDLEWOOD COMMUNITY - cont...
land route (major system) to lead excess water away, but contained
numerous sags on the streets where water ponded. The problem was
further aggravated by the connection of roof drains and foundation drains
by gravity to the storm sewer. Flooding and structural damage to
basements thus occurred from the surcharge in the sewers when the runoff
exceeded the capacity of the storm sewer.
Our basic recommendation was to control the amount of storm water that
would be permitted to enter the existing storm sewer system (minor
system) so that the sewers will not be surcharged during times when
the runoff exceeds the capacity of the sewer. In order to effectively
control storm water entry to the storm sewer system, most downspouts
within the drainage area should be disconnected from the sewer and allowed
to discharge to the surface, utilizing precast concrete splash pads.
In this way, by placing regulators in the catchbasins, effective inlet
control would be obtained.
In order to avoid restrictors with small outlets, we proposed to reduce
the number required by sealing off many of the catchbasins not located
in roadway sags. The restrictors are individually sized to prevent
only flows up to, but not exceeding the full capacity of the particular
sewer segment.
During times when runoff will exceed the capacity of the restricted
catchbasins, the excess runoff will accummulate in the sags until it
overtops the crest and then continue to flow downstream. If ponding
depths at the sags exceed an acceptable level, an inverted siphon
(see Fig. 3) would be installed to transport the excess storm water
downstream.
In this manner a major system was developed for the area, making full
use of detention storage in the roadways, yet assuring that the storm
116
-------
BRIDLEWOOD COMMUNITY - cont...
sewer system would not surcharge. The area fortunately benefits from
having open space locations suitable for additional surface detention,
with the result that only one small underground storage tank is required.
The runoff in excess of the roadway detention will be conveyed to two
low points in the area, from which it will flow through inverted siphons
to a dry pond to be located on lands owned by the Hydro Electric Power
Commission (see Fig. 4).
MODEL SELECTION
The minor system servicing the area consists of a simple tree type
branching system, with no complex hydraulic structures. We based our
initial design on the premise that downstream flood relief works were
also to be initiated, such that the study area would be permitted to
discharge at the gravity capacity of the downstream trunk sewer. The
above two factors, along with the criteria that no storm sewer surcharging
could be tolerated due to the connection of foundation drains, led us to
select the Illinois Urban Drainage Area Simulator (ILLUDAS). Another
capability of the ILLUDAS model which further enforced its suitability
for this application was that this model can effectively simulate down-
spouts discharging to the ground.
The ILLUDAS model is based on the British Road Research Laboratory (RRL)
model developed in the early sixties. Evaluation of the model suitability
for urban areas in the United States was studied by Stall and Terstriep
in the early seventies. Their findings indicated the RRL method
provided an accurate method for predicting runoff from paved areas, but
could not be recommended for all urban areas, unless runoff component
from grassed areas was provided. The ILLUDAS model contains a grassed
runoff component.
Each sub-basin may be discretized to three runoff producing areas;
1) paved areas directly connected to the storm drainage system; 2) grassed
117
-------
BRIDLEWOOD COMMUNITY - cont...
MODEL SELECTION - cont...
areas; 3) supplemental paved areas (paved areas discharging to ground
surface). Hydrographs from the three runoff producing areas are
developed for each sub-basin, combined and then become input into the
drainage network at a particular point and routed downstream.
The model also contains other features which allow the modeller to input
the soil type and antecedent moisture conditions. The model is also
particularly suited for determining detention storage volumes, or outflow
rate to make maximum use of available detention storage. Preliminary
or final conduit sizing may be determined by the model, by executing it
in the design mode. When the model is in the evaluation mode, flows
exceeding the capacity of the conduit are stored at the upstream node.
MODELLING
A number of modifications were made to the ILLUDAS model to suite the
objectives of our particular applications. These changes were made in
consultation with M. L. Terstriep, co-author of the ILLUDAS model.
In order to simulate the effects of the roadway storage at the low
points, the model was adjusted to account for storage in the rising limb
of the hydrograph (see Fig. 5). The model was run in the design mode
with one system representing both the sewer and the roadway system.
Thus hydrographs were developed for the entire drainage system. The
allowable sewer systems flow capacity was then subtrated from the
hydrographs at each particular location, and the resultant peak road flow
was obtained. Flow depths and ponding times for 2, 5, 25 and 50 year
storms were then calculated manually and recorded (see Figs. 6, 7. 8, 9).
In order to simulate the dry pond in the H.E.P.C. lands, the allowable
118
-------
BRIDLEWOOD COMMUNITY - cont...
MODELLING - cont...
flow in the conduit was specified, thus the excess storm water was stored.
With the conventional approach that ILLUDAS uses for storage (see Fig. 5)
the specified discharge rate will be in effect until the storage volume
is dissipated. In our application, the stored water is in off-line
storage and will be released back into the system at a pre-selected release
rate. In order to simulate this, the storage option was modified to allow
for off-line detention storage as shown on Fig. 5.
The versatility and the relative low cost of using the ILLUDAS model allowed
us to perform numerous computer runs for varying conditions and parameters,
such as antecedent moisture conditions and soil classifications. We were
thus readily able to evaluate the sensitivity of the system and arrive
at the most cost-effective solution.
Although the Borough of Scarborough only requires sewers to accommodate
5 year design storms, our recommended solution has been designed to
accommodate 50 year storms. The cost of the works which will be beneficial
to 650 homes has been estimated to be about $400,000.
PARK ROYAL
The Park Royal area within the City of Mississauga, located adjacent to
Metropolitan Toronto, has experienced basement flooding during heavy
rainstorms in the past four years. This area is served by a separate
storm sewer system.
After an extensive study, it was concluded that the flooding problems
basically originated from the fact that the foundation drains for many
119
-------
PARK ROYAL - cont...
of the homes in the area were connected to both storm and sanitary
sewers, thus causing storm water to enter the sanitary system during
times when the surcharge level would reach the level of the foundation
drains. This in turn would cause the sanitary sewers to be overloaded,
resulting in back-ups in basements.
To alleviate the problem, it would be necessary to either disconnect
all foundation drains from the storm sewer system or apply the inlet
control method. Since the first method would not be practical, we
selected the latter.
Similar to the Bridlewood project, the recommended solution also incorporates
regulators in selected catchbasins, with others sealed off. However,
downspouts were left unrestricted.
Open spaces in the area were utilized for detention, by restricting storm
sewer capacity downstream of the detention facilities and diverting the
excess flows along with the road flows to the dry detention basins.
A ten year level of protection was required by the City of Mississauga,
but the system will accommodate higher intensity storms if a significant
portion of the downspouts can be disconnected.
MODEL SELECTION
In order to establish the cause of the flooding, which was not readily
apparent, we had to develop several hypotheses. Each of these was then
checked against actual recorded flooding conditions. To prove the
effect of foundation drains being connected to both storm and sanitary
sewers, a model with surcharge capability was required. The HVM model
was selected for the analysis of the system under existing conditions.
Since the solution will result in elimination of surcharge, the Illudas
model was used in the design of the relief works.
120
-------
PARK ROYAL - cont...
COMPUTER MODELLING
Modelling of the storm and sanitary sewer interconnection for the number
of homes where such connections had been indicated, produced inflow
rates into the sanitary sewer capable of surcharging the sanitary sewer
to the levels that had been reported in the field.
The proposed flood relief works were modelled with the ILLUDAS program
assuming the roadway and sewer flows to be one system. Once hydrographs
had been established, then the allowable storm sewer flows, accounting
for roof downspout contribution, were subtracted resulting in the expected
road flows. The detention storage basins were modelled with the ILLUDAS
modified storage option and storage volumes established.
ESTIMATED COST OF WORKS
The cost of utilizing the inlet control method to solve the flooding
problem in the Park Royal area is estimated to be about $350,000. This
will provide protection for a 10 year design storm for about 1200 homes.
121
-------
ro
paul ihci! associaf.es limited
tomwlling *ngine«"t
)ttv«. Brann>a!cQ. Ontario
ROOF WATER DISCHARGED
TO GROUND
OR FLOW RESTRICTOR
INSTALLED IN
DOWNSPOUT —
STORM WATER
DETENTKDN
TANK
INLET CONTROL METHOD WITH DETENTION
FIG.i
-------
STORM WATER
INFLOW-
STORM WATER
INFLOW
OUTFLOW
(REGULATEDI
HYDROBRAKE
REGULATOR
-R
pout ihcll cmociotes limited
, Onl<
coniitlimg
700 &o£mt
ill/
^/ X/" N/"\y"sx/ N^ 'N/J...->i^...]>vr..'NX>X/r\v
STORM WATER
INFLOW
DETENTION TANK
FIG. 2
123
-------
STREET UMS ^
jul theif ossodotes limiti
, | ^j 700 Dotmoiol Dfivt
PROFILE
INVERTED
FIG. 3
-------
LEGEND
PONDING AREAS
HIGH POINTS
DIVISION BETWEEN
MAJOR SYSTEMS
poul ihcil ossocioles limited
FIG. 4
125
-------
APPLICATION OF STORAGE OPTION IN THE
ILLUDAS MODEL
STORAGE VOLUME SPECIFIED
OR
ALLOWABLE OUTFLOW
RATE SPECIFIED
OUTFLOW
HYDROGRAPH
TIME
CONVENTIONAL ILLUDAS STORAGE OPTION
INFLOW
HYDROGRAPH
SPECIFIED ROAD
STORAGE VOLUME
HYDROGRAPH
MODIFICATION
TIME
FOR ROAD STORAGE
OFF-LINE STORAGE VOLUME
DISCHARGED AT SPECIFIED RATE
ALLOWABLE OUTFLOW
SPECIFIED
OUTFLOW HYDROGRAPH
TIME
MODIFICATION FOR OFF-LINE DETENTION STORAGE
LJD 1 Pou' *^e'' associates limited
__ e£mw>t,ng tngW.pen
L ^lJ 70a Fflimorot D'.»f. B.omolea.Ontario
FIG. 5
126
-------
(ORIGINAL
pcul thell oisociotcs limited
tcotuitirtg trtifnetit
700 tjlfnofal Dn»«.6fomoi*o,Ofliofw
FIG. 6
127
-------
OCPIM or CLOW (MtM%)
PONDlNC Tint
'
/
6.5
26 t
^
^
1 7
,
/
^
nL
.
/
*
26
o
e
£
&
UliAtiS DRT K*l° c
i' POMOM: TIM
/
V ^
Mytytwirf-i.
DfPTH H«H)
5 YEAR STORM
(ORIGINAL DESiGN)
poul llieil aisociolcs limiled
FIG. 7
128
-------
Alt "I
ALT 2
(wilh ulOOf'l*
lonk)
^^
1
V
4 !
»
Ji
i
— 1«-
^•- -•'-*—
^
VX:|
\
=a.oooc
J
(ORIGINAL DESIGN!
paul thet! ossociolcs limiferf
FIG. e
129
-------
DCP1H or now |.»«i>e«l
PONDING OOTM Ucf.«i)
PONCWS IIMC |m.,,U|,,)
PROPOSED SII1ION
|f S|l5J PHY POtlO MI'IH ()„()
|pomuc TI
50 YEAR STOF3M,
(OKIGINAL IXSIGN)
paid Iheil ossociolcs limited
t&fltwJlmg vngtrM-vri
700 ^olmon)l DiKv.fifontoka.Ontefio
FIG, 9
130
-------
ANALYSIS OF DETENTION BASIN SYSTEMS
by
Russell G. Mein
Department of Environmental Engineering Sciences
University of Florida
Gainesville, Florida, 32611
(on leave from
Department of Civil Engineering
Monash University
Clayton, Victoria
Australia 3168
Presented at
SWMM Users Group Meeting
13-14 November 1978
Annapolis, Maryland
131
-------
ANALYSIS OF DETENTION BASIN SYSTEMS
Russell G. Mein
Dept. of Environmental Engineering Sciences, University of Florida, Gaines-
ville, Florida 32611.
(on leave from Dept. of Civil Engineering, Monash University, Melbourne,
Australia 3168)
INTRODUCTION
In recent years there has been an increased interest in the use of detention
basins as a means of reducing peak flows on rivers and streams. This
interest is focused particularity on urban or urbanizing catchments where
the increased volumes and peaks of flood flows, combined with the practice
of building in the flood plain, can create unacceptable risks to both life
and property.
The most common type of basin consists of an embankment placed across the
stream with an uncontrolled conduit outlet through the bottom. The tempor-
ary storage thus created has the effect of both attenuating and delaying
the peak of the flood hydrographs passing through it. An overflow spill-
way set near the top of the embankment is required to safely discharge those
large floods which would otherwise exceed the basin capacity.
Despite the reduction in peak flows due to the construction of a detention
basin it does not inevitably follow that flood damages and levels will be
reduced downstream. It sometimes happens that false confidence in the
protection given by a basin gives rise to increased development in the region
downstream. Consequently, for floods larger than the basin design flood,
damage caused far exceeds that which would have occured had the basin not
been built. In some cases the result is that average annual damages are in
fact increased. Another possibility is that the delay of the flood peak
which results from routing the hydrograph through the basin storage can lead
to an undesirable interaction with downstream tributrary hydrographs (McCuen,
132
-------
1975; Morris, 1976).
This paper briefly reviews the method of design of detention basins and
describes a model suitable for the analysis and simulation of multiple basin
systems. An example application to an urban system is given, although the
method and concepts apply equally to rural catchments.
ESSENTIAL CONCEPTS
The main features of a detention basin are shown in Figure 1. The basin
storage is created by placing an embankment across the stream. The level
of the outlet conduit can be placed so as to provide permanent storage, or
left at the level of the basin floor so that the basin is usually dry;
multiple uses of basins such as for recreation and/or wildlife purposes are
common.
A typical stage-discharge curve for a basin is shown in Figure 2. The initial
portion of the curve is for weir flow over the horizontal pipe inlet changing
at some value of stage into full pipe flow. (The transition point is diffi-
cult to predict, and sometimes special devices are used to induce pipe flow
to occur early.) The pipe flow characteristic in which the discharge
varies only slowly with change in basin water level is a feature of detention
basins and produces a flat-topped outflow hydrograph as shown in Figure 3.
For the basin design flood the basin should just fill to the level of the
overflow spillway (Figure 1).
For floods larger than the design flood part of the outflow is passed over
the spillway. The essential characteristic of this spillway as shown in
Figure 2 is the rapid increase in discharge for small increase in stage. A
standard overflow spillway, or a glory hole spillway operating in the weir
flow range, is suitable. Figure 3 shows the rapid changes in discharge
caused when floods exceed the design basin capacity—these rapid changes can
be a hazard downstream and should be minimized as much as possible by limit-
ing the effective spillway length.
133
-------
Design Storage
Emergency
Spillway
Dead Storage
Figure 1 Schematic diagram of a typ-
ical detention basin showing
essential features.
o
Discharge
Figure 2 Stage-discharge curve for a
detention basin with fea-
tures as in Figure 1.
A
Outflow
Time
(a)
Figure 3 Typical inflow and outflow hydrographs for a detention basin.
(a) for design storm (b) for extreme storm.
134
-------
Three design variables interact with one another in determining basin
dimensions:
(i) recurrence interval of design flood,
(ii) peak outflow discharge through the pipe (which occurs when basin
is full), and
(iii) basin storage capacity.
Once any two of these are chosen, the third one is automatically fixed.
Thus for a given basin size and design flood recurrence interval, the per-
cent reduction in peak flow possible (and therefore the outlet capacity) is
fixed by the requirement that the basin just fill during the passage of the
design flood.
The term "recurrence interval" is used in this paper in connection with
flood peak flows (or levels) downstream of the basin site. For example, a
20 year recurrence interval flow is that flow which is equalled or exceeded
on the average once every 20 years. If no basin is present this definition
presents no problems. If, however, a detention basin is introduced to con-
trol floods up to a 20 year recurrence interval it means that, on the aver-
age, the basin will fill or overflow (over the spillway) once every 20
years and all smaller floods will pass through the pipe outlet. The inflow
flood corresponding to the basin-just-filling condition is not the same
flood which caused 20 year peak flows prior to its construction; it will
have come from a longer duration storm, have more volume, and a lower peak.
That is, the basin inflow flood need not necessarily have a 20 year peak nor
a 20 year flood volume; it is the peak outflow which is the important
parameter.
BASIN DESIGN PROCEDURE
The design procedure for a basin, whether it be a basin in isolation or one
basin of many in a system, involves several trial calculations. This is
because the duration of the critical design storm for the chosen design
recurrence interval is not known in advance; it is required to route design
hydrographs for a range of durations through the basin and to select the
135
-------
"worst" case.
The design calculations are usually for either (i) the basin capacity
required to achieve a f^iven peak flow downstream: here, the design variables
are the overflow spillway level and the diameter and/or number of outlet
pipes, or (ii) the outlet size for a basin whose capacity is fixed (by
topography) ; here the design variables are the diameter and/or number of
outlet pipes. In both of these cases an iterative procedure is necessary.
A similar procedure is required for the sizing of the overflow spillway.
Extreme floods of various durations are routed through the basin for various
spillway sizes to determine some optimum combination of spillway cost vs.
embankment cost for critical storms. The required embankment height is
then determined by adding freeboard to the peak water level reached.
Finally the effect of each basin on the flow peaks downstream should be
investigated. This will determine not only the effectiveness of the
proposed basin but also whether any adverse effects would be caused by its
construction.
Table 1 summarizes the steps required in the final design procedure.
Table 1. SUMMARY OF BASIN DESIGN PROCEDURE.
Basin capacity and outlet sizes to be
determined such that allowable peak
flow downstream is not exceeded.
For capacity fixed.
outlet works.
Design to size
(i) Compute design inflow hydrogr.aph
for a range of storm durations for
the required recurrence interval.
(ii) For each storm in (i) adjust the
diameter and/or number of outlet
pipes such that allowable peak
flow downstream is not exceeded.
Note the maximum water level and
basin capacity required.
(i) Compute design inflow hydrographs
for a range of storm durations for
the required recurrence interval.
(ii) For each storm in (i) adjust the
diameter and/or number of outlet
pipes until the basin just fills
during the passage of the flood
hydrograph.
136
-------
Table 1. SUMMARY OF BASIN DESIGN PROCEDURE (Continued)
Basin capacity and outlet sizes to be
determined such that allowable peak
flow downstream is not exceeded.
For capacity fixed.
outlet works.
Design to size
(iii) Design basin capacity is the
largest required in (ii). This
gives the level of the overflow
spillway. Size of outlet works
is obtained from the same routing
calculation that gave the design
capacity.
(iv) Design overflow spillway dimen-
sions determined by routing ex-
treme floods for a range of dura-
tions through the basin designed
in (iii) and selecting "worst"
case.
(v) Check effects of basin on flows
downstream.
(iii) Size of outflow works is obtain-
ed from worst case in (ii).
(i.e. the largest capacity outflow
required)
(iv) Design overflow spillway dimen-
sions determined by routing
extreme floods for a range of
storm durations through the basin
designed in (iii) and selecting
"worst" case.
(v) Check effects of basin on flows
downstream.
BASIN DESIGN METHODS
The final design of basins linked in series and/or parallel in a multi-
basin systems requires the computation of inflow hydrographs at many points,
the routing of hydrographs through the basins, and the routing of the out-
flow hydrographs downstream (with addition of tributrary hydrographs where
appropriate).
i
A suitable mathematical catchment model which allows for temporal and areal
variation of rainfall, for areal variation of losses, and for calculation
of hydrographs at any point on a, catchment is based on the development
Laurenson model (Laurenson, 1964) described for rural catchments by Mein
et al.(1974). In this raodel the catchment is divided into subcatchments
along watershed lines; the rainfall-excess for each is computed from the
temporal pattern of the nearest recording rain gage and the depth as found
from the isohyetal map of the storm event. Catchment storage effects are
137
-------
simulated by routing flows between subcatchments through simple nonlinear
storage elements. Determination of the parameters for these storages is
discussed elsewhere (Mein et al., 1974; Laurenson and Mein, 1978).
Subsequent work toward the application of the model to urban areas (Mein
and Woodhouse, 1977; Crouch and Mein, 1978) has led to the current version
of the model (Laurenson and Mein, 1978) which is being increasingly used
by the engineering profession in Australia. It is a general model which is
suitable for the estimation of design flood hydrographs from storm data for
rural, urban, or partly urban catchments. A detention basin design sub-
routine is also included and the program runs interactively to allow the
user to design the basin system on-line.
EXAMPLE APPLICATION OF MODEL
Gardiners Creek is situated in the eastern suburbs of Melbourne, Australia,
approximately 12 km east of the,city center. Just below the junction with
2
Scotchmans Creek the catchment area is 64 km and is fully urbanized, being
mainly residential. As can be seen in Figure 4 there are ten detention
basins on the catchment, some in series with one another, others in parallel.
One, Lake Road, uses the upper portion of Blackburn Lake as the detention
basin; the others are normally dry. All were constructed in the period 1963-
1968. Storage sizes vary from 0.05-1.0 (106 m3).
Rainfall is recorded by 15 to 30 standard rainguages and by five pluviometers
throughout the catchment area. Runoff is not measured directly, but the
stage above some dry weather flow level is recorded on charts at each basin.
Rating curves, based on theoretical equations for flow through the conduits
and over the overflow spillways, were provided by the Melbourne and Metro-
politan Board of Works.
After inspection of the rainfall and runoff records three storm events were
chosen for analysis. The May 1974 storm was used for fitting the model; two
others (September 1975, December 1975) were reserved for testing it.
138
-------
• Pluviometer
SCALE
10OO 0
ZkM.
Figure 4 Gardiners Creek catchment in Melbourne (Australia) showing
location of detention basins.
139
-------
Following processing of the charts and rainfall data, the hydrograph corres-
ponding to the stage-hydrograph at each basin was able to be deduced using the
theoretical rating curves. It was then possible to compute the implied losses
for the catchment area of each basin by subtracting the runoff from the
rainfall. Some anomalies immediately became apparent; blockage, or partial
blockage, of at least two basin outlets for some events meant that the
derived outflow hydrograph was incorrect and the deduced volume of outflow
was too high. A slower than expected transition to pipe flow also affected
the rating curves in at least one basin. There was no way of correcting the
hydrographs concerned.
The results of the fit and one of the test runs (results for the other test
storm were similar)are shown in Figures 5 and 6 respectively. It is not
proposed to discuss in detail here the performance of the model for each
basin. Some of the poor agreement between model prediction and the "observed"
hydrograph can be explained by partial blockage of the basin outlet. The
Lerne St. basin was affected for each storm and in turn influenced the inflow
hydrograph to the Glenvalley Road basin downstream. The Middleborough Road
basin outlet was partially clogged with debris'during the May 1974 storm.
The Waverley Road basin developed a vortex which prevented full pipe flow
for at least the May 1974 storm; this in turn affected the inflow to the
Huntingdale Road basin downstream. All things considered, the model was
judged to be performing acceptably within the limits of the data. A point
worth keeping in mind, however, is that the basins provide a large damping
effect which could mask small errors in the generation of the inflow hydro-
graphs .
Once the model has been fitted and tested it can be used to examine the effect
of the detention basins on the catchment outflow hydrograph for any storm.
For example, for the May 1974 storm the basins reduced the peak flow at the
catchment outlet by 18%. Within the catchment the peak reductions achieved
were variable and as high as 70%. It would also be a simple matter to
examine the effect of each basin on flows at any particular point for any
storm. Thus, once the model is fitted and tested, it is a powerful tool for
140
-------
Glenvalley Rd.
- Mason's Rd.
Cornwall St.
Lake St.
0 5 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 35 40
TIME (MRS)
Kinkoro St.
BASIN FUll 10
Mlddleborough Rd.
Waverley Rd.
10 15 20 25 30 35 10 15 20 25 30 35 40 10 15 20 25 30
TIME (MRS)
\ Huntingdon
I Rd.
I
10 15 20 25 30 35
Figure 5 Performance of the model after fitting for May 1974 storm.
Recorded hydrographs are shown by dashed lines.
Glenvalley Rd.
0 0 5 10 15 0 5 10 15 0 5 TO 15 20 D 5 10 15 0 5 10 15 20 25
Middleborough Rd.
10
I Kinkora St.
ic*
E5
Eley Rd.
Waverley Rd.
0 5 10 15
TIME (HRS)
0 5 TO 15 0 5 10 15 20 25 0 5 10 15 20
Huntlngdal*
Rd.
0 5 TO 15 20
Figure 6 Results for test storm (September 1975). Dashed lines are recorded
hydrographs.
141
-------
for analysis of the basin system.
GENERALIZED RESULTS
The model described above was developed for application to any particular
catchment. Having demonstrated that the model is suitable, it can be used
to look at general characteristics of basins in series and in parallel
(Mein and Woodhouse, 1977).
Basins in series
The effect on the outflow hydrograph of a single storage of size S, of two
storages of size S/2 in series, and of three size S/3 in series, were com-
pared for a hypothetical catchment. The results generally agreed with the
theoretical analysis of Wycoff and Singh (1976) and show:
(i) that for a downstream point a single basin reduces the peak discharge
by more than the same storage quantity distributed over two basins in
series, and that two basins are more effective than three. Multiple
basin systems can of course, reduce the peaks over intermediate stream
sections;
(ii) the effect of a detention basin is greatest at its outlet, and diminish-
es fairly rapidly downstream as additional area contributes.
Basins in parallel
McCuen (1974) and Morris (1976) have both reported calculations which showed
that the addition of detention basins in particular cases would have increased
the peak flows in downstream reaches by a few percent. Perhaps this problem
is best illustrated by a hypothetical example. Consider two confluent streams
on which the hydrographs from a storm are shown separately in Figure 7 as
A and B. The constuction of a detention basin on one stream could modify
the hydrograph A to hydrograph A' with a reduced peak but with an extended
tail. The combination of A' and B then would give a higher peak flow at
the stream junction than did the combination of A and B. Thus the detention
basin has increased the peak flows downstream of the junction. [It is worth
142
-------
Time
Figure 7. Showing how a detention basin (output A') can attenuate and delay
an inlet hydrograph A so that it now interacts with hydrograph B on a down-
stream junction.
noting here that catchments which display separated response hydrographs as
depicted by A and B are very uncommon].
Many configurations of single basins and of basins in parallel were tried
in the model in an attempt to define the conditions under which the down-
stream flow peaks could be increased. In only one case did an increase
result, and even then it was only by 2%.
It was concluded from these studies that it is certainly unusual for the
addition of a basin to increase the flood peaks downstream. It is never-
theless possible that a particular basin or combination of basins could have
this effect; thus it is important that multibasin systems be fully analysed
to determine the individual contribution (or otherwise) of each basin to
flood mitigation. With the model described above these checks can easily
be made.
RAPID PRELIMINARY PROCEDURES
Table 1 shows that a large number of routings for various combinations
of storm duration are necessary for basin design at each potential site.
For preliminary studies of single basin systems a number of simple methods
have been proposed.
143
-------
Gulp (1948) proposed the formula:
(Qo/Qi) = 1.25 - 1.5(Vs/Vi) + 0.063
where Qo = peak outflow rate
Qi = peak inflow rate (same units as Qo)
Vs = volume of storage of basin
Vi = volume of inflow hydrograph (same units as Vs)
This formula is applicable for basins with a conduit outlet operating in the
pipe flow range and for which a triangular inflow hydrograph is a satis-
factory approximation. It also shows quite clearly the fixed relationship
between recurrence interval of the design flood (given by Qi and Vi), the
basin volume (Vs), and the peak outflow discharge (Qi), as discussed above.
In the author's experience it gives quite good estimates of basin capacity.
Gould (1968) published a formula based on similar assumptions concerning
the inflow hydrograph but extended the analysis to include the duration of
the design storm. The results were presented in chart form to enable rapid
estimates to be made.
For basins on small catchments rapid procedures may be adequate for final
design. Poertner (1974) recommends use of the Rational formula to compute
inflow hydrographs for basins less than 20 acres (8 ha ) in area; a simple
outflow assumption enables the storage requirement to be quickly determined.
For larger catchments however, the Rational method is not satisfactory to
compute hydrographs for basin design.
For multiple-basin systems the problem is much less suited to simplifying
assumptions. Attempts have .been made (Wycoff and Singh, 1976; Abt and
Grigg, 1978) but have been limited to the case of detention reservoirs in
series. It is probable that a final design model which calculates and
routes hydrographs is necessary for most multiple-basin analysis.
144
-------
CONCLUSION
Design of, and investigation of the effectiveness of, detention basin systems
is a complicated procedure not amenable to analysis by simple procedures
(except perhaps for single basins on small catchments). A distributed model,
such as the one used in this paper, is necessary to compute and route the
hydrographs through the system. With such a tool the examination of the
influence of each basin, or group of basins, on flood flows at any point
is readily performed.
Few generalizations about systems of detention basins can be made. Just two
remarks will be made here based on the use of the model:
(i) A single storage is more effective in reducing peak flows at a
point than is a series of basins with the same combined storage
capacity. The effect of a basin on peak reduction diminishes
downstream as more area contributes.
(ii) It is unusual for a basin or for basins in parallel to cause
higher flow peaks at a point downstream. Nevertheless, each
particular case should be investigated for this effect using a
variety of storms.
ACKNOWLEDGEMENTS
Application of the model to the Gardiners Creek catchment was done by
Mr. M. P. Woodhouse. Data for this catchment were supplied by the Melbourne
and Metropolitan Board of Works.
REFERENCES
1. ABT, S.R. and GRIGG, N.S.,(1978), "An approximate method for sizing
detention reservoirs," Water Resources Bulletin, AWRA, Vol. 14, No. 4,
pp. 956-965.
2. CROUCH, G.I. and MEIN, R.G. (1978), "Application of the Laurenson Runoff-
Routing Model to urban areas", The Institution of Engineers, Australia,
Hydrology Symposium, Camberra, September 1978.
145
-------
3. GULP, M.M. (1948), "The effect of spillway storage on the design of
upstream reservoirs," Agricultural Engineering, Vol. 29, pp 344-346.
4. GOULD, B.S. (1967), "Economical stormwater detention basin design," Aust.
Civil Eng. and Const., March 1967, pp 69-77.
5. LAURENSON, E.M. (1964), " A catchment storage model for runoff routing,"
Journal of Hydrology, Vol. 2, pp 141-163.
6. LAURENSON, E.M. and MEIN, R.G. (1978), "Design Flood Estimation using
runoff routing - User manual,"Dept. of Civil Engineering, Monash University,
Australia, January 1978.
7. McCUEN, R.H. (1974), "A regional approach to stormwater detention,"
Geophysical Research Letters, A.G.U., Vol. 1, No. 7, pp 321-322.
8. McCUEN, R.H. (1975), "Flood runoff from urban areas," Maryland Water
Resources Center, Maryland, Tech. Rpt. No. 33, 70 pp.
9. MEIN, R.G., LAURENSON., E.M. and McMAHON, T.A., (1974), "Simple non-linear
model for flood estimation." Journal of the Hydraulics Division, A.S.C.E.,
Vol. 100, No. HY11, pp 1507-1518.
10. MEIN, R.G. and WOODHOUSE, M.P. (1977), "Design of retarding basin
systems," The Institution of Engineers, Australia, Hydrology Symposium,
Brisbane, June 1977.
11. MORRIS, K.J., (1976), "Mathematical modelling of rivers," The Institution
of Engineers, Australia, Hydrology Symposium, Sydney. 1976, pp 45-49.
12. POERTNER, H.G. (1974), "Practices in the detention of urban stormwater
runoff," Report for Office of Water Resources Research,Amer. Pub. Works.
Assoc. Special Report No. 43, Chicago.
13. WYCOFF, R.L. and SINGH, U.P., (1976), "Prelimary design of small flood
detention reservoirs." Water Resources Bulletin, A.W.R.A., Vol. 12, No. 2,
pp 337-349.
146
-------
CSO FACILITIES PLANNING IN CINCINNATI USING SWMM
(A CASE STUDY)
by Jeffrey D. Sharon, Project Engineer
Havens and Emerson, Inc., Cleveland, Ohio
INTRODUCTION
The U.S. EPA's 201 Facilities Planning criteria for combined sewer
systems require a detailed analysis of overflow volumes and pollutant
loads into the receiving streams. This paper presents a case history of
excerpts of a facility planning effort where the EPA SWMM was used as a
primary tool for examining the system of combined sewers in Cincinnati.
The facilities plan was prepared by Havens and Emerson, Inc. for the
Cincinnati Metropolitan Sewer District. The first discussion presented in
the paper is the calibration of the SWMM, which includes a sensitivity
analysis and plots of SWMM-simulated hydrographs versus observed field data
for three distincly different test catchments. Next, the method used for
simulating the hydraulics of the system's various regulators and overflows
is briefly discussed. And finally, a description of the estimation of
overflow pollutant loads into the receiving streams is presented.
STUDY AREA
The combined sewer system within the study area drains approximately 60,000
acres and serves nearly 500,000 people. The system discharges to three major
watercourses (Mill Creek, Little Miami River, Ohio River) and some smaller
tributaries through nearly 200 overflows. Numerous types of overflows exist
ranging from float operated gates to simple diversion structures with
or without weirs. Intercepted flow is conveyed to either of two major waste-
water treatment plants which discharge effluent into the Ohio River.
147
-------
CALIBRATION OF THE SWMM
Before setting up the model and conducting detailed simulations on
the catchments in the study area, the model had to be calibrated using a
comparison of field gagings with model output. Field measurements were
made on several test catchments chosen to represent the study area. This
included measurements of discharge versus time at the outlets to the chosen
test basins as well as measurements of accumulating rainfall depths versus
time at raingages located in or near each test basin. Runoff hydrographs
and rainfall hyetographs were the results of plots of these data.
The model was initially run using the measured hyetographs along with
a first attempt at input parameters (default values were used whenever
possible). Various input parameters were then adjusted until the predicted
results corresponded closely with the field gagings. Only after the model-
predicted results corresponded closely with field data for various types of
catchments under varying rainfall conditions was the model deemed to be
verified sufficiently for extrapolated use over the entire study area.
The first step in the calibration procedure was the sensitivity
analysis. The sensitivity analysis is a study of the model's response
to the variance of certain parameters which are used as input. The
primary purpose of the sensitivity analysis is to permit the user
to become better acquainted with the model and establish the parameters
to which the prediction of the runoff phenomena is most sensitive. This is
particularly important when certain data are unobtainable or unreliable
since it assists the user in making a decision as to whether he may
148
-------
proceed with simulation or must obtain more precise data. A successful
sensitivity analysis leads to the calibration of the model and its
subsequent use for simulation.
The sensitivity analysis for this study was conducted to confirm
the results obtained in several previous analyses of the SWMM program
and to ensure that such results applied to the ongoing study. Three
previous studies Jj *• J' *- J have concluded concurrently that the percent
imperviousness of a catchment is the most sensitive variable in the process
of calibration. Other parameters were found to be ranked in a somewhat
different order from one report to the next.
Graham et al ranked the parameters most affecting the program in
descending order as follows: watershed imperviousness, pervious area
infiltration rates, watershed length to width ratio, overland flow
slope, pervious area coefficient of roughness, impervious area detention
depths. Other parameters were found to have a negligible effect.
f 21
Wisner et al ' stated that, "When studying the sensitivity of the
SWMM - simulated runoff peak flows, the imperviousness was found to have
the greatest impact, followed by the catchment width. The testing of
the SWMM on several Canadian watersheds indicated that for the remaining
runoff quantity parameters, the default values stored in the model may
be used."
The Ontario Ministry of the Environment's Research Report No.
47^ ' concluded similarly: "The percentage of imperviousness had the
most significant effect on the surface runoff of all the parameters
tested. Both the peak flow and volume of runoff were almost directly
149
-------
proportional to the amount of impervious area. The remaining parameters
were ranked in order of their decreasing influence: width of overland
flow, infiltration rates, retention depths, roughness coefficients and
ground slopes. The SWMM default values for these parameters lie
on the relatively flat portions of the sensitivity curves, and appear to
be reasonable estimates when no actual site measurements exist."
Most of the parameters that can be changed for the sensitivity
analysis are utilized in the Runoff Block. However, a successful cali-
bration must also include a look at Transport Block as well in order to
ensure an accurate portrayal of routing delays through the sewer system
as flows from numerous subcatchments are combined. Consequently, catch-
ments of 3 distinct sizes were analyzed in the calibration procedures.
These include the Grasselli (42.2 acres), Sunset (562 acres), and Bloody
Run (2,347 acres) catchments.
It was the Grasselli Catchment which was used for most of the detailed
sensitivity analysis. Grasselli, a typical residential catchment of
uniform land use, was chosen since it could be subdivided into perfectly
rectangular subcatchments with relatively uniform slope conditions
exhibited by each. This ensured that the characteristic width and
weighted slope calculations would be calculated using the precise
(4)
methodology outlined in the User's Manual and the Short Course
Materials without any estimations. Thus, two of the calibrated
parameters were accurately fixed, thereby simplifying the original sensi-
tivity analysis. Grasselli is also a small watershed which made computer
runs relatively short and inexpensive for the detailed analysis.
150
-------
Figures 1 through 6 present a synopsis of the sensitivity analysis
performed on Grasselli. The July 15, 1976 rainfall was chosen for this
study since it was a rainfall of relatively short duration (minimizing
computer expense) and field measured runoff rates corresponded closely
with the model-predicted flows.
Figure 1 addresses the variability of flow prediction with different
discretization schemes. In the original SWMM run, each subcatchment
discharged directly into Transport Block sewer elements through which
the routing takes place. The second run replaced Transport sewer elements
with Runoff Block gutter/pipe elements as receptacles for subcatchment
overland flow runoff. No appreciable change in routing occurred. In the
third run, the original Transport Block sewer elements were again utilized,
this time in conjunction with Runoff Block gutter/pipes added upstream to
receive subcatchment flows producing a much more discrete simulation. Again,
no appreciable change in the hydrograph occurred. From this, it may be con-
cluded that either the Transport elements or Runoff pipes may be used with the
same degree of accuracy and that further discretization by adding gutter/
pipes upstream of Transport elements is unnecessary. This may not be true,
however, for more complicated catchments.
Figure 2 displays the effects of changing ground surface infiltration
rates. Here, infiltration rates were doubled from a maximum and minimum
infiltration rate of 5.00 inches/hour and .52 inches/hour, respectively,
to a maximum and minimum of 6.00 inches/hour and 1.04 inches/hour,
respectively. Thus, a wide variation in infiltration rates is represented.
Only minor variance occurs at the beginning and end of the hydrograph.
The peak is essentially unchanged. Although infiltration rate variability
151
-------
18
—. 15
*
O
12
ORIGINAL SWMM RUN
GUTTER/PIPES REPLACING
TRANSPORT ELEMENTS
GUTTER/PIPES ADDED UPSTREAM
OF TRANSPORT CLEMENTS
I7OO
taoo
1900
TIME
2OOO 2100
GRASSELLI CATCHMENT
RAINFALL DATE
JULY 15, (976
FIGURE I
SENSITIVITY ANALYSIS
OF SWMM FEATURES
152
-------
16
O
H el-
1700
ORIS1NAL SWMM RUN
- INFILTRATION RATES DOUBLED
1800 I9OO
TIME
ZOOO 2100
SRASSELLI CATCHMENT
RAINFALL DATE
JULY 15, 1976
FIGURE 2
SENSITIVITY ANALYSIS
OF SWMM FEATURES
153
-------
ORIGINAL SWMM RUN
IMPERVIOUS SURFACE
STORAGE CHANGED TO .124
1700
1800 BOO
TIME
200C 2IOO
6RASSELLI CATCHMENT
RAINFALL DATE
JULY 15, 1976
FIGURE 3
SENSITIVITY ANALYSIS
OF SWMM FEATURES
154
-------
ORIGINAL SWMM RUN
IMPERVIOUS RESISTANCE
FACTOR CHANGED TO .02
MANNINGS "N" CHANGED
TO .022
ITOO
800 1900
TIME
2000
28OO
6RASSELLI CATCHMENT
RAINFALL DATE
JULY 15, 1976
FIGURE 4
SENSITIVITY ANALYSIS
OF SWMM FEATURES
155
-------
to.
16
14
IZ -
*
O
ORIGINAL SWMM RUN
10% LESS RAINFALL
1700
1800 1900
TIME
2000
2100
2200
6RASSELLI CATCHMENT
RAINFALL DATE
JULY 15, 1976
FIGURE 5
SENSITIVITY ANALYSIS
OF SWMM FEATURES
156
-------
25
21
12
ORIGINAL SWMM RUN
45% IMPERVIOUSNESS
INSTEAD OF 30%
j_
ITOO
I8OO
I9CX)
TIME
SRASSELL! CATCHMENT
RAINFALL DATE
JULY 13, 1976
2000 2100 2200
FIGURE 6
SENSITIVITY ANALYSIS
OF SWMM FEATURES
157
-------
may have a greater effect in areas characterized by flatter slopes or a
greater pervious percentage, it did not effect this study since the
majority of land in Cincinnati is of an impervious characterization
nearly equal to or greater than Grasselli and is featured by a highly
dissected topography.
In Figure 3, a comparison is made between runs using an impervious
area surface storage of 0.062 inches and doubling it to 0.124 inches.
Here, absolutely no effect on peak runoff is shown and only minimal
variations are apparent at lower flow rates. The variation apparently
only occurs when rainfall accumulations are not great enough to completely
fill all available impervious area surface storage.
It is apparent when looking at Figure 4 that if the impervious area
resistance factor is changed from the program default value of .013 to
.02, no significant change exists in the hydrograph definition, even
though this change represents a significantly rougher overall surface.
When Manning's "n" for the sewer elements were changed, however, from
.015 to .022 the peak flow rate dropped significantly, being greater than
five percent. This is due to a delay in the routing through the sewer
elements downstream to the outlet and would be even more significant
when analyzed on a larger watershed.
Data on Figure 5 shows that rainfall input significantly affects runoff
quantities. Here, a ten percent reduction of rainfall intensities throughout
the duration of the storm results in approximately ten percent less runoff
volume and a reduction of the peak flow rate by ten percent.
158
-------
As was mentioned in the literature, the most significant parameter was
surface imperviousness. Figure 6 shows hydrograph variability when the imper-
vious percentage is increased by fifteen percent from an overall thirty percent
to forty-five percent. Results show a significant change. Peak flow rates
increase from 21 cfs to nearly 27 cfs and volumes also increase significantly.
This verifies the previous findings that the percent imperviousness is a
very significant parameter, if not the most significant in the calibration
procedure.
Ignoring for a moment the characteristic width and weighted slope,
all calibratable parameters in addition to those analyzed in the previously
discussed figures were deemed insignificant. The pervious area resistance
factor would be of less significant value than the impervious area resistance
factor since it controls less flow. Pervious area surface storage would reason-
ably not be any more significant than the impervious area surface storage
by the same manner of reasoning. If the maximum and minimum rates of infil-
tration are not significantly affecting runoff, then it would be safe to
assume that the decay rate of infiltration would also be of minimal consequence.
All other parameters are those which are based on actual measurements such
as catchment acreage, etc. These were, of course, not included in the cali-
bration procedure.
Based on the results of the sensitivity analysis made so far, it
was concluded that as long as correct rainfall data were input, the only
parameter affecting runoff calculations significantly enough to warrant
change in the calibrating procedure was the impervious percentage. This
premise is cognizant of the importance of Manning's roughness coefficient
159 -
-------
for the pipes through which routing occurs. The Manning roughness
coefficient, however, should not normally be subject to wide variation.
This intermediate conclusion, based up to now only on the Grasselli
data, concurs with the results of the cited literature.
The percent impervious values were originally measured directly
from aerial photographs. The area of all impervious surfaces (roofs,
streets, sidewalks, driveways, etc.) were measured and divided by the
total surface area of each subcatchment. Initial results for each of
the rainfalls plotted, plus others that are not displayed in this report,
predicted flow rates consistently higher than observed field measurements
indicated. Basic shapes of the hydrographs were, however, very close to
the shapes of the measured data. This indicated that the discretization
procedure, routing flows through the pipe network, and input rainfall
intensities were realistic. Subsequent runs were made lowering the
impervious percentages. Finally, by lowering impervious values by five
to ten percent, it was found that peaks and volumes correlated.
Comparisons of measured field data and SWMM simulations at Grasselli
are shown in Figures 7, 8, and 9 for rainfall events dated June 18 and
19, 1976, June 24, 1976, and July 27, 1976. The first two plots were
the result of rainfall events lasting over six hours, whereas the last
plot represents the runoff results for a shorter duration rainfall. SWMM
predictions are the result of using default values for all parameters
except the impervious percentage. Routing was achieved through the
Transport Block without any Runoff Block gutter/pipe elements. Correlation
of SWMM results with observed data was excellent.
160
-------
en
M
0>
30
O
?5 1
s sn fc
rn
^8
r-S
3
o
c
30
SWMM RESULTS
FIELD DATA
OIOO
-------
SWMM RESULTS
FIELD DATA
-------
SWMM RESULTS
FIELD DATA
6RASSELLI CATCHMENT
RAINFALL DATE
JULY IT, i»TS
0«90
FIGURE 9
COMPARISON OF SWMM
RESULTS VS. FIELD DATA
163
-------
Next, the Bloody Run and Sunset Catchments were analyzed. The
purpose of testing the model on these catchments was twofold. A further
analysis could be made with respect to the imperviousness calibration
and the model could be checked for the correct procedure for calculating
the characteristic widths and weighted slopes of. subcatchments of a
more complex nature than the subcatchments in Grasselli. In both these
catchments, subcatchments were of a much more irregular shape and non-
uniform slope than those of Grasselli.
Calculations of characteristic widths and weighted slopes strictly
(4)
followed the procedures set forth in-the SWMM User's Manual v and the
Short Course Material . The predicted hydrographs showed a similar
shape to that of the field data except that volumes and flow rates were
somewhat greater. Impervious percentages were then reduced by five
percent and the model predicted much more accurately. Figures 10 and 11
show a sample of the results using the rainfall of June 3, 1976 at Sunset
and the rainfall of September 26, 1976 at Bloody Run.
The model runs on these larger catchments led to two final con-
clusions. The characteristic width and weighted slope calculations,
although variable, may be calculated adequately by precisely following
f4")
the procedures set forth in the SWMM User's Manual and the Short
Course Material . Also, a percent impervious value of approximately
five percent less than the measured value consistently yielded the best
simulation results. This latter phenomenun can probably be best explained
by the argument that, although impervious, certain areas may react as
though they are pervious. For example, runoff from sidewalks and driveways
may, at times, run onto pervious areas instead of into street gutters and
164
-------
SWMM RESULTS
FIELD DATA
noo
1420
I54O
1600
SUNSET CATCHMENT
RAINFALL DATE
JUNE 3, 1976
FIGURE 10
COMPARISON OF SWMM
RESULTS VS. FIELD DATA
-------
320
280
240
SWMM RESULTS
FIELD DATA
203O
2100
BLOODY RUN
RAINFALL DATE
SEPT. 26, 1976
FIGURE II
2130 2200 2230 COMPARISON OF SWMM
RESULTS VS. FIELD DATA
TIME
166
-------
drains; and garage roofs and occasional house roofs may, at times, drain
out onto splash pads which discharge also onto pervious areas.
Based on this premise and actual measurements made from aerial
photos, Figure 12 was devised. It is intended for use as a simplistic
calculation procedure for estimating impervious percentages from dwelling
unit densities. Three curves are displayed in the figure. The results
of numerous measurements from aerial photos were fitted to the curve on
the right. Here, all driveways, walks and garage roofs were included as
impervious. The curve on the left side of the figure represents the
remaining impervious percentage when all driveways, walks, and garage
roofs were excluded. The median of the two curves represents the line
which fits runoff simulations best and is equivalent to assuming that
half of the driveways, walks and garage roofs act as impervious areas
when calculating runoff, while the other half, draining onto pervious
areas act as pervious areas when calculating runoff. The curve is
intended to be used exclusively for residential areas. Direct measurements
must still be made to ascertain impervious percentages from areas
with other types of land use.
Finally, the model was verified based on comparisons of simulated hydro-
graphs with field data for additional rainfall events at the three test catch-
ments using the percent imperviousness plot displayed in Figure 12.
SIMULATION OF THE HYDRAULICS OF OVERFLOW STRUCTURES
Only the most simple types of overflows can be simulated with the
SWMM via flow dividers. These are confined to two situations: (1) where
167
-------
ING UNI
DW
20 40
% IMeeRVIOUSNESS
6O
FIGURE 12
% IMPERVIOUSNESS vs. DWELLING UNITS/ACRE
168
-------
inflow is not diverted until it reaches a specified value and when the
inflow exceeds that value the non-diverted flow remains constant and
surplus flow is diverted, and (2) where a simple weir-type diversion
structure can be modeled as long as flow rate and depth of flow into the
structures can be linearly related. ,
Three general types of overflows were found to exist in the study
area: Type "A", Type "C", and uncontrolled.
The Type "A" overflow is a float and gate controlled
regulator. As inflow increases due to runoff during a
rainfall event, the float rises in an adjacent chamber
connected to the trunk sewer via a "tell tale" pipe. The
float, in turn, lowers the gate as it rises ultimately
shutting of all flows into the interceptor at some pre-
determined level in the trunk sewer. At this point, all
trunk flows overflow into the receiving stream.
The Type "C" overflow is also a float and gate controlled
regulator. In this case, the float is activated by the
intercepted flow after it passes through the gate and
before it is discharged to the branch pipe. As the float
rises, the gate aperture decreases. However, this regulator
is designed so that the gate never completely closes, thereby
allowing a predetermined quantity through the gate regardless
of the rate of flow coming into the regulator via the trunk
sewer. All additional flow is overflowed into the receiving
stream.
Uncontrolled overflows do not utilize a gate and float
mechanism. Flows are diverted into a branch pipe to the
interceptor with or without the implementation of a weir
in the trunk sewer. At some point, as flow increases
overflow begins. The intercepted flow will continue to
increase as long as inflow increases.
It was found that very few of the overflows encountered could be
accurately modeled with the SWMM. : The placement of outlet pipes which
discharge to the interceptors and the frequently encountered situation
where no weir exists presents many situations where even the uncontrolled
169
-------
overflows are too complicated for the SWMM. There were no instances
found where float and gate controlled regulators could be modeled.
Also, numerous circumstances were encountered where overflow devices
operated in conjunction with each other.
Because the facilities planning effort revolves primarily around
the generation of pollutant loads discharged into the receiving streams,
the quantification of overflow volumes is vitally important. Consequently,
approximation via the SWMM subroutines was not considered to be an adequate
representation, and a separate regulator model was developed. The purpose of
this model was to accept SWMM output hydrographs as inputs into each
device and produce specific overflow and interception hyclrographs.
The regulator model that was developed contains two distinct
sections. The first section calculates and produces a table which dis-
plays the characteristics of the overflow device such as is presented
in Table 1. For a series of depth/diameter ratios, influent trunk flows
and corresponding intercepted flows and overflows are produced. This
gives the user a quick picture of the operation of the device. The
second section of the regulator model accepts the influent hydrograph
(generated by the SWMM for this particular study) and produces intercepted
and overflow hydrographs via a step-by-step reading and interpolation
of the graph developed in the first section.
It is important to realistically describe the present conditions
of each overflow device. Physical details from drawings can hypothetically
predict the outflow for inflow hydrographs. However, quite often changes
170
-------
TABLE 1
iMPLE OF REGULATOR
Trunk Flow
cfs
0.01
0.03
0.04
0.12
0.20
0.27
0.34
0.42
0.49
0.56
1.44
2.30
3.67
5.01
8.64
13.12
13.19
23.41
28.43
32.77
33.11
HYDRAULICS TABLE
Intercepted Flow
cfs
0.01
0.03
0.04
0.12
0.20
0.27
0.34
0.42
0.49
0.56
1.44
2.30
3.59
3.62
3.67
3.73
3.78
3.83
3.89
3.94
3.99
, ._. r— . .„.. Overflow
Depth/Diameter cfs cfs cfs
•01 "."A 0.01 0.00
•°2 0.03 0.00
0.00
•°4 0.12 0.12 0.00
•°5 0.20 0.20 0.00
-06 0.27 0.27 0.00
• 07 0.34 Ci T.A. 0.00
0.00
-09 0.49 n.4Q 0.00
0.00
0.00
.20 2.30 2.30 0.00
.25 3.67 3.59 0.08
.30 5.01 3.62 1.40
•40 8.64 3.67 4.96
.50 13.12 3.73 9.40
•60 13.19 3.78 14.41
.70 23.41 3.83 19.56
.80 28.43 3.89 24.54
.90 32.77 3.94 28.83
1.00 33.11 3.99 29.12
171
-------
have been made over the years since the facilities have been built, or
lack of maintenance may create situations quite different from that
which would be hypothesized in the office. Consequently, on-site inspection
of the devices is essential. This would include physical measurements
as well as gagings of influent and effluent flow rates.
A METHOD OF ESTIMATING OVERFLOW LOADS
One of the primary goals of this study was the assessment of combined
sewer overflow loads and a search for cost-effective means of the removal
of these loads from receiving streams. These goals can best be achieved
by examining the time-related response of pollutant loading rates at
combined sewer overflows for various rainfall events. This is accomplished
by developing relationships of pollutant concentrations varying with time.
The multiplication of the pollutant concentrations by the rate of
runoff, which is in the form of a SWMM-generated hydrograph, will yield
the pollutograph, a plot displaying the loading rate of pollutants
versus time, for any give rainfall.
The time and monetary constraints of the study warranted limiting
the number of pollutant parameters to be analyzed. It was decided that
five-day BOD and suspended solids would be the two parameters to be studied.
These two parameters were chosen since they have been used most frequently
and are, therefore, the most familiar. They are indicative of the two
major constituents of water quality namely, solids load and oxygen demand.
172
-------
The SWMM model has the capability of simulating the wash-off of
pollutants which have collected on catchment surfaces since the previous
rainfall event and adding these loads to the sanitary loads already
flowing through the combined sewers. The output of the simulation may
be in the form of pollutographs at any location that is desired through-
out the system. However, this portion of the model has not been well
documented and several reports^ ' ' ' ^ have discussed the difficulties
involved in achieving satisfactory simulation. As discussed by J. Marsalek
: "The SWMM model runoff quality simulations were found to be less
satisfactory. Though the insufficient data prevent drawing any firm con-
clusions, it appears that the quality subroutine is not readily applicable to
all urban catchments. The SWMM quality simulations should be treated with
great caution, particularly if used for a selection of urban runoff
control alternatives, or policy enforcement. It may require another one
or two years of data collection before the SWMM quality subroutine can
be fully evaluated for the feasibility of application on Canadian urban
catchments." The basic reason for the difficulty of simulating runoff
quality by the SWMM model is due to an insufficient data base regarding
the accumulation of surface pollutants which have accumulated since the
previous rainfall. Little data has been gathered to date and approximations
are difficult due to the veritable plethora of parameters affecting this
phenomenon. Other models which simulate runoff quality are based
on the same general input.
Much more data have been collected over recent years regarding
concentrations of pollutants at the outlets of combined sewer systems.
Therefore, it was decided to use this much broader data base to develop
173
-------
plots of pollutant concentrations varying with time and to forego
computer simulation. This approach was developed using a coordinated
effort between a sampling program conducted in three catchments in the
MSD jurisdiction and a literature review.
The sampling program resulted in data being primarily collected at the
three locations - Grasselli, Sunset and Bloody Run. As was previously
mentioned, these three catchments vary considerably in size and land use,
but are characteristic of the study area. Automatic sequential samplers
were employed and set to sample approximately 10 minutes after dry weather
flow rates were exceeded. Samples were taken in 10 minute intervals and flow
rates were obtained simultaneously.
Concentrations versus time were plotted from the field data for
numerous rainfall events at each of the three sampling stations. Averages
of concentrations at each 10 minute interval were calculated and plotted.
Then, a comparison of the average plots from the three gaged areas was
made (see Figures 13 and 14). The abscissa (time coordinate) of each
average plot was first established based on time elapsed since the start
of the rainfall. A second series of plots was also developed in which
the time coordinate was established based on coordinating the time at
which the runoff peak occurred. The second series of plots yielded a
more accurate coordination between the gages. This becomes the logical
conclusion if one considers the variation in sizes of the three water-
sheds (42 to 2,580 acres) and the resulting variation in time elapsed
from the start of the rainfall to the occurrence of the hydrograph peak.
By aligning the abscissas of the three plots with respect to the hydro-
graph peak one is, in fact, adjusting the plots so that the various
174
-------
-J
(Jl
250
200
-50
PEAK
100
TIME (mm) FROM RUNOFF PEAK DISCHARGE
150
FIGURE 13
I
AV6. BOO5 CONCENTRATIONS
(CINN. SAMPLING STATIONS)
-------
IOOO
eoo
60O
40O
0.
VI
3
ZOO
0
-50
LEGEND
— GRASSELU
SUNSET
BLOODY RUN
/
PEAK SO
TIME (mm) FROM PEAK RUNOFF DISCHARGE
100
ISO
FIGURE 14
AVG. S.S. CONCENTRATIONS
(CINN. SAMPLING STATIONS)
-------
phases of runoff concur best. This is considered to be the appropriate
approach since the catchments to which the conclusions of the analysis
are to be applied vary greatly in area, drainage configurations, and a
resultant time of concentration.
Figures 13 and 14 illustrate that concentrations of both BODr and
o
suspended solids are higher during the early part of the runoff and
decrease considerably as runoff continues. During the beginning of a
rainfall, the solids which have accumulated on the surfaces of the
catchment and have been deposited in the drainage elements, are washed
off and scoured resulting in the relatively high concentrations. Even-
tually, most of these pollutants have been washed away and the concentrations
of samples taken as the runoff continues are much lower. This confirms
the already well-documented "first-flush" effect of stormwater runoff.
The plots also illustrate a considerable variation in data due to both
variations in catchment and to rainfall characteristics.
A literature review was conducted to supplement and verify the
results of the sampling program. Literature data, in order to be useful,
had to be able to depict the simultaneous concentrations of pollutants
and flow rates at small time intervals throughout the duration of
runoff for a rainfall event. Small time intervals (preferably 5 to 10
minutes) are necessary in order to establish an accurate portrayal of
the variations in pollutant concentrations throughout the event. A
thorough literature review revealed that published data generally lacks
one or more of the basic ingredients needed for this analysis.
Through communications with Dr. Wayne Huber, University of Florida
Department of Environmental Engineering Services, it was learned that a
177
-------
study was proceeding under contract with the U. S. EPA by the University of
Florida, the purpose of which was to collect and study rainfall-runoff-
quality data that is available throughout the United States and Canada.
It was discovered that the data that was collected and tabulated conformed
to the aforementioned criteria and was available via a computer tape. Upon
receipt of the 9 track tape that was supplied, the data pertaining to com-
bined sewer catchments were printed and tabulated. A description of the
study areas for which data had been supplied revealed that seven catch-
ments were drained by combined sewer systems. Data from all seven of these
catchments were analyzed. The seven catchments include a 134 acre residen-
tial catchment in Lancaster, Pennsylvania, a 27 acre commercial catchment
in Seattle, Washington, and 5 catchments in San Francisco ranging in size
from 168 to 3,400 acres and varying in land use from primarily residential
to primarily industrial.
As with the field data collected in Cincinnati, the data were plotted
and average curves were developed for each catchment. The average curves
were then plotted on one common graph and compared. Again, comparisons us-
ing the time scale (abscissa) relative to the occurrence of the hydrograph
peak yielded closer comparisons than with plots using a time scale based
on time since the start of the rainfall event. This correlation can be
explained by the identical argument that was used for comparing the field
data. The plots of these average curves are shown in Figures 15 and 16.
Examination of these figures reveals that the results were concurrent with
the results of the field data. The first flush is clearly depicted as is
a noticeable scatter from one catchment to the next. Within each catch-
ment, plots also varied considerably with varying characteristics of
rainfall intensities.
178
-------
200
100
LEGEND
LANCASTER, PA.
SEATTLE, WASH
SAN FRANCISCO jCATCHMENTNo.fi)
SAN FRANCISCO ( CATCHMENT No. 7 )
-SO
PEAK 50
TIME ( mm I FROM RUNOFF PEAK DlSCHARCE
IOO
1
ISO
FIGURE 15
AVG. B005 CONCENTRATIONS
( UNIV. OF FLORIDA )
-------
00
o
BOO
600
O
400
200
0
-50
LEGEND
LANCASTER, PA.
SEATTLE, MASH
SAN FRANCISCO (CATCHMENT No.'t 1,2S3I
SAN FRANCISCO (CATCHMENT No. 6)
SAN FRANCISCO (CATCHMENT No 7 )
PEAK 50
TIME (mm ) FROM RUNOFF PEAK DISCHARGE
100
—i
ISO
FIGURE 16
AVG. S.S. CONCENTRATION
( UNIV. OF FLORJDA )
-------
The only distinct deviation between the University of Florida data
and the Cincinnati field data was that concentrations were noticeably
lower at the beginning of the event in the University of Florida data for both
the BOD5 and the suspended solids than those recorded during the occurrence
of the runoff peak. Analysis of the rainfall events for the University of
Florida data revealed that they were generally characterized by a longer
duration and a lower intensity relative to the Cincinnati data, therefore,
having a slower impact on the first-flush of surface pollutants and
storm drain scour. Consequently, samples taken immediately after the
beginning of the rainfall usually did not represent first-flush. Samples
taken in Cincinnati, with rainfall represented by greater intensity,
but shorter durations, almost always represented a first-flush effect with
the first sample taken.
f91
A recent publication by the U.S. EPA discusses quality charac-
teristics of first flushes and extended overflows obtained in Milwaukee,
Wisconsin. Ranges are given for BOD and suspended solids of between
170 to 182 mg/1 and 330 to 848 mg/1, respectively for the first-flush
and between 26 to 53 mg/1 and 113 to 174 mg/1, respectively for extended
overflow. These data were compared with the previous two sources for a
final comparison.
Figures 17 and 18 compare the ranges presented in the U.S. EPA
document with the ranges tabulated from the University of Florida data
and the Cincinnati field data. Both figures illustrate the wide range
of values that were obtained in all three sources. This is particularly
evident when looking at the suspended solids range during the first flush.
The ranges of values for extended overflows are markedly smaller
181
-------
CO
fO
250
LEGEND
CINCINNATI
~~—~Z UNIVERSITY OF FtOHIO*
US EPA
-50
PEAK
SO IOO
TIME I mm I FROM PEAK RUNOFF DISCHARGE
ISO
FIGURE 17
COMPARISON OF RANGES
OF BOD5 DATA
-------
CO
U)
IOOO
LEGEND
CINCINNATI
UNIVERSITY OF FLORIDA
US EPA
-50
PEAK 50
TIME (m,n I FROM PEAK RUNOFF DISCHARGE
100
ISO
FIGURE 18
COMPARISON OF RANGES OF
SUSPENDED SOLIDS DATA
-------
and correlate more closely between each of the sources. However, even
though the ranges are considerable for each source, the same general
pattern is apparent showing a maximum concentration at first-flush,
before and up to the time of peak runoff, with a gradual dilution until
concentrations remain constant throughout the remainder of the time of
runoff.
With the analysis of the data presented thus far, it is evident that
concentrations of BOD and suspended solids are subject to a wide range of
variations depending upon many different parameters. It would take a con-
siderable effort to adequately analyze these data and tabulate the effects
of each in order to develop an accurate pollutograph for every possible con-
dition. Primary parameters that would have to be analyzed include catchment
size, the shape of the drainage configuration, ground slopes, land use, catch
basin details, fallout of dust particles carried via air currents related to
the time elapsed since the last wash-off, rainfall intensities and rainfall
duration. Other parameters, not mentioned here, may also have significant
effects. This is a task that has not been accomplished by anyone to date.
It is, therefore, important to recognize the approximations that have been
made in this analysis when reviewing the conclusions to this water quality
section. However, even though the conclusions are but a rough approxima-
tion, they were based on the most complete data-base available and would
fulfill the objectives of this study to the limits of the present state-of-
the-art.
Since a range of values will not satisfy the requirements for calculat-
ing the pollutographs, a specific concentration at each time interval had
to be developed. The product of the pollutant concentration and the rate
of runoff for each time step will yield the desired pollutograph. The best
184
-------
attempt at determining the final plot of concentration versus time was
achieved by plotting the final averages from the three sources from which
a representative curve was selected. This is shown in Figures 19 and 20.
The resultant representative curve for both BOD and suspended solids
Z5
was drawn from an approximate average of values among the three sources. In
both cases, the first-flush concentrations were assumed to be constant from
the beginning of rainfall until approximately 10 minutes prior to the runoff
peak. These constant values correlate closely with the average value of the
f91
ranges set forth in the U.S. EPA publication^ ' which concurred with the field
data collected in Cincinnati. This was thought to be a better fit than averag-
ing the University of Florida data since the nature of this study is to analyze
runoff from critical rainfall events which would not be of the same nature
as most the events for which data was collected by the University of Florida
(i.e., a greater rainfall intensity at the commencement of the event would
create higher initial pollutant concentrations).
The curve during the period between the first-flush and the extended
overflow was developed as an approximate average of the University of Florida
data and the Cincinnati field data. Minor oscillations were neglected. The
extended overflow was a fit between the three sources and correlates closely
with the average of the ranges presented by the U.S. EPA.
The abscissa of these figures may come under some criticism. Varying
durations of runoff would affect the length of time the first-flush occurs
as well as the period of time elapsed until concentrations remain constant
during extended overflow. However, more detailed data are needed to make
any further analysis. In most cases, the representative curve is intended
to be applied to larger catchments, which would depict the most critical
185
-------
CO
(00
LEGEND
AVERAGE CINCINNATI FIELD DATA
AVERAGE UNIV. OF FLORIDA DATA
AVERAGE US EPA DATA
REPRESENTATIVE CURVE
-50
PEAK 50 100
TIME I mm I FROM RUNOFF PEAK DISCHARGE
ISO
FIGURE 19
REPRESENTATIVE CURVE
FOR B005
-------
00
1000
j,J6END
- AVERAGE CINCINNATI FIELD DATA
— AVERAGE UNIV OF FLORIDA DATA
•• AVERAGE U.S EPA DATA
REPRESENTATIVE CURVE
-SO
PEAK
TIME (mm I FROM RUNOFF PEAK DISCHARGE
FIGURE 20
REPRESENTATIVE CURVE
FOR SUSPENDED SOLIDS
-------
cases, and the time to peak runoff and extended overflow is closely repre-
sented by the time coordinate of the plot. It is certainly within the range
of approximations carried forth in this analysis.
The final product of this analysis was a plot of BOD and suspended
O
solids versus time. It is the result of analyzing the most complete data
base presently available, considering the time and monetary constraints of
the study which place limitations on the amount of work to be done in the
field. The final representative curves were combined with overflow hydro-
graphs at key locations throughout the combined sewer system resulting in
overflow pollutographs (loading rates versus time). The pollutographs
were then used as a final tool for developing total discharges of pollut-
ants from the system and analyzing alternatives available to reduce exces-
sive discharge.
CONCLUSION
The SWMM served as the basic mathematical tool for the analysis of
the Cincinnati combined sewer system. It was used as a simple hydrologic
and hydraulic simulator upon which was built a more complicated hydraulic
analysis of the combined sewer overflow structures and a qualitative assess-
ment based on a literature search and a comprehensive in-field sampling
program. The model calibrated quickly and was verified via repeated com-
parisons of output and field-observed hydrographs. Measured impervious
percentages were altered so as to more accurately depict true runoff situ-
ations. The overall analysis procedure provided the basics for the quanti-
fication of pollutant loads generated in various alternatives leading to
the development of the most cost-effective solution for the Facilities Plan.
188
-------
REFERENCES
1. Graham, P. H., Costello, L.S., and Mallon, H.J., "Estimation of
Imperviousness and Specific Curb Length for Forecasting Stormwater
Quality and Quantity", Journal Water Pollution Control Federation,
Vol. 46, No. 4, April 1974, pp. 717-725.
2. Wisner, P.E., Marsalek, J., Perks, A.R., Belore, H.S., "Interfacing
Urban Runoff Models", ASCE Environmental Engineering Division,
Specialty Conference on Environmental Engineering Research, Develop-
ment and Design, July 20-23, 1975, University of Florida, Gainesville,
Florida.
3. Proctor and Redfern, Ltd. and J. F. MacLaren, Ltd., "Stormwater
Management Model Study, Volume I", Ontario Ministry of the Environ-
ment Research Report No. 47, September, 1976.
4. Huber, et al., University of Florida, "Storm Water Management Model,
User's Manual, Version II", EPA-670/2-75-017. Environmental Protec-
tion Agency, March 1975.
5. Short Course Materials Sponsored by the U.S. EPA in cooperation with
the University of Massachusetts, "Applications of Stormwater Manage-
ment Models", July 28 - August 1, 1975.
6. "Testing of the Stormwater Management Model of the U.S. EPA",
Marsalek, J., Hydraulics and Research Division, Canada Centre for
Inland Waters, Burlington, Ontario. A paper presented at the U.S.
EPA Conference on Modeling and Simulation, Cincinnati, Ohio, April 20-
22, 1976.
7. J. D. Sharon and J. A. Gutzwiller, University of Cincinnati,1972,
"Verification and Testing of the EPA Storm Water Management Model".
8. Wisner, P.E., Roake, A.F., Ashamalla, A.F., "Application of STORM
and SWMM for Assessment of Urban Drainage Alternatives in Canada".
A paper presented at the U.S. EPA Conference on Modeling and Simu-
lation, Cincinnati, Ohio, April 20-22, 1976.
9. Lager, J.H., and Smith, W.G., Metcalf and Eddy, Inc., "Urban Storm-
water Management and Technology: An Assessment", EPA-670/2-74-040,
Environmental Protection Technology Series, December 1974.
189
-------
SWMM USERS GROUP MEETING
(Annapolis, Maryland, November 13-11*, 19780
SWMM USAGE IN FACILITIES PLANNING
Abstract. The USEPA sponsored Stormwater Management Model (SWMM) devel'
oped in 1970 has been used extensively in Facilities Planning to simu-
late rainfall occurrences on urban drainage areas ranging from 5 acres
to 1,600 acres. Resulting runoff hydrographs and pollutographs (show-
ing suspended solids and BOD versus time) have been used to analyze
the effects of rainfall on the quality and quantity of flows into
collection systems, overflows, and flooding.
Case studies are cited which explain the various ways in which
SWMM has been used in engineering studies, designs, and resulting or
intended construction.
Extensive experience with the SWMM under a wide range of condi-
tions has indicated that it is a valuable tool which can be used by
urban communities of virtually every size to predict runoff occur-
rences for design rainfalls and actual rainfalls. Quantification of
the runoff and its effects provide valuable data which can be used
for engineering studies, design., and management decisions.
The authors are G.D. Cole, Senior Engineer, L.W. Varner, Engineer,
and J.W. Shutt, Engineer, Floyd G. Browne and Associates, Limited,
Consulting Engineers - Planners. Marion, Ohio.
190
-------
Introduction
Floyd G. Browne and Associates, Limited, has used the SWMM exten-
sively on a variety of projects ranging from villages which are primar-
ily residential in nature to moderately sized cities with greater than
50,000 people and having substantial industry. SWMM was originally
used during the facilities planning for development of data and eval
uation of alternatives for transporting and treating the overflows
from combined sewer systems. Modeled results have produced valuable
information on the dynamic occurrences resulting in various drainage
systems under selected storm conditions. Updated versions of the
SWMM are constantly used as they become available with each version
normally providing better modeled results. Improvements made in the
SWMM hydraulic calculations (Release II, updated September 1976) have
provided a tool for preliminary design of hydraulic structures. Table
No. 1 lists the combined sewer system projects which this paper will
discuss in detai1.
191
-------
TABLE NO. 1
LOCATION AND DESCRIPTION OF
SYSTEMS ANALYZED BY SWMM
1970 Variation in
Corporate Acres In Acres Size of Individual
System Population" Corporate Limits Modeled Areas, Acres
Lima, Ohio 53,731* *»,236 3,^35 2-172
Bucyrus, Ohio 13,111 2,300 1,638 11-lfSl
Hicksvi1le,
Ohio 3,650 900 708 5-50
Marion, Ohio 38,6^6 2,500 2,Qkk 8-58
*U.S. Department of Commerce, Bureau of the Census. 1970 Census of
Population, Advance Report.
192
-------
Lima, Ohio
Lima, Ohio is located in northwest Ohio approximately 50 miles
southwest of Toledo and has a population of approximately 53,800; it is
divided by the Ottawa River. Approximately 3,600 acres of the City are
served by a combined sanitary and storm drainage system; an additional
3,500 acres on the periphery of the City have separated sanitary sewers
and storm drains.
Problems of overflow from the combined system into the Ottawa
River prompted a 1970 study to determine the necessary pollution a-
batement management necessary to alleviate this condition. In addi-
tion, during times of high river flow, water backed into the combined
system via diversion structures and created a problem.
Because of the extremely high estimated cost of complete separa-
tion; the advance in technology as the result of increased research
and development; and upgrading of stream quality criteria by the Fed-
eral and State authorities the City of Lima officials directed that
the objective of the 1970 study for reducing pollution load in the
Ottawa River be expanded. To be included in the increase was the devel-
opment of a phased program with priority being given to the most cost-
effective elements of the solution to achieve maximum pollution re-
duction based on National goals, including develop an alternative
solution that would be less costly than complete separation.
193
-------
Due to the numerous cross-connections and associated complexity
of the City's combined system it was decided that a computer simula-
tion modeling program would provide the most effective overall analy-
sis of the systems' flow characteristics during any given storm.
Modeling results of real storm occurrences were compared with field-
monitored quantity and quality of flows to establish a correlation be-
tween the model and the actual system operation.
The watershed area for the 1970 study was 3,6^ acres and was di-
vided into subcatchments varying from 2 to 172 acres. Initially, the
entire study area was modeled in a single program; however, St was
decided to divide the drainage area into five main drainage areas
so that the system could be modeled in more detail.
Figure 1 shows the relationship between monitored and computed
results for flow occurring during the 1970 study at the downstream
end of a major drainage area. This rainfall was moderate and major
surcharging was not believed to have occurred; therefore, the Trans-
port Block was able to adequately route the flows and interpretation
to compensate for surcharging was not necessary.
194
-------
CJf
13:00 14:00 15:00
16:00
00
HOUR OF DAY
FIG. \ Flow vs time at Collett - Spring for March 1, 1972 storm - Lima, Ohio
-------
Table No. 2 Is a summary of runoff results from the SWMM runs for
some representative areas in the City. The 1327 acre modeled area was
approximately 63% multi-family residential and 3U industrial. Infil-
tration rates were estimated to vary from 3 iph maximum to 0.5 iph
minimum. Land slopCvaried from 0.2% to 35% and averaged 1.8%.
The street cleaning frequency used was once every ]k days with one pass,
and 10 days were used as the interval prior to the storm during which
accumulated rainfall was less than 1 inch.
TABLE NO. 1
RUNOFF DATA AS MODELED BY SWHM
LIMA, OHIO
Rainfall
Intensity _^ Flow Pollutants
For 1 Hr. Rainfall Percent Percent Percent Suspended Solids Washoff (Ib/acre)
iph c.f. Runoff Infi1. Storage Residential Commerical Industrial
1.23 5,972,880 65.2 32.6 2.2 5^.0 — 104.8
1.6** 7,899,700 71.7 26.7 1.6 55-9 — 68.0
2.00 9,633,800 75-9 22.8 1.3 51.2 — 65.7
i
Generally, the Runoff Block of the SWMM indicated that for a 2 year-1
hour rainfall of 1.23 inches, runoff would be approximately 65 percent-
Suspended solids washed off of the drainage area varied from 105 Ib/acre
for industrial land use to approximately 5** Ib/acre for residential areas.
The process of street sweeping is given adjusted weight in the cal-
culation of suspended solids. Because of the importance of this particu-
lar maintenance item, verification of the SWMM "internal assumptions" re-
garding street sweeping were made.
196
-------
Typical of the design considerations that were investigated in this
study was the selection of a design rainfall. It has long been taken
for granted that the first flush of storm flow from combined sewers con-
tains high concentrations of suspended solids and BOD. However, a search
of the available published data provided virtually no specific design
data on critical rainfalls. Accordingly, the runoff and washout of
solids and BOD were calculated by the SWMM for the Lima system for
numerous standard design rainfalls applicable to Lima. These were
run through the computer program to determine the most critical storm
durations and intensities for Lima.
The calculations confirmed that extra solids and BOD are "flushed"
from the system early in the runoff period by all storms. However, it
appears from the study that the flushing rate by a 30 minute rainfall
is lower than that produced by longer or shorter ones. This means that
more storage will be required to catch the washout from a 30 minute
storm than if ft were washed out by longer or shorter rainfalls. This
phenomenon relates to the time of travel in the sewers; the shorter, more
intense storms do not produce enough runoff to overload the collector
sewers, so the solids are carried out rapidly. The runoff from the
longer, less intense storms can also be carried by the sewers and the
solids are washed out in a normal manner. The 30 minute storm, on the
other hand, produces more runoff than the collector sewers can handle
which causes surface flooding, thereby delaying the washout procedures.
On the basis of this observation and reasoning, the 30 minute storm was
accepted as the most critical for the Lima sewer system.
197
-------
This does not mean that the 30 minute storm would be the most cri-
tical for all sewerage systems; the critical duration of rainfall would
vary with (1) the capacity of the pipes with relationship to the area
served, (2) the nature of the ground surface which affects the time of
concentration and (3) the size of the community.
Most facilities are designed to operate at their maximum capacity
at some predetermined frequency. Since the capacity of storm drainage
facilities is influenced by rainfall, the frequency at which the design
storm will occur is significant. It is not practical to design for the
"ultimate" storm and the five year frequency is the most commonly used
standard for the design of storm for the design of storm drainage facil-
ities. The City of Lima uses the five year storm for the design of
storm drainage facilities when their overload will not case basement
flooding. The five*year frequency 30 minute rainfall for the Lima area
as published in the U.S. Department of Agriculture publication, Number
204, is an intensity of 2.6 inches per hour or a total depth of T.3 inches
and this was used as the "design storm".
This study resulted in the recommendations of a two phase program
for abatement of pollution in the Ottawa River caused by combined sewer-
age and storm water overflows. The Phase I project provides for con-
trolled in-system storage with subsequent treatment of combined flows at
the WWTP, while Phase II provides for detention structures at the over-
flow points.
These phases are explained in more detail as follows:
198
-------
IPhase I
1. Continue; the current program to upgrade the quality of treatment
by the Wastewater treatment plant and increase its capacity in
order /to process controlled combined flow. The necessary capa-
cities would be tertiary treatment for an average daily flow of
18.6 mgd with a peak rate of 33 mgd, and secondary treatment to
53
2. Construct a new 5^-inch intercepting sewer and lift station to
raute 53 mgd of flow to the WWTP.
3- Construct overflow and gate devices to program and control re-
lease of the combined flows to the intercepting sewer, the
river, and the WWTP.
4. Provide a monitoring and automated control system to coordi-
nate overflows, gate devices, the lift station and the WWTP.
Phase I I
1. Install screens and pumps at the overflow points to lift the
flow expected from the design storm at that point.
2. Construct storage structures on the river bank at the overflow
points to hold the pumped flow. Equip the basins with mech-
anisms to remove the settled solids. Include overflow weirs
199
-------
in the storage structures to permit the escape of excess flow
to the river. Provide for returning stored liquid and settled
solids to the intercepting sewer after the storm flows recede.
Figure 2 shows a comparison of the BOD and S;S that are expected to be
delivered to the WWTP through the existing system, a separation program,
the Phase 1 facilities and the Phase II facilities. It was concluded
that the City can reduce the escape of SS and BOD to the Ottawa River to
one-half its present level with Phase I and essentially eliminate all
escapes with Phase II.
-------
EXISTING FACILITIES
To
Treatment
5,682,600
pounds
To
Treatment
6,54-1,200
pounds
To
Treatment
8,64-6,300
pounds
SEPARATION
3,627,300
pounds
To
Treatment
4,090,000
pounds.
To
Treatment
4)988,600
pounds
PHASE ONE PROJECTS
To
Treatment
1,520.200
pounds
pounds
1,167,100
pounds
268,300
pounds
271,000
pounds
PHASE TWO PROJECTS
To
Treatment
9,912,900
pounds
To
Treatment
5,206,800
pounds
FIGURE 2
201
-------
The construction of Phase I was started in 1975- During the period
of construction it became apparent that additional studies of the hy-
draulic effects in the system created by various rainfall intensities
was needed to provide adequate information for automation of the system.
For this analysis the Water Resources Engineers transport block of SWMM
was used and rainfall intensities ranging from 0.3 tph to k.k iph were
analyzed to set the control parameters.
It is anticipated that Phase I will'be completely operational with-
in the next 6 months. As for Phase II, it is presently being reviewed
through the facilities planning process to determine if it is still
cost-effective by today's standards.
202
-------
Bucyrus, Ohio
The City of Bucyrus is located in Crawford County, approximately
twenty miles west of Mansfield. The City had a population of 13,111
persons in 1970 and the estimated current population is approximately
13,300 persons. The
corporate limits encompass approximately 2300 acres.
As part of an NPDES permit complianse requirement, this City has
completed a Facilities Plan, of which an Infiltration/Inflow Analysis
is an integral part.
Alternative techniques for abating the pollution of the Sandusky
River from combined sewer overflows were examined during the facilities
planning stage. Alternatives considered were:
1. Provide separate sanitary and storm drainage systems.
2. Provide on-site treatment facilities to treat excessive
combined flow.
3. Store excessive combined flow in aerated lagoons for sub-
sequent treatment by the main wastewater treatment plant.
k. Construct on-site storage structures to detain the "first-
flush of combined flow for subsequent treatment by the
main wastewater treatment plant.
The final alternative proved to be the most cost-effective technique,
203
-------
During the Facilities Planning stage an earlier version of the
SWMM was used to predict the dynamic occurrences resulting from inflow.
Data from that analysis were used to develop a preliminary design of an
intercepting sewer and detention structure system for combined sewer
1
overflow reduction. The purpose of that effort was to demonstrate the
feasibility of the proposed system.
A detailed analysis of the percent capture of the various annual
rainfall events and, therefore, percent capture of pollutants, assuming
various detention structure capacities was conducted. Then a compari-
son of the cost per pound of pollutant removed by the Intercepting
sewer-detention structure-wastewater treatment plant system was made.
The cost-benefit analysis showed that a detention capacity equivalent
to the runoff resulting from a 0.3-inch rainfall is the most cost-
effect i ve.
It was decided that the intercepting sewer system should be sized
to allow for future expansion of the detention structures if subsequent
monitoring indicates that the receiving stream would benefit from such
an enlargement. A 2 year-1 hour (1.23 inch) rainfall was selected as
the design event for the intercepting sewer systems. This will allow
transport of nearly all combined flow to the detention structure sites,
and will insure economical expansion of the detention structures should
that prove necessary.
204
-------
The existing combined sewer system interceptor operates under sur-
charged conditions and tiHe proposed system is designed to operate under a
s.imilar surcharged condition. However, the SWMM version used for pre-
1 iminary analysis could not account for surcharging so its usefulness
was somewhat limited. With the development of an improved model that
could account for surcharging (Release II, updated September, 1976), it
was decided to model the proposed interceptor-detention structure sys-
''$
tern again.
The preliminary design of the interceptors was based on the peak
1*
flows occurring at each of the 2h overflow locations. Also, these flows
were assumed to be occurring und'er surcharge conditions resulting from
the> maximum water level in the tv/o detention structures. This is ob-
viously a very conservative design since it is not likely that all flow
pealks will occur simultaneously or that they would occur against the
maximum water level in the detention structures. However, there was no
adequate method available to account for all contingencies prior to the
development of SWMM. It is likely that a design based on the "worst case" a
205
-------
described above would have been recommended.
With the advent of SWMM, however, it is possible to design a less
conservative system with the assurance that it will function propeirly
and not introduce additional system flooding problems. SWMM was run
initially with the pipessized using the preliminary design assumptions.
The output (flow through each pipe and ^ater level at each manhol fe for
each time step) was then analyzed for 'compliance with design objectives.
The 0.3-inch rainfall was used to cher;k the detention structure capacity
and to minimize the possibility that no overflows would occur as a
result of that event (0.6-inches/hour for 1/2 hour). The 1.23 ?nr;h-l
hour rainfall was used to check pipei capacities and that no increase In
water levels in the existing collecting sewers as a result of the pro-
posed system. An increased surcharge level in the existing intercepting
sewer was allowed as long as the cjround level was not reached and the
water level in the existing collectors remained unaffected, since this
would simply assure maintenance of maximum flows through the wastewater
treatment plant at all times.
Table No. 3 shows a comparison of preliminary and final sewer sizes
and lengths. Notice the shift to small diameter pipes that re-suited
from our SWMM analysis of this complex interceptor-detention structure
system. The estimated construction cost for the interceptor system was
reeuifed in
••MMiflbBM a savings of 2 5 percent as a result of the use of SWMM
during the design phase of the project.
206
-------
TABLE NO. 3
SEWER DIAMETERS AND LENGTHS
Diameter
(inches)
18
20
30
36
42
48
54
60
72
78
Prel iminary
Length Size Percent of
(ft) Range Total Length
-
350 0-30 10.7
850
1,350
_
3,805 36-60 67.5
370
2,060
1,175 760 21.8
1,280
Fi nal
Length Size Percent of
(ft) Range Total Length
1,125
500 0-30 42.2
2,930
1,075
400
1,810 36-60 57.8
1,410
1,560
76 0.0
-
Total
11,240''
10,810*
"Minor changes in the total length are the result of adjustments in
alignment after route surveys were completed.
Hicksville, Ohio
The Village of Hicksville is located in the southwestern part of
Defiance County, in the northwest corner of Ohio. The population of the
Village is approximately 3&50 and it has a land area of about 900 acres.
207
-------
This Village was modeled with the SWMM as part of the EPA grant
program facilities planning process. The purpose was to analyse the
effectiveness of the existing combined sewer system for transporting
storm runoff and to evaluate overflow detention and treatment alter-
natives.
Hicksvilie's combined sewer system is arranged to carry most of the
runoff south through town toward Mill Creek. In dry weather the waste-
water flows into the major intercepting sewers and to the treatment plant,
In wet weather, flow in excess of the treatment plant capacity is di-
verted to Mill Creek. A special problem at Hicksvtlle is that a large
area of undeveloped land to the west and to the north of the Village
drains over land to Hicksvilie's combined sewer system and through it
to Mill Creek. This problem was caused by the practice of covering over
of Mill Creek as the Village developed. Now, runoff causes overloading
of Hicksvilie's combined sewer system and ponding in several locations
along the western and northern edges of the Village.
Hicksville is primarily residential in nature, however, there is
some significant industrial development. Of 58? acres modeled, k$.Q%
is residential, 13-1% is commercial, }^.k% is industrial, and 22.5% is
undeveloped land. Land slope varied from 0.1% to 3-0% and averaged
Q.3%.
Rainfall intensities of 0.1, 0.3, 0.7 and 1.25 iph for 1 hour dura-
tion were modeled with SWMM to determine the resulting runoff and wash-
off of suspended solids.
208
-------
The data in Table No. 4 indicate that the percentage of runoff in-
creases with increased intensity. Also, for each land use the suspended
solids washoff increases substantially with increases in rainfall inten-
sity. Industrial and commercial areas have higher increases in washoff
with increased rainfall intensity than do residential or undeveloped
lands.
TABLE NO. 4
RUNOFF DATA AS MODELED BY SWMM
mCKSVILLE, QHJO
Rainfall
Intensity Flow Pol lutants __________
For 1 Hr. Rainfall Percent Percent Percent Suspended Solids Washoff (Ib/acre)
t ph c.f. Runoff Infi1. Storage Res id. Comm. Indus. Undev.
0.1 212,270 23-3 55.9 20.8 0.36 1.76 2.32 0.39
0.3 636,832 37-1 55.9 7-0 3.7 16.1 23.9 3-6
0.7 1,485,870 41.1 55.9 3-0 18.1 79.7 116.8 16.7
1.25 2,653,350 47.7 50.6 1.7 37-6 190.0 260.7 51.4
209
-------
Alternatives evaluated based on the SWMM data were sewer separation,
primary treatment and chlorination of the overflows, and detention of the
first flush produced by rainfall for later treatment at the wastewater
treatment plant. The SWMM data for the four rainfalls modeled were
plotted to establish flow and pollutant quantities versus rainfall amount.
Analysis of rainfall records established the annual frequency of rainfall
events of various depths in the Hicksville area. From these two data
sources it was possible to estimate the average annual overflow and
pollutant loading to the receiving stream under each alternative.
Detention of the first flush proved to be most cost-effective.
Additional analysis was used to establish the most cost-effective de-
tention volume. The ratio of the annual cost of the detention facilities
which is composed of both amortization and 0 £ M costs to the pounds of
SS and BOD removed per year (cost-benefit ratio) was used for the evalua--
tion. Table No. 5 shows these ratios for four different detention
volumes. Selection of the cost-effective volume was based upon compari-
son to the wastewater treatment plant removal costs of $0.^5/lb for SS
and $O.Wlb for BOD. A favorable cost-benefit ratio exists for both SS
and BOD up to 110,000 c.f. of detention volume; above that level the BOD
cost-benefit ratio for detention is less than that for the treatment
plant. Therefore, 110,000 c.f. of detention was selected.
21Q
-------
TABLE NO. 5
COST - BENEFIT OF VARIOUS
fjRST- FLlJSH DETENTION VOLUME
Detention Annual
Volume Cost Pojlutant Removal(Ib.) Cost of Removal($/lb)
(cf) ($) S$ BOD SS BOD
85,000 65,000 510,000 200,000 0.13 0.33
110,000 80,000 540,000 210,000 0.15
155,000 100,000 510,000 215,000 0.18 0.47
340,000 205,000 720,000 230,000 0.28 0.89
Marion, Ohio
in
The City of Marion is central lyAMar ion County. U.S. 23 bypasses
the city to the east in a north-south direction, 50 miles north of Col-
umbus in the central part of Ohio. The City is moderately industrialized
and has a population of about 38,700. A new 10.5 mgd wastewater treat-
ment facility intercepting sewer and storm drainage projects have re-
cently been constructed and a grant condition for these facilities was
that the City study and correct the overflow and bypassing problems at
various points in their combined sewer system. SWMM was used as a tool
to assist in the evaluation during that study. Approximately 2,250
acres were studied. Land slopes varied from 0.11 to 1.7% and averaged
211
-------
0.651. Soil infiltration rates were estimated to vary from 3 iph maximum
to approximately 0.5 Iph minimum. Rainfall intensities of 0.2, 0.5, 0.9
and 1.23 iph for a 1 hour duration were used in the SWMM to determine
the resulting runoff and washoff of suspended solids.
Completion of construction of the sewer projects resulted in addi-
tional first flush flows being transported to the Improved wastewater
treatment facility where they receive treatment. When these improvements
are made, modeled overflows from a 2 year-! hour design rainfall indicate
that an improved overflow quality would result which Is comparable to
the quality expected from a storm drainage system If the City's system
functioned as a truely separated system. A summary and comparison of
the overflows expected from a separated system and from the existing
system as being modified with the new projects is given In Table No. 6.
Coliform counts, as modeled by SWMM, in the overflows from the
existing system modified with current construction projects are comparable
to those counts expected in storm drainage. Increased conform counts
in the overflows are primarily a result of organic waste matter which
Is washed into the drainage system.
212
-------
^Marion's system functions as a combined system, the overflow qualities
will have been greatly improved with the new construction projects and
additional treatment of these overflows is not expected to be cost-effec-
tive since the overflow to be treated has been modeled as having a quality
essentially the same as stormwater.
TABLE NO. 6
EXPECTED FROM A 2 YEAR-1 HOUR RAINFALL
OVERFLOW QUANTITIES AND QUALITIES
CITY OF MARION, OHIO
STORM SYSTEM
DRAINAGE AREA
POLLUTANT DISCHARGE
FROM STORM DRAIN
J£ A SEPARATE STORM
DRAIN EXISTED, LBS.
ACRES SS
OVERFLOW FROM EXISTING
SYSTEM MODIFIED WITH CURRENT
CONSTRUCTION PROJECTS
POLLUTANT, LBS. FLOW, GALLONS
SS BOD
COLUMBIA DITCH 1,168 26,900 1,600 26,600 2,000 17,500,000
SILVER ST. &
NILES ST. 8?6 35,400 2,100 37,200 2,800 16,500,000
FOREST LAWN BLVD.
& MT. VERNON AVE. 201 4,900
350
4,900 410 3JOO.OOO
TOTAL
2,245 67,200
4,050 68,700 5,210 37,100,000
213
-------
Cone I us ions
The Storm Water Management Model can be used as a predictive model
for studying various aspects of urban runoff. Both quantity and quali-
ty can be predicted on a efynamic basis and when interpreted properly can
be valuable parameters by which the system can be evaluated.
The accuracy of the model is obviously only as good as the data
which is fed into it as input. Various Studies have been done which
evaluate the input parameters according to their effect on the output.
Using these as guidelines, it Is possible to place the proper emphasis
on the input so that the most precise information is obtained for the
most effectual parameter.
Runoff quantities and qualities for various land uses can vary
widely with the type of development. For example, newer industrial
areas with large green spaces will not contribute as much per acre as
an older industrial area with little or no green space. These parameters
are important to consider for input to the SWMM and each modeled system
should be individually defined with the best parameters possible.
214
-------
References
1. Floyd G. Browne and Associates, Limited. Stormwater Overflow
Study for Lima, Ohio. 1973.
2. Floyd G. Browne and Associates, Limited. Overflow and Bypass
Study for City of Marion, Ohio. 1975.
3- Floyd G. Browne and Associates, Limited. Faci1ities Plan for
Bucyrus, Ohio. 1975.
4. Floyd G. Browne and Associates, Limited. Faci1ities Plan for
Hicksville, Ohio. 1976.
5- Floyd G. Browne and Associates, Limited. Infi1tration/Inflow
Analysis Report for Bluffton, Ohio 1975.
6. Floyd G. Browne and Associates, Limited. Infi1tration/lnflow
Analysis Report for Woodvi1le, Ohio. 1975-
7. Estimation of Imperviousness and specific Curb Length for
Forecasting Sotrmwater Quality and Quantity. Graham, Philip H.,
et. al. Journal WPCF, Vol. 46, No. 4, p. 717. April, 1974.
8. U.S. Environmental Protection Agency, Storm Water Management
Model, Volume I - Final Report. USEPA Report No. 11024 DOC
07/71. July, 1971.
9. Storm Water Management Model, Volume II - Verification and Testing.
215
-------
EPA Report No. 11024 DOC 08/71. August, 1971.
10. Storm Water Management Model, Users' Manual - Version II. EPA
Report No. 670/2-75-017- March, 1975.
11. Storm Water Management Model, Volume IV - Program Listing.
EPA Report No. 1102^4 DOC 10/71. October, 1971.
12. Storm Water Management Model, Refinements, Testing and Decision
Making. Prepared for the City of Lancaster, Pennsylvania, and
the environmental Protection Agency with Federal Grant No. 11023
GSC. Prepared by James P. Heaney, Wayne C. Huber, et.al., Depart-
ment of Environmental Sciences, University of Florida. June, 1973-
13- Storm Water Management Model User's Manual - Version II.
EPA-670/2-75-017- March 1975, updated September 1976.
216
-------
A PRE- AND POST-PROCESSING PROGRAM PACKAGE FOR
THE STORM-WATER MANAGEMENT MODEL
by
William James
Department of Civil Engineering
McMaster University
Hamilton, Ontario
Canada L8S 4L7
phone (416) 527
217
-------
A PRE- AND POST-PROCESSING PROGRAM PACKAGE FOR
THE STORM-WATER MANAGEMENT MODEL
*
by W. James
Abstract: A new package called FASTSWMM has been developed and used by
the author in Sweden and Canada. The basic program in the package is
the standard SWMM program, distributed by the U.S. E.P.A. and is
maintained by updating the basic SWMM modules in accordance with the
latest modifications and corrections issued by the original authors.
This basic program has in no way been modified by the present author.
The processes included in the package are:
a) solicit and accept user-directed initial SWMM input in
conversational free-format mode, usually from a remote terminal
b) solicit and accept design-oriented input data (if any) from
input devices or units as directed by the user for the evolving
design
c) check the validity of the input and return values of certain
dependent parameters to the user terminal
d) convert from metric units to imperial or U.S. units required by
SWMM, if necessary, and reconvert to metric data if required
e) submit the blocks of SWMM in the order required
f) return selected output from the SWMM output file to the user
terminal
g) re-submit the SWMM blocks as directed and return selected
output, and repeat as required
h) save and list input/output files as directed by the user
Steps (e)-(h) are written in simple systems control statements and
should be reasonably easily adapted to different systems. Steps (a)-(d)
are handled by FASTSWMM programs. The package has been carefully
designed to minimize user errors and reduce total design turn-around
times. The paper describes the urban drainage design problem and
FASTSWMM design criteria. The FASTSWMM package evidently considerably
reduces the complexity of SWMM job submission; users are able to focus
on the simple hydrologic, hydraulic and water quality processes
modelled. No knowledge of FORTRAN formats, or systems control language
is required of users. As such it is useful.in an instructional/training
environment and the paper describes such courses given by the author.
Dept. of Civil Engineering, McMaster University, Hamilton, Ontario,
L8S 4L7. Phone (416) 527-6944.
218
-------
INTRODUCTION
Urban drainage problems arise from both the quantity and the
quality of runoff. So far as flood flows are concerned, urban runoff
models are essentially similar to general hydrological models such as
the Stanford Watershed Model (see Figure 1).
In practice there is little difference between flood control
management and urban drainage management - we have to control excess
runoff with minimum damage and cost.
The textbook approach to runoff control has been described by
Linsley in several publications and this introduction is adapted from
his work. Typically the approach requires determination of a flood
frequency curve for the existing drainage condition and a flood
frequency curve for conditions after completion of the proposed runoff
control systems. These curves can be converted into damage frequency
curves by assuming a relation between peak flow and flood damage. The
area under the curves is the average annual damage in the existing
condition and average annual damage after improvements have been made.
The difference between these two damage figures represents the benefits
of the flood control or drainage project and can be compared with the
costs of the project for an economic evaluation. In many urban
situations the damage is little more than nuisance and a decision is
made rather arbitrarily to limit the probability of this nuisance to
some acceptable level [Linsley and Crawford, undated report].
In dealing with pollution from urban storm runoff, the magnitude
and frequency of the pollutant loads should also be known. A
determination should be made of an acceptable level of pollution
frequency given some information on the magnitude of the polluting load.
In other words, . the approach should be similar to that for flow
quantities. Of course, rare storm events are not necessarily
determining; long dry periods may be more significant.
Consequently, it is necessary in urban hydrology to define the
probability of peak flows, and, where storage is being considered, the
volume characteristics of the streamflow as well. Unfortunately, it is
common to define flow frequency by selecting a rain hyetograph and
converting this to a flow hydrograph with a discrete event simulation
model. But the most intense rainfall in a year does not necessarily
produce the maximum peak or the maximum runoff volume [Linsley and
Crawford]. A short duration high-intensity rainfall may produce a very
large peak flow but with a low runoff volume such that it will be
severely reduced by available storage. On the other hand, modest
rainfalls extending over many hours, or even days, may produce a large
volume of runoff which could fill a storage reservoir. The exact effect
will of course depend on the amount of storage and its outlet
capacities.
219
-------
The General Nature of Urban Drainage Design
It is important to discuss the reality of urban drainage design:
it is an improvisation, a provisional lash-up. It is not possible to
design a drainage scheme that works properly, or one that achieves a
truly satisfactory performance. There is no storage facility that has
no unfortunate or undesirable effects. The requirements for good
drainage design are generally irreconcilable, and all designs are in a
sense failures. Compromise implies a degree of failure, for example,
durability vis-a-vis low first-cost.
The influence of economy in drainage is everywhere apparent; the
characteristics of a "functional" design are usually derived from the
requirements of economy and not of use.
The principle ways of reducing costs are:
a) to design and to construct the project as quickly as possible,
b) to use simple construction techniques,
c) to use as few operations as possible,
d) to use as little skill as possible in the work force.
e) to use as little material as possible,
f) to use the cheapest materials available,
g) to use the most easily worked materials,
h) to use the cheapest energy available.
The general design procedure in urban flood control may be thought
of as a series of nested loops as shown in Figure 2 [Ontario MOE, 1977].
The urban drainage designer has limits set upon his choice:
1. The drainage facility being designed must correctly embody the
essential principle or the fundamental use of the facility - a
detention pond, for example, will accumulate fluctuating
quantities of water. We cannot bluff our way around it.
2. The drainage components must be geometrically related as
required in extent and position to each other and to the
overall system. They will not disappear at some future data.
3- The components must be strong enough to resist the required
hydrostatic and erosive forces, and big enough to convey the
flow.
4. Access must be provided, for construction and maintenance.
5. Cost must be acceptable, at least to the majority.
6. Appearance must acceptably accord with the environment.
7. Side effects must be acceptable to those affected.
Hence, drainage design always involves making trial assumptions
based on experience; it is a matter of trial and error and relies
heavily on memory. A totally inexperienced person cannot design a good
drainage scheme, even when using computer models.
Of course computer models are clever devices, but users sometimes
seem to claim that the model will show us the correct and cheapest
answer. However, there is no optimal design. Even if we were able to
220
-------
deliberately pre-determine mathematically the solution that would
actually turn out to be the most economical in long-term practice, the
public would probably still want something cheaper or more striking.
CURRENT URBAN DRAINAGE DESIGN PRACTICE
Fortunately Canadian design practice has been reviewed and
described recently in several publications [MacLaren, J.F., 1975;
Proctor and Redfern et al, 1976; Marsalek, J., 1976; Environment Canada,
1976; Ontario MOE, 1977]. This chapter is an adaptation and
amalgamation of these publications.
The design of storage is now generally agreed to be possible only
through the synthesis of hydrographs (storage is not necessarily
concentrated in a reservoir but may be distributed over different
elements of the watershed, such as roads, parking lots, roofs and
elements of the sewer network).
It has recently become common to consider two extreme drainage
systems in one urban area. One is a minor system consisting of closed
and open conduits and the other is a major system, which is the route
followed by flood or runoff waters when the minor system is inadequate.
Traditional urban drainage studies considered only the minor system.
Nowadays both extremes are analyzed in a drainage project. Typical
minor systems are usually intended to have sufficient capacity to
collect and transport runoff from a storm that might be expected to
occur only once in a 2 to 10 year period. Return periods from 25 to 100
years might be considered for the major system [J.F. MacLaren Ltd.,
1975]. Detention ponds which are part of the minor system are designed
with consideration of possible effects of major floods. The same
methodology for the hydrologic simulation should be applied to both
systems.
SWMM flow simulation has proven to be accurate in a large number of
applications for areas of greatly differing size, land use and sewer
system configuration, and over a range - of meteorological conditions.
Consequently SWMM has been recommended for the simulation of flows,
where required, in drainage planning and design studies, although
particular advantages offered by other models in certain specific
situations should not be overlooked. For example, STORM may be useful
for the screening and comparison of alternatives and for the isolation
of critical events. [Proctor and Redfern, 1976].
In the planning stage of the design, various land-use alternatives,
drainage systems and the resultant pollutional impacts on the receiving
waters are evaluated [Marsalek, 1976]. Typically, at this stage only
limited information describing the watershed is available and,
consequently, a detailed runoff simulation would not be feasible or
appropriate. At the same time, it is important to establish the
probability of occurrence of runoff events of specific magnitudes. The
SWMM model could be applied as a spatially lumped model without a
significant sacrifice in the accuracy of simulation [Environment Canada,
1976]. Reduction of the number of elements in the RUNOFF and TRANSPORT
221
-------
blocks to a minimum results in a considerable saving in data reduction
time and in computer time requirements.
Hydrographs and pollutographs are generated by SWMM for the
different runoff (or overflow) control alternatives. At the completion
of the planning stage, the user has a good indication of the nature of
the runoff or overflow problem in the study area and also has a feeling
for the effectiveness of various runoff/overflow control measures for a
number of critical rainfall/runoff events. The information obtained
during this phase is at a level where the relative effects and
magnitudes are more important than the absolute values required for
subsequent design purposes.
In the next stage of the design, the design of drainage system and
pollution control facilities is carried out, as well as a detailed study
of probable impacts in the receiving waters. Consequently, it is
desirable to produce accurate hydrographs and pollutographs for selected
events using a calibrated, detailed simulation model. At this level,
SWMM would be used at a high level of discretization. Surcharging
becomes very important when analyzing methods to reduce flooding in an
existing sewer system of insufficient capacity, or when evaluating the
response of a drainage system to a storm of a frequency lower than the
design frequency.
At present there are few suitable continuous surcharging simulation
models available for the kind of urban drainage design described in all
of the previous pages. One possibility is to use a tried-and-tested
model such as SWMM in a flexible manner. The development of FASTSWMM is
an attempt to provide rapid and easy use of the SWMM package in this
kind of design environment. It is an attempt to introduce new users to
complicated programs such as SWMM without having to learn software or
programming related material. Hopefully the FASTSWMM philosophy will be
extended to include other proven hydrologic and hydraulic packages.
DESCRIPTION OF THE FASTWMM PROGRAM
FASTSWMM makes it possible to run parts of SWMM from a terminal in
a pseudo-conversational mode. FASTSWMM was originally developed by the
writer in Sweden and known as SWESWMM. The program is presently
available on the CDC Cyber computers at Multiple Access, Toronto and
through the TELENET system to almost all users in North America.
FASTSWMM comprises three parts: PRESWMM, SWMM or ZAPSWMM, and
POSTSWMM (see Figure 3).
The basic processes included in the package are:
a) solicit and accept user-directed basic SWMM input in conversational
free-format mode, usually from a remote terminal
b) solicit and accept design-oriented input data (if any) from input
devices or units as directed during execution by the user
222
-------
c) check the validity of the input and return values of certain
default and dependent parameters to the user terminal
d) convert from metric units to the British Units required by SWMM, if
necessary
e) submit the blocks of SWMM in the order required
f) return selected output from the SWMM output file to the user
terminal, and reconvert to metric data if required
g) calculate simple indicators of the correlation between the
calculated and observed (if any) hydrograph, and printer-plot both
hydrographs.
h) re-submit the SWMM blocks as directed and return selected output,
and repeat as required
i) save and list input/output files as directed by the user.
Steps (a)-(d) are handled by the pre-processor, steps (f)-(g) by the
post-processor. The whole procedure has been carefully designed to
minimize user errors and reduce total design turn-around times. The
FASTSWMM package evidently considerably reduces the difficulties of SWMM
job submission; users focus on the simple hydrologic, hydraulic and
water quality processes modelled. No knowledge of FORTRAN formats,
variable names or systems control language is required of users.
Figure 3 also indicates how the package is set up for a multi-
processor environment: the pre- and post-processor reside on a demand
processing CPU while SWMM resides on a big number-crunching batch
processing CPU. The general philosophy from the point of view of the
user is shown in Figure 4.
SWMM is the main part of the program. FASTSWMM was originally
tested on a stripped down version of the Canadian version of SWMM
(CANSWMM), called ZAPSWMM containing only the flow hydrograph parts of
the RUNOFF block. ZAPSWMM does not contain routines for snowmelt, water
quality or plotting of the results and requires only 48 K..,. words of
core. SWMM or ZAPSWMM is automatically submitted to a central processor
unit as a remote batch job by the package.
PRESWMM works in an interactive mode and prepares the input for
SWMM. PRESWMM makes it possible for the user to enter his input in free
format, and in metric units.
POSTSWMM also works interactively. It returns selected results to
the user and reconverts the data to metric units. It also computes the
linear correlation coefficient between the computed hdyrograph and the
observed hydrograph (if any), and printer plots the two hydrographs.
After the normal log-in procedure, FASTSWMM is activated from the
terminal by the command USE. FASTSWMM.
223
-------
From then on, control passes to the program. The user only has to
enter data as required by the program. However, the user may optionally
take over control by entering a command instead of data.
The following are the valid commands:
BACKSTEP (start this step again)
DEADSTART
END
CHANGE (change an alterable variable)
HELP
In a similar way to the HYMO program (Williams and Hann, 1973) and
the HEC programs, commands are expressed in at least the first 3 columns
of the data line, and columns 4 through 80 are used for numeric data and
keywords. The commands are started in the first space and may be
curtailed after the third letter; only 3 letters are necessary to
identify the command.
The data can be written in any format, but at least one blank space
or comma must be left between data items. A decimal is required for
numbers containing fractions, but not for whole numbers. Keywords can
be written with the data to describe individual data items. When
entering data one should first leave 3 spaces blank (these spaces are
for the commands). Spaces or commas are used to separate data. When
all data corresponding to one prompt are entered, the line ends with a
"carriage return".
For example four of the first few prompts are as follows:
i) IS YOUR SWMM DATA METRIC OR FPS ...
? FPS (or METRIC)
ii) 4. ENTER I/O ASSIGNMENTS ...
? CL (if the runoff block only is to be used)
iii) 5. ENTER SCRATCH TAPE ASSIGNMENTS ...
? _L (necessary for the runoff block)
iv) 6. ENTER BLOCK TO BE CALLED ...
? RUNOFF
When prompts for entering data are printed at the terminal they are
followed by a reference to "CARD" n. The necessary "CARD" descriptions
are the same as in the SWMM documentation.
Where computed flows are to be compared with observed flows the
following should be noted:
a) Observed flows must be stored on a file on the user's account.
224
-------
b) Computed flows and observed flows must be in the same units.
c) It is necessary for the response to the prompt (concerning observed
flows)
AT WHAT LOCATION ...
to specify a location for which flows are being computed by the
program.
To make a run with revised data, the user proceeds as follows:
1) Get the file NNNINPX (NNN = user's initials, X = case number) from
permanent storage (NNINPX is produced by a normal FASTSWM run).
2) Change data using the normal EDIT processor.
3) Enter USE. FASTSWNM as before, but do not specify that a new file
is to be written.
DESIGN CRITERIA FOR INTERACTIVE DIALOGUE
It has been our experience that the design time, including user
transaction time and program turnaround time, are greatly reduced when
the interactive approach to computing is taken [James and Robinson,
1978]. It has been found that concise, unambiguous man-machine dialogue
promotes a reduction in user transaction time, minimizes the potential
for error, increases user confidence and encourages investigation of
different design situations. It was also found that transmitting only
key output to the terminal enhances convergence to a final design.
More significant than computing cost is the cost of the designer's
time. For example, for a batch program that requires one day's effort,
depending on the designer's capability, a savings of about 85% of
turnaround time can be achieved using an interactive package [James and
Zachar, 1974]. The costs of that time difference may be quite
significant to consultant and client alike, especially from remote small
offices.
The importance of making the numerical model work for the user
cannot be stressed too greatly. If the method of. communicating with the
computer is complicated and exacting, or the dialogue ambiguous, the
positive aspects of computing will be nullified. There are three levels
at which attention should be paid to user psychology - the functional
level, the procedural level £nd the syntactical level [Robinson, 1977].
At the functional level, it must be made clear which functions are
to be carried out by the designer and which functions by the computer.
A brief introduction at the beginning of the program is important. In
FASTSWMM the user is informed that he must supply data whenever three
dots (...) appear. He is made aware that the computer will do all the
225
-------
prompting. The user can devote his time to the design problem.
The relative merits of batch and demand processing warrant
discussion. The topic was converted briefly in an earlier report
[Proctor et al, 1976]. In batch processing a run is submitted to the
computer through a device such as a card-reader or keyboard terminal,
and the job is then run and the output returned to the user either
within an hour or so, or at some later time, with no communication
between the user and the program from the time of submission to the
receipt of results. Demand computing is initiated through a
teletype-type terminal, but the program will halt at certain
predetermined points to permit the user to make decisions before the run
is continued. One further term should be noted, i.e. conversational
computing. This is similar to demand computing, but specific prompts
are written into the program in an effort to guide the user in making
his decisions. Whether batch or demand computing should be used depends
on whether or not it is important for the engineer to make decisions,
independent of the computer, in the design function which he is
attempting to carry out.
It is of course normally possible to access SWMM files via keyboard
terminals using existing job control software, i.e. without use of
FASTSWMM. This could permit an engineer to set up a major run in a
remote batch mode and then, having completed a first run, modify certain
input parameters via a keyboard terminal, retain the rest of the data
unchanged, and initiate another run. Many users of SWMM use this method
[Proctor and Redfern, 1976]. Use of FASTSWMM seems particularly
attractive to neophyte users of SWMM. This paper presents the total
urban design problem, in the context of SWMM, so that the reader may
himself decide whether a program such as FASTSWMM would be useful to him
in his circumstances. But there seems to be little doubt that the
philosophy is useful in an instructional/training environment.
INSTRUCTIONAL VALUE OF FASTSWMM
Perhaps the most important attribute of conversational pre- and
post-processors is that their use facilitates rapid assimilation of the
SWMM program without learning effort being expended on variable names or
input formats.
This makes it very easy to provide instructional courses on SWMM
using FASTSWMM. Such a course has now being given at two locations in
Sweden and in Toronto, each course lasting 2, to 3 days.
The main point of the course is to give the participants "hands-on"
experience with a small part of the SWMM program. In other words the
participants are expected to work from actual plans for a real urban
area, discretize the areas themselves, prepare the input data, apply an
actual recorded storm, and run the flow hydrograph part of the SWMM
runoff block many times to gain a "feel" for the model.
Because of the complexity of the SWMM model, it is only possible in
the time available to make a detailed study of hydrograph generation
226
-------
using the runoff block; participants are expected to have a good idea of
the runoff block processes by the end of the course. It is expected
that this experience will help the participants to attempt the other M
or 5 blocks. Use of the FASTSWMM package has proven especially
beneficial when the central processor unit is some distance away.
Completion of the course is deemed to include satisfactory
attendance at the lectures and workshops, reasonable completion of the
quizzes, and submission of a final report summarising achievements in
the workshops.
The course workshop objectives are as follows:
Workshop 1: Select a problem area.
Study maps.
Mark off problem area divide.
Select subcatchments.
Measure areas and lengths.
Abstract data.
Select a design storm.
Enter rough data by means of PRESWMM in the FASTSWMM
package.
Edit data file.
Submit a SWMM job.
Redesign the pipe network QT_
Select another level of discretization .or
Select parameters for sensitivity tests.
Edit the input file.
Submit SWMM jobs.
Write final report.
Reading materials have been prepared and are assigned to the lectures as
shown in Table 1. The contents of the lectures are also described in
that table, and the reference materials are listed at the end of this
paper.
Workshop 2:
Workshop 3:
227
-------
CONCLUSIONS
Most users are faced with a number of realities:
(a) They cannot afford to develop and/or maintain an independent urban
runoff model.
(b) They do not own a computer large enough to run a good urban runoff
model.
(c) They have good telephone access to such computers and can afford a
portable terminal.
(d) They are discouraged by the long learning-time required to become
familiar with good urban runoff models.
(e) They would like to reduce the total design turnaround times
associated with batch computing.
A program called FASTSWMM has been developed to cope with this
environment. The main design principles were that:
(a) It should communicate directly with a standard SWMM.
(b) SWMM must be easily updated locally on receipt of updates from the
original authors.
(c) SWMM should run without long delays because of excessive memory
requirements.
(d) FASTSWMM should operate optionally in SI units but talk to FASTSWMM
in the required FPS units.
(e) FASTSWMM should function on portable slow-speed terminals in a
conversational manner.
(f) It should be easy to include other models such as STORM.
Thus no code has been inserted into SWMM except that required to
make isolated statements transportable (about half-a-dozen statements).
The operations of FASTSWMM are:
1. CONVERSATIONAL INPUT
2. PROVIDE ROUGH DATA CHECKS
3. SI/FPS CONVERSION
4. ARCHIVE INPUT
5. CONNECT INPUT AND OBSERVED FLOWDATA FILES
6. RUN SWMM (& OTHER MODELS)
228
-------
7. ACCESS OUTPUT
8. FPS/SI CONVERSION
9. PRINT SUMMARY OUTPUT
10. PROVIDE SIMPLE OUTPUT ANALYSIS
11. FULL REGULAR BATCH LISTINGS
ACKNOWLEDGEMENTS
The writer spent a sabbatical year (July 1977-June 1978) in Sweden,
the first half at the Institute for Water Resources Research (TVRL) at
Lund Institute of Technology (LTH), and the second with a similar group
at Lulea University. This work would not have been possible without the
help of Gunnar Lindh, Lennart Jonsson and especially Jan Falk, at TVRL,
and of Gunnar Peterson at the LTH Computer Centre. Help from Lars
Bengtsson at Lulea is gratefully acknowledged.
Sabbatical leave from McMaster University is also duly acknowledged
here, and the privilege of being in Sweden for a year.
229
-------
EVAPO-
TRANSPI RATION,
LOSSES
TOPO, PHYSICAL
DATA
PRECIP, TEMP
OTHER DATA
RESERVOIR
ROUTING (IF ANY)
FIGURE 1: A GENERAL HYDROLOGICAL MODEL
(ADAPTED FROM LINSLEY AND CRAWFORD)
DEFINE PROBLEM
- SCOPE, SPACE, TIME, OBJECTIVE
ESTABLISH DECISION AND EVALUATION CRITERIA
-WATER USE, ECONOMIC, SAFETY
DEFINE SYSTEM COMPONENTS
-COLLECT DATA
-BUILD MODELS
FORMULATE ALTERNATIVES
FIND ACCEPTABLE ALTERNATIVES
- APPLY MODELS TO TEST OPTIONS
- APPLY DECISION CRITERIA
IMPLEMENT PLAN
• NEXT LEVEL OF PLANNING, DESIGN OPERATION
FIGURE 2: URBAN DRAINAGE DESIGN SCHEME
(ADAPTED FROM THE M.O.E. DRAINAGE MANUAL)
230
-------
»
POSTSWMM
1
L.
1
Interactive
Submit
Interactive
FIGURE 3: FASTSWMM FILES
USER
[TERMINAL
long distance link
TIME
SHARING
SYSTEM
neophytes
II
FASTSWMM
sophisticates
I
I
I
I
I _
i
SEGMENTED
STANDARD
SWMM
OTHER MODELS
BATCH
LISTINGS
OUTPUT
FILES
FIGURE 4: ACCESS TO FASTSWMM
231
-------
Lecture 1:
Lecture 2:
Lecture 3:
Lecture 4:
Lecture 5:
Lecture 6:
Lecture 7:
TABLE 1
Introduction
Ref: Developing Models Vol. 1
Course outline reviewed. Various hydrological programs
and their inter-relationship. Program development and
program use. Comparison between batch and interactive
computing.
Accessing the'Computer
Ref: Developing Models Vol. 2
Using the terminals. General description of the
operating system. Operation of the demand-processing
system. Editing files at the terminal. Typical FASTSWMM
session.
Introduction to SWMM
Ref: SWMM documentation
General arrangement of the SWMM package. Segmentation of
the program. Sources of information. The EXEC, COMBINE,
RUNOFF, TRANSPORT, STORAGE and RECEIVE blocks.
RUNOFF block flow hvdrographs
Ref: SWMM documentation
Discretization of the catchments. Setting up the
drainage schematization. Surface runoff hydrology.
Infiltration and losses. Surface runoff algorithm.
Routing through the drainage network.
FASTSWMM .job submission
Ref: FASTSWMM Users Guide
The FASTSWMM concept. The input routine. Files
associated with FASTSWMM jobs. Editing the input file.
Repeated job submission. Accessing the complete batch
output listing.
RUNOFF block capabilities
Ref: Water Resources
The Snowmelt algorithm. Quality aspects. Calibration of
the flow hydrographs. Design storms.
Continuous SWMM
Ref: SWMM3 Interim documentation
Continuous SWMM simulation. Snowmelt modeling. The
EXTRAN block.
Last Session: Course Wrao-Up
Ref: All Course Materials
Review of whole course. Assessment of course.
Suggestions for improvements. Suggestions for follow-up
course. Suggestions for further work.
232
-------
REFERENCES
Environment Canada/Ontario Ministry of the Environment, "Storm Water
Management Model Study", Vol. II Users Manual, October, 1976.
* James, W-, "Developing and using Computer Simulation Models of
Hydrologioal Systems", 2 Vol. Report, approx. 200 pp., available from
the author, September 1978 (vol. 1) Nov. 1977 (vol. 2).
James, W., and Zachar, P.P., "Large Interactive Simulation Packages
in Environmental Engineering", 1974 International Conference on
Frontiers in Education, IEEE Conference pub. No. 115, pp. 156-162.
James, W., and Robinson, M., "A conversational program for design of
pipe networks for all climatic regions in Canada", CSCE conference on
Computer Applications in Municipal Engineering, pp. 207-221.
* James, W., and Larsson, R., "FASTSWMM User's Guide", Hamilton,
October 1978.
* James, W., Water Resources Engineering and Design for Ontario
Engineers, McMaster University, Sept. 1978.
Linsley, R.K., and Crawford, N.H., "Continuous Simulation Models in
Urban Hydrology", Unpub. undated report by Hydrocomp. Inc.
Marsalek, J., "Urban Hydrological Modelling and Catchment Research in
Canada", CCIW Tech. Bull. No. 98, June 1976.
MacLaren, James, F. Ltd., "Review of Canadian Design Practice and
Comparison of Urban Hydrologic Models", Environment Canada/Ontario
Ministry of the Environment, Research Report No. 26, Oct. 1975.
Ontario Ministry of the Environment and Environment Canada, "Manual
of Practice on Urban Drainage", Draft No. 3, Mar. 1977-
Proctor and Redfern Ltd., and J.F. MacLaren Ltd., "Storm'water
Management Model Study", Vol. 1, Ontario MOE/EC Research Report No.
m, Sept. 1976.
Robinson, M., "Feasibility of Interactive Design of Water
Distribution Systems", McMaster Univ. M.Eng. Thesis, August 1977-
Williams, J.R., and Hann, R.W., "HYMO Users Manual", U.S. Dept. of
Agric., pub. no. ARS-S-9, May 1973.
* Huber, W., et al, "SWMM Interim Documentation", Nov. 1977.
* Reading materials for the SWMM course.
233
-------
List of Attendees
SWMM Users Group Meeting
November 13-14, 1978
Maqbool Ahmad
8th Floor, Century Place
9803 - 102A Avenue
Edmonton, Alberta
T5J 3A3
William M. Alley
U.S. Geological Survey
Federal Center
Box 25046 Ms 415
Lakewood, Colorado 80255
Mark Alpert
Metcalf & Eddy
1800 K Street, N.W.
Suite 720
Washington, D.C. 20006
James A. Anderson
Urban Science Applications, Inc.
1027 Fisher Building
Detroit, Michigan 48202
John C. Anderson
Gore & Storrie Limited
Suite 700
331 Cooper Street
Ottawa, Ontario
K2P OG5
Albert T. Bain
524 Penn Ayr Road
Camp Hill, Pennsylvania 17011
Tom Barnwell
U.S. EPA
College Station Road
Athens, Georgia 30605
Robert Baumgartner
Raffi Bedrosyan
Borough of North York, Public Storks
5100 Yonge Street
Willowdale, Ontario M2N 5V7
Canada
Dilip Bhargava
O'Brien & Gere Engineers, Inc.
Room 410
Central Operations Facility Bldg.
5000 Overlook Avenue, S.W.
Washington, D.C. 20032
Craig Bishop
James F.McLaren
435 McNicoll Avenue
Willowdale, Ontario
Canada
Clive Bright
c/o Templeton Engineering Co.
966 Waverlet Street
Winnipeg, Manitoba, Canada
R3T 4M5
George Burns
Millar Engineering
520 Gardiner's Road
Kingston, Ontario
Canada
Tas Candaras
Paul Theil Associate
700 Balmoral Drive
Bramalea, Ontario
Judity B. Carberry
356 Dupont Hall
University of Delaware
Newark, Delaware 19711
234
-------
David J. Carleo
c/o O'Brien & Gere Engineers, Inc.
1304 Buckley Road
Syracuse, New York 13221
John Charowsky
Underwood McLellan (1977) Ltd.
1479 Buffalo Place
Winnipeg, Manitoba, Canada
R3T 1L7
Ray Dever
Gannett, Fleming
P.O. Box 1963
Harrisburg, Pennsylvania 17105
Jim DeZolt
N.Y.S. Dept of Environmental
Conservation
50 Wolf Road
Room 324
AlJDany, New York 12233
Sherman R. Ellis
U.S. Geological Survey
Federal Center
Box 25046 MS 415
Lakewood, Colorado 80225
Richard Field
Chief, Storm & Combined Sewer Section
U.S. EPA
Woodbridge Avenue, Building 10
Edison, New Jersey 08817
Richard Gietz
Regional Municipality of
Ottawa - Carlton
222 Queen Street
Ottawa, Ontario
KIP 5Z3
Rodolfo Gutierrez
D.C. Dept Environmental Services
Hydraulic Control Branch
5000 Overlook Avenue, S.W.
Washington, D.C. 20032
John Haapala
Boeing Computer Services
565 Andover Park West
Tukwila, Washington 98188
Bob Hawkins
Ten - Ech
515 Park Avenue
Louisville, Kentucky 40208
Jay Huang
Hazen and Sawyer
360 Lexington Avenue
N.Y.C., New York 10017
Wayne Huber
Dept of Environmental Engineer
University of Florida
Gainesville, Florida 32611
Allen J. Ikalainen
Systems Analysis Branch
EPA Region 1
JFK Federal Building
Boston, Massachusetts 02203
Dr. Bill James
McMaster University
Dept of Civil Engineering
Hamilton, Ontario
L8S 4L7
Robert C. Johanson
Hydrocamp Inc.
1502 Page Mill Road
Palo Alto 94304
David Jones
c/o O'Brien & Gere Engineers Inc.
1304 Buckley Road
Syracuse, New York 13221
Jochen Kuhner
Meta Systems
10 Holworthy
Cambridge, Massachusetts 02138
235
-------
F. Ivan Lorant
c/o M.M. Dillon Ltd.
50 Holly Street
Tornoto, Canada
Dr. Wu-Seng Lung
Wapora, Inc.
6900 Wisconsin Avenue, N.W.
Washington, D.C. 20015
William G. Lynard
Metcalf & Eddy, Inc.
1029 Corporation Way
Palo Alto, California 94303
David J. Lystrcm
U.S. Geological Survey
Mail Stop 415
Reston, Virginia 22092
Barry MacBride
James F. MacLaren Ltd.
1240 Portage Avenue
Winnipeg Monitoba, Canada
R2M 4L3
Mario Marques
Federal Highway Admin.
400 7th Street, S.W.
Washington, D.C. 20590
Richard McKuen
Dept of Civil Engineering
University of Maryland
College Park, Maryland 20742
M.B. McPherson
ASCE, 23 Watson Street
Marblehead, Massachusetts 01945
Russell Mein Dept of Env. Engineer
University of Florida
Gainesville, Florida 32611
Zdenek Novak
Ontario Ministry of the Environment
135 St. Clair, West
Tornoto, Ontario
Emmet Owens
Stearns and Wheler
10 Albany Street
Cazenovia, New York 13035
Hal Pascoe
City of Ottawa
City Hall
111 Sussex Drive
Ottawa, Ontario, Canada
KIN 5A1
Alan Perks
Proctor & Redfern
75 Eglinton Avenue, East
Tornoto, Ontario
M4P 1H3
Herbert C. Preul
639 Baldwin Hall; #71
University of Cincinnati
Cincinnati, Ohio 45221
Robert M. Ragan
Dept. of Civil Engineering
University of Maryland
College Park, Maryland 20740
Steve Roesch
Annapolis Field Office/EPA
Riva Road
Annapolis, Maryland 21401
Larry A. Roesner
Water Resources Engineers
8001 Forbes Place
Springfield, Virginia 22151
William H. Sammons
Soil Conser. Service, USDA
Federal Center Bldg. #1 Rm 269
Hyattsville, Maryland 20782
236
-------
Patty Sedney Bryan Weber
Non-Point Sources Branch, EPA Underwood McLellan (1977) Ltd.
(WH-554) 1479 Buffalo Place
Washington, D.C. 20460 Winnipeg, Manitoba, Canada
R3T 1L7
Jeff Sharon
Havens & Emerson, Inc. Charlie Woo
1300 E. 9th Street, Bond Ct. Bldg. FHWA - DOT
Cleveland, Ohio 44114 2300 Piitmit Drive #712
Falls Church, Virginia 22043
John Siitmonds
City of Ottawa
City Hall
111 Sussex Drive
Ottawa, Ontario, Canada
KIN 5A1
David W. Smith
Gore & Storrie Limited
Suite 700
331 Cooper Street
Ottawa, Ontario
K2P OG5
Reirihard Sprenger
c/o Tenpleton Engineering Co.
966 Waverley Street
Winnipeg, Manitoba
R3T 4M5
Catherine L. Struss
Clark, Dietz and Assoc. Engineers
211 N. Race
Urbana, Illinois 61801
Paul Theil
Paul Theil Associate
700 Balmoral Drive
Bramalea, Ontario
Larry W. Varner
Floyd G. Browne & Assoc., Ltd.
181 S. Main St.
P.O. Box 587
Marion, Ohio 43302
237
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
2.
3. RECIPIENT'S ACCESS!ON> NO.
4. TITLE AND SUBTITLE
Proceedings
Stonnwater Management Model (SWMM)
User's Group Meeting, November 13-14, 1978
5. REPORT DATE
November 1978
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Mi ted by Harry C. Torno
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Research and Development (RD-682)
Washington, D.C. 20460
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
N/A
12. SPONSORING AGENCY NAME AND ADDRESS
same
13. TYPE OF REPORT AND PERIOD COVERED
PE OF R
Final
14. SPONSORING AGENCY CODE
EPA 600/2
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report includes ten papers, on various model - related topics, presented
at the semi-annual joint U.S. - Canadian Stormwater Management Model (SWMM) Users
Group Meeting, held 13-14 November 1978 in Annapolis, Maryland.
Topics covered include a description of the new SWMM Storage/Treatment program,
several papers on stormwater retention, a methodology for evaluating agricultural
BMP's, two papers on model applications in facility planning, and other papers on
model applications.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS C. COS AT I Field/Group
Mathematical Models
Combined Sewers
Runoff
Hydrology
Urban Hydrologic
Models
Urban Hydrology
Combined Sewer Overflows
Stormwater Runoff
13 B
3. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 222O-1 (9-73)
*U& GOVERNMENT PRINTING OFFICE: 1979 620-007/3744 1-3
------- |