United States
Environmental Protection
Agency
Environmental Research
Laboratory
Athens GA 30605
EPA-600/9-80-064
December 1980
Research and Development
Proceedings
Stormwater
Management Model
(SWMM)
Users Group Meeting
June 19-20, 1980
-------
EPA 600/9-80-064
December 1980
PROCEEDINGS
STORMWATER MANAGEMENT MODEL (SWMM)
USERS GROUP MEETING
19-20 June 1980
Project Officer
Harry C. Torno
Science Advisory Board
U.S. Environmental Protection Agency
Washington, D.C. 20460
ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
ATHENS, GEORGIA 30613
-------
DISCLAIMER
This report has been reviewed by the Environmental Research Laboratory,
U.S. Environmental Protection Agency, Athens, Georgia, and approved for
publication. Approval does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
-------
FOREMORD
A major function of the research and development programs of the U.S.
Environmental Protection Agency is to effectively and expeditiously transfer
technology developed by those programs to the user community. 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 by EPA and 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.
David W. Duttweiler
Director
Environmental Research Laboratory
Athens, Georgia
iii
-------
ABSTRACT
This report includes eleven papers on topics related to the development
and application of computer-based mathematical models for water quantity and
quality management presented at the semi-annual meeting of the Joint U.S.-
Canadian Stormwater Management Model (SWMM) Users Group held 19-20 June 1980
in Tronto, Ontario, Canada.
Topics covered include descriptions of three urban runoff models; a
discussion of the use of the Soil Conservation Service TR-55 model; applica-
tions of several models in planning, analysis and design; and a discussion
of kinematic design storms.
-------
CONTENTS
Page
Foreword iii
Abstract iv
COMBINED SEWER SYSTEM ANALYSIS USING STORM AND SWMM FOR THE CITY 1
OF CORNWALL
J.C. Anderson
CONSIDERATIONS REGARDING THE APPLICATION OF SCS TR-55 PROCEDURES 23
FOR RUNOFF COMPUTATIONS
P. Wisner, S. Gupta, and A. Kassem
A SIMPLIFIED STORMWATER QUANTITY AND QUALITY MODEL 45
S. Sarikelle and Y.T. Chuang
METHODOLOGY FOR 'LUMPED1 SWMM MODELLING 64
M. Ahmad
CHARACTERIZATION, MAGNITUDE AND IMPACT OF URBAN RUNOFF IN THE GRAND 80
RIVER BASIN
S.W. Singer and S.K. So
DEVELOPMENT OF AN URBAN HIGHWAY STORM DRAINAGE MODEL BASED ON SWMM 121
R.J. Dever and L.A. Roesner
KINEMATIC DESIGN STORMS INCORPORATING SPATIAL AND TIME AVERAGING 133
W. James and J.J. Drake
HYDROGRAPH SYNTHESIS BY THE HNV-SBUH METHOD UTILIZING A PROGRAMMABLE 150
CALCULATOR
B.L. Golding
DETENTION LAKE APPLICATION IN MASTER DRAINAQE PLANNING 178
A.T.K. Fok and S.H, Tan
ALTERNATIVE URBAN FLOOD RELIEF MEASURES: A CASE STUDY, CITY OF REGINA, 200
SASKATCHEWAN, CANADA
A.M. Candaras
A LONG-TERM DATA BASE FOR THE INVESTIGATION OF URBAN RUNOFF POLLUTION 212
W.F, Getger
List of Attendees 234
-------
COMBINED SEWER SYSTEM ANALYSIS
USING STORM AND SWMM
FOR
THE CITY OF CORNWALL
by
John C. Anderson, P.Eng.
Gore & Storrie Limited
Suite 700
331 Cooper Street
Ottawa, Ontario, Canada
INTRODUCTION
The City of Cornwall, located on the shore of the St. Lawrence River, approximately
75 miles upstream of the City of Montreal, comprises a population of approximately
46,000 people and as such, is one of the major urban centres along the river
throughout its length from Lake Ontario to the Atlantic Ocean.
The developed area of the City comprises a total area of approximately 4250 acres.
Of this, some 1785 acres are serviced with combined sewers.
The age of the sewer system is quite variable and some sewers date back almost
100 years. The orientation of the original sewer system was such that all sewers
discharged directly, without treatment, to the St. Lawrence River. During the
period of the early 1960's however, these sewers were intercepted and flows
-------
directed to a new treatment facility, providing primary treatment before discharge
of the treated effluent to the river.
The layout of the sewer system and the combined and separate sewer areas are
shown on Figure No. 1.
The flow from the various combined sewers are intercepted through regulator
chambers, designed to intercept 2i times dry-weather flow. All flows exceeding
this amount are discharged directly through the overflow sewers to the river. A
total of 8 points of overflow are provided.
PROBLEM DEFINITION
The sewage treatment facilities are designed to provide treatment for a total
flow of 8.25 million imperial gallons per day. When the plant was originally
commissioned, average flow rates to the plant were in the order of approximately
7 million imperial gallons per day. However, these flows increased fairly
rapidly over the next few years along with the growth in the service area, to a
present day average daily flow rate of approximately 11 million imperial gallons
per day and a dry-weather flow rate of approximately 9 million imperial gallons
per day. As such, the plant is presently operating in an overloaded condition.
Although the assimilative capacity of the St. Lawrence River as a receiving body
is relatively large, high bacteria levels have been measured at various swimming
beaches along the river shore, downstream from the City.
In view of these problems, it was apparent that some plant expansion was required.
Beacuse of the nature of the system, however, the problem became:
2
-------
-------
What size of plant expansion would best suit both present
day and some future projected need
What additional control techniques could be implemented
to manage the system and minimize the impact of combined
sewer overflow loadings to the receiving stream.
SYSTEM ANALYSIS
The approach to the analysis was to use model simulation techniques to assist in
analysis and in this respect, the Storage Treatment Overflow Model "STORM" was
employed as the primary tool.
This model was used to:
Screen a number of control alternatives
Develop statistics with respect to system operation for
various levels of both storage and treatment capacities.
Subsequently, the Storm Water Management Model "SWMM" was used to assess
selected control alternatives under individual event operation.
The first step in the application of the models was to undertake some calibration
and subsequent testing of the calibration, both with respect to quantity and
quality of flow. For this purpose, a number of typical sub-catchment watershed
areas were identified both large and small, and data was collected over the
period of approximately 1 summer season. In addition, both flow and quality data
were available from the water pollution control plant records for assessment of
both dry weather and average daily flow conditions.
-------
Quantity and Quality data required is summarized as follows
QUANTITY
1
Combined
Sewers
1
i
D.W.F. Storm
Flow
San. Infil.
1
Sanitary
Sewers
1
I
San.
Flow
Dry
Weather
i
Infil.
1
Wet
Weather
QUALITY
Dry Weather
Flow
Storm
Flow
For the sanitary and infiltration components of dry-weather flow, rates were
calculated from the measured hydrographs, using essentially the minimum night
time flow condition, when the majority of the flow is attributed, particularly in
the smaller areas, to infiltration. Average sanitary components were also
compared to water consumption and favourable correlation was obtained.
Specific major industrial sources were identified separately, along with
associated flow rates and sewage strengths.
Table No. 1 following is a comparison of both simulated and measured flow and
quality data, considering:
-------
- Dry Weather Flow
- Average Daily Flow
- Annual Overflow Volume
- Sewage Strengths
Input to the Storm Model included both quantity and quality information relating
to dry-weather flow for domestic and industrial purposes, as well as infiltration
and extraneous flows. The model then subsequently produced predictions of average
daily flow rates, along with associated quality parameters and annual overflow
volumes.
It is noted on examination of Table No. 1, that both the predicted average
daily flow volumes and associated pollutant concentrations are within a reason-
able range of measured values. It is also noted, however, that the volume of
overflow predicted is significantly lower than the volume of overflow actually
measured.
Further examination revealed this to be attributed to two problems:
The first problem related to the predicted volumes of overflow for spring runoff
conditions. The calibration of the Storm Model was a relatively straight-
forward process for summertime occurrences and for this purpose, calibration was
undertaken by adjusting the average runoff coefficients for both pervious and
impervious areas, until a reasonable fit was obtained. Subsequent testing of the
calibration on additional storm events indicated good and reasonable correlation.
The results of testing on 2 separate storm events are shown on Figure Nos. 2 and 3.
Because the operation of the system was investigated on an average annual basis,
as well as seasonal variations, and in view of the geographic location of the
6
-------
City of Cornwall, it was necessary to utilize also, the snowmelt computation
capabilities of the model. However, because of frozen ground conditions during
the spring runoff season, it was necessary to consider increased runoff
coefficients during this period of time, over and above those values applied to
TABLE NO. 1
CITY OF CORNWALL
WATER POLLUTION CONTROL PLANT
Measured vs. Modelled Values
Recorded (WPCP)
1969
1970
1971 (avg.)
1972 (avg.)
1973 (avg.)
1974 (avg.)
1975 (avg.)
1976 (avg.)
1977 (avg.)
1978 (avg.)
1979 (avg.)
Average Values
(1974 - 1979)
Avg. Year
Precip-
itation
(in.)
-
-
-
-
46.3
36.1
34.7
46.0
47.3
36.4
37.8
STORM MODEL INPUT
DWF ONLY*
Res./Comm - Domestic
Ind.
Infiltr. + Extraneous
Average
STORM MODEL OUTPUT
for Avg. Precip. Year**
DWF
Flow
TSgdT
-
-
-
-
-
8.7
8.3
9.1
9.7
8.2
8.5
8.8
2.6
.9
5.4
8.9
-
Average
Daily
Flow
(mgd)
6.9
7.4
8.4
9.8
11.0
10.4
10.5
11.1
12.2
11.1
10.7
11.0
-
-
10.9
Brookdale
Weir
Overflow
(mg/year)
-
-
-
-
165
128
73
178
358
560
298
265
-
-
68.5
Influent Pollutant
Conc.(mg/L)
S.S.
140
120
155
185
160
160
171
186
221
121
110
161
376
1130
-
224
197
B.O.D.
52
80
80
no
140
141
177
125
127
134
106
135
283
763
-
160
130
P
-
-
5.8
5.2
3.2
3.6
3.7
3.6
4.6
4.4
3.7
3.9
11.0
17.4
-
4.9
4.3
* Refers to Res., Comm. , Indust,. & Infiltration Flows Combined during dry weather
periods indicating average daily dry weather flows and average quality concentrations.
** Indicates the average quality resulting from the DWF combined with runoff due to
precipitation for an average of 10 years precipitation records.
-------
the sunnier season. This was particularly important for pervious areas of the
>
watershed where it was found necessary to increase the runoff coefficient from
.04 to .44 - a factor of 10.
This experience, along with other projects of a similar nature, indicate the
need for a variable coefficient and/or soil complex curve number input into the
model.
24
FIGURE NO. 2
RAINFALL HYETOORAPH
6 8 10 12
14
16 18
20 22
24
O
*
o -
z -
4 -
6 -
8
40
30 -
20 -
10
0
U
O.I
o.z *
- 0.3
i 1
_!
i i
i i i i
-
LEGEND
O.W.F.
MEASURED HYDROGRAPH
STORM HYDROGRAPH
1.0
0
w
0.5
' '
r
!
i
^
vT'1 j
ii i i i i i i i
ZJ
n
r
i
\
\
\
U
^
' iii.
u_
.
24
10 12 14
MAY 9,1978
TIME
16
16
20
22
24
CITY OF CORNWALL
REPORT ON COMBINED SEWER SYSTEM ANALYSIS
TESTING OF CALIBRATED STORM
AMELIA ST. COMBINED SEWER CATCHMENT
8
-------
In addition, it was found necessary also, to decrease the melt coefficient
somewhat, measured as inches per day per degree Fahrenheit from .07 to .055.
This modification was necessary as the simulated melt was occurring too early
and decreasing the melt coefficient both delayed the peak and increased the total
duration of the melting time.
FIGURE NO. 3
RAINFALL HYETOGRAPH
12 14 16 18 20 22 24 2 4 6 8 10 12 14 16 18 20 22 24 2 4 6 8 10 12
!"
6 -
80 -
60 -
40 -
20 -
- o. i
0.2
- 0.3
3.0
2.5
2.0
1.5 X
1.0
0.5
LEGEND
- O.W.F.
MEASURED HYDROGRAPH
STORM HYDROGRAPH
ra.
_£_!
I
MI
i'i\A
u
12 14 16 IS 20 22 24 2 4 6 8 10 12 14 16 18 20 22 24 2 4 6 8 10 12
SEPT 12,1977 SEPT. 13,1977 SEPT. 14, 1977
TIME
CITY OF CORNWALL
REPORT ON COMBINED SEWER SYSTEM ANALYSIS
TESTING OF CALIBRATED STORM
AUGUSTUS ST. COMBINED SEWER CATCHMENT
9
-------
The second problem related to the calibration of the overflow weir where combined
sewer overflow events are measured. It was found that the calibration of the
overflow weir required significant modification as readings were continuously
high.
After the modifications as outlined above, a simulation of the system operation
was undertaken over the period of 1 year, comparing both volume of overflow and
number of events to those actually measured after recalculating overflows on the
basis of the new weir calibration. The results of this comparison are shown in
Table No. 2 and it is noted that reasonable correlation, both in terms of volume
and number of events was obtained.
The simulated volume was somewhat less than that actually measured by about 6%,
whereas the total number of events was in fact greater than that measured by some
27%. It is noted, however, that the number of events simulated included for
various months', a number of minor overflow events, which were not actually recorded
but rather were taken up by natural storage within the system.
CONTROL ALTERNATIVES
A listing of various control alternatives was produced and these included various
levels of treatment plant capacity expansion, in-system storage, diversion of
storm flows from the combined sewer system, improvements in street sweeping
practices and industrial flow control or abatement.
10
-------
A preliminary screening was undertaken of these various alternatives on the basis
of present day dry-weather flow rates. The results of this screening are
summarized in Table No. 3. The screening was undertaken by comparing the
effectiveness of each alternative in comparison with the present day system,
TABLE HO. 2
CITY OF CORNWALL
COMBINED SEUER ANALYSIS
1 Year Comparison - Actual vs. Simulated
(May 1978 - April 1979)
Month
May
June
July
August
September
October
November
December
Janua-ry
February
March
April (20th)
Total
Measured - Brookdale Overflow
Volume (M.G.)
4.423
.929
.714
7.309
1.699
4.309
1.94
-
10.641
.207
54.174
12.943
99.288
No. of Events
6
4
3
5
2
3
2
-
2
1
14
5
47
S.T.O.R.M. Predictions
Volume (M.G.)
2.73
1.51
1.06
8.49
2.42
5.00
3.64
-
8.95
4.25
43.32
12.28
93.65
No. of Events
5
6
3
7*
7*
3
2
-
5*
2
13
7
60
* includes a number of minor overflow events simulated
but not actually measured
11
-------
IX)
TABLE NO. 3
COMPARISON LISTING OF CONTROL ALTERNATIVES
1
2
3
4
3
6
7
a
9
10
II
12
13
14
IS
16
STORM WATER
MANAGEMENT
CONTROL SCHEME
240 MOO HYDRAULIC
PLANT EXPANSION-
HYDRAULIC
U5C OF E ISTING TRUNK SEWEH *
STORAGE
DIVERT 4JS AC. STORU TO
FLY CREEK
IMPROVED STREET SWEEPING
INDUSIHIAL FiOW CLIMINAIEO
INDUSTRIAL FLOW 0 BYLAW LIMITS
FXANT EXPANSIUN- Z4 O MGO »
t STORAGC IN EXISTING /DUNK
PI Alir EXPANSION- 24 O MGO *
» STORAGE IN EXISTING TRUNK
I IMPROVED STREET SWEEPING
PLANT EXPANSION- 24 0 MGO
» SIUHAUE IN EXISTING TRUNK
» IMPROVCD STREET SWEEPING
« MOUSlrtlAL FLOW(i BYLAW LIMIYS
» STORAGE IN EXISTING TRUNK
* STORAGE IN tXlillNG TflUNK
t DIVERT 4S3AC YO FLY CREEK
f STORAGE IN CIISIING TRUNK
» IMPROVE 0 STREET SWEEPING
PLANT EXPANSION - 16 MGD *
» 5IOHAIU (N EXISTING tRUNK *
» INDUSTRIAL FLOWA BYLAW LIMITS
TOTAL ANNUAL LOADING
(FROM WATERSHED)
FlW
651.6
651.6
651.6
651.6
664.9
651.6
595.6
651.6
651.6
564.9
651.6
651.6
651.6
564.9
651.6
651.6
L8S.I06
7.611
7.611
7.611
7.611
6.756
7.495
3.967
5.190
7.611
6.756
7.495
5.102
7.611
6.756
7.495
5.11)5
L BS ilO6
5.262
5.262
5.262
5.262
4.606
4.906
2.515
3.551
5.262
4.606
4.901
3.198
5.262
4.601
4.901
3.191
p
L8S«IO*
0.175
0.175
0.175
0.175
0.158
0.166
0.104
0.175
0.175
0.158
0.168
0.168
0.175
0.156
0.168
0.168
OVERFLOW- ANNUAL LOADING
FL
FT'.IO'
27.21
27.21
16.61
16.68
24.71
27.21
26.20
27.21
16.68
13.17
16.68
16. 611
9.05
5.47
u.05
9.05
BW
NO OF
EVENTS
50
50
33
19
46
49
49
49
19
17
19
19
15
13
15
15
LBSllO3
171
171
111.6
87.7
133.8
110.1
139.3
151.6
87.7
61.6
52.6
42.5
52.8
31.3
29.2
11.4
LBS.io1
305.4
305.4
179.3
115.2
240.9
191.0
272.2
288.5
115.2
78.6
74.8
67.6
62.8
34.2
39.5
35.9
LBS.,05
5.61
5.61
3.3Z
2.30
4.41
3.72
4.91
5.61
2.30
1.57
1.58
1.58
1.26
0.68
0.83
1.58
FLOW TO
F,'.,0«
624.4
624.4
635.0
634.9
540.2
624.4
569.6
624.4
634.9
551.7
634.9
634.9
642.5
559.4
642.5
642.5
LBSllO6
7.440
7.440
7.499
7.523
6.622
7.385
3.828
5.038
7.523
6.694
7.442
5.060
7.558
6.725
7.466
5.091
W.P.C p.
as.io'
4.957
4.957
5.083
5.147
4.365
4.715
2.243
3.263
5.147
4.527
4.831
3.120
5.199
4.572
4.867
3.162
as no6
0.169
0.169
0.172
0.173
0.154
0.164
0.099
0.169
0.173
0.156
0.166
0.166
0.174
0.157
0.167
0.166
WRCR-EFFLUENT DISCHARGE
Y'.IO'
624.4
624.4
635.0
634.9
540.2
624.4
569.6
624.4
634.9
551.7
634.9
634.9
642.5
559.4
642.5
642.5
.s.io'
(251)
5.580
(801)
1.488
(80J)
1.500
1251)
5.642
1251)-
4.967
Tzsry
5.539
f25*r
2.871
(25*1
3.779
(SOU)
1.505
(80%)
1.339
(Kill
1.488
TaoiT
1.012
1801)
1.512
(801)
1.345
(801)
1.493
(80%
1.018
as.io*
(101)
4.461
(301)
3.470
(30»)
3.556
(10»)
4.632
~noi)
3.929
(101)
4.244
TITS)
2.019
(101)
2.936
(30J)
3.603
(30J)
3.169
IJU1I
3.382
(301)
2.19
(301
3.639
(301
3.200
(30T
3.40
(30X
2.21
S.IO*
(ZOI)
.135
(801)
.034
(801)
.034
(201)
.138
(201)
.123
~rzusi
.131
~tzon
.079
(20IJ
.135
(ami
.035
(80S
.031
(bUl
.033
(801]
.033
(801)
.035
(801
.031
(801
.033
(80t
.03
NET
TU'
51.6
51.6
51.6
651.6
564.9
651.6
595.8
651.6
651.6
564.9
651.6
651.6
651.6
564.9
651.6
651.6
LOAD
BS.IO'
.751
.659
.612
5.730
5.100
5.649
3.101
3.930
1.592
1.400
1.541
1.054
1.564
1.376
1.522
1.030
TO RIVER
BS.IO*
.766
3.775
3.379
4.747
4.169
4.435
2.291
3.225
3.718
3.248
3.434
2.259
3.702
3.234
3.44
2.24
asiio*
1406
0396
0372
.1400
.1273
.1351
.0842
.1406
.0369
.0329
.0349
.0349
.0360
.032
.034
.034
/.REDUCTION FROM
PRESENT SYSTEM
71
72
0.4
11
2
48
32
72
76
73
82
73
76
74
82
BOD
21
29
0.4
13
7
52
32
22
32
28
53
22
32
28
53
F
72
M
0.4
9
4
40
74
77
75
75
74
77
76
76
-------
considering as a yardstick, the reduction in total load discharged to the
receiving stream, considering both treatment plant effluent discharge and
combined sewer. Examination of this preliminary evaluation indicates that the
most promising combination of alternatives is some combination of storage and
treatment.
The Storm Model was then utilized further in order to assess various levels of
storage and treatment, first on the basis of present dry weather flow conditions
(i.e. 9.5 MIGD) and then secondly, on a future increased dry weather flow. The
lowest level of storage considered was that level of storage naturally available
within the Riverfront Interceptor, above the natural depth of flow for dry
weather flow conditions. A family of curves was then produced, showing the
relationship of varying levels of treatment rates in conjunction with storage
volumes, both as a function of capital cost and as a .function of degree of control,
measured in terms of runoff from the watershed directed through storage and
treatment. It is important also to note that the cost functions are based on the
premise that:
a) The available storage in the Riverfront Interceptor sewer in the
amount of 2.2 million gallons, is available and can be provided
at minimum cost.
b) The cost of treatment expansion only, above the present capacity
of 8.25 MIGD is included in the analysis. In other words, it is
assumed that the existing facility has already been amortized.
Figure No. 4 shows that an increase in treatment capacity to a 24 MIGD rate in
conjunction with utilization of the available trunk interceptor capacity will
13
-------
provide a degree of treatment for approximately 67% of the stormwater runoff. As
the dry weather flow increases, this level of control decreases to approximately
62%, as noted on Figure No. 5.
FIGURE NO. 4
STORAGE
10s m'
10
DRY WEATHER FLOW - 9.53 mod
DEGREE OF RUNOFF
VOLUME CONTROL
50
100 150
I03 mVd
TREATMENT
CITY OF CORNWALL
REPORT ON COMBINED SEWER SYSTEM ANALYSIS
STORAGE TREATMENT COMBINATION
PRESENT FLOW EFFECTIVENESS
14
200
ZbO
-------
It is noted on both of these figures, that although the selected combinations do
not fall on the theoretical "least cost" line, the 24 MIGD treatment rate is a
function of the present plant size in terms of settling tank units and as such,
is a realistic size of expansion to consider. At the same time, it can readily
FIGURE NO. 5
STORAGE
DRY WEATHER FLOW = 12.0 mad
DEGREE OF RUNOFF
VOLUME CONTROL
I 1
100 ISO
I03m'/d
TREATMENT
CITY OF CORNWALL
REPORT ON COMBINED SEWER SYSTEM ANALYSIS
STORAGE TREATMENT COMBINATION
FUTURE FLOW EFFECTIVENESS
15
200
250
-------
be seen that if desired, higher levels of control can be most economically
provided by additional storage in the system.
It is important to consider also the trend in the "level of control", or the
number and volume of combined sewer overflow events as development takes place
within the City and normal dry weather flow rates increase. For comparison
purposes, a future condition dry weather flow of 12.0 MIGD was assessed, which
represents an increase in present day development of about 25%.
Considering the downstream beaches as a focal point of concern, the seasonal
impact of these overflows is also important.
Figure No. 6 illustrates the projected trend in both total volume and number of
events of combined sewage overflows considering:
A 24 MIGD treatment rate
Existing available interceptor sewer storage.
It is noted that, during the critical summer season, the overflow occurrences
will average approximately 9-10 under present day conditions and this will
increase to 11 or 12 overflows as dry weather flows increase to 12 MIGD in the
future. The associated volumes of overflow will also increase from about
25 million gallons under present day flow conditions to about 30 million gallons
for the future 12 MIGD dry weather flow condition.
For comparison purposes, Table No. 4 shows the control effectiveness of a 24 MIGD
treatment capacity in combination with storage, as opposed to a 36 MIGD treatment
rate without storage.
16
-------
It is noted that the lower treatment rate in conjunction with storage provides a
higher level of control with respect to total number of events but a slightly
FIGURE NO. 6
TOO
600
500
9.
o
u! 400-
ir
UJ
u_ n
° o
^ 300-
>
200
100
LEGEND '
VOLUME
|40 EVENTS
24 m«d TREATMENT RATE
325,000 ft. INTERCEPTOR STORAGE
120
100
20
40
10
mgd
12
14
I
45
1 1
50 55
I03 mVd
DRY WEATHER FLOW
1
60
I
65
15
35
30
25
5
o
a:
ui
15
10
5
70
CITY OF CORNWALL
REPORT ON COMBINED SEWER SYSTEM ANALYSIS
PROPOSED STORAGE TREATMENT COMBINATION
VOLUME /NUMBER OVERFLOWS FOR INCREASING DRY WEATHER FLOWS
17
-------
TABLE NO. 4
OVERFLOW EVENTS
a) Summertime (June-September)
b) Remainder of year (October-May)
c) Total
OVERFLOW VOLUMES
a) Summertime (June-September)
b) Remainder of year (October-May)
c) Total
24 MIGD Treatment
Trunk Sewer Storage
10
II.
22
25 MG
95 MG
36 MIGD
Treatment
16
20i
36
35 MG
75 MG
120 MG
110 MG
lower level of control considering total volume. This is accounted for by the
fact that the treatment-storage combination totally contains more of the smaller
volume events which create overflow for the 36 MIGD treatment rate alone.
The storage treatment combination does not provide as effective a control during
periods of spring runoff and hence, the volume of overflow is somewhat higher for
this control scheme.
PRINCIPLE OF OPERATION
Figure No. 7 is a profile of the interceptor sewer showing the various points of
interception. As a principle of operation, to effect maximum control, it is most
desirable that the points of overflow be centralized to perhaps one or two
locations. In the event then, that higher levels of control are deemed necessary,
additional storage can more readily and effectively be provided and chlorination
and/or diffusion facilities may also be readily installed.
18
-------
140
CITY OF CORNWALL
PROFILE OF RIVER FRONT INTERCEPTOR
,.,. *
SCALE: VERT. ,.. . |0.
FIGURE NO. 7
-------
The most appropriate location for such centralized facilities would appear to be
at the upstream end of the interceptor sewer at Brookdale Avenue at which point,
approximately 72% of the combined sewer area is tributary to the interceptor.
Considering tributary areas, flows, sewer capacities and storage volumes, an
initial conclusion was that all five easterly overflows could be closed and the
entire wet weather flow intercepted. The tributary areas and flows at Pitt
Street and particularly Amelia Street, however, indicated that more analys.is of
the hydraulic consequences of closing these two overflows was warranted.
Further analysis was undertaken using the extended transport block of the Storm
Water Management Model - SWMM - in order to assess the hydraulic operation of the
system under individual event operation.
As a starting point, two historical storms were chosen with total rainfall volumes
in the order of 1.5 inches, which is about equivalent to a 2 year return event in
terms of total volume. The two storms chosen were:
1.51 inches over a duration of 4 hours
1.43 inches over a duration of 9 hours.
The shorter duration storm naturally produced larger peak flow rates. As an
initial step, the analysis assumed total interception at all five easterly
overflow locations and three times dry weather flow rates at both Pitt Street and
Amelia Street.
The hydraulic performance of the system operation for the shorter duration storm
is shown on Figure No. 7. The lower water surface profile is at the time of
20
-------
flow occurrence in the Brookdale Avenue trunk and consequently, the time of peak
overflow rate. It is noted at this time, the available storage has not yet been
fully utilized. The higher water surface (hydraulic grade line) profile is at a
later time period during the storm when peak flows have subsided. At this time
storage is fully utilized. The problem with this operation is that maximum
utilization of storage to contain "first flush" effects has not been achieved.
The second storm event analyzed was of equivalent volume but longer duration.
Because of this, peak flow rates were also significantly lower and in fact, no
overflow was created at the Brookdale Avenue chamber. The profile shows the
maximum hydraulic grade line profile in the interceptor sewer for this event
also and it is noted that full storage utilization was achieved for this event.
The next step in the analysis procedure will be to undertake a sensitivity analysis
considering both various levels of "design storms" and various levels of available
storage in order to determine:
a) Proper and realistic levels of interception for both
the Pitt Street and Amelia Street overflows.
b) Frequency of overflows associated with such intercepting
capacity at each of these two locations.
c) Maximum utilization of available storage.
This procedure is now underway.
21
-------
In summary, the use of such modelling techniques provides for a study of this
nature, a tool which allows analysis of a number of alternatives in such detail
as would not otherwise be realistically possible and as such, permits selection
of appropriate control schemes considering both costs and effectiveness. At the
same time, analysis of the system operation under real storm event conditions
allows the analyst to see how the system will react, pick out potential problem
areas and make the necessary adjustments in deriving the best cost effective
scheme.
22
-------
CONSIDERATIONS REGARDING THE APPLICATION OF
SCS TR-55 PROCEDURES FOR RUNOFF COMPUTATIONS
12 3
P. Wisner, S. Gupta/ and A. Kassem
INTRODUCTION
The U.S. Soil Conservation Service (SCS) Technical
Release Number 55 "TR-55" (12) published in 1975 presents
simplified tabular and graphical methods for estimating run-
off volumes, peak discharges and discharge hydrographs for
runoff computations. The method which has been applied mainly
in the U.S. (8) has also been recommended for application in
Canada (7).
The TR-55 does not provide any comparisons with real
measurements or more sophisticated computations. Mention is
made that the methods for computing peak discharges and hydro-
graphs using time of concentration (T ) and travel time (T )
t*> L»
are approximations of the detailed hydrograph analysis, SCS TR-20
(Computer Program for Project Formulation Hydrology).
The implementation of SWM has led to the need for
more sophisticated computations than the Rational Method (RM).
1,2,3 - Professor, Grad. Student and Research Associate,
respectively. Department of Civil Engineering,
University of Ottawa, Ottawa, Ontario, Canada.
23
-------
Many "simplified techniques" were developed and are popular
for various reasons. In a previous SWMM users group meeting
in Montreal, we have indicated the need for standardization
and testing of these methods and comparison, with more sophis-
ticated ones. In the Gainsville meeting, we assessed the RM
based on comparisons with SWMM (18).
The intent of the paper is to present a critical
review of the SCS TR-55 computational procedures and data
requirements. A comparative assessment of peak discharge
and hydrograph computations with other widely used models
are presented. The comparisons are carried out using
several hypothetical watersheds. Several limitations of
the procedures are also discussed. Some of the results are
taken from other detailed studies conducted by the writers
(5,17).
A CRITICAL REVIEW OF THE SCS TR-55 RUNOFF COMPUTATIONS
Figure 1 is a flow chart of the TR-55 procedures
for runoff computations. The following is a discussion of
various steps and/or input requirements in the procedure.
(1) Meteorological Input:
The computation of peak discharges for urban and
rural areas using either the graphical or the tabular methods
in TR-55 are based on a 24-hour SCS Type II storm distribution,
The distribution is not documented in the release. As shown
24
-------
in Figure 2, a "Chicago" type storm distribution as well as
real storms may be more "peaky" than the SCS Type II Storm
(16)f where peakier storms may be more critical for the analysis
of small watersheds. Therefore in urbanized watersheds, where
So Ms type and
impervious ratio
Select frequency of
24-hr
storm
SCS
Type
II
Select CN based
on area weighted
average
Determine 24-hr
rainfall volume from
IDF curves
Compute total runoff
volume from
Table 2-1
Peak flow computations
based on slope and area
(Appendix D)
Read peak discharge
from Appendix D
,, Yes
Compute adjustment
factors FHLM anc'
F,MP from
Figures 4-1 and 4-2
Multiply peak dis-
charges by adjust-
ment factors to
obtain modified
peak flows
Tabular
based on
(Figure
i
method
5-i,
r
Compute peak flow
from Figure 5-2
^
No
f
Percent of hydraulic
length modified due
to urbanization
Predevelopment
peak discharge
obtained
Watershed and channel
characteristics
Compute Tc for each
subarea and Tt for
each routing reach
Routing of flows
based on Tc and
Tt (Table 5-3)
Obtain routed
flows at outlet
using Table 5~3
FIG. 1 FLOW CHART FOR SCS TH-55 RUNOFF COMPUTATIONS
25
-------
short duration intense rainfalls are critical for storm sewer
design, the meteorological input for SCS TR-55 method is not
appropriate and may give lower peak flows.
(2) Runoff From Small Storms:
Total runoff volumes computed by SCS TR-55 for CN =
60 to CN = 90 for small rainfalls are shown in Figure 3. Run-
off measurements (10) from 2 residential watersheds (see Table 1)
are also shown in Figure 3. It can be seen that the runoff
volumes obtained from real measurements are greater than runoff
volume computed by TR-55 for CN = 90 which corresponds to an
area with a high degree of impervidusness. It can be seen there-
fore that the procedures in SCS TR-55 which recommends the
Real Storm
Chicago Distribution
24 Hour SCS Type V
20 40 60 80
CUMULATIVE PERCENT OF TIME
100
FIG. 2 RAINFALL TIME DISTRIBUTION CURVES FOR A REAL,
CHICAGO , AND SCS TYPE II DISTRIBUTION
26
-------
e 2
2
5
SCS TR-S5
Rod MaasuramenK
NORTHWOOO
GRAY HAVEN
01234
Ralnfill Volume (P), In.
- 3 RELATIONSHIP BETWEEN RAINFALL AND RUNOFF:
TR-S5 AND REAL MEASUREMENTS
TABLE 1
WATERSHED CHARACTERISTICS FOR REAL MEASUREMENTS
Name of Watershed
Gray Haven
Northwood
Drainage Area (acres)
23.3
47.4
Imp.
Ratio (%)
52
68
selection of CN based on averaging requires the consideration of
significant increases of CN for small rainfalls (say less than
3 in.). This finding was substantiated by a recent study (1)
on several watersheds in Texas where calibrated CN's were found
to be higher than those calculated from the manual.
An examination of the tables contained in TR-55 for
the computation of runoff volume as a function of total rain-
fall and hydrologic soil complex number (CN) shows that for
27
-------
CN varying from 75 to 85 (typical for Ontario conditions), the
minimum rainfall required to generate (say) 0.1 in. of runoff
would vary from 1.0 in. to 1.3 in. Thus a 1/2-year, 3 hour
storm for Metro Toronto with a rainfall of 1.2 in. would yield
0.07 in. of runoff from a watershed with a CN = 75.
(3) Variations of Runoff Curve Numbers Within a Watershed:
According to the SCS TR-55 the composite CN for any
watershed is calculated by means of a weighted average based on
area. Runoff is not calculated separately for pervious and
impervious areas as is done in several other models and as a
result, the assumption discussed above becomes critical. This
may be easily verified by means of a simple example. An urban
residential watershed with an imperviousness ratio of 30 per-
cent and 1/3 acre lot size and hydrologic soil group B would
be assigned a CN = 72. The minimum precipitation required to
generate a runoff of (say) 0.1 in. corresponding to this CN is
approximately 1.5 in. For the same area however, the runoff
computed by a model such as SWMM for the same rainfall would be
approximately 0.5 in. It is also interesting to note that
worked examples in the TR-55 report generally utilize storms of
high return frequencies ranging from 1/50-yr. to 1/100-yr. with
rainfall volumes of the order of 5.0 to 6.0 in., which do not
correspond to conditions currently considered in most Canadian
studies. The method was also used in the Maryland Pond Design
Manual for low (1/2-yr) frequency storms (8).
28
-------
A comparison of runoff volume coefficient computed
by several procedures and degrees of imperviousness are shown
in Table 2. These results are based on either simulations that
have been carried out or computations based on information
available in the literature. It can be seen that for small
rainfalls the TR-55 tends to underestimate runoff volumes.
The low C corresponding to Desbordes (2) measurements was
probably caused by insufficient data for areas having low
imperviousness ratios.
TABLE 2
COMPARISON OF RUNOFF VOLUME COEFFICIENTS
Imperviousness
CO
30
60
90
Runoff Volume Coefficient, Cy
MODELS
SWMM
(Rainfall
Volume =
1.22 in.)
0.29
0.57
0.86
STORM2
(Annual
Rainfall)
0.38
0.60
0.83
SCS TR-55
(Rainfall
Volume =
1.22 in.)
0.14
0.32
0.63
DESBORDES1
(Real
Measurements
Small
Rainfalls)
0.18
0.49
0.79
Cv -1.01 ((IMP/100 - 0.12)) Reference 7
In STORM used for individual storm events
29
-------
(4) Adjustment of Computed Peak Flows:
One of the TR-55 methods (Appendix D of the manual)
for computing and modifying computed peak discharges, includes
two adjustment factors namely an imperviousness factor and a
hydraulic length modification factor. Various users may work
with different magnitudes for the last factor.
Other factors are watershed slope adjustment factor
(WSAF), swamp and pond area adjustment factor (SPF) and water-
shed shape factor (WSF). These adjustment factors account for
watershed slopes, swamp and ponded areas and shapes for situa-
tions other than those for which the graphs have been constructed,
Of the three adjustment factors in Appendix E in the
release, watershed slope and shape factors are easy to apply
once the average watershed slope and shape are known. However,
the pond and swamp area adjustment factor requires some assess-
ment of flow paths to determine whether a significant amount
of the flow from the watershed passes through such areas and
a study of topographic maps to locate such areas.
(5) Development of Tabular Discharges:
It can be shown that the relationship for peak dis-
charges against TC shown in Figure 5-2 and Table 5-3 in TR-55
can be plotted as a straight line on a log-log scale as shown
in Figure 4 and can be represented by the following relation-
ship:
30
-------
3980
T 0.62
c
(1)
where q = peak flow in cfs/sq. mi/in, of runoff,
T « time of concentration.
w
For a watershed of drainage area of A sq. mi. and a 24-hr.
rainfall of R in., the peak runoff rate is given by:
3980AR
0.62
(2)
Equation 2 is analagous to a generalized relationship for peak
runoff rate for a triangular hydrograph.
The tabular discharges in the technical release
describe a unique hydrograph profile, viz., one obtained with
a 24-hour SCS Type II storm. The routed hydrographs in terms
3:0
2.7S
2.5
2.25
2.0
Time (mnl
\
.5
1.0 1.5
Log TC (mtn)
2.0
2.5
3.0
FIG. 4 RELATIONSHIP BETWEEN PEAK FLOW AND T : TR-5S
C
31
-------
of travel time (T ) therefore cannot be used for any other storm
profile or for other hydrographs.
(6) Alternative Procedures:
Although peak discharges from any given watershed can
be obtained either in terms of time of concentration (TC) or in
terms of area, slope and CM, there is no clear recommendation as
to the adequacy of the two procedures given in the manual.
This brief review indicates that the SCS TR-55 pro-
cedures cannot be applied for certain conditions for example
short duration rainfalls, low imperviousness ratios etc. and
that flows may be underestimated. So if peak flows are to be
computed by TR-55, caution is required.
3.2 PEAK DISCHARGES; COMPARISON OF METHODS CONTAINED IN TR-55
There are two procedures for computing peak flows
given in TR-55. One of the methods is based on the time of
concentration for a watershed and the peak flows are given in
terms of csm/in. while the other one is based on watershed area
and slopes, and peak flows are given in terms of cfs"./in. of
runoff.
In order to compare the results obtained by the two
methods, several watersheds ranging in size from 10 acres to
160 acres undergoing urbanization were considered. The water-
sheds in the predevelopment stage are considered to have
32
-------
uniform runoff characteristics and a CN = 75. For post develop-
ment condition CN = 83 for a residential area of 1/4 acre lot
size corresponding to an imperviousness ratio of 38% and hydro-
logic soil group (C). The 24-hr, rainfall for a 1/25 year storm
for Oakville, Ontario (P24 = 3.31 in.) was used for the com-
parison.
The results and the watershed characteristics are
shown in Table 3. It can be seen that different results are
obtained by the two methods. The computations based on T
C
give higher peak rates of runoff than those based on area
and slopes. The percentage increase in peak flow using the
computations based on T vary with area. For example, for
C
10 acres the increase is 8% whereas for 160 acres it is 55%.
It therefore appears that the difference in peak flows would
increase with the area. However, for urbanized conditions, the
method based on area and slope give larger flows for areas up to
60 acres. For areas larger than 60 acres the method based on
T computes flows which are once again larger but the percentage
C
increase is much less. For example, for 160 acres the difference
is only 22%.
3.3 COMPARISON WITH OTHER RUNOFF COMPUTATION MODELS
It is not possible to compare the results from dis-
charge computations with SCS TR-55 against real measurements
since it contains an implicit assumption; the use of a 24-hr.
SCS Type II storm. However, it is possible to compare TR-55
33
-------
TABLE 3
COHPAKISOH OF PEAK FLOW CONFUTATIONS BV TR-55 METHODS
w
^ 3
'I
5 -
* 9
3
O rH
3 2 "i
|'a
** i
i !~
Owl
*j *"*
a * "3
£ «5 S
.
I 3
1|;
ri «i
H aa a
|g§
BASED ON APPENDICES D «
Watershed
Area
Acres
(sq. >i.)
(1)
10 (.016)
60(.094>
160(.2S)
10(.OU)
60(.094)
160(.25)
Factor
laperviouanasa
(2)
Hot
Required
"
ii
1.17
1.17
1.17
Factor
Hydraulic
Length
Modified
(3)
Hot
Required
i
ii
1.43
1.43
1.43
U/S Slope
Factor
(4)
1.21
1.26
1.29
1.21
1.26
1.29
e
W/S Shape
Factor
(5)
1.0
1.0
1.0
1.0
1.0
1.0
T 'a are computed baaed on overland flow velocity of 2
conditions and
a velocity
5 ft /second
and 7.S win
-------
results with hydrograph models which have already been tested
where the 24-hr. SCS storm may be simulated. Two models,
namely HYMO for rural conditions (14) and SWMM for urbanized
conditions (9) were selected for this purpose.
For rural conditions, the same hypothetical watersheds
(10 acres to 160 acres) were used for which TR-55 computations
were performed. An average CN of 75 was used to reflect low
imperviousness ratios. This is based on an earlier study (15)
where it was found that infiltration losses with HYMO and SWMM
(with default values) are close for imperviousness CN = 98 and
pervious CN = 75. Table 4 shows the peak flows computed by the
two models. It is seen that although for very small areas (up to
10 acres) and rural conditions, the difference in flows is
small, the difference increases sharply for larger areas. For
example, for an area of 160 acres the difference is as high
as 55 percent.
For urbanized conditions, where 38 percent of the
watershed is impervious, peak rates of runoff and hydrographs
computed by TR-55 were compared with those simulated by SWMM.
These results are shown on Table 4 and Figure 5. The following
may be observed:
(1) The peak rates of runoff computed by TR-55 (T
c
method) is consistently lower than those computed
by SWMM up to 160 acres;
35
-------
TABLE 4
COMPARISON OF PEAK FLOWS: SCS TR-55 (BASED ON T ), HYMO AND SWMM
Watershed
Area
(acres)
(1)
10
60
160
PEAK FLOWS (CFS)
PREDEVELOPMENT CONDITIONS
SCS TR-55
(Runoff =1.18 In.)
(2)
12.3
52.0
106.2
HYMO
(Runoff =1.10 In.)
(3)
13.5
71.9
166.9
POST DEVELOPMENT CONDITIONS
SCS TR-55
(Runoff =1.68 In.)
22.9
114.5
256.2
SWMM*
(Runoff =1.66 In.)
27.9
132.8
290.6
Co
cn
Impervlousness ratio = 38 percent.
-------
(2) For small areas up to 60 acres the difference is
of the order of 20 percent and appears to decrease
for larger areas.
It may be mentioned however, that some of the input
data to TR-55 (CN for example) can be modified in order to
obtain flows which would be much closer to those simulated
by SWMM.
140
130
120
110
100
90
80
TO
60
SO
40
30'
20'
10'
0
60 Acra HypottMthul Aral
SCS TB-SS
SWMM
11.0
12.0
1Z.5
13.0
13.5
FIG.5 COMPARISON OF HYOROGRAPHS : SCS TR-55 (TABULAR METHOD) AND SWMM
37
-------
3.4 CHANNEL ROUTING
The tables contained in the TR-55 manual for routing
runoff hydrographs through stream channels are based on the
convex method (13). The routing equation in the convex method
is:
02 = (1-C) 01 + C II (3)
where I is the inflow to the reach, 0 is the outflow from the
reach and C is the storage coefficient given by
C = At/k (4)
where At is the wave travel time and also the required routing
interval, k is the reach travel time for a selected steady flow
discharge of a water particle through the reach; subscripts 1
and 2 refer to the beginning and end of time interval. More
details on the convex method can be found in the SCS National
Handbook (11).
For the purpose of assessing routing by the TR-55
tables, a hypothetical system formed by a watershed and a
channel (Figure 6) was used. The watershed has an area of
0.4 sq. mi. and CN = 80. For a rainfall of 6.0 inches and
time of concentration of 1.0 hr, the runoff hydrograph ob-
tained from the TR-55 tables (for travel time = 0.0) is routed
along the channel by means of: (a) the TR-55 tables, (b) the
convex method, and (c) the WRE-TRANSPORT Model (4,6), a
38
-------
hydraulic routing model used extensively in Canada and the
U.S. It was shown by the first and third authors that routing
by the WRE-TRANSPORT model is accurate when compared with the
method of characteristics (5). The routed hydrographs by the
three methods are compared in Table 5 and shown in Figure 6
at two locations: 15,643 ft. and 29,683 ft. downstream of the
watershed outlet.
TABLE 5
ASSESSMENT OF ROUTING BY THE TR-55 TABLES -
COMPARISON WITH THE CONVEX METHOD AND THE
WRE-TRANSPORT MODEL
~*>>«VH^ Routing
^**1**^^ Method
Hydrograph ^^^s,^^
Location ^*s^N(i^
M-l
m
4J
M-l
00
CM
Peak Discharge
Qp (cfs)
Error in Qp*
Time to Peak
(hr)
Peak Discharge
QP (cfs)
Error in Qp*
Time to Peak
(hr)
WRE -Transport
461
-
12.75
450
-
13.10
Convex.
465
0.9%
12.91
453
0.7%
13.38
SCS TR-55
429
6.9%
12.90
396
12.0%
13.50
la comparison with WRE-Transport Model
39
-------
It can be seen that the TR-55 routing tables tend to
underestimate flows. The error increases with travel time,
that is with the length of channel. A direct application of
the convex method/ on the other hand, gives more accurate
results.
MATERSMEO
A .0.4 «
CN.H
I -tt> In.
T . 1JJ fir.
L,-15.B43 It.
CHANNEL
SB« .002
n s .013
WMIX .V
U. M.040 tt.
J. JL OUTLET
(I) WRE-TRANSPORT VS. SCS-TBS5
HTDROGRAPHS AT X = 15.643 FT.
Mlo-
WBE-TRANSPORT
-«-«-SCS-TRS5
11 12 13 M
Tim*
-------
Although a complete assessment of these tables would
require more testing under a variety of conditions, there is
an apparent major limitation of the tables, that is the use of
a hydrograph profile corresponding to 24-hr. Type II storm.
The routing tables may be attractive because of their
simplicity. The above test shows however that the convex method,
which is as simple to use as the TR-55 tables, is more accurate
and does not have any limitation regarding the hydrograph pro-
file.
SUMMARY AND CONCLUSIONS
(1) Users of the SCS TR-55 procedures should be aware
of the limitations introduced by the meteorological
input assumptions and the methodology for calculation
of infiltration and other losses, viz.;
(a) a particular distribution and a specific
(24-hours) duration design storm;
(b) weighted averaging of curve numbers (CN) for
pervious and impervious areas which may lead
to overestimation of rainfall losses correspond-
ing to 1/2 to 1/10-year storms.
(2) The two procedures for peak discharge computations
in TR-55 lead to different results.
41
-------
(3) The second method giving peak flows in terms of areas
and slopes seems to significantly underestimate peak
flows.
(4) For the same meteorological input and rainfall losses,
simulations carried out by the "time of concentration"
method underestimated the peak flows for areas up to
160 acres as compared to SWMM.
(5) The tabular procedures in the SCS TR-55 for routing
runoff hydrographs through stream channels seem to
underestimate the flows and can be used only in con-
junction with a 24-hr., Type II storm.
(6) Direct use of the convex method seems to be more
accurate and can be used for any hydrograph profile.
FINAL REMARKS
The philosophy of the TR-55 seems to be that storm-
water management computations can be carried out by a step
by step method without giving the practitioners too much
information regarding the limitations of the method, testing
with meansurements or other models etc. while such an approach
makes the method relatively attractive, it may lead to consider-
able errors. On the other hand if properly applied some of its
features are quite attractive.
The main purpose of this paper is to invite discussion
regarding the use of such "black box" approaches whether it
42
-------
applies to desk top or computerized techniques. It is also felt
that agencies in charge with implementation of SWM and flood
control studies should carefully review methodologies and be
aware of their limitations.
REFERENCES
1. Altman, D.G., Espey, W.H., and Feldman, A.D., "Investiga-
of Soil Conservation Service Urban Hydrology
Techniques", Paper presented at the Canadian
Hydrology Symposium, Toronto, May 26-27, 1980,
13 pp.
2. Desbordes, M., "Estimation des Coefficients de Ruisselle-
ment Urbains", Centre Beige d1Etude et de
Documentation des Eaux, No. 376, Mars, 1975, pp.
106-110.
3. Heaney, J.P., et al. "Storm Water Management Model Level I
Preliminary Screening Procedures", EPA-600/2-76-
275, Oct., 1976, 77 pp.
4. Kassem, A.M., Roesner, L.A., and Wisner, P.E., "Updated
Documentation of the WRE-IMPSWM Transport Model
(January 1979 Version)", (Unpublished Report),
Dept. of Civil Engineering, University Of Ottawa,
Ottawa, Ont., Canada, October 1979.
5. Kassem, A.M., and Wisner, P.E.,"Assessment of Routing
Techniques in Storm Water Modelling (Unpublished
Report), Dept. of Civil Engineering, University
of Ottawa, Ottawa, Ont., Canada, June 1980.
6. Kibler, D.F., Monser, J.R., and Roesner, L.A., "San
Francisco Stormwater Model-User's Manual and
Program Documentation", Water Resources Engineers,
California, 1974.
7. "Manual of Practice on Urban Drainage", Draft No. 3,
Canada-Ontario Agreement on Great Lakes Water
Quality, Ministry of Environment, Ontario,
March 1977.
43
-------
8. Maryland Association of Soil Conservation Districts,
"Stormwater Management Pond Design Manual",
Maryland, Nov., 1978, 131 pp.
9. Metcalf & Eddy, Inc., University of Florida, and Water
Resources Engineers, Storm Water Management Model,
EPA, Wash., D.C., 1971.
10. Queens University Urban Runoff Study, Volume II, Supple-
mentary Information on Computer Programs and
Rainfall/Runoff Data Used in the Study.
11. Soil Conservation Service: Section 17, "National Engineer-
ing Handbook, U.S. Department of Agriculture, 1972,
12. U.S. Department of Agriculture, "Technical Release No. 55,
Urban Hydrology for Small Watersheds"/ Soil Con-
servation Service, Jan., 1975.
13. Viessman, W., Knapp, J., Lewis, G., and Harbaugh, T.,
Introduction to Hydrology", IEP - A Dunn
Donnelley Publisher, New York, 1977.
14. Williams, J.R., and Hann, R.W., "HYMO: Problem Oriented
Computer Language for Hydrologic Modelling",
USDA, ARS, May 1973, 76 pp.
15. Wisner, P., and Cheung, P., "Application of the HYMO Model
for Runoff Control Studies, Comparison with SWMM,
Limitations and Recommendations", IMPSWM Progress
Report No. 2, University of Ottawa, Aug., 1979,
29 pp.
16. Wisner, P.E., and Gupta, S., "Preliminary Considerations
on Selection of Design Storms'" IMPSWM Progress
Report No. 4, University of Ottawa, Aug., 1979,
31 pp.
17. Wisner, P., and Gupta, S., "An Assessment of the SCS
Procedures for Discharge Computations", IMPSWM
Progress Report No. 7, University of Ottawa,
Jan., 1980, 28 pp.
18. Wisner, P., Kassem, A., and Cheung, P., "Comparison of
Design Peak Flows Calculated by the Rational
Method and EPA-SWM Model", SWMM User's Group
Meeting, Gainsville, Florida, January 1980.
44
-------
A SIMPLIFIED STORMWATER QUANITY AND QUALITY MODEL
By
1 2
S. Sarikelle and Yu-Tang Chuang
INTRODUCTION
Urban stormwater management has drawn much attention in recent years due
to problems associated with quality of storm runoff and its detrimental effects
on receiving waters. Studies have shown that pollution carried by stormwater
discharging untreated and highly polluted street washings into receiving waters
exceed those discharges from secondary municipal treatment plant effluents (1,2,3).
Indeed, in many cases, the quality of receiving water is more likely to be
governed by waste produced in urban areas which is flushed away by runoff.
Due to the very complex nature of the urban rainfall-runoff quality process,
it is recognized that the utilization of simulation models provide an efficient
means for the investigation of the various aspects in urban drainage systems.
Over the years, various computerized mathematical models such as U.S. E.P.A.
Stormwater Management Model, University of Cincinnati Urban Runoff Model, and
Hydrocomp Simulation Program, etc. (4) have been developed. Most of the existing
models, however, have complicated structure, require extensive input data, and
do not incorporate the qualitative aspect of the urban runoff phenomena efficiently.
Therefore it is felt that there is a need to develop a quantity and quality simu-
lation model which utilizes simplified rainfall runoff relationship and yields
results which are comparable to those obtained by the more comprehensive and
complex models.
The study presented herein summarizes the development of a-quantity and
quality urban runoff simulation model with simple input requirements to be used
in the planning and analysis of stormwater systems. The model entitled MLSURM,
an acronym for a Modified Linearized Subhydrographs Urban Runoff Model, consists
of a quantity submodel and a quality submodel.
Professor, Department of Civil Engineering, The University of Akron, Akron,
OH 44325.
Research Associate, Department of Civil Engineering, The University of Akron,
Akron, OH 44325.
45
-------
Quantity Submodel
The quantity submodel basically represents the simulation of runoff for a
given area which results from a rainfall event. The method is based on a simpli-
fied concept (5,6) where hydrographs are generated for each subarea in accordance
with the duration of rain and the time of concentration. Three cases of linear-
ized subhydrographs are assumed for each subcatchment.
In Case I, tr = tc ; that is, storm duration equals the time of concentra-
tion for the subbasin. As shown in Figure 1, the peak runoff occurs when the
total flow from the subbasin contributes to the inlet the peak runoff rate is
defined by:
q = C i A (1)
where:
q = peak runoff rate (ftVsec)
C = runoff coefficient
i = intensity of rainfall (inches/hour)
A = area of the subbasin (acres)
The runoff rate of the rising and receding limbs are determined by assuming
a linear relationship between the runoff and time although it is known that they
are of a nonlinear nature. The linear assumption is realistic since the subbasins
are small the nonlinear variation of the subhydrograph can also be assumed to be
small. Therefore, for t £ t , the runoff rate q. at time t is given by:
qt = C i A £- (2)
L*
For t > tr, the runoff rate is calculated by:
t + t - t
qt = C i A (-C ^ ) (3)
c
The time base of the subhydrograph t. is computed by:
*b - \ + *c = 2tr (*>
The volume of runoff BV resulting from the storm is then calculated by:
V = C i A tr (5)
In Case II, the storm duration is assumed to be greater than the time of
concentration for the subbasin, that is tr>tc. Thus, after a time period equal
to the time of concentration, the peak runoff rate is reached and remains constant
46
-------
re
S- c
C
i- «M
(0 C
Q£
t * storm duration
i = storm intensity
O
3
V^C
qp = C i A
*b = V^c
V = C i A t
Time After Start of Rainfall, t
Figure 1 Case I Linearized Subhydrograph
47
-------
until the rainfall stops. Then, the runoff recedes to zero in a time period whicn
equals to t . As shown in Figure 2, a trapezoidal shape subhydrograph is pro-
posed. Thecrunoff rate of the subhydrograph is given by:
For t < tr, qt = C i A -£- (6)
c t tc
For t £ t < t, q = C i A = q (7)
C I- r
t + t - t
For t > t , q = C i A (-£ ^ ) (8)
**
The time base of the subhydrograph is given by:
The runoff volume is then computed by:
V = C i A tr (10)
In Case III, t < t , that is the time of concentration for the subbasin
is greater than the StorrrTduration. Thus, the equilibrium runoff rate is not
reached when the storm ceases. The peak runoff rate is calculated by:
t
qD = C i A ^ (11)
K c
After the storm stops, the runoff rate will remain constant until a time t is
reached, then the runoff recedes to zero (7). The value of t depends on pthe
values of t and t and is given by P
i \*
tp = 0.4 tr + 0.6 tc (12)
Therefore, the runoff rate of the subhydrograph is given by
For t < tr , qt = C i A -^- (13)
_t
For tr £ t £ tp, qt = C i A:-£- (14)
*r *h - l
For t > t , qt = C i A T1- ( p ) . . . (15)
48
-------
>i
f +*
* «f~
(O (A
C 0)
t- 4->
-------
where:
tb = 0.6 tr + 1-4 tc . .
The runoff volume is determined by:
V = C i A tr '
A schematic representation of this case is shown in Figure 3.
The runoff coefficient C represents the abstractions, or "losses", between
rainfall and runoff generated from a particular subcatchment. It is noted that
the abstractions decrease in magnitude as the duration of the storm increases.
In the development of MLSURM model, Hoad's runoff coefficients (8) as shown in
Figure 4 are adopted.
The time of concentraion for a subbasin is equivalent to the inlet time,
this is the time required for the surface runoff to flow from the most remote
point of the subbasin to the inlet of the subbasin. Various methods of estimating
time of concentration for a drainage area have been proposed. For overland flow,
a formula based on kinematic wave theory (9) was used in the model. The equation
is given by the following relationship:
L°-6 N°-6
t = 0.928 iu-4 S"''3 (18)
where
t = time of concentration (minutes)
L*
L = length of overland flow (feet)
N = Manning's roughness coefficient
i = intensity of excess rainfall (inches/hour)
S = average overland slope (foot per foot)
For large subbasins with street curbs, the time of concentration is obtained
by summing the overland time and the time of travel in the gutter. The overland
time for the impervious area or the pervious area is determined by using the dis-
tance from upstream portion of the impervious area or pervious area to the gutter
as overland length. Based on overland time, an initial overland hydrograph con-
tributing to the gutter is obtained by the linearized subhydrograph procedure.
Gutter flow travel time computed using the initial overland hydrograph is then
included in the time of concentration to develop the inlet hydrograph.
In application where sewered areas are considered, the hydrograph resulting
from each subcatchment must be routed through the sewer system to obtain the out-
50
-------
-a
-------
en
o
»
4->
c
*
u
ex.
o
o
in
r-
o
o
in
in
ru
o
o
Hoad: C =
t + 8
Impervious area
Hoad: C =
0.5t
t + 15
Improved pervious area
",. a .. r _ 0.3t
Hoad. C - irr-2Q
Sandy pervious area
ah. oo sb.oo 75.00
Duration of Rainfall, t, Minutes
100.00
o
m
~o
o
ro
125.00
150.00
Figure 4
Runoff Coefficients vs Duration of Rainfall and
Area Characteristics (21)
-------
fall hydrograph. The "time-offset" routing procedure developed by Tholin (10; is
utilized. The travel time in the sewer is determined by assuming uniform flow
condition in the sewer pipe. The inflow hydrograph is then offset according to
the travel time. In this fashion the routed hydrographs are then utilized to
develop system hydrographs at various points in the sewer network.
Quality Submodel
The development of the quality submodel centers on the determination of
pollution levels in the stormwater. Surface pollutants consist of street litter
and dustfall that accumulate on the ground and street surfaces prior to a storm.
When the storm occurs, the accumulated materials on the surface are dissolved by
rain. As rainfall continues, surface runoff begins to wash off the pollutants.
The impact of the raindrops on the relatively rough surfaces provides a high level
of turbulence which tends to accelerate the pollutant removal process.
Keeping in mind the complexity of the pollutant removal process, an attempt
is made to develop a relationship that would be able to describe such phenomenon
and aid in the proper simulation of the pollutant wash-off from the basin surfaces.
The relation is given as follows:
dP_- k
dt k
(19)
where:
fir = rate of washout of pollutant (pounds/hour)
P = amount of pollutant remaining on the surface (pounds)
q = rate of runoff (inches/hour)
S = overland slope (ft/ft)
y = depth of flow (ft)
K = proportionality constant
Integrating equation (19) yields
where:
P = p e /y
Kt+At t e
.(20)
t At
P.
t
At
Av
amount of pollutant remaining on the surface at time
t+At (pounds)
amount of pollutant on the surface at time t (pounds)
time interval (hour)
incremental runoff volume during At (inches)
53
-------
y = depth of runoff flow (ft)
S = overland slope (ft/ft)
K = proportionality constant
Thus, the rate of surface pollutant washout Mg in pounds per hour is:
Ms = (Pt - Pt+At)/ At = Pt (1 - e * )/ At . - - (21)
Equation (21) gives the amount of surface runoff pollutograph corresponding to
the inlet hydrograph. The equation is applied successively, the value of Pt+At
which is determined at the end of the current interval becomes the value
of Pt at the beginning of the next interval.
Equation (21) attempts to model the surface pollutant removal process with
the following restrictions:
1. For catchment subjected to identical rainfall input, the one
with steeper overland slope would result in faster pollutant
removal rate.
2. For a particular catchment where overland slope is fixed, the
motive force of washing out the pollutant then depends on the
ratio, AV//y. The higher this ratio, the more significant the
motive force becomes in washing out the surface pollutant. This
relationship indicates the effect of incremental runoff volume
and the effect of the depth upon pollutant removal efficiency.
In applications to watersheds with sewer systems, the pollutographs entering
into the sewer must be routed to yield the outfall pollutograph. The routing
procedure similar to that used in the SWMM model (1.1) is incorporated in the model
developed herein.
Model Description
The model structure basically follows the concepts presented in the preceding
sections. A generalized flow chart that shows the basic quantity and quality simu-
lation algorithm is shown in Figure 5. The model is divided into a main driving
program and 43 subprograms. The object-time dimension feature of FORTRAN IV is
utilized to allocate the adjustable dimension variables into a single one-dimen-
sional array. The size of the array is declared in the main program. Advantages
in the program's memory storage allocation make it compatible with standard
compilers so that computer systems comparable to that of the IBM 370 or UNIVAC 1108
can be used.
Application of Model
The model was applied to two typical urban drainage areas with measured
runoff data corresponding to various gaged storm events.
54
-------
f START )
1. Assign Storage for S Array
2. Define MSOS
3. List Input Card Images
YES
Read in the Case Study
Title and Control Parameters
QUANTITY SIMULATIONS
1. Read in and Allocate Data
2. Develop Inlet Hydrographs
3. Route Inlet Hydrographs
Through Sewer Systems
Quality
Simulations
Desired?
Print
Hydrograph Output
QUALITY SIMULATIONS
1. Read In and Allocate Data
2. Develop Inlet Pollutographs
3. Route Inlet Hydrographs
Through Sewer Systems
4. Print Hydrograph and
Pollutograph Output
Figure 5 Generalized Flowchart of MLSURM Model
55
-------
The first drainage area selected is the Oakdale Avenue Basin located in a
residential section about six miles northwest of downtown Chicago. The 12.9 acre
basin is composed of 7.05 acres of pervious area and 5.85 acres of impervious
area. Due to the small size of the basin, the entire drainage area is treated
as a single catchment in the computer simulation instead of being subdivided
into individual subcatchments. Rainfall and runoff data have been recorded since
1959 by the City of Chicago's Bureau of Engineering (12).
The simulation of the storm occuring on July 2, 1960 is shown in Figure 6.
In comparing the simulated hydrographs to the recorded hydrographs, the time to
peaks and the peak discharges from the MLSURM model yielded fairly good agreement.
The second application of the model is the Mortimer Avenue Basin in Toronto,
Canada. The basin is located about four miles to the northeast of downtown
Toronto in the Borough of East York. The 383 acre catchment has been subdivided
into 33 subcatchments. A study by M.M.Dillon Limited Consulting Engineers (13)
generated gaged rainfall data and runoff data for this basin. The runoff hydro-
graphs, the suspended solid pollutgraphs and levels of BOD were also determined.
However, no precise information about the amount of pollutant accumulated on
the surface prior to a storm was available. Therefore, the accumulation rates
of the dust and dirt, pollutant content of the dust and dirt,and the pollutant
removal rate constant K were calibrated by comparing the simulation results
against recorded values. The calibrated value of these parameters are shown in
Table I. The simulation results are presented in Figures 7 and 8. As can be
seen from the results, both the peak discharges and the time synchronization of
the MLSURM simulated hydrographs are in good agreement with the recorded hydro-
graphs. In the simulation of pollutographs, the results obtained by the MLSURM
model also compare well to the recorded pollutographs.
Conclusions
An urban runoff model entitled MLSURM is developed for the simulation of
storm water quantity and quality. The following conclusions are reached based on
the test application of the model:
1. The model simulated stormwater hydrograph well for small urban
catchments with little calibration effort. The successful
simulation of hydrographs for large watersheds is also achieved
if the watersheds are subdivided into small catchments.
2. Satisfactory pollutograph simulations were also obtained by
using calibrated parameter values.
The model may be used as an alternative to the more comprehensive and complex
models in the planning and analysis of stormwater systems. It incorporates simple
but physically realistic parameters in the simulation of runoff of stormwater
systems. However, the model should be applied to more watersheds including urban,
suburban, and rural areas where runoff quantity and quality data are available.
56
-------
i CM
inr
e
o
en
u.
u
o
_i
u.
20. oo
-------
Table I Calibrated Accumulation Rates of Dust and Dirt,
Pollutant Contents in Dust and Dirt,
Pollutant Removal Constant K
Land Use
Single family, residential
Commercial
Undeveloped or Park
Accumulation Rate of Dust and Dirt
(Pounds/Day/100 ft - curb)
1.4
6.6
3.0
Land Use
Single family, residential
Commercial
Undeveloped or Park
Pollutant Removal
Constant K
Milligram pollutant/gram Dust and Dirt
Suspended Solids BOD
160
160
160
Suspended Solids
10.
35.
30.8
35.
BOD
15.
58
-------
--- Recorded
sww si«
HLSURH simulated
ik.w ih.toiV.« rt.Mtk.ooiif.zo IM.\OissTtom. w
TIME (HINUTES)
- R
i '*
f
I
\
oo ik.«
\
\
\
«7.« rt.to ik.oo its.w m.«a IM.M rtz.to
TIHE (MINUTES)
Figure 7 Results from Mortimer Avenue Basin, Toronto, Canada
Storm of June 30, 1976
59
-------
U
a
S 8.
H 8
I «
Recorded
St#t1 simulated
HLSUHM
7k, M »k.oo
TIME miHUTESI
tki.m
TIME MINUTES)
Figure 7 (continued) Results from Mortimer Avenue Basin,
Toronto, Canada
Storm of June 30, 1976
60
-------
"liToo
il.w ak. «o
iV.io ik.H tfc.oo
TIHC CHIHUTCS)
iic.zb ib«.«o
iz. to
fV.M 7i.»0 A.Mtls.M
TIHC ININUTES)
Figures Results from Mortimer Avenue Basin, Toronto, Canada
Storm of July 31, 1976
61
-------
- Recorded
jlmullted
N.SUAM simulated
iTio5/io RTiio ToTwSTw tlslotSTw tks.w
TIME IHINUTES)
ir^r
TIHE MINUTES)
iisl
o
Figure 8 (continued) Results from Mortimer Avenue Basin
Toronto, Canada
Storm of July 31, 1976
62
-------
REFERENCES
1. E.P.A. Technology Series, "Characterization and Treatment of Urban Land
Runoff," EPA-670/2-74-096, December 1974.
2. E.P.A. Technology Series, "Nationwide Evaluation of Combined Sewer Over-
flow and Urban Stormwater Discharges," EPA-600/2-77-064, Three volumes.
3. Whipple, W., Hunter, J.V., Yu, S.L., "Unrecorded Pollution from Urban
Runoff," Journ. Water Poll. Control Fed., 46, 873 (1974).
4. E.P.A. Technology Series," Assessment of Mathematical Models for Storm and
Combined Sewer Management," EPA-600/2-76-175a, August 1976.
5. Chien, J.S., Saigal, K.K., "Urban Runoff by Linearized Subhydrographic
Method," Journal of the Hydraulics Division, ASCE, Vol. 100, No. HY8,
Proc. Paper 10766, August 1974, pp. 1141-1157.
6. Sarikelle, S., Chien, J.S., French, G.L., "Development of Linearized
Subhydrographs Urban Runoff Model," International Symposium on Urban Storm
Water Management, University of Kentucky, Lexington, Kentucky, July 26-
27, 1978.
7. Henderson, F.M., Wooding, R.A., "Overland Flow and Groundwater from a
Steady Rainfall of Finite Duration," J. Geophys. Res. 69, No. 8, pp. 1531-
1540, April 1964.
8. Fair, G.M., Geyers, J.C., Okun, D.A., Water and Wastewater Engineering,
Vol. I, John Wiley & Sons, Inc., New York, 1966.
9. Ragan, R.N., Durn, J.O., "Kinematic Wave Nomograph for Time of Concentra-
tion," Journal of the Hydraulics Division, ASCE, Vol. 98, No. HY10,
Proc. Paper 9275, October 1972, pp. 1765-1771.
10. Tholin, A.L., Keifer, C.J., "The Hydrology of Urban Runoff," Transactions,
ASCE, Vol. 125, Paper No. 3061, 1960, pp. 1308-1355.
11. E.P.A. Technology Series, "Stormwater Management Model Vol. I, Final
Report," EPA 11024 DOC 07/71.
12. Tucker, L.S., "Oakdale Gaging Installation, Chicago - Instrumentation and
Data," ASCE Urban Water Resources Research Program, Technical Memorandum
No. 2, August 5, 1968.
13. M.M.Dillon Limited Consulting Engineers, "Draft Report on the Storm Water
Management Model Verification Study, February 1978.
63
-------
METHODOLOGY FOR 'LUMPED' SWMM MODELLING
by
M. AHMAD1
INTRODUCTION
The purpose of this paper is to present a systematic methodology for
lumping or aggregating urban drainage areas when using the Storm Water
Management Model (SWMM) to simulate the rainfall-runoff process. In a
lumped model the study area is discretized into large subcatchments (i.e.
coarse discretization) and as such the spatial details of hydrologic
characteristics and the internal drainage system of the area are not
explicitly modelled. This approach of using coarse discretization in
watershed modelling studies considerably reduces the costs of setting up
and running the SWMM simulation. However, it also implies that modelling
accuracy will suffer unless appropriate steps are taken to account for the
effect of the omitted internal details. The main objective of the
methodology presented herein is to overcome the above limitation
associated with the concept of "lumped modelling".
The growing popularity of SWMM among planners and decision makers over the
past few years has expedited the need for developing simplified procedures
for lumped simulation. Although use of lumped simulation techniques can
considerably reduce the overall modelling costs, selection of a particular
discretization level depends entirely on the objective and the nature of
the study under consideration. For instance, if detailed information on
the hydraulic performance of all major conduits in a basin is required,
then it would be necessary to use a detailed discretization scheme.
Drainage Systems Upgrading Engineer, Edmonton Water and Sanitation,
12220-Stony Plain Road, Edmonton, Alberta, T5N 3M9
64
-------
Conversely, if the purpose of the study is to develop a hydrograph at the
outlet of a catchment, then a coarse discretization through the use of a
few larger subcatchments and conduits would be more appropriate.
Edmonton Water and Sanitation (Drainage Engineering Section) has been
using SWMM and other models extensively since 1975 for analyzing the
existing combined and separated storm trunk systems (Ref. 1). These
models are being used mainly for designing relief sewers and for drainage
planning purposes. The existing trunk sewer systems in the city serve a
total area of over 50,000 acres. By the end of this year, hydraulic
analysis of all the major trunk systems would essentially be completed and
all data would be stored on the computer for future planning and sewer
updating purposes. When analyzing large and complex trunk systems such as
those existing in Edmonton, it is essential to employ simplified modelling
techniques which reduce the amount of effort required in data preparation
and also the simulation costs.
The author, therefore developed the simplified methodology given in this
paper for analyzing the existing trunk systems using lumped modelling
approach. As a major part of this methodolgy, the concept of equivalent
gutter was introduced in RUNOFF block simulation to compensate for the
eliminated conduit storage existing within the lumped catchment. In
addition, a set of generalized curves relating in-system conduit storage
to impervious area were developed using relevant data from new residential
and idustrial subdivisions in Edmonton. Similarly, curves relating the
drainage area to the peak flow for a range of imperviousness values were
also generated. A systematic step-by-step procedure that uses these
curves to determine the overland flow "width" parameter and the dimensions
of the representative equivalent gutter appropriate for the lumped
catchment was formulated.
The lumping methodology presented in this paper was tested against
detailed simulations using rainfall and flow measurements for three
recorded storm events for the Norwood test area. Modelling results
employing lumped and detailed discretization schemes, respectively, were
65
-------
also compared for the 5 - year design storm using the catchment data for
Fulton Drive basin. Comparisons of detailed and lumped simulations for
both test areas were found to be reasonably good.
PREVIOUS WORK
The concept of lumping in computer modelling work is not a new one.
Literature contains many references on lumped modelling. An excellent
reference on this subject is the Canadian Storm Water Management Model
Study conducted jointly by J.F. MacLaren Ltd. and Proctor and Redfern Ltd.
(Ref. 2). A brief summary of the findings of previous work (Refs. 3,4,5
& 6) related to lumped simulation, is presented in the above Canadian SWMM
Report. While the previous studies indicated the feasibility of lumped
simulation, they did not evolve and present a systematic procedure or
methodology for lumped modelling.
The Canadian SWMM study suggested two alternate methods for introducing
additional storage into the lumped model so as to essentially compensate
for the unmodelled conduit storage. The study showed that the designed
attenuation and the time delay of the hydrograph, which are critical for
the accuracy of the lumped simulation, could be achieved either by
increasing the length of the aggregated conduit or by reducing the
overland flow "width" of the "lumped" catchment. The study concluded that
it was possible to conduct an accurate lumped simulation for a particular
catchment by using average values for parameters describing surface runoff
characteristics of the lumped catchment (ie. infiltration, detention
depth, ground slope, Manning's 'n'). However, the study also indicated
that it was not possible to determine the appropriate value of overland
flow "width" for the lumped catchment without first conducting a detailed
simulation. Similarly, the length of the equivalent gutter needed to
produce consistent results from lumped simulation could not be determined
without conducting a detailed simulation. As a result, the study was
unable to recommend a generalized procedure for the lumped simulation.
66
-------
METHODOLOGY FOR LUMPED SIMULATION
Previous work has shown that an accurate lumped simulation can be carried
out either by reducing the value of overland flow "width" parameter in
RUNOFF block or by selecting a "reasonably" large equivalent conduit to
compensate for the omitted conduit storage in the lumped catchment (Ref.
2). Results of a previously conducted RUNOFF block sensitivity analysis
study (Ref. 7) indicated that while it was difficult to relate the value
of "width" parameter to the eliminated in-system conudit storage, it was
much easier to determine the size and length of the equivalent gutter for
a lumped catchment. Therefore, a simplified procedure was formulated to
estimate the section dimensions and the length of the equivalent gutter
for general use in a lumped simulation. Based on the results of our
sensitivity analysis, a relationship between catchment "width" and
drainage area was derived as illustrated in Figure 1. "Width" values in
UI
cc.
FIGURE I
DRAINAGE AREA
VS
OVERLAND FLOW WIDTH
30 40 SO SO
DRAINAGE AREA IN ACRES
feet/acre appear not to be very sensitive to changes in the drainage area
size. Further, it was discovered that the peak flow was not sensitive to
the changes in "width" especially at the higher "width" values as shown in
Figure 2. The total runoff volume was also found to be quite insensitive
to the variation in "width" as shown in Figure 2. Using the results of
67
-------
,j
z
^80000-1
gTOOOO
^SOOOO
s
NOTE
FOLLOWIN8 TEST DATA USED
DRAINAGE AREA ' 33.3 Ac
IMPERVIOUSNESS - 4S %
SUaCATCHMENT WIDTH " 44«2 FEET
PEAK DISCHAME
FIGURE 2
SUBCATCHMENT OVERLAND
FLOW WIDTH
V3
PEAK DISCHARGE
AW FLOW VOLUME
W/Z
.2*
1000 2000 SOOO 4OOO MOO «000 TOOO 6000 9000 10000
SUBCATCHMENT WIDTH IN FEET
this sensitivity analysis, a simplified lumping procedure described below
was developed. This step-by-^tep procedure can be used to determine the
necessary input parameters for a lumped catchment. For simplicity, a
rectangular equivalent gutter is suggested for lumped simulation.
Lumping Procedure
1. Determine the average values of various parameters describing surface
runoff characteristics including imperviousness ratio, "width",
infiltration, detention depth, ground slope and Manning's 'n'.
(value of "width" can be estimated using the curve shown in Figure
1).
2. Calculate total length of each sewer size in feet.
3. Calculate available conduit storage in cubic feet for each sewer size
by multiplying the cross-sectional area of the sewer and the total
length computed above in step 2.
4. Calculate total available storage by adding storage values for all
sewer sizes.
68
-------
5. Estimate the 5-year design peak inflow (or any other flow as
required) to the equivalent gutter. For the City of Edmonton, the
5-year design storm flows can be estimated from the generalized
curves shown in Figure 3. These curves were developed using the
DRAINAGE AREA IN ACRES
results of RUNOFF block simulations for a number of newly developed
residential and industrial areas. These curves show a relationship
between drainage area and the peak flow for 3 different imperviouness
ranges for the 5-year design storm. The peak flow values used for
these curves are pre-routing flows. For example, curve 1 (Figure 3)
represents a relationship between drainage area (acres) and peak
discharge (cfs) for catchments with impervioness ratio ranging from
15% to 30%.
6. Select an appropriate width 'b1 for the equivalent gutter
(rectangular) based on the peak inflow estimated in step 5. By using
Figure 4 (which shows relationships between peak inflow and gutter
depth for a number of selected widths), determine ilepth of'flow 'd'
in the gutter for the peak inflow obtained in step 5. (Equivalent
gutter is designed at a slope of 0.5% with a Manning's 'n1 value of
0.018.)
7. Determine the length of the equivalent gutter in feet by dividing
the total storage volume computed in step 4 by the cross-sectional
area of the gutter (b x d).
69
-------
MOTE Gurreft DESKIWD AT A SLOPE
OFOS% WITH »«««W6'S V OF O.OH
PEAK OiSCHAflGE IN C.F.S.
It is evident from the above procedure that in order to estimate in-system
conduit storage accurately, one has to spend a considerable amount of
time, especially if a large basin is involved. Also, the above method
cannot be applied directly to the presently undeveloped areas. Therefore,
to overcome these problems, two generalized curves relating the in-system
conduit storage to the impervious drainage areas for predominantly
residential and industrial areas, respectivley, were developed using
regression analysis techniques. Sewer system data for the newly developed
residential and industrial subdivisions were used for the statistical
analysis. Figure 5 shows the curves for both residential and industrial
areas. Storm sewer system in these areas have been designed to carry the
5-year design storm flows. The total amount of in-system conduit storage
can be estimated using curves shown in Figure 5 instead of going through
IMPERVIOUS AREA IN ACRES
70
-------
steps 2 to 4 as suggested in the lumping procedure. Again, curves shown
in Figures 3 and 5 are valid only for Edmonton conditions and their use
for other cities is not recommended.
TESTING AND APPLICATION OF LUMPING METHODOLGY
Fulton Drive Test Catchment
The proposed lumping methodology was tested using Fulton Drive test
catchment data for the 5-year design storm conditions. A 38.91 acre
subarea with an average imperviousness of 34.12% was selected for
comparison purposes. Schematics of the test area and the sewer system are
shown in Figures 6 and 7, respectively. The test area was divided into 12
subcatchments for a detailed simulation and input data was prepared using
the procedure given in the SWMM User's Manual (Ref. 3). Equivalent
parameters were then estimated using the lumping procedure described
earlier to allow the entire 38.91 acre catchment to be modelled as a
501 ) SUBCATCHMENT NUMBER
FIGURE s
SCHEMATIC OF FULTON OR.
DETAILED AREA
SUBCATCHMENT BOUNDARIES
71
-------
704
f-
" i i <
M
<£>
= 0 o ffi S
m 2 in O &
55 in in o to in 10
24"~ 24" ~ 24" "" 2t" ~ 18" ~ 15" ~ 15"
Vo, K ,0^8
*" o o o o to
m* t i r £ i in
8T ll" " l«" " 15" ~ IZ'
702
«r « PJ o
o o Q o in
in in K ,n __ «,
(^ * « l
* ,.- " IB" "12" '
FIGURE 7
SCHEMATIC OF FULTON DR.
DETAILED AREA SEWER SYSTEM
AS MODELLED
I02A AVE.
l-
o
DIA AVE.
501 NODE NUMBER
single catchment. Equivalent parameters were also estimated by dividing
the entire area into 3 subcatchments. SWMM simulations were carried out
using the 5-year design storm for three cases where the study area was
represented by a single lumped catchment, 3 subcatchments and 12
subcatchments respectively. For detailed simulation with 12
subcatchments, all sewers were modelled in the TRANSPORT block. Lumped
simulations with one catchment and with 3 subcatchments were carried out
using only the RUNOFF block.
The results of the simulations are summarized in Table 1 and the simulated
hydrographs are shown in Figure 8. Detailed and lumped simulations
compare reasonably well for the tested conditions. However, it should be
noted that the simulated peak flow rate with single lumped catchment was
about 20% higher than that with detailed discretization using 12
subcatchments, when both simulations were made with a 5-year design storm
(Chicago type) discretized into average 5 minute intensities. As
illustrated in Figure 8, improved simulation results were obtained for the
lumped simulation when the same design storm was discretized by averaging
the highest peak intensities over 15 minutes (which happens to be the time
of concentration of the catchment). Difference in simulation results with
3 subcatchments and 12 subcatchments was insignificant. Therefore, it can
be concluded that as part of the above lumping procedure, the design storm
72
-------
TABLE 1
FULTON DRIVE TEST CATCHMENT
LUMPED SIMULATION
(5-YEAR DESIGN STORM)
Number of
Sub catchments
Hyetograph
Discretization
Interval
(minutes)
Overland Flow
Width (Feet)
Peak Flow
(cfs)
Total Runoff
Volume
(cu. ft.)
1
1
3
12
5
15
5
5
4,900
4,900
4,900
9,800
37.5
32.0
33.0
31.2
69,162
69,162
68,946
68,958
45
40
FULTON DRIVE TEST CATCHMENT
COMPARISON OF
SIMULATED HYOROGRAPHS
FOR VARIOUS LEVELS
OF DISCRETIZATION WITH
THE S-YEAR DESIGN STORM
I SUBCATCHMENT (LUMPED SIMULATION)
WITH'5 MINUTE RAINFALL AVERAGINS
3 SUBCATCHMENTS (LUMPED SIMULATION)
I SUBCATCHMENT (LUMPED SIMULATION)
WITH 15 MINUTE RAINFALL AVERAGING
12 SUBCATCHMENTS (DETAILED SIMULATION)
NOTE-
FOLLOWING TEST DATA USED
SUBCATCHMENT AREA = 38.91 Ac.
IMPERVIOUSNES8 - 34.12 %
GROUND SLOPE = 0.007
EQUIVALENT GUTTER LENGTH
FOR LUMPED CATCHMENT = 1211 FEET
0=30 I'OO ('-SO 2=00 2-30 3*00 3>30 4=00 4=30 8=00 830
TIME IN HOURS
73
-------
should be discretized by averaging the highest intensities occurring
during an interval of time approximating the estimated time of
concentration of the catchment under consideration.
Norwood Test Area
The lumping mthodology was also tested using data for the Norwood area.
This is an old residential neighbourhood in which relief sewers were
constructed in 1976-77. A 32.6 acre subcatchment in this area was
selected for the lumped simulation testing purposes. Imperviousness of
the area is approximately 26%. A schematic of the test area is shown in
Figure 9 and the existing storm sewer sytem is illustrated in Figure 10.
The test area is drained by a 36" storm sewer which discharges into an
111" storm trunk along lllth Avenue.
Detailed SWMM simulations were carried out by discretizing the test area
into 19 subcatchments ranging in size from 1 acre to 3 acres and by using
IGURE 9
SCHEMATIC OF NORWOOD
PARTIALLY SEPARATED
AREA BOUNDARIES
74
-------
(£ ift K
112 AVE.
>-110 = CONDUIT NUMBER
113 AVE.
NORWOOD
Q DROP STRUCTURE
II NODE NUMBER
FIGURE 10
SCHEMATIC OF NORWOOD
SEPARATE SEWER SYSTEM
AS MODELLED
input data prepration procedures as outlined in the SWMM User's Manual
(Ref. 3). The input parameters for lumped SWMM simulations were estimated
using the lumping methodology presented in this paper. Rainfall and flow
data for the three recorded storm events (July 10th and llth, 1978 and
June 14th, 1979) were used to compare the recorded hydrographs with those
computed by SWMM for detailed and lumped simulations, respectively.
Lumped simulations were carried out using both 5 and 15 minutes hyetograph
discretization levels as described earlier for the Fulton Drive area. The
recorded hydrographs are compared with lumped and detailed simulation
hydrographs in Figures 11, 12, and 13. A very good comparison is found
with 15 minute hyetograph descritization for the tested storm events. The
results of these simulations are summarized in Table 2. These results
clearly indicate that by properly applying the lumping methodology, it is
possible to achieve an accurate lumped simulation. Furthermore, it
confirms that input rainfall hyetograph should always be discretized
according to the catchment size. For instance, if the estimated time of
concentration of an area is 20 minutes, then the input rainfall hyetograph
should be discretized by averaging the highest rainfall intensities over a
time interval of 20 minutes.
This lumping methodology which is also summarized in Figure 14 has been
applied successfully in the City of Edmonton for analyzing the existing
75
-------
rot TX .HUT to. «T« i
t*. 01 oo oroo
MOO arvt 0*00
IfrflO 1«>00
I -
.
(j) I 9MCATCHMCHT (lUWKV *NULArtCMI
VTTH S IMHUTC UMTALL MEMtHI
0
SSL.
reuoMi TEST Dm «e*c utn
TOTAL in*. «n IMTHI *«* i M* '
««« ««!*>«» »_.
J MM MtUlATCB
HTMMDAM)
1 TM Jam M, VT« STOBM
TIME IN HOURS
TABLE 2
NORWOOD TEST AREA
LUMPED SIMULATION
(RECORDED STORMS)
Storm Number of
Subcatchments
1978 07 10
1978 07 10
1978 07 10
1978 07 11
1978 07 11
1978 07 11
1979 06 14
1979 06 14
1979 06 14
19
1
1
19
1
1
19
1
1
Hyetograph Overland Flow Peak Flow
Discretization Width (Feet) (cfs)
Interval
(Minutes)
5
15
5
5
15
5
5
15
5
14,433
4,458
4,458
14,433
4,458
4,458
14,433
4,458
4,458
Rec.
14.0
14.0
14.0
21.2
21.2
21.2
8.8
8.8
8.8
Sim.
15.3
15.4
16.2
21.2
17.6
17.5
10.3
9.5
10.6
Total Simulated
Runoff Volume
(cu. ft.)
33,500 '
35,100
35,100
92,500
88,500
88,500
26,500
26,900
25,900
76
-------
FIGURE 14
METHODOLOGY FOR LUMPED
SWMM RUNOFF SIMULATION
FOR URBAN AREAS
DETERMINE AVERAGE VALUES FOR IMPERVIOUSNESS.
WIDTH.DETENTION STORAGE,GROUND SLOPE AND
MANNING'S X FOR LUMPED CATCHMENT
ESTIMATE TOTAL IN-SYSTEM CONDUIT STORAGE
VOLUME USING FIGURE 5
ESTIMATE PEAK FLOW RATE USING FIGURE 3
ESTIMATE DIMENSIONS (WIDTH AND DEPTH) OF
EO.GUTTER USING FIGURE 4 AND CALCULATE
CROSS-SECTIONAL AREA
CALCULATE LENGTH OF EO. GUTTER
TOTAL STORAGE VOLUME
" CROSS-SECTIONAL AREA
DISCRETIZE INPUT RAINFALL HYETOGRAPH BASED
ON TIME OF CONCENTRATION
PREPARE RUNOFF BLOCK INPUT DATA
MAKE SWMM SIMULATION
77
-------
storm trunk systems as well as for designing new storm trunks. At
present, studies are underway to develop similar procedures for larger
developing drainage areas.
CONCLUSIONS AND RECOMMENDATIONS
The simplified lumping procedure described in this paper can be applied
for analysing and designing of storm sewer systems in urban basins without
using detailed SWMM simulations. A significant reduction both in the
amount of effort required in input data prepration and in the overall
simulation costs can be achieved by employing this lumping methodology.
Generalized curves relating impervious drainage area and in-system conduit
storage similar to those given in this paper for Edmonton can be developed
for other cities. Further studies should be undertaken to develop
simplified methods for other SWMM applications such as storage volume
estimation for stormwater lakes, runoff peak and volume computations for
pre-development conditions, etc.
REFERENCES
1. Ahmad, M. "Stormwater Modelling Applications in the City of Edmonton",
paper presented at SWMM Users Group Meeting, Gainesville, Florida,
January 10-11, 1980.
2. Proctor and Redfern Ltd., and James F. MacLaren Ltd., "Storm Water
Management Model Study", Ontario Ministry of the Environment, 3
Volumes, 1976.
3. Metcalf & Eddy, Inc, University of Florida, and Water Resources
Engineers, Inc." Storm Water Management Model", U.S. EPA Report 11024
DOC 07/71, 4 Volumes, October 1971.
4. Keeps, D.P. and R.G. Mein, " An Independant Evaluation of Three Urban
Storm Water Models", Monash University Civil Engineering, Res. Report
No. 4, 1973.
78
-------
5. Shubinski, R.P., and L.A. Roesner. "Linked Process Routing Models",
Paper presented at American Geophysical Union Annual Spring Meeting,
Washington, D.C., April 16-20, 1973.
6. Jewell, T.K. et al, "Application and Testing of the EPA Storm Water
Management Model to Green Field, Massachusetts". In "Short Course on
Applications of Storm Water Management Models" sponsored by the
V.S.E.P.P. and the University of Massachusetts, August 19-23, 1974.
7. "RUNOFF block sensitivity Analysis Study", In-house study by Edmonton
Water and Sanitation, 1978.
79
-------
CHARACTERIZATION, MAGNITUDE AND
IMPACT OF URBAN RUNOFF IN
THE GRAND RIVER BASIN
by
S.N. SINGER and S.K. SO
Water Resources Branch
Ministry of the Environment
INTRODUCTION
The general goals of the Grand River Basin Water Management Study are:
1- to develop viable water management options needed to plan for, and
encourage, the integrated use of water and land resources, within the
Grand River Basin.
2- to identify the necessary trade-offs to achieve protection against
flooding, acceptable disposal and transport of waste effluents.
3- to provide adequate supplies of good quality water to meet water
supply, aesthetic, fish, wildlife and recreation desires and needs.
4- to ensure a productive and fulfilling environment for the people of the
basin.
In summary, the three water management objectives of the Grand River Basin
Study are to reduce flood damages, to provide adequate water supply and to
maintain an acceptable water quality.
80
-------
Water quality constitutes an important component of the study. The key
elements of water quality investigation are:
1- to determine existing water quality conditions and relate them to
various water uses.
2- to identify the type, magnitude, relative significance and impact of
pollutants from point and nonpoint (urban and rural) sources on the
water quality of the river.
3- to develop water quality management programs to preserve areas of high
water quality and to upgrade areas where poorer water quality exists.
Urban stormwater runoff has been recognized as a potential major
contributor of pollution to the Grand River. Therefore, investigation of
pollution from urban sources became an integral part of the basin's water
quality assessment program. Because costs associated with the abatement of
urban stormwater pollution range in the tens of millions of dollars it was
important to assess the impact of this pollution on the river and to
determine its significance.
This paper contains a description of the Grand River Basin and its major
urban centres and an overview of urban data collection programs.
Significant results to date related to the characterization of urban
runoff, magnitude of pollution loads and their impact on the Grand River
are presented.
DESCRIPTION OF THE GRAND RIVER BASIN
The Grand River Basin is located in southwestern Ontario between longitudes
79° 30' and 80° 57' W, and latitudes 42° 51' and 44° 13' N. The
basin occupies the central part of a peninsula bounded on the north by
Georgian Bay, on the west by Lake Huron, on the south by Lake Erie and on
the east by Lake Ontario (Figure 1). The basin has an area of about 6,700
81
-------
GRAND RIVER
WATERSHED
Figure 1. Location of the Grand River Basin in Southwestern Ontario.
82
-------
2
km , a length of about 290 km and a width which varies between 5 and 75
km.
The headwaters of the Grand River rise in a massive swampy upland south of
Georgian Bay at an elevation of approximately 526 m above mean sea level
(msl). The river flows in a southerly direction until it reaches the Town
of Paris. From there it follows a southeasterly direction to discharge
into Lake Erie at Port Maitland at an elevation of 174 m above msl. The
Conestogo, Speed and Nith rivers are the three major tributaries which join
the main stem in the middle portion of the basin. The Conestogo River
drains the northwestern portion of the basin with the Speed and Nith rivers
draining the eastern and western portions of the basin, respectively. In
the upper part of the basin, the Grand and its tributaries flow in
previously formed glacial spillway channels. In the lower part, below the
City of Brantford, the river has scoured its own channel across glacial
lake deposits of silt and clay.
The drainage basin is characterized by a temperate climate that receives a
moderating influence from the nearby Great Lakes System. The long-term
mean annual temperatures vary from 6°C in the headwaters to 9°C at Lake
Erie. The long-term mean annual precipitation varies from 84 cm (178 cm
snow) in the lower reaches to 88 cm (127 cm snow) in the upper reaches of
the basin.
The mean annual flow at the outlet of the river is estimated to be 64
o
m /s which corresponds to a mean annual runoff of 30 cm of
precipitation. Peak flows range from 500 to 1400 m3/s. In general, peak
flows occur during the spring melt period. The highest flow on record,
however, occurred as a result of Hurricane Hazel in November of 1954.
LAND USE, POPULATION AND MAJOR URBAN CENTRES
The Grand River Basin has been developed extensively for urban and
agricultural uses which comprise 3% and 75% of the total basin area,
respectively. Wooded and/or idle areas account for approximately 19% of
83
-------
the basin area and the remaining 3% lies in other uses. Urban uses are
predominant in the central portion of the basin where the cities of
Kitchener, Waterloo, Guelph, Cambridge and Brantford are located (Figure 2).
The population of the basin is approximately 545,000 and is primarily
concentrated in the above mentioned five large urban centres. This,
however, was not always the case. In 1921, 43% of the basin's population
was rural and only 47.5% lived in the five urban centres (Table 1). By
1976, however, the proportion of rural population dropped to about 20%
whereas the proportion of population in the five urban centres increased to
72%. If these trends continue through subsequent years, it can be expected
that the urban population by the year 2001 might increase by about 200,000
to 300,000 people. This could increase pollution loadings to streams from
urban runoff, further impair water quality conditions and tax available
water supplies.
At present, water quality problems in the upper or northern third of the
basin, where agriculture is predominant, are related primarily to erosion.
Soils and nutrients are carried into the river causing turbidity, nutrient
enrichment and limited localized algal growth. The algal density (bottom
cover) varies from 0-80% coverage of the streambed (Figure 3). With the
exception of localized aquatic growth, water quality problems in this part
of the basin are usually limited in extent and do not have significant
detrimental impact on river uses.
The heavily urbanized and industrialized central third of the basin is
significantly affected by the discharge of treated domestic and industrial
wastes and urban stormwater runoff. Two of the major problems that have
been identified through the Grand River Study are: 1) the reduction of
dissolved oxygen in areas downstream from the major municipalities caused
by municipal discharges of oxygen-consuming wastes; and 2) profuse algae
and plant growths stimulated by nutrient inputs from point and nonpoint
sources.
84
-------
TABLE 1 POPULATION IN THE GRAND RIVER BASIN
CO
en
Population Projections for 2001**
Cities
Kitchener
Waterloo
Cambridge
Guelph
Brantford
Total
% of watershed population
Average Annual Growth Rate %
Incorporated Towns and Villages
Total Population
% of watershed population
Average Annual Growth Rate %
Rural Areas (including
unincorporated rural hamlets)
Total population
% of watershed population
Average Annual Growth Rate %
Total watershed population
Average Annual Growth Rate %
1921
21,763
5,883
21,416
18,128
29,440
96,630
47.5
18,589
9.1
88,204
43.4
203,423
1941
35,657
9,025
25,108
23,273
31,948
125,011
53.0
1.3
20,818
8.8
.6
89,795
38.0
.09
235,624
.74
1966
91,376
29,770
51,482
49,497
58,395
280,520
68.2
3.3
35,961
8.7
2.2
95,118
23.1
.23
411,599
2.3
1976
131,801
49,972
71,482
70,374
66,930
390,599
71.6
3.4
43,559
8.0
2.1
111,185
20.4
1.5
535,051
2.8
Low
200,795
86,461
46,815
90,250
85,833
588,159
69.8
1.8*
58,896
7.3
1.57*
200,947
22.9
1.81*
877,137
1.92*
Medium
227,486
92,131
124,474
115,456
94,982
649,529
N/A
2.3*
69,114
N/A
2.27*
N/A
N/A
N/A
* Growth rates apply for the years between 1976 and 2001.
** Population projections based upon available data, March 7-79.
-------
QF LAND
IN AGRICULTURE
POPULATION
> 100,000
100,000 - SO.
to.ooo
O 9.999 - 3.000
<3,000
' ""' i~
Figure 2.' Population and Land Use Distribution in the Grand River Basin.
86
-------
Figure 3. Aquatic Plants Density and Percent Coverage of Streambed in the
Grand River.
87
-------
The lands of the lower third of the basin and the Nith River, a major
tributary draining the western portion of the watershed, are primarily used
for agriculture and the problems encountered there are quite similar to
those identified earlier for the northern third of the watershed (Jeffs et
al, 1978).
The land use distribution and drainage systems in the five major cities in
the basin are described in the following sections.
Brantford - The City of Brantford is located on the Grand River
approximately 91 km north of Lake Erie. The City has a developed area of
3214 ha. and it is served by separate sanitary and storm sewer systems.
Approximately 30% of the city area is drained via the storm sewer system
directly to the Grand River, 35% to Fairchild's Creek, 30% to Mohawk Creek,
3% to Paper Mi.ll Creek and 2% to D'Aubigny Creek. The land uses are:
residential 52.6%, commercial 4.9%, institutional 4.5%, industrial 16.0%
and open space* 21.7%.
Brantford is the centre of agricultural machinery manufacturing in Canada
and it produces a wide range of goods from chemicals, building products,
pulp products, refrigeration units and construction machinery to school
buses and transport trailers.
Cambridge - The City of Cambridge was created in 1973 as a result of the
amalgamation of the towns of Hespeler, Preston and Gait. Hespeler lies 7.3
km upstream on the Speed River while Preston is located at the confluence
of the Speed and Grand Rivers and Gait lies 6.3 km downstream on the Grand
River. Cambridge has a total developed area of 3675 ha. The land uses
are: residential 52.5%, commercial 10.3%, institutional 4.5%, industrial
14.0% and open space 18.7%.
* parks, cemeteries ... etc.
88
-------
Each of the three amalgamated towns has its own separate storm sewer
system. The storm sewer system in Hespeler discharges to the Speed River
whereas in Preston it discharges to both the Speed River (50%) and the
Grand River (50%). Approximately 30% of storm runoff from Gait discharges
to Gait Creek and the remaining 70% to the Grand River.
Guelph - Guelph is located at the confluence of the Speed and Eramosa
Rivers. The city has a developed area of 3080 ha which has been divided
into the following land uses: residential 50.5%, commercial 3.5%,
institutional 12.8%, industrial 13.6% and open space 19.6%.
The City has completely separate sanitary and storm sewer systems. The
storm sewer system discharges into the Speed River (70%) and Eramosa River
(30%).
Modern stormwater management concepts are being implemented in some of the
developing subdivisions in the city. Ponds are used in three new
residential subdivisions for sedimentation and ground water recharge;
another pond is under construction in a developing industrial zone.
Kitchener - Waterloo - The twin cities of Kitchener and Waterloo are
located in the central part of the Grand River Basin. The cities have a
combined developed area of 7864 ha which have the following land uses:
residential 39.7%, commercial 5.9%, institutional 2.9%, industrial 14.7%
and open space 36.8%.
The City of Waterloo is served by separate sanitary and storm sewer
systems. The storm sewer system discharges at several locations along
Laurel Creek and its tributaries. The City of Kitchener has essentially
separate sanitary and storm sewer systems except in the old section of the
City where house foundation drains are connected to the sanitary sewers
because the storm sewers are placed at inadequate depths. Schneider Creek
and its tributary, Montgomery Creek, receive about 80% of Kitchener's
stormwater runoff while the remaining 20% drains directly to the Grand
89
-------
River. The Kitchener- Waterloo urban area has been heavily
industrialized. Major economic activities include furniture, automotive,
tire and shoe manufacturing, meat packing and other industries.
DATA COLLECTION PROGRAMS
1 - MOE - COA Program - A data collection program on urban runoff in the
Grand River Basin was initiated by the Ontario Ministry of the Environment
(MOE) and supported by the Canada-Ontario Agreement on Great Lakes Water
Quality (COA) for a duration of two years (1975-1976). Under this program,
two urban catchments (the North and West catchments) were established in
the City of Guelph and instrumented for quantity (precipitation and runoff)
and quality monitoring.
The objectives of the program were to investigate the variation of quantity
and quality of stormwater runoff; to document the difficulties encountered
in the selection and instrumentation of representative urban catchments
and, to investigate the applicability of the Storage, Treatment, Overflow
and Runoff Model (STORM) for prediction of urban runoff quantity and
quality.
The North Catchment is located at the northern limits of the City of Guelph
(Figure 4). The catchment is a relatively new suburban division consisting
primarily of single family residential land use (82.9%) with minor areas of
multiple family residential (3.5%), commercial and institutional (6.2%) and
open space (3.3%). The area of the catchment is 58.8 ha with a total
imperviousness of 39.0%.
The West Catchment (Figure 4) is adjacent to the downtown core of the City
of Guelph. The catchment (234.8 ha) represents a range of various land
uses, old and recently developed, consisting of single and multiple family
residential (44.9%), comnercial (3.1%), industrial 29.5% and open space
22.5%. The total imperviousness of the catchment was estimated at 32.3%
(Novak, in press).
90
-------
^K\ wTr\\f"jm\
"II^T'll nl'," ' "«.' i X.,, ..^^'"'aa-^1*"^'' r V-Q
.r= i z.'» iiitj-.-.^ .^X ~f^~~'|^\V
^Hr-iipjiiittMiiiLX P V
L^Joi^iaz1 I \
- City Linits
Developed Area
North Catchment
Vest Catchment
Catchment Gauges
(Precipitation,
Flov, Flow
Ouality)
Federal Gauges
(Precipitation)
L'pstreem and
Downstream City
Flov; Ounlity
Location of the Northern and Western Urban Catchments
and Quantity and Quality Monitoring Stations in the City
of Guelph (after Novak, 1979).
V'ater Pollution
Control Plant
91
-------
Precipitation in both catchments was measured using two Leopold and Stevens
tipping bucket rain gauges of capacity of 0.254 mm. Flow measurements were
made at the outlet of both catchments and samples were collected manually
(during the first phase of the program) and using a manually activated
automatic sampler (during the second phase of the program). Samples were
analysed for the following parameters:
five-day biochemical oxygen demand,
chemical oxygen demand,
solids (total and suspended),
phosphorus (total and soluble),
nitrogen forms (free ammonia, Kjeldahl, nitrite and nitrate),
iron, lead, phenols, chloride, conductivity and coliforms
(total and fecal) and streptococcus sp.
During this program a total of 14 events were monitored in the North
Catchment and 13 events in the West Catchment. Unfortunately, only four
events in the North Catchment and two in the West Catchment were completely
monitored in terms of precipitation, flow (hydrograph) and quality
(pollutograph). The remaining events were partially monitored. This
program clearly indicated that automatic sampling stations are essential
for the successful conduct of any urban runoff quality investigation.
2 - PLUARG Program - A second data collection program on urban runoff was
established in the Grand River Basin under the Pollution from Land Use
Activities Reference Group (PLUARG) during the period (1975-1977). The
objectives of this program were to determine the sources of pollutants
within urban areas, estimate their magnitude in terms of unit-area
loadings, determine their relative significance and investigate the nature
of their transport from urban areas.
Three urban areas were selected in the Grand River Basin for the study,
namely: the megalopolis of Kitchener-Waterloo-Cambridge, the City of Guelph
and the Town of New Hamburg. In addition, two small urban watersheds were
92
-------
monitored, namely: Schneider Creek and Montgomery Creek. Monitoring
stations were established upstream and downstream of the urban areas for
the purpose of collecting flow and quality information for pollutant
loading estimates (Figure 5). The difference between the pollutant load
measured at the outlet (downstream) station, and the inflow (upstream)
stations was considered to be the net pollutant load from the study area.
The load from the urban sources was then estimated by subtracting measured
point sources and other non-urban diffuse sources from the net pollutant
load (O'Neill, 1979).
3 - The Grand River Basin Water Management Study Program - During the Grand
River Basin Water Management Study monitoring of the quantity and quality
of urban runoff continued in the Schneider Creek and Montgomery Creek
watersheds. Flow data were obtained at the outlets of both watersheds and
stream water samples were collected using two automatic samplers. Details
of flow measurements, .sample collection techniques, handling and analytical
procedures are discussed in detail in the methodology report (Onn, 1980).
Schneider Creek drains the western portion of the City of Kitchener (Figure
5). The drainage area is 3577 ha and consists of 60% urban, 35%
agricultural and 5% wooded land. The major land uses in the urban area
are: residential (primarily single family dwellings) 42%, commercial 5%,
industrial 4%, recreational 8%, and transportation 1%. The watershed has
an estimated population of 74,000 and is serviced by separate storm and
sanitary sewer systems. The storm sewer system discharges at several
locations along Schneider Creek.
Montgomery Creek drains the eastern portion of the City of Kitchener. The
watershed has an area of 958 ha which consists of 96% urban and 4% wooded
land. The major land uses in the urban area are: residential 64%,
recreational 13%, comuercial 12%, transportation 6% and industrial 1%. The
watershed has a population of 58,000 and is serviced by separate storm and
sanitary sewer systems. The storm sewer system empties into Montgomery
Creek at several locations.
93
-------
A SEWACC TREATMeNT PLANT
O TOWN or VILLAGE
Figure 5. Urban Areas in the Grand River Basin Monitored Under the PLUARG Program.
-------
During the monitoring period a total of 22 hydrographs and 10 pollutographs
were obtained for Schneider Creek and 14 hydrographs and 4 pollutographs
for Montgomery Creek. The pollutographs, based on a sampling frequency
ranging from 15 to 30 minutes, were produced for suspended solids, five-day
biochemical oxygen demand, total phosphorus, filtered reactive phoshorus,
total Kjeldahl nitrogen, filtered ammonia, chloride and lead.
In addition, at the time of initiation of the Grand River Basin Water
Management Study in 1975, a network of seven continuous monitoring stations
was established to monitor.important water quality parameters in the
central megalopolis area of the basin (Figure 6). Dissolved oxygen and
temperature were monitored continuously at five stations using EIL
(Electronic Instruments Ltd.) instrument systems. At two other locations,
NERA (New England Research Associates Inc.) instrument systems recorded
dissolved oxygen, temperature, conductivity, phi and redox potential at
half-hourly intervals (Draper et al, 1980). Data collected at these
stations were used to calibrate and verify the Grand River Simulation Model
(GRSM) and to assess the impacts of various pollution sources including
urban runoff on the water quality of the Grand River.
Characteristics of Urban Runoff
Since the PLUARG monitoring program was established to collect flow
quantity and quality data for the estimation of annual pollutant loads,
intensive sampling was not conducted at the monitoring stations during each
storm event. As such, detailed runoff quality data are available only for
a few storm events which occurred in the sampling period. For this reason,
the characterization of urban runoff from the four test catchments relied
primarily upon data generated from the other two monitoring programs. A
statistical summary of the runoff quality data obtained is presented in
Table 2.
Table 2 indicates that the quality data collected in each of the four test
catchments span several orders of magnitude. These wide variations are in
agreement with the characteristics of urban runoff reported for other
95
-------
Scale: ums 6.H Km
Figure 6. Continuous Monitoring Network in the Grand River Basin.
96
-------
TABLE 2 STATISTICAL SUMMARY OF URBAN RUNOFF QUALITY DATA
5-day
Suspended Biochemical Total Total Lead Zinc Chloride
Solids Oxygen Demand Phosphorus Nitrogen
1.
2.
3.
4.
SCHNEIDER CREEK WATERSHED
Mean mg/L
Standard Deviation, mg/L
Range, mg/L
Number of Samples
MONTGOMERY CREEK WATERSHED
Mean, mg/L
Standard Deviation, mg/L
Range, mg/L
Number of Samples
NORTH CATCHMENT, GUELPH
Mean, mg/L
Range, mg/L
WEST CATCHMENT, GUELPH
Mean, mg/L
Range, mg/L
267
643
5-4791
128
81
65
19-445
87
77
10-1090
195
5-756
3.9
2.5
0.6-13.0
128
3.1
2.2
0.2-9.5
87
10.2
0.2-60
13.9
0.2-95
0.66
1.10
0.07-9.60
127
0.19
0.10
0.06-0.64
86
0.20
0.04-1.60
0.35
0.03-2.40
3.04
2.19
1.21-20.15
124
2.50
1.21
0.96-7.10
86
2.30
0.40-5.30
3.70
0.2-3.4
0.08
0.07
0.01-0.35
117
0.14
0.09
0.01-0.40
72
_
-
_
0.01-0.65
0.78
0.71
0.02-3.20
117
0.18
0.15
0.03-1.00
73
_
-
_
-
66
36
10-175
124
61
45
8-188
86
_
1-68
_
0-383
-------
cities in Ontario (Weatherbe and Novak, 1977) and can be attributed to the
large number of factors known to affect the quality of urban runoff.
It can be seen in Table 2 that the concentrations of suspended solids and
total phosphorus were highest in the runoff from the Schneider Creek
watershed. The high suspended solids concentration values can be
attributed to the construction activities in progress in the watershed
during the monitoring period while the elevated concentration of phosphorus
can be explained by the attachment of phosphorus on sediment particles.
For undetermined reasons, the concentration of five-day biochemical oxygen
demand was higher for the two test catchments in Guelph than in the
Schneider Creek and Montgomery Creek watershed.
A comparison between the dry-weather and wet-weather data indicates that
the concentrations of the monitored water quality parameters increased when
streamflow increased. However, only weak correlations were found during
storm events between flow rates and the concentrations of the monitored
water quality parameters. Typical runoff hydrographs and pollutographs
obtained from the Schneider Creek and Montgomery Creek watersheds are shown
in Figures 7 and 8. Specifically, the first flush phenomenon was not
observed in either watershed though it was noted in the two smaller
catchments in Guelph. Apparently the quick hydro!ogic response of a small
catchment favours the occurrence of the first flush phenomenon.
A correlation analysis was performed using the wet-weather water quality
data collected from the Schneider Creek and Montgomery Creek watersheds.
Relatively strong correlation (coefficient of correlation r = 0.80) was
found between suspended solids and total phosphorus but not among the other
water quality parameters.
The effects of total event precipitation, total runoff volume and length of
the antecedent dry period on event pollutant loads were analyzed using data
collected in the Schneider Creek watershed. Total event precipitation and
total runoff volume were found to be the most dominant factors which
determine the event pollutant loads.
98
-------
Precipi ta tion
millireters
4.0 J
Flew, m /s
3.0
10
2.0
1.0
SCHNEIDER CREEK
25-26/10/1978
SUSPENDED SOLIDS (SS)
-f TOTAL NITROGEN (TN)
4
SS TP BOD
250 .5 10
200 .4 8
TOTAL PHOSPHORUS (TP)
5-DAY BIOCHEMICAL
,»%>BxbXYGEN DEMAND (BOD)
50 .1
16:00
18:00
20:00
22:00 24:00
2:00
4:00
FIGURE 7. Hydrograph and Pollutographs of Suspended Solids, 5-Day Biochemical Oxygen Demand,
Total Phosphorus and Total Nitrogen as Measured at the Outlet of Schneider Creek
on 25-26AO/1978.
TN mg/1
5
150 .3 6
100 .2 4
-------
Precipitatiorf
millimeters
4.0
Flow, m /s
3.0
o
o
2.01
1.0 J
MONTGOMERY CREEK
25-26/10/1978
FLOW
5-DAY BIOCHEMICAL
OXYGEN DEMAND (BOD)
V
^SUSPENDED SOLIDS (SS)
SS TP BOD TN mg/1
I- 200 .484
TOTAL
TOTAL NITROGEN (TN)
r 150 .363
h 100 .242
50 .1 2 1
16:00 18:00
20:00
22:00 24:00
2:00
4:00
FIGURE 8. Hydrograph and Pollutographs of Suspended Solids, 5-Day Biochemical Oxygen Demand,
Total Phosphorus and Total Nitrogen as Measured at the Outlet of Montgomery Creek
on 25-26/10/1978.
-------
Simulation of Urban Runoff Quantity and Quality
The two objectives of urban runoff simulation are:
1. to provide input data to the GRSM model in terms of urban runoff
volumes and pollutant loads on an event basis from the cities of
Brantford, Cambridge, Guelph, Kitchener and Waterloo.
2. to extrapolate urban runoff pollutant loads for the populations
projected for the five cities to the years 2001 and 2031.
An evaluation of several urban runoff models with respect to the above
objectives lead to the selection of the STORM model (U.S. Army Corps of
Engineers, (1977). STORM can simulate both the quantity and quality of
urban runoff. It is designed for use with many years of continuous hourly
precipitation records but can be used for individual storm events. The
model employs an accounting scheme that, for each storm event, allocates
runoff volumes to storage and treatment and notes those volumes exceeding
storage or treatment capacities. Water quality is handled as a function of
hourly runoff rates, with generated quantities of pollutants allocated to
storage, treatment or release into receiving waters. Statistics are
generated for each event and collectively for all events processed,
including average annual values for runoff and pollutant loadings. The
results of model calibration and verification, with respect to both runoff
quantity and quality, are presented in the following sections.
Runoff Quantity
Fourteen storm events recorded in the Montgomery Creek watershed were used
to calibrate the STORM model with respect to total event runoff volume.
The results are illustrated in Figure 9.
The simulation results compare well with the measured event runoff
volumes. About 79% of the deviations - the difference between simulated
101
-------
and measured runoff volumes - are less than 25% of the measured runoff
volume. Most of the large deviations are associated with minor storm
events for which measurement errors are large relative to the event runoff
volume.
O)
e
3
3
cC.
OJ
+->
ra
13
GO
1 2 3
Measured Runoff Volume, nm
Figure 9. STORM Model Calibration with respect to Event Runoff
Volume, Montgomery Creek Watershed.
102
-------
The coefficients of imperviousness assigned to the various urban land uses
were identified to be the most significant factors which determine the
event runoff volume. In order to obtain a good agreement between simulated
and measured event runoff volumes, it was found necessary to use relatively
low imperviousness values. The coefficient of imperviousness was reduced
to 15% for residential land use. This low imperviousness value appears to
be reasonable since most of the roof leaders in the Montgomery Creek
watershed are not connected directly to the storm sewer system.
For model verification, the calibrated model for the Montgomery Creek
watershed was used to simulate the quantity of urban runoff from the
Schneider Creek Catchment. The calibrated model was adjusted to reflect
the characteristics and urban land use distribution in the Schneider Creek
watershed. The simulation results (Figure 10) are in close agreement with
the measured event runoff volumes. About 50% of the deviations are less
than 25% of the measured runoff volume.
Runoff Quality
Although runoff quality data were collected for most of the storm events
recorded in the Montgomery Creek watershed (14 events), complete
pollutographs are available for only four events. These events were used
to calibrate the STORM model in terms of event pollutant loads. The
results are given in Table 3.
The results are acceptable, particularly when one considers the simplistic
approach used by the STORM model to simulate the quality of urban runoff.
In most cases, the simulated event pollutant loads are within + 50% of the
measured loads.
During model calibration, it was found that the simulated event pollutant
load depends primarily on three sets of parameters: the pollutant
accumulation rates, the characteristics of dust and dirt and the washoff
exponent. The effect of the washoff exponent on the shape of the simulated
pollutograph, however, was found to be small.
103
-------
The calibrated model for the Montgomery Creek watershed was used to
simulate the quality of urban runoff from the Schneider Creek watershed.
As mentioned earlier, the model was adjusted to reflect the characteristics
and urban land use distribution in the Schneider Creek watershed. The
measured and simulated event pollutant loads are compared in Table 4.
-------
TABLE 3
STORM MODEL CALIBRATION WITH RESPECT TO EVENT
POLLUTANT LOADS, MONTGOMERY CREEK CATCHMENT
o
en
Date Storm
Event
Occurred
3/10/78
5/10/78
25/10/78
23/11/78
TABLE 4
Date Storm
Event
Occurred
16/8/78
24/8/78
14/9/78
17/9/78
30/9/78
3/10/78
5-6/10/78
25/10/78
13/11/78
23/11/78
Suspended
kg
Solids
Measured Simulated
1909
176
1619
2768
Suspended
kg
Measured
906
2033
18383
372254
1684
21924
1843
5750
12473
21268
2849
296
1362
2889
Solids
Simulated
2711
1800
12781
104926
670
8786
2513
5613
7522
6950
5-day Biochemical
Oxygen Demand Total Phosphorus
kg kg
Total
Measured Simulated Measured Simulated Measured
109
8
78
26
STORM MODEL
POLLUTANT
178
17
93
167
5.3
0.6
3.3
5.4
7.0
0.8
3.3
7.1
120
32
66
158
Nitrogen
kg
Simulated
151
15
73
151
VERIFICATION WITH RESPECT TO EVENT
LOADS, SCHNEIDER CREEK CATCHMENT
5-day Biochemical
Oxygen Demand
kg
Measured
72
106
254
1001
47
453
24
171
488
71
Total Phosphorus
kg
Total
Simulated Measured Simulated Measured
186
120
703
5006
39
489
132
325
478
384
6.3
5.5
43.3
857.0
2.6
48.2
4.7
10.5
26.6
30.2
6.7
4.4
31.6
258.8
1.7
21.6
6.2
13.9
18.6
17.1
112
104
390
3592
46
434
106
182
392
376
Nitrogen
kg
Simulated
146
97
661
5307
35
459
130
295
399
361
-------
On the whole, the simulated event pollutant loads compare well with the
measured loads; both are of the same order of magnitude. The results are
particularly good for total phosphorus and total nitrogen; the simulated
event loads are within + 50% of the measured loads. The results are not as
good for suspended solids and five-day biochemical oxygen demand.
Long-Term Simulation
The STORM model was used to simulate the urban runoff and associated
pollution loads for the cities of Brantford, Cambridge, Guelph, Kitchener
and Waterloo for a 20-year period (1956-1975). For each city, the model
was adjusted to reflect the characteristics and urban land use
distribution. Parameter values for the pollutant accumulation rates, the
characteristics of dust and dirt and the washoff exponent were the same as
used in the calibrated model. A statistical summary of the simulation
results is presented in Table 5. Simulated urban runoff volumes and loads
for the five cities were used as input to 6RSM model for impact assessment.
TABLE 5
STATISTICAL SUMMARY OF SIMULATION RESULTS
(1956 - 1975)
Brantford Cambridge Guelph Kitchener Waterloo
Area, ha
Runoff as Fraction
of Precipitation
Pollutant Loads,
kg/ha/yr
Suspended Solids
5-day Biochemical
Oxygen Demand
Total Phosphorus
Total Nitrogen
7943
0.17
9080
0.19
7611
0.19
12954
0.15
6477
0.15
92
4.8
0.23
9.5
108
5.6
0.27
11.1
87
4.5
0.22
9.0
75
3.9
0.18
7.7
75
3.9
0.18
7.7
106
-------
A comparison between the annual pollution loads of suspended solids, total
phosphorus and total nitrogen from the five urban centres and from
agricultural and sewage treatment plants within the Grand River Basin
indicates that the urban percentage contribution is small (2%-6%).
IMPACT OF URBAN STORMUATER RUNOFF ON THE GRAND RIVER
Untreated stormwater runoff may contribute a significant portion of the
total pollution load entering receiving waters on an annual basis, and are
often significant on a shock-load basis during wet events.
When pollutants from urban runoff are discharged into receiving waters,
they may affect the water quality in several ways. Some of their effects
are immediate such as bacteria contamination. Others are long-term effects
such as nutrient enrichment which may lead to eutrophication.
Receiving waters such as streams, lakes and estuaries differ in the manner
in which they react to similar pollutant loadings. Further, the types,
extents and rates of water quality processes that occur in water bodies are
controlled by the immediate physical environment as defined by climate and
physiography.
The response of a receiving water to an introduced waste load depends also
on its initial state. Thus, the particular response of the receiver under
different initial states is basically a matter of defining the appropriate
boundary conditions at the time the waste load is imposed.
The impacts of urban runoff can be also viewed in terms of major pollutants
(oxygen demanding materials, suspended solids and associated contaminants,
nutrients, heavy metals and bacteria) and their specific effects on the
various uses of the receiving water (Singer, 1979). For example, suspended
solids which find their way into a river may be deposited and become
sediment. Sediment is a nuisance if-deposited in navigation channels and
it can reduce the capacity of drainage ways to carry high flows. Bacterial
loadings from urban areas may constitute a health hazard and result in
107
-------
restrictions on swimming and other recreational activities.
All the above considerations indicate that the question of impacts of urban
runoff on receiving waters is an issue which is dependent on specific local
conditions. For the purpose of the Grand River Basin Water Management
study it was decided to investigate the urban runoff impacts on the river
in terms of dissolved oxygen (DO).
The strategy was:
1 - to develop a continuous dynamic water quality model capable of
accepting inputs from point and nonpoint (urban and rural) sources
and predicting the water quality parameters (mainly DO) under
various flow regimes, sewage treatment levels and meteorological
conditions.
2 - to collect continuous data on DO, temperature and other quality
parameters in the Grand River Basin in order to calibrate and
verify the model.
3 - to use the model in the assessment of the impacts of various
inputs including urban runoff and to evaluate water management
alternatives in the megalopolis area of the Grand River Basin.
The Grand River Simulation Model (GRSM)
This is a dynamic model which utilizes O'Connor and DiToro's formulation
for the calculation of DO in river systems. The model accounts for the
deficits of DO caused by carbonaceous (BOD) and nitrogenous (NOD) oxygen
demand, benthic oxygen demand as well as the replenishment of oxygen due to
reaeration. Oxygen production and uptake (photosynthesis and respiration)
as well as the day to day and seasonal growth, death and washout of three
types of attached aquatic plants are calculated using an ecological
subroutine (ECOL) (Kwong et _al_, 1979).
108
-------
The water quality parameters include DO, BOD, NOD, nitrate, suspended
solids and total phosphorus. The dynamic model simulates the effects of
sewage treatment plants effluent and urban runoff under different flow
conditions. The urban loadings are calculated by STORM and input to the
dynamic model at five nodes representing the five urban areas: Brantford,
Cambridge, Guelph, Kitchener and Waterloo, (Figure 11). Output of the
model provides DO concentrations at each 2-hour time step.
GRSM Model Calibration and Verification
The GRSM model was calibrated using actual meteorological data, river
hydrology, STP effluents and upstream boundary conditions (quality and
quantity) for the year 1976, from June to September. The simulated DO
concentrations were compared with the continuous monitoring data for reach
No.8 at Glen Christie below Guelph, reach No.12 at Preston and reach No.16
at Glen Morris below Gait STP. Model parameters were adjusted until good
agreement was obtained between observed and simulated DO values (Figures
12, 13 and 14).
Actual survey data of 1977 were then used to verify the dynamic model. The
verification results indicate that the model is capable of reproducing the
observed daily minimum DO within 4-1 mg/L approximately 80% of the time. DO
frequency distributions at specified levels within a month as a percentage
of total time were generally predicted within 10 percentage points of the
total time. In general, the accuracy of the predicted minimum DO
concentrations is much better than the predicted maximum (Kwong, 1980).
Evaluation of Impact of Urban Runoff
The impact of urban runoff on the Grand River was evaluated by running the
GRSM model twice for the period June-September, 1976 The first run
included the urban input from the five cities whereas in the second run,
the impact of this input was nullified by altering the quality of the
stormwater input. This was done to maintain the same flow patterns for the
109
-------
/ UaLicr\oo STP
G-R7
STP ,
GRli
IH
15
16
Ni'it> River ,g
19
20
21
22 J
ft
fc
12 n 10 9 & 7 6
Gilt STP
G22
Par/s STP
STP
/Y4 DO
Figure 11. Grand River Water Quality Simulation Model
- Geometry of River System.
110
-------
c
CJ
C71
X
o
-O
O
5 '
4 '
^Simulated
...AJ
1976
June
July
August
September
Figure 12. A Comparison between Observed and Simulated Daily Minimum Dissolved Oxygen
Concentrations for Reach No.C at Glen Christie Below Guelph on the Speed River.
-------
June
July
August
September
Figure 13.. A Comparison Between Observed and Simulated Daily Minimum Dissolved Oxygen
' Concentrations for Reach No.12 at Preston on the Speed River.
-------
CO
QJ
CD
Ol
£ 3
o
5 2
.Simulated
June
July
August
September
Figure 14. A Comparison Between Observed and Simulated Daily Minimum Dissolved Oxygen
Concentrations for Reach No.16 at Glen Morris Below Gait STP on the Main Grand River.
-------
two runs. Negation of the quality effects of stormwater was achieved by
setting the concentrations of BOD, NOD, N03, suspended solids and total
phosphorus in urban runoff to minimum values of 0.1 mg/L. A comparison of
the results of the two simulation runs (Table 6) indicates that the percent
of time in DO violations on the Speed River decreased by a few percentage
points when urban runoff was excluded. The improvements in the daily
minimum DO concentrations (Figures 15 and 16) are minor and average from
0.5 to 1.0 mg/L. The two critical reaches on the Speed River, Reaches No.
7 and 8 were still in violation for 28.1% to 55.9% of the time during the
entire simulation period.
The main Grand displayed opposite trends in comparison with the Speed
River. The critical reaches, Reach No. 5 (below Kitchener) and Reaches No.
13 - 16, show evidence of an increase in the percent of time in DO
violations with the removal of urban input. The worst conditions occurred
in Reaches No. 5 and 13 where violations practically doubled. Neverthe-
less, the changes in the daily minimum DO concentrations averaged less than
1 mg/L (Figure 17). Also, the percent of time in violation for all the
reaches on the main Grand ranged from 1.1% to 18.0% for the entire
simulation period.
The improvement of the in-stream DO regime of the Speed River with the
removal of the urban input is probably due to the reduction of total oxygen
demanding load which had a positive effect due to the limited assimilative
capacity of this river.
The response of the main Grand to the removal of urban input is hard to
explain. It shows that the urban input is having a positive effect on the
river under present conditions. A possible explanation of this phenomenon
is that the main Grand is light limited and the biomass growth is affected
by the available light. By reducing the amount of suspended solids in the
urban runoff (second run), the turbidity of the river was reduced and more
light was allowed to penetrate to plant depth, thus resulting in more
biomass growth and lower DO concentrations. It should be noted, however,
114
-------
TABLE 6
A COMPARISON BETWEEN THE RESULTS OF TWO SIMULATION RUNS (WITH AND WITHOUT URBAN RUNOFF)
IN TERMS OF PERCENT TIME IN DO VIOLATION FOR 21 REACHES ON THE GRAND RIVER
(JUNE-SEPTEMBER, 1976)
REACH
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
JUNE
Run No. 1
0.8
0.0
0.0
0.0
3.6
0.0
30.3
50.0
2.5
21.1
1.7
7.5
1.1
1.4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Run No. 2
0.8
0.0
0.0
0.0
3.6
0.0
28.1
48.6
1.4
19.2
1.1
8.3
1.7
1.4
0,0
0.0
0.0
0.0
0.0
0.0
0.0
JULY
Run No. 1
0.0
0.0
0.0
0.0
4.8
0.0
36.8
48.7
7.3
8.6
4.3
3.0
3.2
3.0
2.2
0.0
0.0
0.0
0.0
0.0
0.0
Run No.
0.0
0.0
0.0
0.0
5.9
0.0
31.5
41.7
2.7
7.5
1.3
1.1
3.8
4.6
2.4
1.3
0.0
0.0
0.0
0.0
0.0
AUGUST
2 Run No.
0.0
0.0
0.0
0.0
5.4
0.0
43.5
56.5
2.7
0.3
0.0
0.0
4.8
9.4
7.0
0.8
0.0
0.0
0.0
0.0
0.0
1 Run No. 2
0.0
0.0
0.0
0.0
10.2
0.0
43.5
55.9
5.1
6.5
0.0
0.0
10.2
18.0
11.0
4.8
0.0
0.0
0.0
0.0
0.0
SEPTEMBER
Run No.l
0.0
0.0
0.0
0.0
0.0
0.0
29.2
44.2
3.9
2.2
1.9
0.0
0.0
0.8
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Run No. 2
0.0
0.0
0.0
0.0
1.4
0.0
28.6
38.3
1.1
3.6
0.0
0.0
0.3
2.2
0.6
0.0
0.0
0.0
0.0
0.0
0.0
-------
en
c
CJ
x
o
T3
ttJ
O
i/l
6 '
5 '
4 '
3
2
Without Urban Runoff
June
July
August
September
Figure J.5. The Impact of Urban Runoff on Simulated Daily Minimum Dissolved Oxygen Concentrations
for Reach No.8 at Glen Christie Below Guelph on the Speed River.
-------
8
7 -
QJ
cn
o 4
a;
o 3
Ul
I/I
a _
Without Urban Runoff
HJith Urban Runoff
June
July
August
September
Figure 16. The Impact of Urban Runoff on Simulated Daily Minimum Dissolved Oxygen
Concentrations for Reach Mo.12 at Preston on the Speed River.
-------
CO
c
V
en
x
o
o
I/I
t/1
5 H
With Urban Runoff
Without Urban Runoff
June
July
August
September
Figure 17. The Impact of Urban Runoff on Simulated Daily Minimum Dissolved Oxygen Concentrations
for Reach No.16 at Glen Morris Below Gait STP on the Main Grand River.
-------
that the difference between the two runs in terms of the minimum DO
concentrations is minor.
CONCLUSIONS
The characteristics of urban runoff and the magnitude of the associated
pollution loads from the cities of Brantford, Cambridge, Guelph, Kitchener
and Waterloo are similar to those reported for other cities in Ontario.
The impact of urban runoff from the five cities on the dissolved oxygen
regime in the Grand River is minor. Parts of the Speed River below Guelph
and certain reaches on the main Grand between Kitchener and Brantford
suffer from profuse algae and plant growth during the summer and early fall
period. This results in extremely low dissolved oxygen concentrations
during the night due to the respiration process. High dissolved oxygen
levels are observed during the day as a result of the photosynthesis
process. Respiration and photosynthesis are the two dominant in-stream
processes in this section of the Grand River system. Improvement of the
dissolved oxygen regime would require the control of nutrient input (mainly
phosphorus) from point and non-point sources. The nutrient input from
urban runoff is small relative to agricultural diffuse sources and sewage
treatment plants. Therefore priority for pollution control measures should
be given to those two sources.
REFERENCES
DRAPER, D. W., Gowda, T. P. H., Frank, D. C. 1980. Continuous Monitoring
of Dissolved Oxygen; Grand River Basin Water Management Study, Report
Series, Report No. 11, Ministry of the Environment, Toronto.
JEFFS, D. N. Coutts, G. M., Ralston, J. 6., Smith, A. F. 1979. The Grand
River Basin Water Management Study, An Integrated Decision Aid in
Ontario; a paper presented at the 4th National Hydrotechnical Conf. on
River Basin Management, Vancouver.
119
-------
KWONG, A. (1980). Grand River Assimilation Model - Calibration,
Verification and Application, Water Resources Branch, Ministry of the
Environment, Toronto (Draft).
KWONG, A., WEATHERBE, D., GOWDA, T.P.H. and DRAPER, D., 1979. Water
Quality Assessment Techniques for Grand River Basin Study. Grand River
Basin Water Management Study, Technical Report No.10, Water Resources
Branch, Ministry of the Environment (Draft).
NOVAK, Z. (in press), Stormwater Runoff Investigation and Model Testing
for the City of Guelph; Grand River Basin Water Management Study,
Technical Report Series, Report No. 8, Ministry of the Environment,
Toronto.
O'NEILL, J. E. 1979. Pollution from Urban Land Use in the Grand and
Saugeen Watersheds; International Reference Group on Great Lakes
Pollution from Land Use Activities, Ministry of the Environment, Toronto
ONN, D. 1980. Data Collection Methodology Used in the Study of Pollution
from Land Use Activities in the Grand and Saugeen Watersheds;
International Reference Group on Great Lakes Pollution from Land Use
Activities, Ministry of the Environment, Toronto.
SINGER, S. N. 1979- Evaluation of Impacts of Storm Water Runoff and
Combined Sewer Overflows on Receiving Waters; Ministry of the
Environment, Toronto.
U.S. ARMY CORPS OF ENGINEERS, 1977. Storage, Treatment, Overflow, Runoff
Model; Hydrologic Engineering Centre, Davis, California.
WEATHERBE, D. Novak Z. 1977. Water Quality Aspects of Urban Runoff;
presented at the Conference on Modern Concepts of Urban Drainage
sponsored by the Urban Drainage Subcommittee of the Canada-Ontario
Agreement on Great Lakes Water Quality, Ottawa.
120
-------
DEVELOPMENT OF AN URBAN HIGHWAY
STORM DRAINAGE MODEL BASED ON SWMM
Raymond J. Dever, Jr. and Larry A. Roesner2
BACKGROUND
Water Resources Engineers/Camp Dresser and McKee Inc. is
currently completing development of an Urban Highway Storm Drainage Model
for the Federal Highway Administration, U.S. Department of Transportation.
The general capabilities of the Model include:
Preliminary highway drainage system design, including locating
inlets, sizing pipes, and estimating construction costs;
Hydraulic analysis of highway drainage systems under rainfall
conditions more severe than those used in design; and
t Simulation of the generation and washoff of pollutants in
the highway corridor.
The Model consists of four related but independent modules, as
follows:
Precipitation Module
Hydraulics/Quality Module
Analysis Module
Cost Module
Associate Engineer, Camp Dresser and McKee Inc., Springfield, Va.
Principal Engineer, Camp Dresser and McKee Inc., Springfield, Va.
121
-------
The Precipitation Module can perform a variety of statistical analyses
on long-term hourly precipitation data and generate design storm
hyetographs. The Hydraulics/Quality Module is the basic design tool in
the package. This Module simulates time-varying runoff quantity and
quality, locates stormwater inlets and sizes the conduits of the major
drainage system. The Analysis Module simulates unsteady gradually-varied
flow in the drainage system and can be used to analyze complex hydraulic
conditions, such as surcharge and backwater, that may be encountered
during an extreme storm event. The Cost Module can be used to estimate
construction, operation and maintenance, and total annual costs associated
with the drainage system.
The capabilities of each Module will be discussed below in more
detail, as will the relationships among them. The Hydraulics/Quality
Module employs some of the same formulations that are used in the EPA
SWMM package; therefore, this Module will be highlighted, especially with
regards to new features that have been added to the basic SWMM routines.
OVERVIEW OF THE HIGHWAY DRAINAGE MODEL
The four Modules of the Urban Highway Storm Drainage Model are a
powerful and flexible set of tools for use in drainage system analysis
and design. The computer programs which make up the Model have been
developed with several key features (1). First, each of the hydraulics
programs is fully dynamic. This is in contrast to the static procedure
generally used for highway drainage design, where pipes are sized based
on a single peak runoff flow generated with the rational formula.
Second, the Model has purposely been developed in related but indepen-
dent Modules to maximize the flexibility of the package. The user may
apply as many or as few of the Modules as are appropriate to his
particular design or analysis problem. Third, the programs are designed
to accommodate local drainage practices and design procedures. The
input/output of the programs are in terms familiar to the highway
drainage engineer and can include most inlet, channel, and pipe types
122
-------
available. Also, the programs are structured to allow the use of
whatever design criteria are required locally.
A summary description of each of the four Modules of the Urban
Highway Storm Drainage Model is given below, along with an explanation
of the relationships among the Modules.
Precipitation Module
The Precipitation Module consists of a single computer program
with three major options - two for generating single-peak synthetic
hyetographs and one for performing a variety of statistical analyses
on long-term precipitation data (2). The two options for generating
synthetic hyetographs both rely on a methodology developed by Chen (3)
and require the return frequency, duration, and skew of the desired
hyetograph as input. The first option requires the 10-year, 1-hour;
the 10-year, 24-hour; and the 100-year, 1-hour rainfalls for the user's
study area as input. The second option makes use of an input intensity-
duration-frequency curve of the same return frequency as the desired
hyetograph. Both of these options can generate the rainfall hyetographs
required as input for the dynamic programs of the Hydraulics/Quality
Module.
The third major capability of the Precipitation Module is the
statistical analysis of long-term hourly precipitation records (such as
those available on magnetic tape from the National Oceanic and Atmos-
pheric Administration, U.S. Department of Commerce). Some of the
statistical analyses of which the program is capable are:
Annual series analysis and partial duration series analysis,
including the generation of intensity-duration frequency
curves;
t Frequency of occurrence analysis for such parameters as peak
rainfall intensity per storm event, storm duration, and dry
period duration; and
123
-------
Analysis of storm skew (i.e., the ratio of the time to peak
rainfall intensity of a given storm to the total duration of
the storm).
Each of these analyses may be performed on a seasonal basis as well as
an annual basis, for as many as 12 user-defined seasons.
The statistical analyses that can be performed with the Precipi-
tation Module clearly give the user of single-event stormwater simulation
models better local information than is generally available for developing
the required design storm event and associated conditions. For example,
the dry period frequency of occurrence analysis should greatly assist the
modeler in specifying antecedent conditions for his simulations. The
storm skew analysis should enable the engineer to use a rainfall distri-
bution more representative of his local area than some others routinely
used, such as the Soil Conservation Service Type-II storm distribution
commonly applied to almost the entire continental United States (4).
Hydraulics/Quality Module
The Hydraulics/Quality Module consists of three computer programs -
the Inlet Design Program, the Surface Runoff Program, and the Drainage
Design Program (5). Together, these programs can perform most of the
major computations involved in design of highway drainage systems.
The purpose of the Inlet Design Program is to locate inlets in the
surface runoff conveyance system of the highway right-of-way so as to
maintain hydraulic conditions during the design storm event within
specified criteria. Specifically, the Inlet Design Program simulates
time-varying runoff and routes the runoff flows through surface gutters
or channels. The program then determines the placement of inlets in
gutters required to maintain flow spread within a user-specified maximum
or the placement of inlets in channels to maintain flow depth within a
user-specified maximum. The program also checks that the percentage of
the gutter/channel flow that carries past each inlet to the next gutter/
124
-------
channel section does not exceed a given maximum, again specified by the
user. The routines for computing overland flow and for simulating inlet
hydraulics in this program will be discussed in more detail below.
The Surface Runoff Program simulates time-varying runoff quantity
and quality, routes these flows and pollutants through surface gutters,
channels, and detention basins, and computes inlet hydrographs and
pollutographs. The inlet hydrographs and pollutographs are saved as a
disc or tape file for subsequent use by the Drainage Design Program.
The Surface Runoff Program is similar in most respects to the Runoff
Block of EPA's SWMM. Two significant differences between this program
and the SWMM Runoff Block are described below.
The Drainage Design Program reads the inlet hydrographs and
pollutographs generated by the Surface Runoff Program for the design
storm and sizes the major drainage system accordingly. Specifically,
the program determines the diameter of circular conduits and the
bottom width of trapezoidal channels so that each conduit and channel
flows full at peak flow. The diameters of circular conduits so
determined are rounded up to the nearest commercially-available pipe
size. A complete simulation, including pollutant routing, of the just-
designed system is then automatically performed by the program.
Analysis Module
The Analysis Module consists of a single computer program that
simulates unsteady, gradually-varied flow in the major drainage system
of the highway right-of-way, using inlet hydrographs generated by the
Surface Runoff Program as input. The program is basically a modified
version of the Extended Transport program of the EPA SWMM package. Its
primary purpose is to analyze the performance of the drainage system
under an extreme storm event, a step generally included in the highway
drainage design process. This program will not be discussed further here,
since there are no significant differences between the capabilities of
this program and the Extended Transport program.
125
-------
Cost Estimation Module
The final Module of this package, the Cost Estimation Module, also
consists of a single computer program (6). The purpose of this program is
to estimate the capital costs, operation and maintenance costs, and total
annual costs associated with the construction and maintenance of a highway
drainage system. All of the cost computations are based on unit costs for
materials, installation, and O&M. As part of the cost analysis, the
program also estimates the excavation and backfill associated with
construction of the drainage system; elevation of the highway grade line,
invert elevations of the system conduits and junctions, and sizes of the
conduits and junctions are employed in the excavation and backfill
calculations.
DIFFERENCES FROM SWMM
The Hydraulics/Quality Module described above is closely related
to the EPA SWMM package in its basic computational routines. However,
there are two fundamental differences between SWMM and the Urban Highway
Storm Drainage Model that are of interest here - the inclusion of inlet
hydraulics and the modification of the overland flow routines to allow
cascading of flow over as many as three separate land surface types for
each catchment. Both of these modifications are included in the two
programs concerned with simulation of surface runoff - the Inlet Design
Program and the Surface Runoff Program.
Simulation of Inlet Hydraulics
Both the Inlet Design Program and the Surface Runoff Program are
structured to allow the simulation of six basic inlet types:
Curb Opening Inlet
Depressed Curb Opening Inlet
Grate Inlet
126
-------
Depressed Grate Inlet
Combination (Curb Opening and Grate) Inlet
Depressed Combination Inlet
This is in contrast to the EPA SWMM package in which inlet hydraulics are
not simulated at all. The programs are structured to allow simulation
of fixed-size inlets of the above six types either by empirical equations
built into the program codes or by inlet efficiency curves supplied as
input by the user. At present, only empirical equations for depressed
curb opening inlets are in the programs, but the codes have been
structured to allow easy addition of equations for other inlet types by
the user.
The simulation of inlet hydraulics by means of inlet efficiency
curves proceeds as follows, in either of the two programs mentioned
above. The user supplies as input a group of inlet efficiency curves
for the size and type of inlet in question, as shown in Figure 1.
(Actually, the user supplies the coordinates of points that define the
curves.) The curves give the percentage of gutter or channel flow
intercepted by the inlet as a function of the total gutter or channel
flow at a given point in time. For inlets in gutters, the flow inter-
ception capacity is a function of both the gutter slope and the cross-
slope of the highway surface; thus, a family of curves, one curve for
each of several typical gutter slopes and cross-slopes, must be supplied.
For inlets in channels, the flow interception capacity is a function of
the channel slope; a family of curves for typical channel slopes must be
input by the user. In the example curves of Figure 1, Qj is the flow
intercepted by the inlet, Q is the gutter/channel flow, and Sx is the
cross-slope of the highway.
For a given inlet, the program will select the appropriate inlet
efficiency curve to use based on the slope of the gutter/channel section
where the inlet is located and the cross-slope of the highway, if
appropriate. At each time step, the program then calculates the gutter/
127
-------
QT/Q
0.2-
=1/16
4.0
6.0
8.0 10.0
Q (cfs)
FIGURE 1
INLET EFFICIENCY CURVES FOR 2' X 4'
PARALLEL BAR GRATE INLET WITH GUTTER
SLOPE = 0.02 (7)
128
-------
channel flow, determines the inlet efficiency from the curve, and computes
the flow intercepted by the inlet and the flow carried over to the next
gutter/channel section.
Kinematic Cascades for Overland Flow
In order to account for the peculiar geometry and surface types of
the drainage area associated with the highway right-of-way, it was
decided to modify the overland flow routines used in the Runoff Block of
SWMM. Basically, SWMM employs the kinematic wave approximation for
overland flow, accounting for infiltration by Morton's equation and for
depression storage (8). Each catchment is characterized by a drainage
area, width, slope, roughness coefficient, infiltration type and
depression storage and' drains directly to a surface gutter or channel.
This approach has been modified in the Inlet Design Program and the
Surface Runoff Program to allow runoff to cascade over as many as three
subcatchments for each catchment before reaching a surface gutter or
channel. Each of the three subcatchments can thus have a unique set of
hydrogeometric characteristics, as listed above (with the exception of
the catchment width which must be the same for each subcatchment).
The basic computational difference caused by addition of this
kinematic cascade appears in the continuity equation for overland flow.
Figure 2 summarizes the basic flow computations for a given time step
for a typical subcatchment. In Figure 2, dQ is the depth of flow at the
previous time step, dx is the depth of flow at the current time step, and
d is the maximum depth of depression storage. The continuity or storage
equation is then given by:
At
129
-------
where
Ad =
R =
I =
Q
Q.
A
rainfall intensity during the time step;
infiltration rate during the time step;
outflow from the subcatchment;
inflow from the upstream subcatchment; and
surface area of the subcatchment.
Thus, the basic difference is the addition of the inflow term Qi in this
equation.
The remainder of the solution for overland flow proceeds in
basically the same manner as in SWMM. Infiltration, I, is computed with
RAINFALL INTENSITY
R
INFILTRATION
FIGURE 2
RAINFALL-RUNOFF FROM A
TYPICAL SUBCATCHMENT
130
OUTFLOW
-------
Horton's equation; outflow Q is given by the Manning equation (wide
channel assumption). The continuity equation and the outflow equation are
combined and solved by the Newton-Raphson iterative technique.
SUMMARY
This paper has presented an overview of an Urban Highway Storm
Drainage Model developed for the Federal Highway Administration. The
Model consists of a powerful and flexible set of computer programs for
highway drainage analysis and design. The capabilities of the Model
include statistical analysis of rainfall records, design of drainage
systems including locating inlets and sizing pipes, analysis of drainage
systems under extreme storm events, and simulation of runoff quality
from the highway corridor.
Some of the computer programs described in this paper employ the
same basic solutions for rainfall-runoff and flow routing as the EPA
SWMM package; however, there are some major differences. Two significant
modifications to the SWMM approach have been described in this paper -
the inclusion of the simulation of inlet hydraulics and the addition of
kinematic cascades for overland flow.
The Model is presently being tested on five typical highway sites
around the United States. The user's manuals and documentation reports
for the Model are being completed and should be available in the near
future from the Federal Highway Administration.
ACKNOWLEDGEMENTS
The authors would like to gratefully acknowledge the assistance
and guidance of Dr. Dah-Cheng Woo of the Federal Highway Administration,
who has served as Program Manager for this study.
131
-------
REFERENCES
1. Knepp, A.J. and R.P. Shubinski, "Conceptualization of an Urban
Highway Storm Drainage Model", prepared for the Federal Highway
Administration, Washington, D.C., April, 1977
2. Trotter, R.J. and R.J. Dever, Jr., "Urban Highway Storm Drainage
Model: Precipitation Module User's Manual", prepared for the
Federal Highway Administration, Washington, D.C., June, 1980
3. Chen, C., "Urban Storm Runoff Inlet Hydrograph Study, Vol. 4:
Synthetic Storms for Design of Urban Highway Drainage Facilities",
Utah Water Research Laboratory, Utah State University, May, 1975
4. Soil Conservation Service, U.S. Department of Agriculture,
"A Method for Estimating Volume and Rate of Runoff in Small
Watersheds", SCS-TP-149, April, 1973
5. Schmalz, R.A., A.J. Knepp, and L.A. Roesner, "Urban Highway Storm
Drainage Model: Hydraulic/Quality Design Module", prepared for the
Federal Highway Administration, Washington, D.C., August, 1978
6. Bristol, C.R. and R.J. Dever, Jr., "Urban Highway Storm Drainage
Model: Cost Estimation Module User's Manual", prepared for Federal
Highway Administration, Washington, D.C., May, 1980
7. Burgi, P.H. and D.E. Gober, "Bicycle-Safe Grate Inlets Study,
Volume 1 - Hydraulic and Safety Characteristics of Selected Grate
Inlets on Continuous Grades", Report No. FHWA-RD-77-24, Federal
Highway Administration, Washington, D.C., June, 1977
8. Metcalf and Eddy, Inc., University of Florida, and Water Resources
Engineers, Inc., "Storm Water Management Model, Volume 1 - Final
Report", Report No. 11024DOC07/71, Environmental Protection Agency,
Washington, D.C., July, 1971
132
-------
KINEMATIC DESIGN STORMS INCORPORATING SPATIAL AND TIME AVERAGING
by
W. James and J.J. Drake1
ABSTRACT
The paper explores the use of a numerical
storm model as a pre-processor for a detailed
urban runoff model. The proposed storm model
generates hyetographs for each subcatchment, thus
simulating the spatial and temporal growth and
decay of a system of storm cells as they move
across an urban catchment system.
Traditionally, design storms were and
continued to be developed from statistical
analysis of point rainfall records that include
all types of rainstorms; the methodology was
originally appropriate for flood predictions
based on the rational formula. The resultant
rain distributions are unlike any type of
observed rain storm. This synthetic temporal
distribution is typically applied uniformly
across the catchment and flow hydrographs are
consequently also unlike observed runoff
hydrographs.
Large static cells of uniform rainfall
intensity are rare even in prolonged frontal
events. Convective cells tend to be circular
with a circular rainfall intensity pattern. Rain
cells tend to be elliptical, aligned sub-parallel
to the front and moving sub-parallel to it.
Rainfall is typically most intense near the
leading edge of the cell. Fast moving storms
produce very rapid point-intensity-duration
changes. Statistics of the size and distribution
of rain cells can be obtained most readily from
weather radar studies. The storm model presented
» Professor, Dept. of Civil Engineering and Associate Professor,
Geography Department, McMaster University, Hamilton, Ontario, Canada L8S
4L7
133
-------
in this study is based on the synoptic
observations of rain cells reported in weather
radar literature.
Intensity-duration-frequency curves obtained
from the model are similar to those derived
statistically from the long-term rainfall record.
The model is applied to catchments in the
Hamilton area in Southern Ontario. Results
indicate that storm cell kinematics are
significant, particularly in peak runoff
estimates (for drainage design) and water
pollutant loading estimates, because of the
sensitivity of the time-to-peak and rate-of-rise
of hydrographs and pollutographs. Such storm
models appear to be useful for computer-based
rainfall-runoff studies.
INTRODUCTION
Modern rainfall-runoff models for use in urban hydrology, such as the
Stormwater Management Model (Huber, 1975), allow up to six rainfall
hyetographs, and discretization of the catchrflent into (typically) 40-100
subcatchments. With a little modification the programs can accept 10-20
hyetographs distributed across these subcatchments. It is then a simple
matter to lag and adjust the hyetographs to represent a moving storm tracking
across the catchment in any given direction, and also to account for probable
growth and decay of the spatial size of the storm cell as well as change in
the rainfall intensity distribution across the cells.
In fact it is unlikely that storms will spontaneously occur, grow and die
off while remaining stationary over a typically small urban basin. It can be
shown that translatory storms produce very different runoff hydrographs and
pollutant loadings than those produced by the usually-accepted static storm.
NOT to account for translatory storms may introduce unjustified errors.
134
-------
DESIGN STORMS AS MODEL INPUT
Generally a prescribed return period is specified for a design project
and then a design storm that has the same return period is selected. The
project is designed not to fail when subjected to a calculated flood produced
by the design storm; it is assumed that the capacity of the design drainage
system has a return period equal to that of the design storm.
The validity of this assumption of a linear relationship between some
measure of the runoff hydrograph and a description of the rainfall hyetograph
is of importance (Wenzel and Voorhees, 1978 and 1979) and open to question.
This is especially true where antecedent moisture conditions are significant
and variable and where surcharging is involved. Continuous modelling, though
expensive, is seen to be helpful in this regard, but few continuous models
account for the complex actions of the combined sewer overflow and diversion
structures commonly encountered.
Design storms are usually either developed by a simple statistical
analysis of point rainfall records that include rain of all types, or an
historic storm is used (particularly for rare events) . In the former case the
resulting temporal distribution of rain may be quite unlike the rainstorms
occuring in nature (Urbonas, 1979) and thus result in runoff peaks that can
vary significantly from peak flows calculated statistically from long-term
simulation using recorded rainfall and calibrated models. The use of storm
models based on known synoptic characteristics of storms does not seem to have
been considered a design alternative. Synthetic storms attempt to aggregate
intensity duration stations from many storms of all types, thunderstorms and
135
-------
cyclonic rains, into a single and hence impossible storm event. The general
design hyetograph shape is based on data from many geographic regions, some
involving orographic and other local effects.
A number of studies have been carried out on the various published
methods of synthesizing design storms from rainfall records (see for example,
Marsalek, 1978; Arnell 1978) and these draw conclusions regarding the adequacy
or otherwise of the various synthetic design storm techniques. None of the
studies available to us has accounted for the storm kinematics or dynamics,
notwithstanding the fact that storm movement may significantly affect the
computed catchment response, especially pollutographs, or pollutant loadings
to the receiving waters. Most synthetic design storms are tantamont to an
attempt to replace the several precipitation types and kinematics by a single
simplistic rain hyetograph applied uniformly across a catchment, supposedly
appropriate to all shapes, sizes and kinds of catchment.
A promising path through this jungle has been proposed by Walesh et al.
(1979): hyetographs of major rainfall events are assembled from a long
historic record and applied to a calibrated event model. Rain intensities
thus still relate to actual precipitation types.Unfortunately, once again the
historic record is usually based on independent single point rainfall
observations, incapable of accounting for storm dynamics. Dahlstrom (1978)
has suggested that the spatial variability of precipitation intensity can be
taken into account by a complicated analysis of precipitation records from a
limited number of adjacent stations for urban areas of large size. However,
Dahlstrom evidently does not believe that current knowledge of meso-scale rain
storm characteristics would justify the development of a useful storm model
136
-------
for urban hydrology. Both Dalstrom and Arnell appear to support Walesh's
general approach to the use of many storms from the long-term rainfall record,
but do not caution against overlooking storm kinematics.
A recent paper (Yen and Chow, 1980), although another attempt to apply a
simple, approximate, spatially uniform, and static design hyetograph, does
discuss the effect of the inherent storage and attenuation of a basin on the
selection of the design hyetograph. These authors define a small basin as one
sensitive to high-intensity rainfalls of short durations and sensitive to land
use; inherent storage-channel characteristics do not suppress these
sensitivities.
The problem is not new. Clark (1973) has emphasized the case for
including inherent stochasticity and error in the modelling procedure and has
discussed this relationship to the observation network density. Van Nguyen et
al. (1973) advocate the use of a radar system rather than a dense network of
rain gauges. Both of these studies appear to hint at the need to model moving
storms .
In summary, the trend seems to be increasingly critical of the use of a
simple static, spatially-uniform design storm aggregating all kinds of storm
types, and towards the use of a number of historic storms selected from the
long-term record, or even continuous modelling using the entire long-term
record. There is also increasing interest in better sampling of the rainfall
inputs, but at present the accent is on data analysis rather than modelling of
dynamic storms. Hydrometeorologists have perhaps been too cautious to
advocate the use of storm models incorporating cell kinematics.
137
-------
PRECIPITATION TYPES SIGNIFICANT TO STORMWATER MODELLING
Three types of rainfall are generally recognized: orographic, cyclonic
and convective. In the Hamilton area, orographic uplift over the Niagara
escarpment rarely leads to rainfall although it may generate clouds. Cyclonic
precipitation, associated with frontal systems, has two components: a broad
belt of relatively low intensity rainfall lying along the warm front, and a
narrower belt of relatively high intensity rainfall lying along the following
cold front. Convective precipitation in summer may be caused by differential
local surface heating and generate small intense rain cells, or the cold front
rainband may in fact be composed of a linear set of convective cells
associated with the air mass moving from Lake Erie over the land.
Occasionally, very intense, widespread convective rainfall may be associated
with the incursion of a tropical storm into the area (e.g. Hurricane Hazel in
195*0 .
Adiabatic cooling is the cause of condensation and rainfall, and
vertical transport of humid air masses is a requirement. In convective
precipitation, heated air at the ground expands, reduces in weight, and takes
up increasing quantities of water vapour. The warm moisture-laden air becomes
unstable and pronounced vertical currents are developed. Dynamic cooling then
causes condensation and precipitation in the form of light showers, storms or
thunderstorms of high intensity.
Thunderstorms begin as cumulus clouds characterized by strong updrafts
that reach 25,000 feet. During development of the storm additional moisture
is provided by a considerable horizontal inflow of air. The storm enters a
138
-------
mature stage when the strong updrafts produce precipitation. Gusty surface
winds move outward from the region of rainfall and heavy rainfall occurs for a
period of 15 to 30 minutes. In the final dissipating stage of the storm the
downdrafts predominate and precipitation tails off and ends.
Thunderstorms comprise one or more such cells in varying stages of
development, the life cycle of which is usually completed in an hour or less,
However, such storms tend to be self-propagating by the formation of new cells
and in the Hamilton area generally move from the west (between, say, north
west and south west) at speeds of 35-50 km/hr and in broken lines or bands up
to 80 km in width. Severe storms may produce 5 cm of rain in less than half
an hour, while slowly moving storms may appear to remain in one locality for
an hour or more and produce a total point rainfall as great as 30 cm.
In the Hamilton urban catchment the greatest rainfall rates are
associated with convective precipitation. Although major structures may be
designed on the basis of an exceptional recorded event, urban stormwater
structures are designed on the basis of a composite synthetic storm, derived
from raingauge data that is assumed to begin, peak and end simultaneously over
the whole catchment. In fact, the greatest rates of runoff are usually
associated with a linear set of convection cells containing individuals that
are continually being generated and dissipated, and which moves across the
area.
In summary, rain storms travel in preferred directions; they do not
spontaneously grow and die over one spot, as suggested by current practice in
the analysis of point rainfall data. Cells have substantial speeds and
139
-------
intensity variations across areas typically appropriate to urban runoff
studies (e.g. 5-5000 acres). A substantial body of information is available
and the general characteristics of stormcells can be described.
It is preferable to specify the expected speed and direction of movement
of cells, and even cell size and rainfall intensity distribution, rather than
to assume no speed or direction, and excessively large cells with uniform
rainfall intensities.
Finally, thunderstorms have different characteristics from cyclonic
events. Point rainfall data does not distinguish between rainfall types, and
statistical analyses of rain data includes intensities from all types. Point
rainfall data cannot generally provide information on storm cell kinematics.
THE STORM MODEL
Radar studies of summer rain events resulting from moving clusters of
sub-circular convective cells have shown that the cells are relatively
short-lived (Austin 1967), that they tend to have an exponentially-decreasing
intensity away from the cell centre (Konrad, 1978) and that their statistical
properties can be matched to those of ground-based precipitation records
(Drufaca, 1977). Gupta and Waymire (1979) have proposed a stochastic model of
rainfall from such clusters, but no a priori single-event design storm for the
Hamilton area can be derived from it.
Studies of "line-convection" rainbands associated with extra-tropical
cyclones have shown them to be longer lasting and their structure to be one of
140
-------
sets of extended elliptical cells oriented sub-parallel to the front with a
component of motion along it (Hobbs and Brewer, 1979; James and Browning,
1979). This pattern of rainfall, oriented across the drainage basin and
moving down from head to mouth is apparently common in Hamilton, and forms the
basis of the present model. The form of the model is an infinitely wide
rainband in which the rainfall intensity decays exponentially away from the
line of peak intensity at different rates ahead and behind it. Thus
-k (t -t)
P = P e p tt
o p
where PQ is the instantaneous point peak intensity t is the time-of-peak at a
point and t is time at a point. Statistical studies of rainfall rates before
and after peak rates recorded by raingauges in several parts of North America
indicate that k = 0.54 k( and we have adopted this value. There are therefore
two parameters of the model, P and k (or k ), which can be evaluated from
intensity-frequency-duration curves published for the Hamilton area by
assuming that events of all durations for a given recurrence interval are
embedded in one storm. Figure 1 shows that the model i-f-d curves for which
the parameters have been evaluated with the 10 and 20 minute points from the
observed curves provide a close approximation to those curves. Projected
instantaneous point maximum intensities (PQ), and the values of ^ and k2 are
shown in Table 1. Konrad (1978) and other studies have suggested that
convective cells with similar peak intensities have an intensity
distance-decay exponent of about 0.5 km"1. If this value is assumed to be
correct for linear rainbands in the Hamilton area storm velocities can be
calculated, and these are also shown in Table 1
141
-------
Table 1
Return period
yr
10
50
100
o
mm hr
157
188
196
1
-1
mm
0.19
0. 15
0. 13
2
-1
mm
0.10
0.08
0.07
km hr
15
13
11
-1
The calculated values are similar to those observed for motion of linear
rainbands normal to their orientation on the radar studies. It appears that
increased rainfall for a given duration at longer return periods is due to
both an increased intensity in the storm and to a slower motion of it across
the area in question.
200
10
20 30 40
DURATION (WIN.)
50
60
Figure 1 Intensity-Duration-Frequency Curves Obtained by the Storm Model
142
-------
APPLICATION OF THE MODEL
One a model of a rain cell is proposed it is easy to show the effect of
the necessary time and spatial-averaging in order to produce expected runoff
hydrographs. Fig. 2 shows the instantaneous cross-section of a typical line
convection cell, having a peak intensity of 180 mm/hr.
The effect of storm kinematics is shown in Figure 3. The greater the
cell speed across the raingauge, the sharper the hyetograph. This is
equivalent to the point data recorded by rain gauges and used in design storm
determination. Hence the statistics are significantly affected by storm
velocities.
200
E
Cfl
Z
UJ
Fig. 2. Cross-section of typical line convection cell
143
-------
In order to model runoff it is necessary to produce input hyetographs
averaged over each subcathment or spatially discretized area, and also over
the concomitant basic computational time-step. The effects are shown
respectivly in Figures 4 and 5.
Typical hydrographs for the Chedoke Creek catchment in Hamilton are shown
in Figures 6 and 7 for the rainfall inputs shown in Figure 3.
CONCLUSIONS
Ground-based estimates of areal rainfall show considerably less
attenuation from the maximum point value than do radar-based estimates (e.g.
200
20
40
60 80
MINUTE
100
120
140
Fig. 3. Effect of velocity (5,10,20 km hr~' ) on time-distribution of
instantaneous rain intensity from storm shown in Fig.2
144
-------
Barge et al, 1979), because they do not take into account the translation of a
storm across the area. The implication of these studies for operational urban
hydrology is thus that a given duration-frequency event in a given catchment
may produce a smaller total volume of rain than is-currently assumed from
depth-area curves based on raingauge records, but that this volune may be
distributed in space and time such that it gives rise to a greater peak runoff
and rate-of-rise of the hydrograph.
Thus preliminary analysis has shown that the effects of the kinematics of
storms on urban stormwater hydrographs and pollutographs can be significant to
both modelling and design. More detailed information on the structure and
200
v = 10 km hr'1
At = 5 min
""""UAx = 0 (point rainfall)
JC.
e
E
en
z
UJ
100 -
= 2km
20
40 60
MINUTE
Fig. 4. Effect of basin size (0, 2 km length) on 5 minute rain intensities
from storm shown in Fig. 3. moving across basin at 10 km hr~
145
-------
kinematics of rainfall in the Hamilton urban catchment can be derived from
data from the A.E.S. weather radar that has operated for the past 10 years at
Woodbridge, Ontario.
ACKNOWLEDGEMENTS
The work reported here is part of a joint study on the Chedoke Creek
catchment. Assistance from the Regional Engineering Department of the
Hamilton-Wentworth Regional Municipality is gratefully acknowledged.
The runoff computations for the Chedoke catchment were carried out by Zvi
Shtifter, graduate student in the Department of Civil Engineering at McMaster
University.
200
E
E
CO
z
UJ
H
Z
100 -
I v*-At = o {instantaneous)
80
100
MINUTE
Fig. 5. Effect of time averaging (0,5j10 rain) on average rain intensities over
a 1 km long basin resulting from storm shown in Fig. 3 moving across
basin at 10 km hr.
146
-------
REFERENCES
Arnell, V. (1978): Analysis of Rainfall Data for Use in Design of Storm Sewer
Systems, Proc. Int. Conf. on Urban Storm Drainage, Univ. of Southamton,
April 1978, pp. 71-86.
Austin, P.M. (1967): Application of Radar to Measurement of Surface
Precipitation, Technical Report ECOM 01472-3, Contract DA28-043
AMC-01472(E), M.I.T., Cambridge, MA, 18 p.
Barge, B.L., R.G. Humphries, S.J. Mah and W.K. Kuhnke (1979): Rainfall
Measurements by Weather Radar: Applications to Hydrology, Water Resours.
Res. vol. 15, pp. 1380-1386.
Clark, R.A. (1973): Temporal and Spatial Variability of Storm Rainfall
Patterns and their Relationship to Operational Hydrologic Forecasting,
Fall Annual Meeting of the A.G.U., San Francisco, Dec. 1973.
Dahlstrom, B. (1978): A System for Analysis of Precipitation for Urban Sewer
Design, Proc. Int. Conf. on Urban Storm Drainage, Univ. of Southamton,
April 1978, pp. 100-110.
Drufuca, G. (1977): Radar Derived Statistics on the Structure of
Precipitation Patterns, J. Appl. Meteorol, vol. 16, pp. 1029-1035.
2000-
LJ
o
CE
<
I
O
1000-
Stolic
5 km hr~ up basin
10
20
30
TIME (HOUR)
Fig. 6 Computed runoff for a storm moving up the Chedoke Creek catchment
147
-------
Frederick. R.H., V.A. Myers and E.P. Auciello (1977): Storm Depth-area
Relations from Digitized Radar Returns, Water Resources Res. vol. 13, pp.
675-679.
Gupta, V.K. and E.G. Waymire (1979): A Stochastic Kinematic Study of
Sub-synoptic Space-time Rainfall, Water Resources Res., vol. 15, pp.
637-644.
Hobbs, P.V. and K.R. Biswas (1979): The Cellular Structure of Narrow
Cold-frontal rainbands, Quart. Jour. Royal Heteorol. Soc., vol. 105, pp.
723-727.
Huber, W.C. et al. (1975): Stormwater Management Model User's Manual, pub.
U.S. EPA, EPA-67072-75-017, March 1975.
James, P.K. and K.A. Browning (1979): Mesoscale Structure of Line Convection
at Surface Cold Fronts, Quart. J. Royal Meteorol. Soc., vol. 105, pp.
371-33?.
Konrad, T.G. (1978): Statistical Models of Summer Rainshowers Derived from
Fine-Scale Radar Observations, J. Appl. Meteorol., vol. 17, pp. 171-188.
Marsalek, J. (1978): Synthesized and Historical Storms for Urban Drainage
Design, Proc. Int. Conf. on Urban Storm Drainage, Univ. of Southamton,
April 1978, pp. 87-89.
2000 -
o
UJ
cc
u
1000 -
TIME (HOUR;
6
Fig. 7 Computed runoff for a storm moving down the Chedoke Creek catchment
148
-------
Urbonas, B. (1979): Reliability of Design Storms in Modelling, Proc. Int.
Syrap. on Urban Storm Runoff, Univ. of Kentucky, July 1979, pp. 27-35.
Van Nguyen, V.T., McPherson, M.B.- and Rousselle, J. (1978): Feasibility of
Storm Tracking for Automatic Control of Combined Sewer Systems, ASCE
Urban Water Resources Research Program, Technical Publication #35, Nov.
1978.
Walesh, S.G., Lau, D.H. and Liebman, M.D. (1979): Statistically-based use of
event models, Int. Symp. on Urban Storm Runoff, Univ. of Kentucky, July
1979, PP- 75-81.
Wenzel, H.G. and M.L. Voorhees (1978): Evaluation of the Design Storm
Concept, Fall Meeting of the A.G.U., San Francisco, Dec. 1979.
Wenzel, H.G. and M.L. Voorhees (1979): Sensitivity of Design Storm Frequency,
Spring Meeting of the A.G.U., Washington, May 1979.
Yen, B.C., and Chow, V.T., (1980): Design Hyetographs for Small Drainage
Structures, Proc. A.S.C.E., vol. 106, No. HY6, June 1980, pp. 1055-1076.
149
-------
HYDROGRAPH SYNTHESIS
BY THE HNV-SBUH METHOD
UTILIZING A PROGRAMMABLE CALCULATOR
by
Bernard L. Golding1, P. E., P. Eng.
Introduction:
The determination of the complete flow hydrograph for retention/deten-
tion basin design is now a prerequisite to the formulation of storm water
management plans for most new developments. They are also generally re-
quired for the investigation of existing urban drainage basins. At the pres-
ent time, most engineers use the SCS procedure (the- computation of rainfall-
excess by the SCS rainfall-runoff equation and the application of the com-
puted rainfall-excess increments to the standard SCS unit hydrograph) for
hydrograph computation because of its simplicity and general low cost even
though its deficiencies have now been recognized by many engineers. The
alternative is the use of more sophisticated models (i.e., SWMM, ILLUDAS,
HEC-1, etc.) whose fairly high costs and complexity are not always warranted
considering the benefits achieved.
The purpose of this paper is to acquaint engineers with the Howard
Needles Version of the Santa Barbara Urban Hydrograph Method (HNV-SBUH Method),
a fairly simple, yet easily applied low cost model for computing design flow
hydrographs. Included with this paper are calculator programs and user in-
structions for developing hydrographs by this model utilizing the Hewlett-
Packard HP-97 and HP-67 programmable pocket calculators. A BASIC language
version of these programs for microcomputers and user instructions for their
application is also available. The validity of the model by the simulation
of actual runoff events utilizing the programs listed herein is also demon-
strated.
General:
The HNV-Santa Barbara Urban Hydrograph Method (HNV-SBUH Method) is a
modification of a method originally developed by Mr. James M. Stubchaer,
F. ASCE, of the Santa Barbara County (California) Flood Control and Water
Chief, Water Resources (South), Howard Needles Tammen & Bergendoff, Orlando,
Florida
150
-------
Conservation District (1). The method as described herein computes a hydro-
graph directly without going through any intermediate process, as does the
unit hydrograph method.
The HNV-SBUH Method is in many respects similar to some of the time-
area-concentration curve procedures for hydrograph computation in which an
instantaneous hydrograph in a basin is developed and then routed through an
element of linear storage to determine basin response. However, in the
HNV-SBUH Method, the final design (outflow) hydrograph is obtained by rout-
ing the instantaneous hydrograph for each time period (obtained by multiply-
ing the various incremental rainfall excesses by the entire watershed area
in acres) through an imaginary linear reservoir with a routing constant
equivalent to the time of concentration of the drainage basin. Therefore,
the difficult and time consuming process of preparing a time-area-concen-
tration curve for the basin is eliminated.
Model Description:
A step-by-step description of the basic HHV-SBUH Method is given
below:
1. Runoff depths for each time period are calculated using the following
equations:
(1) Directly Connected Impervious Area Runoff -
R(0) = I P(t) (inches) (1)
(2) Pervious Area Runoff -
R(l) = P(t) (l-i)-f(l-lt) (inches) (2)
(3) Total Runoff Depth -
(R(t) = R(0) + R(l) (inches) (3)
where
P(t) = Rainfall depth during time increment At (inches)
f = Infiltration during time increment At (inches)
lt = Total impervious portion of drainage basin (decimal)
I = Directly connected impervious portion of basin (decimal)
At = Incremental time period (hours, i.e., 0.25, 0.50, etc.)
The directly connected impervious areas (I or DC IA), sometimes referred to
in the literature as the hydraulically effective impervious area, are those
impervious areas where runoff therefrom does not flow over a pervious area
before reaching and entering an element of the drainage system (streets with
curbs, catch basins, storm drains, etc.).
In the derivation of Equation 2 above and as shown on Figure 1, the rain-
fall 6n the pervious area is consid&red to be made up of two parts, the rain-
fall on the nondirectly connected impervious area (lt-l)P(t) which is assumed
to run off uniformly onto and to be distributed uniformly over the pervious
area and the rainfall on the pervious area (l-lt)P(t) which, when added to-
151
-------
gether, gives the equation (l-l)P(t). The runoff from the pervious area is
then obtained by subtracting infiltration f (l-lt) from this total rainfall
or P(t)(l-l)-f(l-lt).
2. The instantaneous hydrograph is then computed by multiplying the total
runoff depth R(t) for each time period t by the drainage basin area A in
acres and dividing by the time increment At in hours.
(4) l(t) = R(t) A/At
(cfs)
As in the Rational Method, the conversion factor 1.008 was ignored.
3. The final design (outflow) hydrograph Q(t) is then obtained by rout-
ing the instantaneous hydrograph I(t) through an imaginary reservoir
with a time delay equal to the time of concentration (Tc) of the
drainage basin. This flood routing may be done by use of the follow-
ing equation which is subsequently derived on Table 1.
(5) Q(t)=Q(t-l)+K[l(t-l(+l(t)-20(t-l)] (cfs)
where
K =
At
2Tc+At
and where T = Time of concentration of the basin (hours).
c
In the HNV-SBUH Method, all of the rain which falls on the basin is
considered runoff except for the first 0.1 inch (depression storage) which
is automatically subtracted from the rainfall in the programs.
Infiltration;
In the programs for the HNV-SBUH Method subsequently listed, infiltra-
tion is computed in accordance with the relationships illustrated on Figure
2 as first described by Holtan and developed and presented by Terstriep
and Stall (2). In this methodology, it is assumed that an Initial (Maximum)
Total imptrvioui Area *
* *
| PERVIOUS AREA]
AREA i (I-UI
1
\ NON-DCIA!
AR£A*(tT-Il
E
A
RID
t i
ijg
REA< [
RIO)
r
AREA' (I-1TI*(IT-1)+I*I
R(n p(D-(i-iTi*p(t)(iT-n-f(i-iTi
R(ll = P-(I-I)-«I-IT>»P(«)-I
152
-------
SW. brim Urb«n Hydro*** Mtathod
Dwivttlon wf Routing Equation
1. Consider a linear reserve with a definite siwage S men that (ha Slorage is oirec,l₯
proportional to ihe omf low A S
-------
infiltration. Standard Antecedent Moisture Conditions for these four types
of soils, as also established by Terstriep and Stall and which are subse-
quently discussed, can be used to determine F. The various factors used to
compute the Standard Infiltration Curves are shown on Figure 3 for each of
the four SCS Standard Hydrologic Soil Groups are listed on Table 2.
10
IT 2
HYDROLOGIC SOIL GROUP
01234
TIME, hours
Figure 3 Standard infiltration curves for bluegrass turf
Facton UMd For Crttti Mir* 1T» Standard
Infiltration Curva For Q'MMd Ann
Hydrolooic soil group USOA designation
Final constant infiltration rate fc, inches pw hour
Initial infiltration rate. fg, inches per hour
Stiape (actor, If, or infiltration curve
Available storage caoacity. 5, in joil mantle,
inches, far four antecedent conditions
Bone dry. condition I
Rather dry, condition 2
Rattier met, condition 3
Saturated, condition 4
Infiltration accumulated. F. in sail manite.
inches, at start olrainlall
Sons dry. condition 1
Raiher dry, condition 2
Rathef wet, condition 3
Saturated, condition 4
0.25
5
Total rainfall during
Bone dry
Rather dry
The four antecedent moisture conditions listed in this table - Bone Dry
(Condition 1), Rather Dry (Condition 2), Rather Wet (Condition 3) and Sat-
urated (Condition 4), are dependent on the total rainfall that occurred dur-
ing the five days preceding the particular storm (Antecedent Moisture Con-
dition) as shown in Table 3.
In the computer programs subsequently described, the infiltration f, at
some particular value of F, is first computed for the particular Antecedent
Moisture Condition by first summing up incremental volumes under the curve
for increments of 0.01 hour, computing a second value of infiltration fz at
incremental time At later and then averaging the two computed infiltration
values. To compute rainfall-excess from the pervious areas, this average
infiltration over the At time period is then applied to that rainfall-excess
increment during which time element a total rainfall of 0.1 inch has fallen
(depression storage).
Theoretically, the 0.1 Initial depression storage should be subtracted
from the rainfall and the infiltration curve applied only at that subsequent
time after rainfall has started, at which the total volume of rainfall and
the total volume of infiltration storage under the infiltration curve are
equal. However, in the programs subsequently listed the infiltraton curve
154
-------
is applied immediately after rainfall begins and the 0.1 depression storage
has been satisfied. Therefore, a storm with small initial amounts of rain-
fall falling over an extended period of time will cause a premature reduction
of the infiltration rate (a decay of the infiltration rate with time), which
will result in higher runoff than would actually be the case. However, since
the infiltration accumulated in the soil matter due to antecedent rainfall
(=F) is very broadly defined and since the pervious area contribution is gen-
erally small, this induced error is normally not significant.
As an alternative to the four standard infiltration curves shown on Figure
3 and as defined on Tables 2 and 3, site specific infiltration data obtained
by means of infiltration testing may be used. Necessary infiltration tests
to determine the initial and final infiltration rates and decay factors are
simple to perform. A recent Federal Highway Administration publication (4)
describes standard procedures for infiltration testing. Antecedent moisture
conditions may be determined by means of rainfall gages.
Required Model Input Data:
In addition to successive rainfall increments,either actual increments for
runoff simulation or those from an arbitrary design storm for design hydro-
graph computation, required inputs to the model are as previously indicated
the total drainage basin area (acres); the total impervious area (decimal);
the directly connected impervious area (decimal);the time of concentration
the basin (hours) and the required soils infiltration information - the
initial infiltration rate f (inches/hour) at the start of the rainfall; the
final infiltration rate f finches/hour) and the water stored in the soil F
(inches) at the start of the rainfall.
Although the total impervious areas of most drainage basins can, in
most cases, be readily determined from aerial photographs, the determination
of the directly connected impervious areas cannot. This determination is
particularly difficult in the case of the roofs of buildings where in some
cases all or parts thereof are directly connected by roof drains and in
other cases (many times in the same block), runoff from roofs drains onto the
grassed areas, either by roof drains or directly from the periphery of the
roof. Therefore, a detailed field inspection of the buildings, driveways,
etc. in the drainage basin and the marking of those directly connected im-
pervious portions on the aerial photographs is generally required. For
larger basins, a sampling technique must generally be used because of the
cost involved. If streamflow records are available, the runoff depths from
small storms on dry watersheds may be used to determine the directly con-
nected impervious area, (ratio of runoff to precipitation), as almost all of
the runoff from small storms will be from those areas. Where no actual data
is available or obtainable such as where hydrographs for a new development
must be computed, values based on similar type developments or literature
values can be used. Recent studies by Miller (5) and Alley and Veenhuis (6)
indicate that the directly connected impervious area constitutes approxi-
mately 50 percent of the total impervious area of single family residential
areas (all lot sizes including trailer courts), 50 to 75 percent of multi-
family residential areas and 75 to 90 percent of commercial areas.
155
-------
The time of concentration of a basin, which by definition is the travel
time of flow from the most remote basin to the point of interest, is calcu-
lated in the usual manner by summing up the initial (overland flow) and chan-
nel travel times. The channel travel times, both in roadside gutters or
swales and in the downstream storm sewer systems or collector channels, are
relatively easy to compute. A simple calculator program for travel time in
storm sewer systems which computes the velocity of each pipe flowing full
(=flowing 1/2 full), the flow travel time in each pipe and then sums up the
flow travel times is available from the author. The determination of the
initial travel time (overland flow time), which in many instances (particular-
ly in the case of small basins) is larger than the channel travel times, is
more difficult to compute. For instance, the pervious areas which generally
constitute much of the flow paths of overland flow may not contribute at all
during the early portion of a rainfall (where f>P(t)). However, based on the
simulations so far undertaken by the writer and as undertaken by Stubchaer
utilizing the original version of the SBUH Method, a nominal initial flow
travel time of some sort is necessary input to the model. This initial flow
travel time may be either computed utilizing one of the overland flow travel
times (time of concentration) equations such as the equation of Ragan and
Durer (7), or estimated. Generally, the writer has used 5 to 10 minutes as
the initial flow travel time from the center of a residential lot to the road-
side gutter or swale.
However, in the HNV-SBUH Method, as in the original SBUH Method as de-
veloped by Stuchaer, or in other methods where the time of concentration is a
principal parameter, such as the Linearized Subhydrographs Model (8), the
determination of the time of concentration is obviously the weakest link.
This is particularly so where trapezoidal or rectangular shaped open chan-
nels (not pipes) are the principal conveyance systems. This is, of course,
the direct result of the fact that the velocity of flow in open channels
increases with depth which in turn reduces the travel time in such systems.
However, the use of the velocity of flow at bank full stage to compute
flow travel time appears to give a fair approximation of such travel time
and has been used by the writer for design purposes.
Also, the HNV-SBUH Method makes the assumption that the flow in
the channel is unrestricted by inadequate conveyance system capacity (i.e.,
inadequate pipe sizes). Where such inadequate system capacity exists, the
time of concentration should be increased to reflect resultant street pond-
ing (storage).
Model Validation
General: Four urban drainage basins of different sizes, locations and char-
acter were selected for use in model validation; hereinafter referred to as
Basin Nos. 1 thru 4. All were well documented as to their area, physical
properties, land use, soils types, etc., and all had excellent, accurate
recorded rainfall-runoff information available including information on
the amount of antecedent rainfall prior to the particular event simulated.
The properties of each of the four basins used in the validation effort are
listed on Table 4 and are described in detail elsewhere. (9) (10) (2)
156
-------
TABLE IP
PROPERTIES OF OMIHME 8 AS I US
h,ln N-*. 22' Contributing Ar.. (Acr«l
(Aero*! Total Imp. DC I A P«rV(ou* Gro
HultlfMtMy 14.7 U.I? 10.* 6.5 k.J Q
3d.) 34.4 15.8 10.5 Id. 6 A
19.5 17.9 38.1 Bl 15
B3J 5J* 332 3 60
n Group B; not ctaiilFlad by M»r*»lek.
Basin No. 1, the USGS's Multifamily Residential Area, in Bade County,
Florida, 14.7 acres in size, consists principally of a very flat area essen-
tially covered by apartment buildings and adjacent parking lots. Runoff from
this site is collected by a Y-shaped (in plan) storm sewer system terminating
in a 48" diameter pipe in which stage was monitored. The streets have no
curbs .or gutters; surface runoff from which is collected in the center of
the streets which are depressed to act as a swale.
The soil of the pervious portion of the basin which constitutes 19.3
percent of the total area consists of Perrine Marl which has a very low in-
filtration rate (Hydrologic Soil Group D).
Basin No. 2, the USGS's Transportation (Highway) Site, in Broward
County, Florida, 58.3 acres in size, consists essentially of a very flat
3,000 foot long segment of a six lane divided highway and adjacent contribu-
tary area. Runoff from this site is collected by a conventional circular
reinforced concrete pipe storm drain system running on one side of the high-
way. The adjacent contributary area consists of a mixture of small com-
mercial, residential and open, undeveloped areas. The base soil consists of
fine sand (Hydrologic Soil Group A) such that runoff is generally less than
20 percent of the rainfall (from the 10.5 acres of directly connected imper-
vious area). However, the time of concentration of the basin (=20 minutes)
generally reflects 15 minutes of flow travel time in the storm sewer system
and 5 minutes of flow overland travel time on a directly connected impervious
area on its extreme end. Certain of the undeveloped portions of the basin
are extremely flat and also extremely rough, overgrown with vegetation and
contain significant depression storage and do not contribute runoff as will
be subsequently described.
Basin No. 3, The Malvern Test Catchment in Burlington, Ontario, 57-6
acres in size, is a gently sloping (1%) single family residential area
drained by a conventional storm sewer system. Input to the storm sewer
system is by means of catch basins located on both sides of the curbed road-
way system and in the swales which drain the backyards of the houses.
As described by Marsalek (10), 19.51 acres of this 57.60 acres basin
is impervious of which area 17.88 acres are directly connected impervious
area; the 1.63 acres of sidewalk area are not directly connected.
As also described by Marsalek the soils are sandy loam with >an initial
infiltration rate f0 of 3.0 inches/hour, a final infiltration rate of fc of
0.52 inch per hour and a shape (decay) factor K of 4 day-
157
-------
Basin No. 4, the USGS's Boneyard Creek Basin, located in Champaign-
Urbana, Illinois, consists of a large essentially fully developed, fairly
flat urban area 2,290 acres in size; This basin contains portions of the
University of Illinois campus, old and new residential areas and a sizable
commercial area. Runoff is collected by a large network of storm sewer
systems which conduct flow into an open channel collector system approximate-
ly 3 miles in length. According to Terstriep and Stall (2), only 1,165
acres of this 2,290 acre basin actually contribute flow (for the intensity
of the storms simulated) of which 830 acres are impervious and of which 534
acres are directly connected. The soils in the basin consist, again ac-
cording to Terstriep and Stall, of silty loams (Hydrologic Soil Group B).
For purposes of simulating runoff from this "reduced" basin an "average"
time of concentration of 60 minutes was used rather than the 72 minutes as
originally determined by Terstriep and Stall for the entire basin (11).
Close proximity enabled a visual inspection of the first two basins.
From this visual inspection it was further concluded that only approximate-
ly 50% of the pervious area of Basin No. 2 (the USGS's Transportation Basin),
and the nondirectly connected impervious area contributing thereto, con-
tributed to runoff from the storm events simulated (0.5" to 2.0" of rain-
fall) because of the large depression storage inherent in the undeveloped,
pervious portions of this basin which estimate, as will, be .subsequently
described, resulted in the best simulation of actual events. From the vis-
ual inspection of Basin No. 1 (the USGS's Multifamily Residential Basin),
it appeared that perhaps 5 to 10 percent of the pervious area and nondirect-
ly connected impervious area contributary thereto might not contribute to
runoff from this basin from the storm events simulated. However, the assump-
tion was made that the entire basin contributed which resulted in satisfac-
tory simulation of actual runoff events so that effect of a reduction in
contributary drainage area was not investigated.
In the simulation of runoff from the first two basins, five minute
duration rainfall increments were used: 1 minute, 2 minute and 10 minute
increments for Basin No. 3; and six minute increments for Basin No. 4.
Basin No. 1: In all, runoff from eight rainfall events on the USGS's
Multifamily Residential Area were simulated; the results of which simula-
tions are shown on Figures 4 thru 11. The rainfall events simulated were
essentially selected at random. However, care was taken to choose rainfall
events with a reasonably wide range of antecedent rainfall conditions. For
all events simulated, a time of concentration of 15 minutes was used in the
simulations; 10 minutes computed pipe flow time assuming at an n=0.012 plus
5 minutes of overland flow time. The results of the simulations; the date
of each storm event simulated; the antecedent moisture condition at the on-
set of the storm; the recorded and simulated peaks and volumes; and the per-
cent errors are listed on Table 5. As can be observed from the figures and
from Table 5, the simulations of runoff from Basin No. 1 were fairly good
in all cases.
In order to determine the sensitivity of the simulated runoff to the
time of concentration, simulations of the storm of 2 June 77 (Storm No. 6),
158
-------
cn
MULTIFAIV1IL.V RESIDENTIAL BASIN
STORM NO. 4 - 11 MAY 77
iToO 1/30 eCKJ
MULTIfiAMILY RESIDENTIAL BASIN
STORM NO. B - 1 JUNE 77
BIMULATEO
I&OO 19:00
TIME
FIQURE 4
FIGURE E
MULTIFAMIUY RESIDENTIAL BASIN
STORM. NO. 6 - a JUNE 77
SIMULATED
RECORDED
MUL.TIFAMILY RESIDENTIAL BASIN
STORM NO. 7-4 JUNE 77
BIIVIUI-ATEO
T
5:OO 5:30
TIME
PIQURE B
FIGURE 7
-------
MULTIFAIVIILY RESIDENTIAL BABIN
STORM IMO. 8 - GISO/77
.SIMULATED
MULTIFAMILY RESIDENTIAL BABIIM
STORM IMO. 9 -1B JULY TT
FIBURE B
O1
o
u.
u
0
MULTIFAMILY RESIDENTIAL. BASIN
STORM IM0.16 - B AUBUST 77
SIIV1ULATED
8*00 8:3O
9OO 9:30
TIME
IO3O M-00
FIOURE 10
MLJLTIKAMILY REaiDENTIAL BASIN
8TOpM NO. 17 - 8 AUGUST 77
17OT) IB'OO
(9-OO 2OOO
TIME
a-oo 2200 23^00
-------
were also made for times of concentration of 12 and 18 minutes. As can be
seen from Table 6, an approximately 5 percent difference in peaks resulted
which differences were not considered significant.
in« 77
ilv 77
ug 77
2.25
0.39
1.12
1.12
1.56
0.53
TABLE 5
HOPEL VALIDATION RESULTS
MULTtFWILY RESIDENTIAL BMIN
Antecedent Recorded SInulatM
folsture Runoff Runoff
Condition (ln.) (|n.)
1.82
0.16
0-71
0.67
1.92
0.26
0.7*
1.15
Recorded Stnwlgted
Paak - Q Peak - Q
tcfs) (cfs)
Z.2J
5,13
0.40
1-31
6-77
IS- 66
19.23
1
-------
TRANSPORTATION BASIN NO. 2
STORM NO. 3S - 31 OCTOBER 7S
SIMULATED
RECORDED
M:00 I|:3Q 12-00
TIME
TRANSPORTATION BABIN NO.
STOPM NO. 44 - 21 MAY 7E
SIMULATED
RECORDED
I3:'30 I4'OO
TIME
FIOURE 12
cn
a
5
tt.
TRANSPORTATION BABIN NO.
BTORM NO. 4B - 22 MAY 76
SIMULATED
TRANSPORTATION BABIN NO 2
BTORM NO. 47 - SB MAY 7B
SIMULATED
I7OO I7;3O
TIME
FIOURE 14
FIBURE 16
-------
TRANSPORTATION BA8IN NO.
STORM NO. 49 - as MAY 7B
SIMULATED
RECORDED
0
' 5
IL
TRANSPORTATION BASIN NO.
STORM NO. 53-11 JUNE 7B
SIMULATED
RECORDED
ZOO 230
330 4«O
FIOURE 16
FIOUHE -17
co
TRANSPORTATION BASIN NO. 8
TORM NO. s? -IB JUNE 76
0
' 5
TRANSPORTATION BASIN NO.
BTORM NO. OB - 23 JUNE 76
SIMULATED
80O 83O 9:OO 9:30 IO:OO IO:iO
TIME
FIOURB 1G
FIOURE ta
-------
O1
TRANSPORTATION BASIN NO. 2
BTORM NO. S9B - B4 JUNE 7B
SIMULATED
TRANSPORTATION BASIN NO.
STORM NO. 97 - 1O IVIAV ~77
BIIVIULATED
2'00 2-30
TIME
TRANSPORTATION BABIN NO. 2
BTORM NO. 1OO - B7 MAV 77
BlMUI-ATeo
TRANaPORTOTION BABIN NO. a
STORM NO. 1O1 - BS MAV 77
-------
TRANSPORTATION BASIN NO. i
QTORM NO. 1Oa - 1 JUNE 77
SIMULATED
TABLE 7
MODEL VALIPATIOM RESULTS
TRANSPORTATION (HIGHWAY) BASIN
Antecedent Recorded Slmolatt
la I nf a II Holsture Runoff1 Runoff1
(In.) Condition (in.) (In.)
Recorded Simulated
Peak - Q Peak - Q
(efs) (cfs)
31 Oct 75
21 May 76
22 May 76
28 May 76
29 Hay 76
it June 76
19 June 76
Z3 June 76
24 June 76
10 May 77
27 Nay 77
Z9 Hay 77
t June 77
,5
.5
.4
.6
.5
.5
3 O.I it
It 0.23
it 0.16
3 0.50
It y. |5
l| 0.16
.40 2 O.Wt
.09 k 0.33
.74 It 0.20
.04 3 0.32
.37 3 0.53
.98 It 0.53
.16 it 0.41
.17 +2
.27 *l
.15
.66 +3
.14
.12 -Z
.47 *
.39 *l
-31 +5
.52 +6
.63 +2
.64 +z
.51 *2
9.85
17-30
9.50
23.69 2
7.22
a. 22
9.85
12.
14.
32.
34.
52.
21.
7
7
3
B
3
.19
.14
.98
.60
.45
.45
.65
.88
.66
.20
-29
.93
6 20.27
TABLE 8
SENSITIVITY ANALYSIS - ANTECEDEKT MOISTURE COW1TIOH
TRANSPORTATION (HIGHWAY) OASIH
STOW NO. 100 - 27 JUNE 77
Simulated
Recorded
r I bur Ing ATM of 3*1.4 Acre*.
Basin No. 3; In all, runoff from five rainfall events on the Malvern Catch-
ment were simulated; the results of which simulation are shown on Figures
25 thru 29.
MAL.VERN CATCHMENT
STORM rjo.1 - es SEPT. 73
RECORDED
e:oo e-30
TIME
165
-------
CT)
MALVERN CATCHMENT
STORM NO. a - Z3 SEPT. 73
MAI-VERN CATCHMENT
STORM NO- B - 89-30, OCT. T3
RECORDED
SIMULATED
17 OO ZIOO
TIME
MALVERN
TDHIVI NO.
CATCHMENT
i - 31 MAV T>
19 30 ZO OO
TIME
MALVERN CATCHMENT
BTORM NO. 7 - 81 JUNE T4
-RECORDED
-------
As can be observed from these Figures, the simulations were gener-
ally fairly good considering that the basin was considered as a whole and
not broken down into subbasins.
For all events simulated, it was assumed because of the full devel-
opment of and steepness of the basin that the total 38.0 acres of pervious
area contributed to flow and that in conformance with the assumption in the
HNV-SBUH Method, that runoff from the nondirectly connected impervious area
(1.63 acres of sidewalks) was supplementary rainfall uniformly applied to
and distributed over the pervious area.
For all events simulated the Marsalek infiltration data was used. In
applying this data, it was assumed that no water had accumulated in the soil
from previous rainfalls (F=0); a Bone Dry (Condition 1) situation existed.
Actually, however, the low initial infiltration rate and large decay factor
inherent in the Marsalek infiltration data was, for all practical purposes,
the equivalent of a constant infiltration rate of 0.52 inch/hour. In all
cases, the assumed Marsalek initial abstraction of 0.02 inch rather than the
standard 0.1 inch of the program was used.
In all cases, a time of concentration of 15 minutes (9 minutes of
pipe flow time at n=0.015 plus 6 minutes of overland flow time) was used.
A sensitivity analysis of the time of concentration for three of the
five events simulated showed that variations of three to four minutes did
not seriously affect the computed peak flows and had little or no effect on
hydrograph shape or volume.
Basin No. 4: In all, runoff from six rainfall events on the Boneyard Creek
Basin were simulated; the results of which simulations are shown on Figures
30 thru 35. For all events simulated, a time of concentration of 60 minutes
and the reduced (actually contributing) areas as determined by Terstriep
and Stall were used.
As can be observed from Figures 30 thru 35, the simulations, except
for the storm of 7/2/65 (Storm No. A-24) as shown on Figure 35, were gen-
erally quite good considering that the basin was taken as a whole (not sub-
divided) . Also, the simulated hydrographs do not include a base flow which
varied from about 2 cfs at the start of the various storms to about 10 cfs
at their conclusion, which would have pushed these hydrographs up and to
the right, thus improving the simulations over that shown. This one excep-
tion to good simulation was attributed to the fact that the particular storm
was of such magnitude (1.9" in 60 minutes) that the capacity of the storm-
water conveyance system was obviously overtaxed (storm sewer suppression)
with resultant significant roadway ponding. Increasing the time of concen-
tration to 80 minutes gave a fair reproduction of the actual recorded hydro-
graph. Terstriep and Stall using ILLUDAS (2) computed a peak flow of 722
cfs for this particular rainfall event, also significantly greater than
that recorded.
167
-------
ft.
H,,
3:OO
BONEYARD CREEK BABIM
STORM NO. A-1 - 10/Pe SO
330 4:00 4:30 5:00
TIME
.
0
O'ocH
5:30 6:00
FIGURE 3O
BONEYARD CREEK BABIN
TORM NO. A-8 - 11.1B BO
.neconoeo
0-
1800 18:30 19:00 19:30 20:00 20:30 21.00 21:30
TIME
Ptoune 31
cn
00
CREEK BASIN
STORM NO. A-17 - 3/B/84
18:00 18:30 19:OO
19:3O 20:OO
TIME
21:00 21:30
Fiaun
ft
S30°-
z
It
200-
BONEYAnD CREEK BABIN
STORM NO. A-S3 - B/BB/SB
\\ RECDHDED
BIIVIUI.ATBD
10:30 11:00
11:30 12:00 12:30 13:00
TIME
13:30 14:00
FIGURE
-------
CT>
to
B.
U.
0 600
Z
B
400
BONEYAHD CREEK SABI1M
STORM IMO. A-B4 - 7/B/GB
BONEYARD CREEK BABIN
STORIVI NO. A-B7 - 4/BO/BB
4:00 4:30
5:00 5:30
TIME
-------
Generally acceptable to good simulation of actual runoff events
by the HNV-SBUH Method has also been achieved by the writer on the flat 85
acre Hartford Avenue Basin, a single family, storm sewered residential area,
and on the 58 acre Brentwood Drive Basin, a fairly steep storm sewered,
mixed residential-commercial area, both in Daytona Beach, Florida and both
gaged as part of the 208 studies in that area; on the 28 acre Lake Eola
Basin in Orlando, Florida, a downtown business-commercial area almost 100%
impervious and drained by a storm sewer system; on the 494 acre Cedar Hills
Basin near Surry, British Columbia, a residential area essentially drained
by open ditches; and on the 97 acre St. Louis Heights Basin, a very steep,
storm sewered, residential area in Hawaii. Site specific infiltration data
was used for simulating runoff from the Cedar Hills and St. Louis Heights
Basins. A simulation of the May 12, 1979 rainfall event over the 28 acre
Lake Eola Basin is shown on Figure 36.
30-,
as-
20-
u.
IL 15-
0
z
EC
10-
5-
COMMERCIAL STUDY AREA
LAKE EOLA BASIN
STORM Of IS MAY 73
DRAINAGE AREA = 2B ACRES
TIME OF CONCENTRATIONS 1O MIN.
RECORDED
SIMULATED
BV HNV-SBUH
METHOD
Iff 00
20:00
TIME
20'30 2MOO
FIQURE 36
Program Descriptions:
Two separate versions for the HNV-SBUH Method are listed at the end
of this paper. The first version requires the manual entrance of each rain-
fall increment. In the second version, all of the rainfall increments are
entered at the start of each program.
Hewlett-Packard (HP) Programs; The first HP program, referred to as the HP-67
version although it will operate on the HP-97, requires the use of and can be
recorded on a single magnetic card. Initially, after the variables are en-
tered and the program started (Press R/S), the infiltration curve is integra-
170
-------
ted at intervals of 0.01 hours until the volume of water already stored in
the soil (=F) is achieved at which time 0.0000000 appears in the display. As
previously stated, each rainfall increment must then be manually entered into
the calculator, which, when restarted each time (Press R/S), will compute the
ordinate of the runoff hydrograph.
The second HP program, referred to as the HP-97 version, or the HNV-
SBUH Method-Automatic Version, although it will also operate on the HP-67,
has been separated into two different parts each one of which must be record-
ed on a separate magnetic card.
In the first part of the program (Card No. 1), the rainfall incre-
ments (a maximum of 26) are placed by the program into Storage Registers Rll-
R19; three each in Storage Registers R11-R18 and two in R19. The maximum
rainfall increment that can be placed in any of the registers is 9.99 inches;
the minimum is 0.01 inch. The rainfall increments so entered can be recorded
on a blank magnetic (data) card for subsequent reuse and thereby avoid the
necessity of the individual reentry of same.
In the second part of the program (Card No. 2) the ordinates of
the runoff hydrograph are computed. Initially, as previously stated, the
area under the infiltration curve is integrated at time intervals of 0.01
hour until the volume of water already stored in the soil when rainfall be-
gins is achieved (=F). At the conclusion of this operation, the calculator
will then automatically start to print the time at the end of each rainfall
(=£At) and the ordinate of the runoff hydrograph Q in cfs at that time. The
program stops when the ordinate of the runoff hydrograph <0.01 (0.01 displayed)
In this version zero (0.0) is used as a flag in the second part of
the program, which precludes the use of zero rainfall increments. In such
cases, rainfall increments of 0.01 must be used which small increments will
not substantially affect the accuracy of the computed hydrographs.
The volume of runoff in inches (area under the hydrograph) can also
be computed by both programs (by pressing fa) which volume is then com-
puted by the equation
12*360Q'£Q'At = Q.99274«ZQ»At
V ~ 43,500'A A
In both HP versions, the shape (decay) factor K in Horton's equa-
tion may be changed from 2 in the programs by deleting 2 from Steps 10 and
65 of the HP-67 program or from Step 188 of Card No. 2 of the HP-97 pro-
gram and placing another value at these same locations.
Detailed User Instructions precede each of the programs listings.
Conclusions;
1. The HNV-SBUH Method with the four standard infiltration curves (for
Soil Groups A, B, C and D) is an effective, easy to use method of
simulating acceptable runoff hydrographs from urban drainage basins.
171
-------
2. Site specific infiltration data would probably have considerably im-
proved the simulations and if possible should be determined and util-
ized.
3. Runoff hydrographs from residential areas or other urban areas with
less than 50% impervious cover (25% directly connected impervious area)
are extremely sensitive to antecedent moisture conditions and the in-
filtration capacity of the soil(s) in the particular basin such that
the use of a more sophisticated model to compute runoff hydrographs
for design purposes would probably not significantly increase the
accuracy of such computed runoff hydrographs such that the use of a
more sophisticated model may not be warranted.
4. Small errors in the time of concentration of a particular basin do not
appear to significantly affect the accuracy of predicted runoff by the
model.
5. Because of the simple input requirements of the HNV-SBUH Method and
the ability to quickly compute runoff hydrographs with it on a pro-
grammable calculator or on a small microcomputer, its use is extremly
economical; hydrographs can be computed in 20 minutes on the program-
mable calculator and in 10 minutes on a microcomputer by an engineer-
ing technician. A calculator program for computing a hydrograph by
the HNV-SBUH Method on an HP-33E or HP-25 calculator is available from
the writer.
6. The design inflow hydrograph from a major design storm over a basin at
the entrance of the hydrograph into the receiving stream computed for
the purpose of determining receiving stream size should reflect
street ponding (storage) by increasing the time of concentration
used to compute such design inflow hydrographs when the convey-
ance system leading to the receiving stream has been designed for
a smaller design storm with resultant pipe sizes too small to
adequately convey runoff from the major design storm or when the
catch basin capacity is insufficient to permit entrance of all
runoff from the major design storm.
172
-------
HP-67
HP-
P«Bt>
co
HNV-SANTA BARBARA UBCAH HYDRDGilAPH METHOD
STCP
K
5.
4^
7.
8.
MSTRUC1HWS
Initialize (cleat all register*)
Input Known Valuea
c) Directly Connected Imoervious Portion
Dralnaae Basin
d) floe Increment
e) Time of Concentration
f) Initial Infiltration Rate
g) Final Infiltration Rate
Start Pragm
Place Flrat Rainfall Increaent in X-ReKleter
Continue to nlace succeaaive Rjiinfall Incre-
enta P(t) In X-Reglater one at a tiatt until
all entered. Ttien auccegaive keroa
To coopute runoff volume (area under bydronrap
To coapute new pydronraph. to to Step 2
m*uT
DATAAJMTS
A (arrra)
T iteclnnl
IL««cl«sl
at (houra
T (houra
fo (in/hr
fc (in/hr
F (inches
P(t),(Wi
Pftl Hal
)
KIV8
L II.
i ": :j to
i ii
I.HTI ( o
ISTO II 1
ISTOII 2
1 II
ISTO |( }
ISTO II t
ISTO II A
ISTO || >
ISTO II c
r'll i
t)l - llws
t)l IU/S
1 II
1 II
i .til .:.:
t ::if/:
i: .11
i n
i .. it. :
i : n
i n
i n
i ii
Lil...
i JI:L
r:..n:::;
L... JL.. .
r ir
L::JI::;
r. n
i n
i n. .
i "ir~::
1 ]L
UJLZJ
L,,.irn
OUTPUT
DATA/UMTS
0.00
0.00000000
Qi(cffl>
O fcfa)
Vol (inch
-67 Program Listing 1 ?«* s <*
STEP KEV EHTRV HEY CODE COMMENTS STEP KEY EMTflV K£V CODE COMMENTS
01
'to
SO
030
40
SO
flit e
Ar.l A
_
STO D
0
1
STO+ft.
an ft
21-K)
X
CHS
RCL 0
X
STO E
0
1
f
STO+9
RCL I
Rf.1 C
erq F
W| f
0
R/s
5TO 7
DSp 2
1
1
tlSTI
RCL (1)
1
0
oX>y
OTO A
era
f mi a
an 1
.
1
a
in Ml
S*y
eitri
il 2S 15
U 11
U 19
51
« lit
8)
OO
01
11 61 OH
|L no.
02
71
42
32 52
3>l 111
71
33 15
8?
00
01
71
11 61 04
Hi at
no.
33 07
?3 n?
of
IS H
34 21
8?
01
00
u ai
22 II
77 13
11 1C If
)3 fl }t,
>t
Al
m
Ji U
38 71
H 1}
81
0
fo-fc)
fo-fc) In RD
K
Kt
:-fc-(fo-fc)e kt
D [-"At
£ f-fc) at In R9
r f-fc) at
f
YES)
c
fi
f, In RC
Pit) In R7
0
(HO)
P(t) In KM
EP(l)
(YES)
(HO)
REQt
A ' ,, ' ,
EO. " £P(t) "
1 it ' Tc
U 54
fo P fc PC
O1??
OM
)70
OM
HO
100
tlO
1
n
_
«TO 7
f IBI R
orf 3
«Tn*fi
m*t "
2{~K)
««x
X
RC1 B
4.
CTA F
1*1 r
4-
^
i
1
ACt_ 1
_
X
1
fin i
.
an 7
X
x>v
-
fx<0
&TO 2
RCL 7
RC1 2
V
a.
CTO \
f LBL 2
WL 7
1CL 2
}(
F LBL 3
ICL 0
C
ICL \
!
;TQ 7
iri A
j,
net «;
fj
V
01
on
CI
33 n7
Jl 11 19
33 £1 FA
4(| 04
02
71
42
32 53
)4 l"l
' 71
74 1?
tl
73 ic
7b '3
At
02
81
nl
V> 01
51
71
01
JL A9
Cf
3& A?
71
35 52
11 71
11 02
U 07
*& n?
71
nl
77 ni
11 25 OZ
M07
02
71
11 25 03
14 00
71
111 01
B!
» "7
}t n£
SI
34 nC
03
11
At
t-EAt In
t
K
Kt
-Kt
e-kt
fc
>8
fj-fc(fo-fc)e-kt
f2 in RE
fl
fl * fj
2
At
f In 4t(-
1
(l-lt)
1
(1-1)
P(t)
f|ti-(i-i
p(t)
p(t)
Kt)-P(t)
l(t)-A
Kt)-R(t)-
Kt) In *,
4fW»_l
rtrtS
9 Q BHt-l) ' P(t)
65 » 57
' lit
3
D fo-fc |E
-------
I'rogrum MM lug 11 p«»» ' <*
STEP KEV ENTRV KEY CODE COMMENTS STEP KEY ENTRY KEV CODC COHHEHTS
120
130
IW
IM
in
_
RCL k
X
RCL 1
+
RCL. 3
t
X
RCL S
STO 5
fO$S
STft+G
fp$.
uti 7
srn &
RCI F
STO C
RCL <;
r-Tft l
f i (ti f
rt
CTQ t
F IRI n
ri y
f CL REG
fpl*
f Ct REG
hRTN
a LBL fa
RCL \
I
0
(1
V
foSs
RCL 0
X
fP2.
(in a
J
^
1
£
£
n
t
1
t
w
hHH
1
1
02
71
l 01
SI
J
tl 01
<; i;?
81
Il_
14 OS
1
nl
i as
11 42
11 fit OO
11 It?
it m
37 nA
^4 l<
J
5
11 ?
1 11
5 nl
C 1 1
on
35 01
31 3
S U
11 M
11 42
1
1
12 2
1
M
5 4;
i II
1 01
01
OO
1
3
00
71
'li
1 00
71
11 4?
tfa no
91
0)
IK
n&
nA
fti
1)1
09
TI
}
"
X B
b
0 1
2
At
2Tc»ot
K-t/2Tc+At
K|l'«
-tlNCNHt
KiVS
/ II I
1 1
cmi :
L_JI
CULL.
f IL..
1." .11"
IA..U .
r.'ii "
r" n ;
h/s I!
L.:.:ii-_
L ;ii ...
Ws'-'ii:"
r 11
i ii
B7S II
i n
i n
k/s II
i.. 11:1
i" .ii"":;
1 , It/DAT*
1. 11 .
1 .11 .
L IL1J
rui_,:i
nun
L .JCZ^
r it..
i. .~ir:;j
i -"ir "
n .
i n
i . ir:::
r n ::
i n "
i n
i n
i n -i
OUTPUt
DATMVMTS
OIWT [1.1
HFU BF|
OUNT (2.),
HEN ftti .
F,
(HINT (Vl
THEN REi
ftC2, KE,
:ou«T (ii.l.
THfH R,.
COUNT,
IHFM df ' c
-------
P««ol] ol
I ^1 HNV-SACTA BADBAjLA UHBAN HYDROGRAPH METHOD - CASH NO. 2
la. ... . . .«...
en
STIR
D
i
WtTMUCnOM^
1. IHTVI KHOUH VALUES
«) DtAIHACE USIN A«Ejk
h) TOTAL IHJEHVIOUS r^fellOII >ASIH
) TI^IZ OF CfflKEHTRATIQH
{> TltTTItT ^H^TIT^ATTIH «'₯
>) f IUL TM9II.TSATTMI ^A^K ^
h> HflfifFUTf' lHTTKTB4TlftK
IMVIS«AI, mwma KjinuK Tun
BEPOU OUTHIT BECIUS.)
to CMPUTI union m«1t f*f "» > TPfoos
10 coMpirrm ireu mnnooispn (pff^s m. HOIST
- MM unmu. MCUMEHTS tot M jgarraaai
r*an
wntr
DATAWMTt
(ACEES)
(DBCIMA
(HRTTXA1
t fHOUBS)
(w«msJ
o rrH/Hai
q flH/Uft)
f^yCHRfil
"
PH)
as
1
NKYS
:: ii i
it
ISTO || o
)bm 1! i
II
U H i
U II 3
km II 4
IBTO H c
Wo II 9
Wo II .
II
Ii II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
f II .
II
B II
II
II
II
II
II
II
11
II
OUTPUT
DATA/UNITS
:...:.
&C-CUBSJ ..
SR
lec i
iSP 9
H/S
TOT T
rrn f
21 It
Ol
"J m
35 W
-11
21 01
01
^
""fffr
01
35-55 00
M 00
-63 00
16 51
16 41
-63 09
51
-21
-21
1
',xlO in
L.OO
1.00 In ED
L.OO
1.
1
P,«10
%
PulO in
1. 00
Ul
Ul
2.00.3.00,4.00. «tc
2.00.3.00,4.00, «tc.
2. .3. ,<
l
PiilO , Pi
ft»10" In"
1.00
3.00,4.00,!
3.00,4.00.!
) 4 '
. , *tc.
*
xlO
Ul
.00. ic.
.00, etc*
t«10" ,1
M
m
0
DW
ioe
IKJ
1
L L
" ".Pi.r, pCr».r. F.,f..».
' r p
lifllPl^'llPl'.Pu S»P|7Pj» fl»Tit?2
jo
B B
90 SO
i ^aa?i»^a* *2S' pzs
T
-------
CT>
HP-97
STEP
010
IW
too
MO
KG
CARD NO, 2
KCY ENTRY
n
rtSIL n
STO >
1
X
STO +9
RCL 9
RCL E
X>Yt
GTOE
RCL B
STO t
0
BTH 4
1
«Tn T
*I HI 4
RCL i
EEX
3
X
INT
1
0
0
x-n?
GTO 6
STO 7
fiCR 1
P^T 1
EEX
3
K
FRC
EEX
3
X
TUT
1
0
n
Y-n?
GTQ f
flTO 7
CS« 1
B/*T 4
KEVCOOE
-*>
nn
m
71 na
" 17
-il
M
01
-35
35-55 09
36 09
36 15
16-34
22 15
36 12
-55
35 15
00
M 09
01
15 it
ii ni
36 45
-23
03
-35
16 34
01
00
. 00
-24
16-41
22. 06
15 07
31 m
7i A*
-n
85
-3S
16 4<
-23
03
-35
16 34
01
nn
nn
16-41
2? 06
35 07
n ai
Program 1 Jsllng 1
COMMENTS f.TE» KEY ENTRY
ut
I -«"kt
* »in RB
at (-0.01)
(f-fc)it
I(f-fc)St In >9
I(£-fc)4t.
F
P>C(f-fc)Qt
(f-fc)
fl
fl In RE
& In R9
P,, P., P,. etc.
RE01
' A ' It ! I
« ^I.Pl.P. ?».F|,P
3 it ' Tc
S3 M
Q f (fo-fc) |c fo
no
070
0
080
100
10
FRC
1
0
X
X-0"
GTO 6
STO 7
GSR 1
ISZI
GTO 4
STO 7
*I.BI.7
C,Ttl 7
*!.*!. 1
DSP 2
Ha. 5
.
A
ST+5
,
1
n
GTO C
1
0
STO 7
*1 MS
an. i
ran A
urn. n
2
RCL 3
i
1
_
KEY COOE
I6r«
01
00
-35
U-41
22 06
3^ 07
23 p|
16 26 46
22 04
21 M
35 07
21 O7
2? H7
2t At
-63 03
Kj-V
l«i_ ij o^
-62
Ol
no
22 11
-62
nl
00
21 J7
\f> 01
23 nft
35 li
5^
11 17
01
36 03
ft-j
*^i-
Page 19 of 2
COMMENTS
Pj, P.. F», "c.
P(t) in R7
P(t) in R5
P«
ISO
fM
-
1-
RCL
2
in *
ft T
X<0?
CTO
2
*1.H1
t
Bid ft
RCL
5
*
STO
RCL
7
&
+
RCL
A
2
X
_
RCL
4.
?
fCl 1
+
-Kt_
ypy
3
*
a
RCL
A
4.
STO
A
P*S
illi
rfr.
Ra.
STn
7
6
nri. n
CTY1 V
iCL.
3
ST-W
RTL
jpHjj
Bel
^
-
A
PBTlt
*
D
*
fi-
16 0^
36 07
-41
-45
tt M
-3.S
-3S
21 ni
«
-35
36 03
-24
35 07
& 06
-55
36 11
.-35
-45
36 04
02
-35
36 ni
-55
36 03
-41
-3S
' 36 11
-SS
35 ^]
lft-^1
35-55 Oft
1A-M
36 07
' 1S flft
36 19
11 1^
36 03
15-55 0^
f* 1T4
3*' ^
-1A
i-D
Rl<0 (1-It)
'(t)
RO-P(t)'I
«(l)-RO«U
P(t)
R(t)-P
't .1
KtMJ
i(t-l)
2
2Q(t-I
fc
2
2Tc
it
2Tc+fit
Ac
2tc«,t
k-At/(
Q(t-l)
Qt-q(t
qt la
IQ(t)
I(t)
Kt) t
f2 b«c
At
ut In.
Eat
Q(t)
»LAI
6 C
< I
t r
y?
B/S
BTH
*TttI^
RCL 3
ST+9
am «l
PBTX
o
PBTX
SPC
RTH
*T fil.ft
fiT+R
itn R
7
CHS
RCLC
RCLD
x
STH
n.T
HTBR
P*fL "
RTH
RCL 3
9
9
2
7
4
K
P?S
RCL 0
Mn ft
FLAQ9
0
1
t
)
KEY CODE
-ft2
no
QL_
16-14
94
?1 13
36 03
35-55 04
36 D9 .
-14
00
-14
16-11
24
71 ftfi
is-ss na
\6 oa
n?
-22
36 13
36 14
-45
-35
24
21 14
}ft~^\
If- 51
Io-iS
24
71 If. 11
36 03
-62
09
09
02
07
04
-35
16-51
36 OO
-35
u. An
COMMCNTB
C
ID R9
EAt
£At in RB
t
kt
-kt
o
c
°~Ec -kt
fit
£o
^
«T STATUS
FLAGS TWO LHSP
o toc§f
i a a"
3 Q 19
OEG ffl' FIX H(
oftAD o sci a
RAO a ENG a
-------
References:
(1) Stubchaer, J.M., "The Santa Barbara Urban Hydrograph Method", Pro-
ceedings, National Symposium on Hydrology and Sediment Control, ORES
Publication College of Engineering, University of Kentucky, Nov. 1975.
(2) Terstriep, M.L. and Stall, J.B., "The Illinois Urban Drainage Area
Simulator ILLUDAS" Illinois State Water Survey, Bulletin 58, State of
Illinois Department of Registration and Education, 1974.
(3) Golding, B.L., Personal communication with M.L. Terstriep, Jan. 18, 1978.
(4) Hannon, J.B., "Underground Disposal of Stormwater Runoff Design Guide-
lines Manual", Federal Highway Administration, U.S. Dept. of Transpor-
tation FHWA-TS-80-218, Washington, D.C., Feb. 1980.
(5) Miller, R.A., "The Hydraulically Effective Impervious Area of an Urban
Basin, Broward County, Florida", Proceedings, International Symposium
on Urban Storm Water Management, UKY BU 116, University of Kentucky,
July 1978.
(6) Alley, W.M. and Veenhuis, J.E., "Determination of Basin Characteristics
for an Urban Distributed Routing, Rainfall-Runoff Model", Proceedings,
SWMM Users Group Meeting, May 24-25, 1979, pp. 1-27.
(7) Ragan, R.M. and Durer, J.O., "Kinetic Wave Nomograph for Times of Con-
centration", Journal of Hydraulics Division, Proceedings ASCE, Vol. 98,
No. HY10, Oct. 1972.
(8) Simsek, S., Chien, J-S, and French, G.L., "Development of Linearized
Subhydrographs Urban Runoff Model", Proceedings, International Sym-
posium on Urban Storm Water Management, UKY BU 116, Univ. of Kentucky,
July 1978.
(9) Miller, R.A., "Characteristics of Four Urbanized Basins in South Florida",
Open File Report 79-694, U.S. Geological Survey, Tallahassee, Florida,
May 1979.
(10) Marsalek, J. , "Malvern Urban Test Catchment, Vol. 1, Ontario Ministry
of the Environment, Pollution Control Branch", Research Report No. 57,
Toronto, Ontario, 1977.
(11) Terstriep, M.L. and Stall, J.B., "Urban Runoff by the Road Research
Laboratory Method", Journal of Hydraulics Division, Proceedings ASCE,
Vol. 95, No. HY 6, Nov. 1969.
177
-------
DETENTION LAKE APPLICATION IN MASTER
DRAINAGE PLANNING
BY
ALAN T,K, FOK, P.ENG,
SR, HYDROTECHNICAL ENGINEER
PROCTOR & REDFERN LIMITED
TORONTO/ CANADA
S,H, TAN
ENGINEER-IN-TRAINING
PROCTOR & REDFERN LIMITED
TORONTO/ CANADA
SUMMARY
This paper illustrates that the Stormwater Lake concept can be applied to obtain
an optimum urban drainage system design. The proposed lake system converts the
existing sloughs into a system of aesthetic urban lakes interconnected by con-
trolled-release pipes. Computer Models, HYMO, STORM and SWMM were used for simu-
lating the various hydro!ogic conditions and testing the hydraulic responses within
the system. The results of modelling were further utilized to optimize the number,
size and layout of the proposed lake system with respect to economics, design,
hydraulic operation and development constraints.
1.0 INTRODUCTION
The application of Stormwater Management concepts has gained increasing popu-
larity in the design and analysis of urban drainage systems, and the City of
Edmonton has taken a leading role in the application of these new techniques
in Canada. A Stormwater Drainage Analysis for the entire north and northeast
Edmonton Development Area of which the present study area (Land Section 34
and its vicinity) is a part, was conducted by James F. MacLaren Ltd. (JFM) in
178
-------
1978 (Ref. 1) and subsequently a paper related to this report was published
by A. Fok et al in 1979 (Ref. 2), The lake optimization analysis presented in
this paper is founded on the results of these previous studies.
The drainage concept developed in the previous studies is based on the con-
version of existing sloughs and depressions into a system of aesthetic urban
lakes interconnected for controlled drainage. The plan also incorporates the
dual drainage concept of major and minor system flow. Storm sewers of conven-
tional design will drain to the lakes to effect runoff control.
As the present study area is proceeding toward development, a final Stormwater
Management Report was required in support of the detailed development plans for
City approval. Proctor Redfern Butler & Krebes Lijnited (PRB&K) was commis-
sioned to conduct this study at the request of the Section 34 owners group.
With the knowledge of the detailed land use and neighbourhood layout for the
area, this study refined and optimized the number, location, and sizing of
Stormwater Management facilities from a land development point-of-view as well
as satisfying the City's criteria for flood protection and operation.
This paper focuses on the optimizing of the proposed lake system within the
study area. However, the original Stormwater Management concepts, which apply
to the whole north and northeast Edmonton development area, are also discussed
in some detail to give a complete illustration of this Stormwater Management
application.
2.0 ASSESSMENT OF ALTERNATIVE STORMWATER MANAGEMENT CONCEPTS
2,1 Description of the North and Northeast Development Area
The total area of the development located in north and northeast Edmonton
(Figure 1) is about 3050 ha. A north-south ridge of land divides the
area into an eastern and a western watershed. The 1350 ha eastern water-
shed drains directly to the North Saskatchewan River. The 1700 ha
western watershed (the 'Lake District1) is dotted with sloughs and drains
southward into the existing Kennedale District Sewer System which ser-
vices a 3200 ha residential area. Stormwater runoff is now temporarily
179
-------
flldgt
Boundary ol Sludy »re«
Boundary of Drainage Arta
K««n*d«l» Or«m»9«
Onwal OMelKxi 61 Nilwal
low
KENKEMlE TRUNK
FIG. 1 NATURAL DRAINAGE CONDITIONS
retained in natural sloughs and depressions throughout the watershed.
However, spring snowmelt and heavy summer storm runoff can cause very
significant flow to the downstream Kennedale System and flooding potential
does exist under present conditions.
The existing watersheds (both east and west) are agricultural or vacant
land. The proposed development will be predominantly residential with
small sectors of light industrial, commercial and Institutional lands.
2.2 Description of Alternative Measures
Since the eastern watershed has natural drainage access to the North
Saskatchewan River, drainage alternatives for this area included only
variations in the location of the main trunk and a comparison between
open channel and sewer conveyance with and without in-system detention
storage facilities.
However, natural drainage in the western watershed is internally complex
due to the sloughs in the central area and externally constrainted by the
capacity limits of the southern (Kennedale) receiving system. Three
basic Stormwater Management alternatives for controlling runoff from this
watershed were distilled from a number of options for more detailed con-
sideration.
180
-------
Alternative 1 - Gravity Flow Eastward Through a Tunnel
This alternative requires the diversion of runoff from its natural
southerly direction eastward to the new drainage system east of the
ridge. Storage facilities are required locally to limit the size of
the tunnel.
Alternative 2 - Pumping Flows Eastward Over the Ridge
This alternative is essentially the same as Alternative 1 except
that pumping facilities are used to lift the flows over the ridge
and eliminate the need for a deep tunnel. It should be noted that
this alternative originated from the fact that a pumping station is
required anyway to lift sanitary flows over the ridge.
Alternative 3 - Controlled Release to the Kennedale Storm Sewer System
This alternative maintains the normal flow direction to the south.
The only possible way of maintaining this natural flow direction with-
out severe impact to the downstream area is to strictly control the
amount of flow entering the Kennedale System. Detention lakes are
therefore required to store the excess runoff from the area. The
concept of Alternative 3 is illustrated in Figure 2.
2.3 Comparison of Alternatives
In order to facilitate the comparisons of the alternatives, the merits
and demerits of each alternative were summarized in Table 1.
It was found that Alternative 3 results in a considerable cost reduction
over the other alternatives due to the elimination of the tunnel and
oversizing of the eastern trunk sewers. Furthermore, Alternative 3
permits the expenditures for drainage to be staged in accordance with
the rate of development. New storage ponds can be constructed as re-
quired to control flows from each new development. Staging is also
possible for the investments in the sanitary sewer system. In addition,
storage ponds have environmental advantages over a system that would pipe
runoff directly into the river. For example, the ponds will improve the
urban aesthetics, enhance groundwater recharge, improve water quality
181
-------
-- -Boundary of Study Am
Saundary ol Drainage Ana
- Existing Kennadala Qrammgm
SysMffl-
-'-" N«w Pipe» ConnecNna 8un«r
Laka (o Existing Konnedele
Svsnm
HB Present Study Am
Swim ol Ponds
existing Trunk 2«w«r
New Trunk Sawar
Kanmdolt Trunk
FIG. 2 SCHEMATIC OF ALTERNATIVE 3 - CONTROLLED RELEASE
I) Conwnttoiul
Relief »pend«tur«
for th* tuMWl.
5) Steolnj
) Pwailne Relief
with Detention
1) Plant!*) capacity can be
staoed.
2) Sanitary and (torn sewn
1n sane allqnaent.
3) Slightly l«sl enpaffilv*
than Alternative 1.
1! Hever** natural rto>
direction.
2) W onerlMd flo* outlet
for Mjor lyitw.
operational
t.
ratxil
1) Controlled
Release vlth
Detention
1) Follow natural drainage
direction.
2) Least expensive.
31 Can be staged.
4) Outlet or overland flow
available.
I) Additional protection
measure to downstream
development required.
182
-------
and provide reservoirs for possible use in irrigating parklands adjacent
to the ponds during dry periods.
Therefore, Alternative 3 was selected as the most acceptable drainage
scheme.
2.4 Storage of the Proposed Lake System (Total System Requirement)
One of the design criteria for the proposed lake system was to provide
high level of flood protection and runoff control within the study area
and in the existing downstream development in the City. As such, the
lake system must provide adequate storage to accommodate runoff volumes
from critical hydrological events. The system was tested for several
conditions as discussed below.
The 100 Year - 6 Hour Duration Storm is the design level of flood
protection required by the City. As seen from Figure 3A, uncontrolled
drainage after development would result in a significant increase In
the peak rate and volume of runoff. However, the proposed lake
storage volume of 10.3 cms-day will reduce outflow south to the
Kennedale Sewer System to a nominal rate of 1-42 cms and no overland
flow will occur. By comparison, the present conditions will produce
large overland flow rates in the order of 42.5 cms for this event.
« The July 14-15, 1937 Storm is the largest storm on record, when 163 mm
rainfall fell in 48 hours. The proposed lake system (10.3 cms-day)
was analysed for this event. It was found that there would not have
been a significant reduction in the peak discharges for this event but
the duration of the overland flow would have been substantially re-
duced (Figure 36).
Furthermore, the return frequency of this event is estimated to be
about 1/250 years. As economic considerations preclude designing for
such an extremely rare occurrence, and since the potential flooding
for the proposed conditions would improve somewhat over the existing
conditions, this storm was not considered for the determination of
the storage requirement.
183
-------
1 d» - 0-028 cmi
1 ac-lt = 0-014 cm»-d*y
Poll-On. With
360 Ac. - FL Pond
FIG. 3a OVERROW HYDROGRAPHS FOR
100-YtAR 6-HOUR DESIGN STORM
FIG. 3b OVERLAND FLOW HYDROCRAPH OF
JULY 14,1937 STORM
i
1
I '
AJ\
I No
-------
The 17 Years of Rainfall Data from 1961 - 1977 contained one of the
largest rainfall months in record (June, 1965). These data were used
to test the system for consecutive small events and snowmen/rainfall
events. It was found that these events are not critical as no over-
flows were predicted for this period from the proposed lake system
(10.3 cms-day design storage). This represents a definite improve-
ment compared to existing condi.ti.ons for which five significant over-
land flow events were predicted for this data (Figure 36).
From these analyses it tan be concluded that the 10.3 cms-day lake storage
system is adequate for the whole 1700 ha Lake District. The next stage
of the study was directed to a better definition of the system with
respect to its components, operation for different hydrologlc conditions
and integration with the downstream system.
3.0 OPTIMIZATION OF THE LAKE SYSTEM
3.1 Development Plan of the Study Area
The present study area (Land Section 34 and its vicinity) consists of a
307 ha drainage area. It is located at the southwest corner of the Lake
District (Figure 1). The general flow pattern and the relevant land use
data which may affect the drainage design are summarized in Figures 4 and
5.
The original 1978 study proposed a total of seven lakes for this area.
Because of the newly adopted maintenance and operation criteria, the
refinement of development plans and cost-benefit considerations, the
number, size and location of the lakes in this area were to be optimized
to fulfill the design criteria as discussed below.
3.2 Design Criteria
In order to optimize the use of the lake system for flood protection and
community benefits, several design criteria were adopted:
1. The lakes should be easily accessible to the community.
185
-------
DATUM! +000-90 m
FK3. 4 PROPOSE ORAMAQE
SYSTEM
>4<<» 97-01 M-07
FK3. t PROPOSED
DEVELOPMENT PLAN
186
-------
2. The lakes are to be located on the existing sloughs and depressions.
3. The size of the lakes should be proportional to the local drainage
area.
4. A balance design is required for the lake system. The outflow from
each lake should be proportional to the drainage area and remain
constant during th6 storm. 'This eliminates the need for oversizing
the downstream lakes and facilitates staging and cost sharing.
5. For storms larger than 1/100 year storm, the sharing flooded area
should be confined to the vicinity of the lakes.
6. The number of lakes should be minimized to provide easier mainten-
ance.
7. The area of the lake should not be less than 1.6 ha at permanent
water level. (P.W.L.)
3.3 Results of Analysis for Various Lake Systems
Five Lake System (Equal Lake Size)
One of the alternative lake systems for the study area is five lakes
(Figure 4). Each lake would be 1.6 ha in size at P.W.L. for a total
area of 8 ha. Results of analysis indicated that the equal sizes of
lakes for tributary areas of different sizes would result in unequal
maximum water depths and variable flow rates in the interconnected
pipes during storms. Thus a balanced design of the lake system for
this alternative cannot be achieved and, this alternative was not
considered further.
* Five Lake System (Balanced Design)
In order to achieve a balanced design, (i.e. a uniform average lake
depth) the size of each lake was made approximately proportional to
its contributing drainage area (Figure 4) with a ta-mget maximum water
depth of 2 m. Preliminary analysis indicated that a total lake sur-
face area of 5.2 ha would be adequate. The resulting hydraulic
response of the system is summarized below:
187
-------
Maximum Depth of Water
Lake No. Size (ha) Runoff Volume (cms-day) Above P.H.L.Cm)
3
5
6
7
8
0.9
1.0
0.33
1.15
1.74
0.30
0.33
0.11
0.38
0.55
2.04
2.04
1.95
2.13
2.10
Total 5.20 1.67
These results show that the 5.2 ha, Five Lake Alternative can pro-
vide adequate 1/100 year flood protection in a balanced system.
However, 4 of the 5 lakes are smaller than the 1.6 ha requirement and
Lake 6 is only 0.33 ha0 Hence, the Five Lake System was not considered
further.
Three Lake System (Balanced Design)
The location of each lake and its drainage sub-watershed for the Three
Lake System are shown in Figure 4. Lake 5 will have a 2.0 ha surface
area while Lakes 7 and 8 have 1.6 ha each. The surface area of Three
Lakes are roughly proportional to their tributary area and, as shown in
Table 2, maximum water depths are quite close in all lakes for each of
the design storms.
TABLE 2
IMPERVIOUSNESS VALUES FOR VARIOUS LAND USES
Roof Area Density % of rmpar-
Types of Land Uses m /unit (units/ ha.) viousness
Rl - Single Family Residential 116 3.24 23
R2 - Semi-detach Residential 116 4,36 34
R2A- Townhouse 93 6.07 34
S3 - 3 Storey Apartment 186 10,12 35
Innovative Housing 93 4.05 23
Commercial 35
School 33
Park 5
Roads,Lakes 2.0G
188
-------
The flow hydrograph and water level variation for the lakes are plotted
for the 100 year - 6 hour and July 14-15, 1937 Storms in Figures 6 and
7 respectively. Since the water levels in all lakes rise at the same
rate during the design storms, a balanced hydraulic response in the
Three Lake System has been achieved.
The basin hydraulic profile in the Three Lake System is plotted in
Figure 8. This figure indicates clearly that the system can handle
the 100 year storm adequately. With the available freeboard, no flooding
is expected to occur even for a storm such as the July 14-15, 1937
storm.
During all of the design storm events under investigation it was found
that the lake connection system was conveying a flow rate approximately
equal to the corresponding pipe capacity and equal to the required con-
trol rates. This further illustrates that a smooth hydraulic response
in the proposed lake system can be achieved.
The Three Lake System therefore has the following benefits and hence
was recommended for the final design:
A) Required flow control and flood protection is achieved with a
smaller lake area.
B) Each lake meets the minimum size requirement of the City.
C) A balanced design is achieved in which lake area and outflow
rates are based on the respective tributary areas.
.4.0 METHODS OF ANALYSIS
4.1 Hydrology
The HYMO Model which is based in the SCS Method, was used in the 1978
Study for simulating single events for design storms.
The Computer Program STORM was also used to simulate a continuous rain-
fall record (1960-1967) in which closely spaced consecutive storms
occurred.
189
-------
OUTFLOW
FIG 6a HYOROGRAPH FOR 100-YEAR 6-HOUR STORM
190
-------
663
£g
660
ML
676
662
«»
660
677
662
660
677
LAKE &
(2-04)
I
I I
" MI-69
LAKE 7
7»B1
UOZ)
r MI M
LAKE B
JZZO)
01 2345676
TIME (hour)
FIG. 6b LAKE LEVEL VARIATION FOR 100-YEAR 6-HOUR STORM
191
-------
OUTFLOW
(emi)
0-31
*
o
i o
LAKE 7
0-OT
LAKE 8
. 0-49
JULY 14
JULY IS
JULY 16
FIG. 7o HYDROGRAPH FOR JULY 14-B, 1937 STORM
192
-------
683
660
PgL
678
682-44
LAKE 5
r
68143 -Mm.WnUr El.
(2-63) Wattf
682
680
§
> .
ui -.
ui 677
681-89
LAKE 7
682 _.
677
'681-83
JULY 14
LAKE 8
r
67*70
'(2>68)
JULY M
FIG. 7h LAKE LEVEL VARIATION FOR JULY 14-15 STORM
193
-------
ELEV. (m)
684 _
LAKE
LAKE 8
68183
682 _
680 _
678 _
676 _
674 _
* EXISTING GRADE 76Hi STREET
, 661 83
LAKE 5
750 mm STORM SEWER
ON TG STREET
67650
,Gr. El«v.
JUL.M/te,l937
100 YEAR
P.W.L
BOTTOM
FIG. 8 THREE LAKE SYSTEM HYDRAULIC PROFILE
-------
4.2
The SWMM Sub-Model Runoff was used in the optimization design. This
model computes overland flow hydrographs for a given storm pattern for
each subcatchment in the drainage area. It allows for the variation in
infiltration before and during runoff occurrence.
The study area was subdivided into numerous tributary subcatchments
ranging from 0.5 to 7 ha in size. The characteristics of each sub-
catchment, such as overland flow length and slope were directly measured
and input into the runoff model. The imperviousness values for various
residential land uses were determined from the size and density of
housing conditions as indicated in Table 3. The distribution of various
imperviousness conditions on the area can be found in Figure 2. The
overall a erage percentage of imperviousness in the whole drainage area
was estimated to be 42%.
Comparison of SWMM and HYMO
Since the two hydro!ogical models had been used at various stages in this
and previous studies, a comparison between SWMM and HYMO was undertaken
to determine the best model for refining the analysis in this study. The
results of the comparison are presented in Table 4 for several design
storm events.
TABLE 3
COMPARISON OF COMPUTED RUNOFF FROM SWMM AND HYMO MODELS
STORM SWMH
frequency
lyeap
5
5
100
Duration
2
6
6
July 14-15, IrtTt
48 hi.
max. 6 ttr.
UK. 24 IT-
July 15, 1917 only
Rain
Volume
Inn)
32.2
44.5
90.2
157.5
65.3
101.6
Peak flow
Itt. 3
7.6
25.5
7-3
7.3
7.1
KUDO f I
Volume
(cna-day)
'0.4S
O.SO
1.70
2.60
1.07
1.57
i.oi ~
Runoff Vol.
Eilfi Vol.
40.0
38.5
53.3
45.0
47.0
43.6
6.0
7.4
23.5
10.7
10.7
19.7
"" iTi
Volune
0,28
0.54
1.80
4.20
1.5S
1.43
Rain Vol.
24.0
34.3
56.3
70.1
«3.4
61.0
W.O
* Imperviousjiess used in SWMH IB t_ 42*
* Complex number used In IIYMO IB i8S
195
-------
TABLE
BASIC HYDRAULICS WITH VARIOUS STORM EVENTS
CTl
Storm Runoff Volume to Lake Max.
Frequency Duration
(Year) (Hours)
A. Infrequent Storms
100 6
July M-lb.193?
B. frequent Stores
1 2
2 2
5 2
10 2
25 2
5 6
S
.656
.991
.082
.096
.174
.249
.451
.231
7 ]
(CBS/
.495
.771
.062
.079
.134
.194
.351
.197
8
day)
.554
.846
.069
.085
.151
.213
.387
.179
Total
1.706
2.608
.213
.262
.459
.656
1.191
.607 ,
5
.636
.836
.072
.089
.164
.239
" .443
.213
Lake Storaq
7 r~a
(cos/day)
.479
.640
.056
.072
.128
.187
.144
.165
.538
.689
.062
.079 .
.144
.207
.380
.182
; Used Max. Hater Depth Above Lake PML Max. Mater Surface Area
Total
1.653
2.165
.190
.239
.436
.633
1.168
.560
5
2.05
2.64
0.30
0.38
(1.68 '
0.90
1.55
.81
T
(«)
2.03
2.56
0.29
0.35
0.66
0.88
1.51
.80
8
2.20
2.68
0.38
0.47
0.80
1.03
1.70
.91
5
2.93
3.23
2. 15
2,17
2.33
2.44
2.73
2.38
(ha
2.52
2.75
1.72
1.76
1.08
4.98
2.2U
1.96
)
2.59
2.81
1.77
1.82
1.96
2.07
2.16
2.01
a.t Lake
IBliT
8.04
8.79
5.64
5.75
6.17
6.49
?.3I
6.17
-------
The analysis showed that both. SWMM and HYMO provided very close results
for the 100 year and 5 year 6 hour duration storms which were adopted
for lake sizing.
However, the SWMM runoff volumes range from 38.5% to 53.3% of the rain-
fall volume for various storm events while the HYMO volumes vary from
24 to 99%. It appears that SWMM provides a more consistent runoff volume
considering the 42% imperviousness of the drainage area.
On July 5, 1937, (2nd day of the 48 hour storm), HYMO generated a runoff
volume of almost 100% of the rainfall. This means that no infiltration
was considered by HYMO on this day« The rainfall record (upper part of
Figure 3B) shows that there was no significant rainfall for almost 9 hours
after the rainfall of the first day and hence recovery of soil infiltra-
tion should be considered. This finding indicates that HYMO is not very
suitable for long duration flow simulation purposes.
As a result of these results, it was concluded that the SWMM Runoff
Model was more suitable for the present system analysis. This conclusion
appears to be consistent with the HYMO Model's origin as a rural, single
event simulator.
However, it cannot be recommended, at this stage to replace HYMO by SWMM
runoff for all other urban drainage analyses. Our experience indicates
that after revising several parameters, (such as TP, K, CN), HYMO can
also provide satisfactory results.
4.3 Hydraulics
Geometry of Lakes
The geometry of the proposed detention lakes have been designed
according to a number of engineering and planning factors. These
include the shape of the existing sloughs and depressions, space
limitations due to street and housing planning, and the location
of the inlet and outlet of storm sewers.
Kidney shaped lakes (Figure 9) are proposed with side slopes of 7:1
197
-------
E 3
STAGE M
STORAGE VOLUME
05 I 0
STORAGE VOLUME (cmt-do»)
FIG. 9 STORAGE CHARACTERISTICS OF TYPICAL LAKES
and surface areas of 1.6 and 200 HA. at permanent water levels (P.W.L.).
The storage characteristics of typical lakes are plotted on Figure 9.
The shape of lakes may be altered in the final design provided there is
no significant change in their hydraulic characteristics as shown on
Figure 9.
Hydraulic Computer Model
The SWMM sub-program, extended transport (EXTRAN), was used to analyze
the design and operation of the detention lake system. This model was
used to route hydrographs through the drainage system (including lakes)
to the point of discharge. The routing technique used is very sophis-
ticated and it gives reliable results. The lakes were modelled as very
large trapezoidal open channels connected by small closed conduits.
The trapezoidal "Lake" elements in the model were 60 m wide with side
slopes of 8.5:lo Within the hydraulic conditions under investigation,
(1.6 and 2.0 HA. surface areas at P.W.L. and less than 3 m water depth
variation) no significant variation in the hydraulic characteristics
198
-------
of the lakes was found. During the hydraulic simulation, no instability
was observed and the accuracy is believed to be within 5%.
A special modelling technique was applied for simulating a hydro brake
concept at the most downstream lake (Lake 8). The flow at the outlet
of the lake system was controlled to a constant rate of 0.45 cms during
all storm events simulated.
REFERENCES
1. James F. MacLaren Ltd. Report "Stormwater Management for Development in
North and Northeast Edmonton", May 1978.
2. A. Fok et al, "Application of Stormwater Management Concepts for a Master
Drainage Planning Study", a paper submitted to Sixth International Sym-
posium on Urban Storm Runoff, July 1979, Kentucky, U.S.A.
199
-------
ALTERNATIVE URBAN FLOOD RELIEF MEASURES
A CASE STUDY
CITY OF REGINA, SASKATCHEWAN, CANADA
A. M. CANDARAS, P. ENG.
Hydrotechnical Projects Engineer
Paul Theil Associates Limited
Bramalea, Ontario Canada
1 - INTRODUCTION
The state of the art of computer simulation in hydrology, specifically
in urban drainage,has advanced significantly in recent years. This
has been a result of the increase in the useage of such models in
studies and design, along with the advancement in the sophistication
and the efficiency of the models.
However in many cases these advanced tools and methodologies are
being used to analyze and design conventional or traditional solutions
to urban drainage problems. Probably the most common solution to
solve combined sewer overflows and flooding is either large scale
sewer separation in older urban areas, or the installation of
larger conduits in areas where a separate system presently exists.
A recent report by the Comptroller General of the United States
(December 28, 1979) stated that "large construction projects to
200
-------
correct combined sewer overflows are too costly". The report stated
that estimates by the Environmental Protection Agency to curb
pollution caused by sewer overflows are $26 bill ion,and $62 billion
to prevent flooding. It noted that progress in stemming pollution
and flooding has been slow, with neither Federal Government nor
local communities capable of supplying the funds for these large
construction projects. The report goes on further to recommend
that new low-cost techniques should be investigated and encouraged
prior to considering costly solutions.
2 - CASE STUDY
A study was recently completed by our firm in which a series of
alternatives to correct flooding problems in an existing urban area
were developed. These ranged from conventional solutions, the
installation of larger piping, to alternative storm water management
measures.
2.1 Background
The City of Regina, Saskatchewan Canada, has been plagued over the
years by frequent flooding of basements and subways within the
Seventh Avenue drainage area. Such events have caused inconvenience
to both home owners and vehicular travel, in addition to the cost
associated with the flooding damage. Other resulting problems which
cannot be as readily assessed are the health dangers when sanitary
sewage backs up into basements, as well as the personal grief
experienced by homeowners and the environmental damages to the
receiving stream during combined sewer overflows at the
pumping station. Furthermore,if any re-development is to proceed
in the downtown core, which is located partially within the study
area, improvements to the existing system will be required.
201
-------
The City of Regina, recognizing these existing problems and the
need to correct them, commissioned a study in the spring of 1978 to
evaluate the present design criteria utilized for storm sewer
design. In summary, the findings of the study concluded that:
(1) Present storm sewers have a one year storm
frequency capacity.
(2) Further engineering studies would be required to
determine the cause of, and the most effective
remedies for the inadequate storm drainage system.
As a result of the latter conclusion, the City of Regina
requested proposals and our firm was engaged to carry out the
required study.
2.1.1 Terms of Reference
The City of Regina did not provide detailed terms of reference
for the study, but to allow maximum freedom of approach
established the following guidelines:
(1) To investigate the cause and the magnitude of the
deficiencies in the existing storm drainage system.
(2) To provide an economical upgrading plan for the system.
(3) To provide preliminary cost estimates to implement
the plan.
2.1.2 Objectives
The objectives of this study were established after review of the
terms of reference and discussions with the City of Regina1s
Departments of Public Works and Engineering. They were as follows:
202
-------
(1) To determine the cause and the magnitude of the
deficiencies of the existing storm and sanitary
sewer network serving the Seventh Avenue drainage
area.
(2) Provide an economical upgrading plan for the system,
achieving at least a five year protection against
basement flooding and reducing the frequency of
street and subway flcodings.
(3) Provide an upgrading plan for the system that
recognizes and accounts for continued development
in the downtown area.
2.2 Description of Study Area
The study area known as the Seventh Avenue drainage area encompasses
approximately 1,040 hectares, as shown in Figure 1. Drainage for
the site is through a separate sewer system, namely storm and sanitary
sewers. The trunk sewers are primarily located along Seventh Avenue
and drain in a westerly direction towards the Wascana Creek. The
storm trunk outlets directly to the Wascana Creek, while the sanitary
trunk sewer is pumped below the creek and is lifted to the sewage
treatment plan. Sewers along minor roadways outlet to the trunk sewers,
and generally have a perpendicular orientation to the trunks. Close
examination of the sewer system in the course of the study indicated
that the system is not truly separate as the name implies. Numerous
interconnections between the storm and sanitary sewers occur throughout
the study area. Common manholes exist throughout the drainage area,
which result in direct storm water inflow during storm conditions
to the sanitary sewer, and ultimately in basement floodings. Further,
approximately 50% to 67% of the roof areas within the residential
and industrial areas respectively are discharged directly to a
sewer which, in the majority of the study area, is the sanitary
sewer, thereby directing additional storm runoff to the sanitary
203
-------
ro
o
\
'-r\
i _ "-^-ss.1.. -\-V? 'l&l^^sbs
. .;- <,.;" s ',"-!rfr.% *\- «- -SJS.VfflKrf
[h^,^aM"\.^^;;
CITY OF REGINA
SEVENTH AVENUE
DRAINAGE AREA
FIGURE I.
ll ilium DRAINAGE BOUNDARY
-------
system. Also the reported frequency of basement and surface flooding
indicates that the sewer network is often overloaded, resulting in
surcharged conditions.
2.3 Computer Simulation
The hydraulic complexities of the existing sewer system dictated
that a sophisticated urban hydrologic-hydraulic computer simulation
program be utilized in this study. Traditional design methods are
incapable of dealing with such complex networks, nor can they be
utilized to evaluate advanced storm water management upgrading
schemes. Thus the Hydrograph Volume Method (HVM) program was selected
for the analysis.
The basic modelling procedure was to segmentize the trunk sewer system
in detail, manhole to manhole, including the common manhole inter-
connections. The relatively simple tree-type configuration of the
local sewers consisting of long straight sections permitted a more
coarse segmentation, while still maintaining the hydraulic character-
istics of the sewers.
Initially the existing sewer system was evaluated through computer
simulation in order to assess its hydraulic performance under actual
and design storm events. The results indicated that the system is
heavily overloaded for storms of a 2-year return frequency with severe
overloading for the 5-year return frequency, with the conveyance
capacity of the system being at a 1-year level or less.
2.4 Evaluation of the Existing System
Considerable storm water overflows were indicated to be occurring
to the sanitary trunk at common manholes. In addition, roof water
contributions were quickly overloading the existing sanitary sewer
along local streets. The present conveyance capacity of the sanitary
trunk is only 2.5 times the design dry weather flow, or alternatively
205
-------
the peak dry weather flow, with three trunk segments requiring some
minor upgrading.
Similarly, the storm sewer system is incapable of conveying the storm
runoff flows, resulting in roadway flooding, and overflows to the
sanitary sewer. In portions of the study area where weeping tiles
are discharged to the storm sewers, surcharging in the storm sewer
would lead to basement heaving, cracking and flooding. Also a
restriction in the storm water flow occurs at the outlet to the
Wascana Creek, both by the limited capacity of the sewer and turbulence
at the junction chamber
2.5 Minor Remedial Works
Having determined that storm water overflows at common manholes and
the restricted outlet to the Wascana Creek are prime contributors to
the flooding problems, the effectiveness of incorporating remedial
work to rectify these problems were evaluated. This was achieved
through the computer simulation of these alternatives.
The results indicated that a significant reduction in the trunk
sanitary flows will result, and backwater levels in the storm sewer
immediately upstream from the Wascana Creek outlet would be reduced.
These remedial works, however, will not eliminate basement floodings
for the 2 and 5-year storms, but will reduce the severity of the
flooding. The effectiveness of these works will increase in
significance for the lesser storms. Also the wet weather flows
reaching the treatment plant will be reduced, thereby lessening
the probability of overflows to the Wascana Creek and the volume
of waste water reaching the treatment plant.
Since these measures were not sufficient to upgrade the sewer
network to an adequate level of service,additional alternatives
were evaluated.
206
-------
2.6 Generation of Alternatives
A series of individual alternative upgrading schemes were generated
for the storm and sanitary system. These were under three general
groupings:
(1) Conventional upgrading; consisting of the replacement
or addition of piping to convey the peak flows for
either the 2-year or 5-year return frequency storms.
(2) Total storage; the provision of storage within the
sewer system, such that flows are limited to the
existing sewer's capacity. The storage would be
provided in off-line and in-line subsurface tanks.
(3) Combination system; conventional upgrading in the trunk
sewers, with storage to prevent flooding on streets
served by local sewers.
All the alternatives included the elimination of common manholes and
the increased storm outlet capacity to the Wascana Creek.
2.6.1 Roof Downspout Disconnection Modifications
Roof downspouts connected to either the storm or sanitary sewers
contribute significantly to the peak flows conveyed by these
sewers. Although volumetrically their total contributary
area may be relative minor to the total storm runoff, their
instantaneous high peaks may cause floodings, or structural
damages. The effectiveness of downspout disconnection was
evaluated, through the resimulation of the above alternatives
with all downspouts assumed to be discharged to the surface.
207
-------
2.7 Evaluation of Alternatives
Through the course of the study, nine individual alternatives were
developed, six being for the storm system and three for the sanitary
system. From this series of individual alternatives, twenty-seven
combination upgrading schemes were developed as shown in Tables 1
and 2.
The findings of the study indicated that conventional solutions,
namely those requiring the replacement and the addition of sewers,
were the most costly and would result in greatest inconvenience and
disturbance during the construction phase. Approximately 45% and
61% of the existing sewer system would require upgrading for the
2 and 5-year storms respectively. Further, the resulting peak flows
outletting to both the sewage pumping station and the Wascana Creek
would be greater than those presently experienced. This will result
in the degradation of the Wascana Creek, through erosion and overflows
at the pumping station.
Alternatives incorporating storage proved to be the most cost-
effective. The total storage alternative requires no major upgrading
to the existing sewer network, other than some minor piping to
convey flows to the storage facilities. These storage facilities
for the most part would be subsurface storage tanks. However,
where surface ponding such as roof tops, parking lots, or open
spaces is feasible, they will be utilized to reduce subsurface
storage requirements. In addition to being the least expensive of
the alternatives, other benefits derived are a reduction of peak
flow to the Wascana Creek and the sewage pumping station,
thereby possibly improving conditions in the Wascana Creek.
The combination system did not result in any significant benefits,
with the high costs associated with upgrading of the trunks negating
any benefits derived from the storage.
208
-------
TABLE 1
ALTERNATIVE UPGRADING SCHEMES
STORM SYSTEM
SANITARY SYSTEM
Conventional
5 Year $41
2 Year $21
Upgradi
.1 M (1)
.2 M (2)
ng
Conventional Upgrading
(all downspouts to storm sewers)
5 Year
2 Year
$41.9 M (3)
$21.8 M (4)
Conventional Upgrading
5 Year $14.4 M (a)
2 Year $12.2 M (b)
Conventional Upgrading
(trunk sewers only)
Storage for Laterals
5 Year $28.2 M (5)
2 Year $15.0 M (6)
Downspout Disconnection
5 Year $0.5 M (c)
2 Year $0.5 M (d)
Conventional Upgrading
(trunk sewers only)
Storage for Laterals
(all downspouts to storm sewers)
5 Year $30.0 M (7)
2 Year $16.7 M (8)
Total Storage
5 Year $19.8 M (9)
2 Year $6.2 M (10)
Total Storage
Total Storage roljpr<;1
(all downspouts to storm sewers;
5 Year
2 Year
$22.5 M (11)
$9.0 M (12)
Total Storage
5 Year $8.1 M (e)
2 Year $5.0 M (f)
All costs in millions of dollars
209
-------
Downspout disconnection is by far the most cost-effective solution
for the upgrading of the existing sanitary sewer. Elimination of
this source of storm water inflow will limit flows in the sanitary
sewer to the peak dry weather flow, which the system has sufficient
capacity to convey.
The recommended solution called for the elimination of all common
manholes and other interconnections, with storm outlet improvements
to the Wascana Creek. Upgrading measures would require that all
downspouts be disconnected with storage being provided within the
storm sewer system.
TABLE 2
COST ESTIMATES FOR VARIOUS UPGRADING SCHEMES
5 Year
5 Year
la
le
3c
5a
5e
7c
9a
9c
lie
Storm *
Sanitary
$55. 5M
$49. 2M
$42. 4M
$42. 6M
$36. 3M
$35. OM
$34. 2M
$27. 9M
$23. OM
2 Year
2 Year
2b
2f
4d
6b
6f
8d
lOb
lOf
12d
Storm +
Sanitary
$33. 4M
$26. 2M
$22. 3M
$27. 2M
$20. OM
$17. 2M
$18. 4M
$11. 2M
$ 9.5M
2 Year
5 Year
2a
2e
4c
6a
6e
8c
lOa
lOe
12c
Storm +
Sanitary
$35. 6M
$29. 3M
$22. 3M
$29. 4M
$23. 1M
$17.2M
$2Q. 6M
$14. 3M
$ 9.5M
210
-------
3 - CONCLUSION
The case study presented here clearly indicates that solutions to
combined sewer overflows and basement floodings do not necessarily
require large costly construction projects. Rather, with some
innovativeness, low cost techniques can provide a much more
cost-effective solution.
211
-------
A LONG-TERM DATA BASE FOR THE
INVESTIGATION OF URBAN RUNOFF POLLUTION
BY
W. F. GEIGER
TECHNICAL UNIVERSITY OF MUNICH
INTRODUCTION
Realistic conclusions concerning receiving water
pollution can not be derived from rainfall-runoff measurements
or simulations of singular events. This is especially true of
intermittent storm runoff and the resulting storm or combined
sewer overflows [l] . The numerous highly interdependent
variables which influence the runoff process, and the random
nature of storm events necessitate the investigation of long-
term records. Further, the rapid variation of precipitation
and runoff necessitates close observation of the runoff process.
This requires continuous and detailed monitoring of the rain-
fall-runoff-overflow process over a time period of several years.
Precipitation, runoff and overflow measurements are usually re-
ported in literature for individual storm events.
This paper describes an instrumentation and monitoring
program collecting rainfall, runoff, outfall and overflow data
continuously in a sequence of closely spaced intervals in two
test catchments: one a combined sewer system of 1340 acres
(542 ha), the other a small separate system of 57 acres (23 ha).
Particular emphasis is placed on the difficulties in establishing
and analyzing such an extensive data base. The project, funded
by the German Research Society, commenced in 1974 [2J . Continuous
monitoring started in 1976 and is ongoing.
212
-------
STUDY OBJECTIVES
The prime objective of this study is to establish a
continuous and detailed data base of rainfall and runoff, both
quantity and quality. On such a basis the following tasks are
planned:
- determination of seasonal and annual totals for urban runoff
quantity and quality and respective combined sewer overflow
and separate system outfall figures;
- definition of antecedent moisture conditions and pollution
potential in the study areag;
- some indication of the influence of areal distribution of
rainfall on the runoff process;
- identification of the different parameters influencing the
runoff process and definition of the basic principles of the
quality mechanisms existent in the rainfall-runoff process;
- description of the runoff process as mathematical algorithms
in a deterministic, parametric, and stochastic way;
- verification of different deterministic, parametric and
stochastic mathematical approaches.
The objectives of this study tie neatly into the
findings and recommendations of last year's design storm seminar,
held in conjunction with the SWMM-Users' Group meeting in
Montreal, which suggested that
- for development of alternatives to the design storm concept
a continuous data base is required;
- further research should concentrate on antecedent moisture
conditions; and
- for consideration of qualitative aspects the amount of
pollutants present at the beginning of a storm event is of
prime importance [3j.
HARLACHING TEST CATCHMENT
Selection Criteria
The criteria for selection of the major test catch-
ment for the study were the following:
- the drainage area had to be large enough to be representative
of an urban combined sewer system;
- for valid correlation between rainfall and runoff multiple
diversions or a large number of intercepting points were not
desirable;
213
-------
- the catchment slopes had to be relatively flat and the site
accessible in order that monitoring equipment could be
installed;
- traffic and public disruption had to be kept to a minimum;
- the site should be a combined sewer overflow in order to
allow for direct derivation of overflow figures from the
data;and
- the site had to be sufficently large to allow for installation
and maintenance of equipment.
The Harlaching drainage area met these criteria and
was selected for sampling.
Catchment Character isties
The Harlaching catchment is part of the Munich
drainage area and has a combined sewer system (Figure 1).
Located at the southern edge of the city, it covers an
area of 1340 acres (542 ha). 39 % of the area is imper-
vious, of which 15 % is roofs and 24 % streets and side-
walks. Land surface slopes vary up to 1.7 %. The land-use
is mainly residential with a population density ranging
from 12 to 80 inhabitants per acre (Figure 2). The area
includes some small commercial sections, two hospitals,
playgrounds, and schools (Figure 3).
The catchment boundaries are clearly defined
with the exception of a few interconnections to the ad-
jacent drainage system located at the northwestern catch-
ment boundary (Figure 4). Most of these interconnections
were completely closed off for the study; four had to be
left partially open in order to prevent local flooding
under severe storm conditions. The southern catchment
boundary is formed by a forest and the western boundary
is formed by a recess in the land surface adjacent to the
Isar River. Changes in the nature of the area as well as
in the number of inhabitants during the course of the in-
vestigation are negligible.
Rainfall Monitoring
There are three rainfall recording stations
within the catchment. Their locations are indicated on
214
-------
Figure 1:
Munich drainage area
indicating the monitored
catchments of Harlaching
and Pullach
MeSgebiet
HARLACHING
Figure 2:
Harlaching residential
area
Figure 4 as triangles. The individual stations are
equippend as follows:
- Harlachinger str./Wetter-
steinstr.
- Am Hollerbusch
- Harlaching hospital
1 HohenpeiBenberg type
Ombrometer
1 Hellmann device
1 Totalizer
1 Hohenpeiflenberg type
Ombrometer
1 Hellmann device
The three types of rainfall recorders just
mentioned may briefly be described as follows:
215
-------
Figure 3:
Harlaching commercial
area
Harlochino drginoqe oreg.
1 A = 542 ho
r - 0.39
^monitormq Btatt
A ram gauge
Figure 4:
Harlaching catchment
241
216
-------
- The Hellmann device records cumulative rainfall depth as
indicated by a float, mounted in a container, which is
situated underneath a catching funnel. The recording cylin-
der advances at a rate of 6 in/hr (16 mm/hr).
- Components of the Ombrometer type are a standardized re-
ceiving container, a drop dispenser that transforms preci-
pitation into drops of uniform size, and a light barrier,
which produces a signal according to the number of drops.
Usually the gauge is equipped with a clock and a digital
printer, which gives results as time, and number of drops
per minute. The rainfall intensity is calculated from a
calibration curve.
- The Totalizer rain gauge stores precipitation and furnishes
precipitation totals.
Under continuous operation, all rainfall recorders
used functioned faultlessly. The reliability of the preci-
pitation measurements, however, was generally influenced by
the cleanliness of the catching funnels. For the Ombrometer
it should be noted that under very high rainfall intensities
a continuous stream is formed rather than individual drops.
This offsets the light barrier drop counting principle, and
this data is therefore unusable for the very high rainfall
intensity range.
For 1976 the rainwater caught in the totalizer has
been analyzed for total suspended solids, COD, total phospho-
rus, Ammonia-nitrogen (NHJ , Nitrite-nitrogen (NO2), Nitrate-
nitrogen (NO.,) , conductivity, and pH.
Runoff Monitoring
All runoff monitoring and sampling was done at the
overflow site, the area's intercepting point, shown on
Figure 4 as a solid circle. The time of concentration to this
point is approximately 40 minutes at a maximum sewer length
of approximately 3 miles (4500 m) , and an average sewer
slope of 0.5 %. The combined sewer overflow structure is
situated underneath a parking lot (Figure 5). Therefore it
was possible to establish the sampling station on top of the
manhole leading down to the structure. The shed holds a
field laboratory and above-ground instrumentation.
217
-------
Figuce 5:
Shed of Harlaching moni-
toring station on top
of combined sewer over-
flow structure
Figure 6 portrays the overflow structure
looking downstream prior to the installation of a working
platform. The runoff peak of a storm event with a^once
per year frequency amounts to about 350 cfs (10 m /s) .
Overflow starts at 20 cfs (0.6 m /s) . Figure 7 shows the working
platform looking upstream toward the inflow of the structure.
For runoff, in addition to flow depth, the following
quality parameters are continuously recorded in 5-minute-inter-
vals: temperature, conductivity and turbidity. The flow is
determined by water levels measured at three different points:
- at the entrance to the overflow;
- at the exit to the conduit leading to the treatment plant;
and
- in the overflow pipe leading to the Isar River.
The water depths are measured by ultrasonic echo sounders
(Figure 8). A frequency of 50 kHz generates a very narrow
beam, which provides for clear measurements, even at a very
low water level. In addition, the devices operate without
touch-contact. The depth-flow relationship was established
by tracer measurements. The water level recorders operate
with a minimum number of failures. The instrument indicators
have to be checked and verified annually.
Wastewater turbidity provides an indication of the
pollution level in general. For tugbidi£y differgnt angles
of dispersion are measured: 0 , 25 , 90 and 135 . The equip-
ment itself functions faultlessly. Up until now, however,
218
-------
Figure 6:
Harlaching overflow structure
prior to the installation of
monitoring equipment
Figure 7:
Harlaching overflow
structure after
installation,'of a working
platform
there have been frequent periods where the turbidity equip-
ment was inoperative due to blockage in the connecting pipes.
The conductivity of the wastewater is a measure of
the content of dissolved, dissociated inorganics and organics.
The conductivity changes significantly during runoff and is
proportional to the content of dissociated inorganic salts.
Knowledge of the salt content and variations thereof in
storm runoff is important for both the operation of treatment
plants and for the living conditions of microorganisms in re-
ceiving waters. Conductivity is measured with a palladium
electrode. The device functions free of maintenance exept that
it is important to clean the sensor twice a week.
219
-------
Figure 8:
Ultrasonic echo sounder for
water depth measurement at
the entrance to the Harla-
ching overflow structure
The temperature is measured over a thermosensor.
With regular maintenance inoperative periods for this sensor
have thus far been avoided.
A small computer was installed in the station in
order to allow for (Figure 9):
- continuous monitoring;
- on-line data recording to enable the immediate control
and review of the gathered data; and
- providing the data in ar form ready for processing to
minimize work and to avoid mistakes.
The central processing unit (CPU) of this computer has a core
of 32 k bytes. Half of the core is occupied by the operating
system, which uses the process language BASEX, an extension of
BASIC. A teletype featuring a keyboard, tape punch and reader
is attached through an interface. Besides monitoring and coordi-
nating the on-line data which were the three flow depths, turbi-
dity, conductivity and temperature, this computer has to:
- check the data for plausiblity and provide warnings in case
of malfunctions;
- record the rainfall data collected next to the monitoring
station with a scanning frequency of one minute; and
- survey and operate the sampler and a pumping system related
to the sampler.
The sampler itself is of special design There is
a continuous sample flow which has the advantage that blockages
in the sampler conduit can be recognized in time and removed
without the loss of samples. For this a flush-back set-up was
220
-------
Figure 9:
Computer and teletype at the
Harlaching monitoring station
installed which washes out the conduit with tap water by re-
versing the pump direction. The flushing is triggered auto-
matically by the computer. The sampler has a capacity of
160 one liter bottles, one liter is approximately 1 quart,
which are accomodated in 20 metal baskets, each with 8
bottles (Figure 10) . The sampler operates by chain advance-
ment. Conservation of the samples is achieved through
cooling to 4 C. The positioning of the bottles under the
filling mechanism occurs automatically (Figure 11). The
position of the bottles in the sampler is recorded through
binary coded mechanical contacts by the computer and is
coordinated to the filling-time. The filling of the bottles
is achieved through free fall, using a flow diverter in the
sample stream. At each sampling time two bottles are filled.
The time and duration' of this action is controlled by the
processor. Sampler overflow is collected in an open channel
and diverted back into the sewer. A circuit-breaker is built
into the drain channel in order to prevent flooding in case
of line blockage.
Under dry weather conditions the time lapse between
two samples amounts to 90 minutes; under storm conditions it
221
-------
Figure 10
Bottle baskets of sampler at
the Harlaching monitoring
station
Figure 11:
Filling mechanism of
sampler at the Harlaching
monitoring station
is 5 to 20 minutes. Storm runoff conditions are recognized
and double checked by the processor using the rainfall
measurements next to the station and the water level in
the conduit leading to the treatment plant.
The sampling interval during a storm event is
controlled by the rate of change of the "turbidity load".
The turbidity load is defined as the product of turbidity
measured at the 0 angle and flow.
222
-------
The laboratory analysis program encompasses six
parameters: total suspended solids, COD, and Kjeldahl-nitro-
gen on all samples, and BOD, total phosphorus and total or-
ganic carbon on every fifth sample. This program has been
in effect since 1977.
For relating runoff to rainfall and quality to
quantity, the exact timing of each sample is of considerable
importance. The sampler, the continuous pollutant recor-
dings, the water level measurements, and the one rainfall
recorder are all exactly synchronized with the aid of the
computer. Exact timing in the other rainfall recorders is
kept with a quartz watch.
For an overview Figure 12 portrays the complete
data collection system of the Harlaching monitoring station
[4J . Shown is the continuous sample flow drawn from the
sewer passing trough the turbidity measurement of 0 angle
and through the sampler. From the continuous sample flow
theQflows for the turbidity measurements of 25 , 90 and
135 angles are branched off. A data flow chart is incor-
porated into this figure as dashed lines: the different
signals from the recorders - three for water depth, four
for turbidity, one each for temperature, conductivity and
rainfall - feed into the computer. The slah-dotted lines
indicate the path of active control information, as for
example to control pump and sampler operation.
PULLACH TEST CATCHMENT
Selection criteria
As mentioned above, measurements are also taken
in a smaller separate system with similar land-use and topo-
graphical characteristics. This was done to obtain infor-
mation on storm runoff without dry weather flow in order to
aid analyzing and separating the individual components of
combined runoff in the other catchment, The Pullach separate
drainage system was selected.
Catchment Characteristics
The Pullach drainage area is located adj'acent
to Munich's southern border and lies on the high bank of
the Isar River, This catchment is 57 acres (23 ha) in
223
-------
ro
ro
Figure 12:
Data collection system of
the Harlaching monitoring
station
VOR- KANAL-
FLUTER NETZ
RE6ENMESSER
g) WASSERSTAND-M
T) LEITFAHIGK6IT
© TEMPERA7UR
[vJvERTEILER
--- MESSDATEN
STEUERDATEN
ABWASSERZUFLUSS
AKUST. ALARM
ANAL. DATEN
DIGIT. DATEN
HANDSTEUER.
KANALNETZ
LABOR
LABORDATEN
LEITFAHIGKEIT
MESSDATEN
PROBENEHMER
PROGRAMME
RECHNER
wastewater inflow
acoustic signal
analog data
digitized data
manual control
sewer system
laboratory
laboratory data
conductivity
measured data flow
sampler
programs
computer
RECHNER ANF.
REGENMESSER
REGENUBERLAUF
STEUERDATEN
TELETYPE
TEMPERATUR
TRttBUNG
VERTEILER
VORFLUTER
WASSERSTAND-M.
ZUSTANDSANZ.
ZUST. WERTE
computer requirements
rainfall recorder
combined sewer overflow
active control information
teletype
temperature
turbidity
divider
receiving waters
flow depth recorder
status indicator
status figures
-------
size (Figure 1). The land-use is residential and includes
a small central commercial section, a playground and a school.
36 % of the area is impervious, of which 18 % is roofs and
backyards, and 18 % streets and sidewalks. Land slopes are
similar to those in Harlaching and the concentration time
ranges from 22 to 12 minutes for rainfall intensities
ranging between. 14 and 1.4 in/hr (10 and 100 1/s ha). A
storm sewer outfall to the Isar River provides the sole
intercepting point in the area. The sampling station is
located at this point (Figure 13). All catchment boundaries
are clearly defined. The groundwater table is approximately
130 ft (40 m) below the surface, ensuring that groundwater
does not infiltrate into the sewer lines. The drainage
system has not undergone any changes significant to the
investigation.
Rainfall Monitoring
In the Pullach drainage area there is one rainfall
recording station with all three types of recorders mentioned
before: a Hellmann device, an Ombrometer, and a Totalizer.
The tiny circle on Figure 13 indicates the location of the
rainfall station within the catchment.
As in the Harlaching catchment the rainwater
caught in the totalizer in 1976 was analyzed for total
suspended solids, COD, total phosphorus, Ammonia -nitro-
gen (NH3), Nitrite-nitrogen (N02), Nitrate-nitrogen (N03),
conductivity and pH.
Figure 13:
Pullach catchment
rain gauge
storm runoff moni-
toring station
225
-------
Runoff Monitoring
The monitoring station of the Pullach storm
sewer system is situated in the stilling section of the
main stormwater collector 250 ft (75 m) behind the drop
into the Isar Valley.The construction of the measuring
station, with installations for direct measurements of
flow depth, conductivity, temperature and pH and for
collecting stormwater samples, is portrayed in Figure 14.
Flow is measured by determining the water depth
at a Venturi, which was built into the channel. An influence
on the data due to downstream backwater in the channel can
not occur, as the outflow of the storm sewer is above the
high-water level of the Isar River. An air bubble method
is utilized in order to determine the water depth. The con-
version of water depths into flow is done according to a
calibration curve. The continuous flow record is plotted
by advancing the velocity of the paper 2.4 in/hr (60 mm/hr).
This yields a resolution of one minute. This measuring
procedure functions quite reliably.
Sl2m/1,7m-HVenturi f|u.
Figure 14;
Schematic view of the
Pullach monitoring station
C conductivity
ES egg shape
pH pH-value
Q flow
T temperature
226
-------
In order to measure conductivity and temperature,
equipment similar to that employed at the Harlaching moni-
toring station was used.
Stormwater samples are collected downstream of the
Venturi flume, insuring complete mixing of the flow. Constant
back-up in the flow section is forced by a sill obstruction,
5.5 in (14 cm) high in order to guarantee sampling at the
starting and final proportions of the storm runoff process
when flow depths are low, and in order to prevent air entry
into the pumpline. The pump was placed on a working platform
where it forced the sampling flow into a sample collection
device above ground surface. The sampler has a capacity of
22 bottles of 1.5 quarts (approx. 1.5 1) each.
In 1977 and 1978 there have been two programs
for sampling: one was to draw mixed samples, that ife
filling only one or a limited small number of sample
bottles per event. Here the sample was automatically
compounded from different samples drawn at variable
spacings, the timing of which varied from 5 to 60
minutes and depended on the flow rate.
The other program was to fill individual sample
bottles, which occurred automatically at time intervals
of 5 to 60 minutes, depending on the flow rate.
The frequent sampling in both catchments resulted
during rainy seasons in a quantity of samples which exceeded
the capacity of laboratory. As one of the major intents of
the data collection in the separate system of Pullach was
to define antecedent moisture and pollutant conditions, and
as these figures also may be derived from the continuous flow
recording and mixed samples, the Pullach monitoring program
was changed to only draw mixed samples - that is where only
one or a few sample bottles are filled per rainfall event.
The samples were analyzed for settleable solids,
total suspended solids, BOD, COD, Ammonia-nitrogen (NH3) ,
Nitrite-nitrogen (N02), Nitrate-nitrogen (N^ ) and total
phosporus.
227
-------
DATA RECORD
Existing Data
The amount of data collected per year is con-
siderable. In 1977 and 1978 approximately one million pieces
of data were collected, and therefore data storage and retrie-
val had to be completely computerized.
The data acquired by the small local computer of
the Harlaching monitoring station are punched on paper tape
and then fed into a data bank established at the University
computer center. All laboratory data are listed and punched
for data bank entry.
Data checks for format and plausibility are
employed prior to finally establishing the data sets. The
reliability of data can be impaired by various factors which
include external influences, and malfunctioning or maladjust-
ment of the measuring equipment. The data correction proved
to be very difficult and is for the most part attainable
only through human intervention. In this case, analog recor-
dings are of the greatest value.
The best overview of the existence of data is pro-
vided by plots showing the number of measurements available
per day or week. Figure 15 and 16, for instance, are an
illustrative example from 1979, when the monitoring station
was out of service for six weeks due to an act of arson.
Figure 15 shows the existence of directly processed data,
figure 16 the existence of laboratory data.
The goal of a truly continuous data base can never
be met. Data gaps of different lengths occurred due to mal-
functioning of individual recorders and, more seriously,
due to malfunctions of the small processor at the measuring
station or of the sampler. Major sources of disruption of
the monitoring program, aside from the above mentioned act
of arson, have been pump failures and motor defects in the
sampler. Statistical analysis of the data gaps showed that
rainfall data are available at an almost continuous basis,
while runoff data include gaps of up to 35 days in 1977,
up to 15 days in 1978 and up to 56 days in 1979. Those
figures refer to the data of the combined sewer system of
Harlaching. In total it is estimated that about 50 % of
228
-------
INHALTSUEBERSICHT GESAM79 1979
» WASSERST.I/2 * TRUEBUNG x REGENWERTE U. W3
Figure 15
Existence
processed
of directly
data
WOCHENNUMMEk
Figure 16:
Existence of laboratory
data
»5T*C-iB*KJ
INHALTSUEBERSICHT EESAM79 1979
' RESTL.LABW
WOCHENNUMMER
AST + CSB + KJ
INHALTSUEBERSICHT GESAM 79
REGENWERTE U. W3
RESTL. LABW
TEMP + LEITF
TRUEBUNG
WASSERST. 1/2
WOCHENNUMMER
ZAHL DER EINTRAEGE PRO WOCHE
TSS and COD and Kjeldahl-nitrogen
content 1979
rainfall data and flow depth 3
other laboratory data
temperature and conductivity
turbidity
flow depth 1 and 2
number of week
number of figures per week
229
-------
the runoff data are recorded, referring to 100 % being a
truly continuous data set.
Data supplement
Still, a complete data base is required for the
purpose of:
- defining annual and seasonal pollutant loads contained
in runoff and overflow;
It is also desired for
- statistical analysis of rainfall-runoff-overflow data
with the aim of defining durations and frequencies;
and for
- performing a stochastic analysis of the rainfall-runoff-
overflow process.
Therefore, the gaps must be filled, an effort which
is being undertaken at present. The seriousness of a gap is.
judged upon by comparison of the gap's duration with the in-
tended recording spacing. Minor gaps are interpolated from
the existing data. It is intended to employ rainfall-runoff
simulations, using the complete rainfall data as input, to
produce runoff figures, both for quantity and quality, to
fill the major gaps. Such substitution may be trusted, as
there are sufficient data available to calibrate rainfall-
runoff models for the test catchments.
At the beginning it was mentioned for the Harlaching
catchment that for high intensity storms there might be more
than the one interception point recorded. This uncertainty
will also be checked by employing rainfall-runoff simulations.
Data in question then may be replaced by simulation results.
LEGAL AND REGULATORY ASPECTS
The study is scheduled to be completed by the end
of 1982 and is quite important in Germany for legal and regu-
latory reasons:
- communities are required by law to furnish detailed master
plans in order to obtain allowances to discharge stormwater
into receiving waters [s] ;
- a law enacted in 1976 requires communities to pay fees for
wastewater loading [e] . This includes stormwater discharged
into receiving waters.
230
-------
This "Waste Water Fee Law" is of primary importance.
It aims for a higher level of water quality and to more justly
distribute the cost burden for the prevention of, removal of,
and compensation for water pollution. The mandatory fee does
not take effect until December last of 1980. After this date
the fee will be levied for so-called pollution damage units.
With respect to stormwater discharges and combined sewer
overflows this law asserts that:
- the number of pollutional units for storm water only which,
via public drainage, is discharged into receiving waters is
equal to 12 % of the total population figure - for instance,
for a city like Munich that is a fee of 3.6 million 0 US/year.
The treatment plant effluent is taxed separately.
- the individual states determine to what degree this number
of pollutional units shall be diminished, if storm water
runoff or combined sewer overflows are retained or treated.
In Northrhein-Westphalia, for example, this specific
regulation reads as follows [7] :
- The number of pollutional damage units for storm water dis-
charges shall be reduced by 50 % if storm runoff from a sepa-
rate system is retained and treated for a time period of at
least 20 minutes. For combined sewer overflows they shall be
reduced if the system is designed so that overflow does not
start until the rainfall rate exceeds .07 in/hr (5 1/s ha) .
- Direct storm runoff is free of charge if it is treated in a
settling tank for a time period of at least 20 minutes. For
combined sewer overflows there is no charge if the system is
designed so that overflow does not start until the rainfall
rate exceeds 2 in/hour (15 1/s ha).
The basic principle on which the procedures are
based to dimension overflows and retention basins properly
is,according to another regulation, the concept of catching
the "first-flush" related pollution [s] . The data on which
this concept is based, were derived from a limited number of
measurements taken in a small combined sewered catchment [9] .
Hence/ the entire body of laws, regulations and
guidelines lacks a well-defined data base. For German
conditions this emphasizes the necessity of a continuous
dense data base as established in this study.
231
-------
SUMMARY AND CONCLUSIONS
The necessity for a continuous and simultaneously
dense urban runoff data base is shown for technical and regu-
latory reasons. Therefore rainfall and runoff quantity and
quality were monitored in two catchments, 57 acres (23 ha) and
1340 acres (542 ha) in size, since 1976. For quality mainly
total suspended solids, BOD, COD, Kjeldahl-nitrogen, total
phosphorus and total organic carbon were analyzed.
The study, sponsored by the German Research Society,
showed the limits of automation in urban runoff monitoring.
Frequent checking of the monitoring equipment and manual compa-
rison of directly computerized data with orginal recordings
proved to be necessary.
The goal of a truly continuous field data base hardly
can be met. Data gaps of different lengths occured mainly due to
malfunction of equipment. It is suggested to fill gaps in the
collected runoff data with simulation results to provide a
complete data base for statistical analysis and derivation of
annual and seasonal runoff-overflow figures.
It is recommended to establish further long-term
monitoring programs to advance the knowledge in urban hydrology,
REFERENCES
[l] Geiger, W.F., 1977. "Continuous Unsteady Simulation
with QQS". in Proceedings Storm Water Management Model
Users Group Meeting. Gainesville, Florida, USA.
[2] Massing, H., 1979. "Progress in Urban Hydrology since
1976 in the Federal Republic of Germany", in International
Symposium on Urban Hydrology. Washington, D.C., USA.
232
-------
[3] Patry, G. and Me Pherson, M.B. , 1979. "The Design
Storm Concept". Proceedings of a Seminar at Ecole
Polytechnique de Montreal, Montreal, Quebec and of
a related Session of the American Geophysical Union,
Spring Annual Meeting, Washington, D.C.Montreal, Canada
GREMU-79/02.
[4] DFG(German Research Society),1979. "Niederschlagsab-
fluB und -beschaffenheit in stadtischen Gebieten,
Zwischenbericht Marz 1979" (Quantity and Quality of
Storm Runoff from Urban Areas, Interim Report March
1979). SFB 81 TPA,4 Interim Report. Munich, Germany.
[5] Federal Republic of Germany, 1976. "Gesetz zur Ordnung
des Wasserhaushalts (WHG)"(Federal Law on Water Resources
Management). BGB1 I pp. 3017 and 3341. Germany.
Federal Republic of Germany, 1976. "Gesetz ueber Ab-
gaben fuer das Einleiten von Abwasser in Gewaesser
(AbwAG)" (Federal Law on taxing wastewater discharges
into receiving waters). BGBl I pp. 2721 and 3007.
Germany.
[7] Northrhein-Westphalia, 1979. "Wassergesetz fuer das
Land Nordrhein-Westfalen (LWG)" (State Law on Water
Resources Management for Northrhein-Westphalia).
GVBl No. 38, p. 488. Northrhein-Westphalia, Germany.
[8] ATV, 1977. "Richtlinien fuer die Bemessung und Ge-
staltung von Regenentlastungen in Mischwasserkanaelen"
(Regulations for the design of combined sewer overflows).
GFA, A 128. St. Augustin, Germany.
[9] Krauth, K,. 1970."Der Abfluss und die Verschmutzung
des Abflusses in Mischwasserkanalisationen bei Regen"
(Wet-Waather Runoff and its pollution in Combined
Sewer Systems). Stuttgarter Berichte zur Siedlungs-
wasserwirtschaft, Band 45. Kommissionsverlag R.
Oldenbourg. Munich, Germany.
233
-------
SWMM Users Group Meeting
June 19-20, 1980
List of Attendees
1. Joseph Adler
Dam as fc Smith
2. Maqbool Ahmad
Edmwtwr tfater- an* Sanitation
2nd Floor, West Chambers
12220 Stony Plain Road
Edmonton, Alberta
T5N 3M9
3. Mark P. Allen
Lozier Engineers
50 Chestnut Plaza
Rochester, New York
4. James A. Anderson
Urban Science Applications Inc.
1027 Fisher Building
Detroit, Michigan
48202
5. John Anderson
Gore & Storrie Limited
Suite 700
331 Cooper St.
Ottawa, Ontario
K2P OG5
6. G. R. Bache
Del can Limited
133 Wynford Drive
Don Mills, Ontario
M3C 1K1
7. S. G. Barber
Land Development Wing
Ministry of Housing
60 Bloor Street West
10th Floor
Toronto, Ontario
M4W 3K7
8. Tom Barnwell
U.S. E.P.A.
Athens, Georgia
9. Darshan Basran
Dept. of .Transport Canada
Toronto (Maiton) Implementation Team
P.O. Box 6003
Toronto AMF, Ontario
L5P IBS
10. Cathy Biggs
Gore & Storrie Limited
1670 Bayview Avenue
Toronto, Ontario
M4G 3C2
11. D. Brierley
Del can Limited
133 Wynford Drive
Toronto, Ontario
M3C 1K1
12. Terry Brueck
EMA Inc.
270 Metro Square Building
St. Paul, MN
55101
13. Brian Bodnaruk
Acres Consulting Services Limited
500 Portage Avenue
Winnipeg, Manitoba
R3C 3Y8
14. Larry Bodnaruk
Northwest Hydraulic Consultants
4823-99 Street
Edmonton, Alberta
15. Herb Bolinger
HNTB
Indianapolis, Indiana
16. William M. Cameron
Hydrology and Monitoring Section
Ministry of the Environment
1 St. Clair Avenue West
Toronto, Ontario
17. A. M. Candaras
Paul Theil Associates Limited
700 Balmoral Drive
Bramalea, Ontario
L6T 1X2
18. R. S. Cebryk
Cumming-Cockburn & Associates Limited
Ste 407-10169 104 St.
Edmonton, Alberta
19. K. L. Chua
Paul Theil Associates Limited
700 Balmoral Drive
Bramalea, Ontario
L6T 1X2
234
-------
20. Yu-Tang Chuang
Dept. of Civil Engineering
University of Akron
Akron, Ohio
4432 J
21. Bill Clarke
Proctor & Redfern
75 Eglinton Avenue East
Toronto, Ontario
22. Donald Cook
Regional Municipality of Niagara
150 Berryman Avenue
P.O. Sox 3025
St. Catherines, Ontario
L2R 7E9
23. Norman Crawford
Hydrocomp Inc.
Palo Alto, CA
24. Arthur Dee
EMA Inc.
270 Metro Square Building
St. Paul, MN
55101
25. Richard N. DeGuida
Calocerinos & Spina
Consulting Engineers
1020 Seventh North Street
Liverpool, New York
13088
26. Ray Dever
Water Resources Engineers
Springfield, Virginia
27. Roger Dickey
HNTB
Indianopolis, Indiana
28. C. Doherty
Dillon Limited
50 Holly Street
Box 219 Stn. K
Toronto, Ontario
M4P 2G5
29. John Drake
McMaster University
30. W. Friesen
City of Reglna
1790 Regina
Reglna, Saskatchewan
,S4P 3C8
31. M. D. Garraway
Kilborn Limited
2200 Lake Shore Blvd. W.
Toronto, Ontario
M8V 1A4
32. W. F. Geiger
Technical University of Munich
8000 Munich 71
Strasslacher Str. 2
Germany
33. Bruce L. George
Associated Engineering Services Limited
13140 St. Albert Trail
Edmonton, Alberta
T5L 4Z8
34. Bernard L. Golding
HNTB
35. Swapan Gupta
University of Ottawa
Dept. of Civil Engineering
University of Ottawa
Ottawa, Ontario
KIN 984
36. Martin Habicht
James F. MacLaren Ltd.
1220 Sheppard Avenue East
Willowdale, Ontario
37. Jim Hagernian
Calocerinos and Spina
Liverpool, New York
38. D. J. Hay
Environment Canada
Ottawa, Ontario
K1A 1C8
39. David M. Heiser
O'Brien & Gere Engineers Inc.
P.O. Box 4873
Syracuse, New York
13221
40. Karl Hefimerich-
Department of Public Works
City of Toronto
41. John Hoddenbagh
Hoddenbagh, Norton & Associates (1980) Limited
Consulting Engineers
306 Dundas Street West
Whitby, Ontario
LIN 2M5
235
-------
42. Wayne Huber
University of Florida
43. W. James
McMaster University
Civil Engineering Dept.
Hamilton, Ontario
L8S 4L7
44. Robert C. Johanson
Dept. of Civil Engineering
University of Pacific
Stockton, California
45. Chih-Chia Ko
Marshall, Mack!in Monaghan
117 Cosburn Avenue, Apt. 105
Toronto, Ontario
46. C. D. Leavens
Paul Theil Associates Limited
700 Balmoral Drive
Bramalea, Ontario
L6T 1X2
47. Weng-Yau Liang
University of Ottawa
Civil Engineering Department
371 Nelson Street
Apt. 1
Ottawa, Ontario
KIN 7S5
48. J. D. Lee
J. D. Lee Engineering Limited
1 Yonge Street, Suite 2206
Toronto, Ontario
M5E 1E5
49. Harry S. Loijens
Rideau River Stormwater Management Study
222 Queen Street, 10th Floor
Ottawa, Ontario
KIP 5V9
50. C. C. Macey
Underwood McLellan Ltd.
1479 Buffalo Place
Winnipeg, Manitoba
R3T 1L7
51. J. Lance Maidlow
Associated Engineering Servcies Limited
13140 St. Albert Trail
Edmonton, Alberta
T5L 4R8
52. Gilles Marchi
Ecole Polytechnique
Civil Engineering Department
P.O. 6079 Station A
Montreal, Quebec
H3C 3A7
53. Dr. Gerald McDonald
County of Monroe
New York
54. C. McEwan
Paul Theil Associates Limited
700 fra^mora} Brive
Bramalea, Ontario
L6T 1X2
55. D. F. McGovern
Oliver, Mangione, McCalla & Associates, Ltd.
1755 Woodward Drive
Ottawa, Ontario
56. Murray B. McPherson
23 Watson Street
Marblehead, Mass.
01945
57. John Moore
Keifer Engineering Inc.
20 N. Wacker Drive
Chicago, 111.
60606
58. Z. Novak
Water Resources Branch
Ministry of the Environment
135 St. Clair Avenue West
Toronto, Ontario
M4V 1P5
59. Helmut Pankratz
Barr Associates
1034 Highway 53 West
R.R. #1
Ancaster, Ontario
L9G 3K9
60. Choon Eng P'ng
University of Ottawa
Civil Engineering Department
18 Henderson Avenue
Ottawa, Ontario
KIN 7P1
236
-------
61. Pierre Purenne
Communaute Urbaine De Montreal
12001 Boul Maurice Duplissis
Montreal, Prov. Quebec
H1C 1V3
62. Robert E. Ragan
Calocerinos & Spina
1020 Seventh North Street
Liverpool, New York
13088
63. Larry Roesner
Water Resources Engineers
Springfield, Virginia
64, J. C. Sinmonds
Corp. City of Ottawa
1355 Bank Street
Ottawa, Ontario
K1H 8K7
65. S. Singer
Water Resources Branch
Ministry of the Environment
135 St. Clair Avenue West
Toronto, Ontario
66. Howard M. Shapiro
Lozier Engineers
50 Chestnut Plaza
Rochester, New York
14604
67. Reinhard Sprenger
1.0. Engineering Co.
966 Waver ley Street
Winnipeg, Manitoba
R3T 4M5
68. R. J. Stoltz
Paul Theil Associates Limited
700 Balmoral Drive
Bramalea, Ontario
L6T 1X2
69. S. H. Tan
Proctor & Redfern Limited
75 Eglinton Avenue East
Toronto, Ontario
70. Harry Torno
U.S. EPA Science Advisory Board (A101M)
Washington, D.C.
20460
71. Kevin Walters
Giffels Associates Limited
30 International Blvd.
Rexdale, Ontario
M9W 5P3
72. 0. Weatherbe
Water Resources Branch
Ministry of the Environment
135 St. Clair Avenue West
Toronto, Ontario
73. W. P. Wolfe
Reid Crowther & Partners Limited
831 Portage Avenue
Winnipeg, Manitoba
R3G ON6
237
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/9-80-064
3. RECIPIENT'S ACCESSIO(*NO.
4. TITLE AND SUBTITLE
Proceedings Stormwater Management Model (SWMM) Users
Group Meeting, 19-20 June 1980
5. REPORT DATE
December 1980 Issuing DAte.
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO,
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Research Laboratory
U.S. Environmental Protection Agency
College Station Road
Athens GA 30613
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection AgencyAthens GA
Environmental Protection Agency
College Station Road
Athens GA 30613
13. IYpE i?F REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/01
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report includes eleven papers on topics related to the development and
application of computer-based mathematical models for water quality and quantity
management presented at the semi-annual meeting of the Joint U.S. Canadian Stormwater
Management Model Users Group held 19-20 June 1980 in Toronto, Ontario, Canada.
Topics covered include descriptions of three urban runoff models; a discussion
of use of the Soil Conservation Service TR-55 model;, applications of several models
in planning, analysis and design; and a discussion of kinematic design storms.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
Simulation
Water Quality
12A
68D
8. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS {ThisReport)
UNCLASSIFIED
21. NO^
PAGES
20. SECURITY CLASS (Thispage)
UNCLASSTFTFD
22. PRICE
EPA Form 2220-1 (9-73)
238
* U.S. GOVERNMENT PRINTING OFFICE: 1981 -757-064/0219
------- |