SYNTHESIS OF TxDOT STORM DRAIN DESIGN

by

David Thompson
Associate Professor
Department of Civil Engineering
Texas Tech University

Xing Fang
Associate Professor
Department of Civil Engineering
Lamar University

and

Gharty-Chhetri Om Bahadur
Research Assistant
Department of Civil Engineering
Lamar University

Report 4553-1

Project Number 0-4553

Research Project Title: Synthesis of Storm Drain Design; Current
Methodologies and Need for Alternatives to the Rational Method

Sponsored by the
Texas Department of Transportation

October 2003

Center for Multidisciplinary Research in Transportation

Department of Civil Engineering
Texas Tech University
P.O. Box 41023
Lubbock, TX 79409-1023
 TABLE OF CONTENTS
Page No.

List of Tables…………………………………………………………………... v
List of Figures………………………………………………………………….. vii

1. INTRODUCTION
1.1 Hydrologic Perspectives of Storm Drainage Systems…….…………… 1
1.2 Background and Scope of the Study…………………………………… 4

2. LITERATURE REVIEW
2.1 Rational Method for Storm Drain Design
2.1.1 Introduction……………………………………….…………… 6
2.1.2 Assumption of Rational Formula………………………………. 6
2.1.3 Rational Formula
2.1.3.1 Runoff Coefficient……………………………………... 7
2.1.3.2 Rainfall Intensity, Rainfall Duration and Time
of Concentration……………………………………. 7
2.1.4 Limitation of Rational Formula……………………………….. 11
2.2 Hydrograph Generation Methods
2.2.1 Hydrograph Development for Inlets…………………………… 12
2.2.2 Modified Rational Method…………………………………….. 13
2.3 Design of Stormwater System
2.3.1 Introduction……………………………………………………. 15
2.3.2 Design Frequency and Spread…………………………………. 16
2.3.3 Curbs and Gutters……………………………………………… 17
2.3.4 Flow in Gutters with Uniform Sections………………………... 18
2.3.5 Flow in Gutters with Composite Sections……………………... 19
2.3.6 Drainage Inlet Design
2.3.6.1 Interception Capacity and Efficiency on
Continuous Grade Inlet………………………………… 21
2.3.6.2 Curb Opening Inlets…………………………………… 22
2.3.6.3 Grate Inlets……………………………………………... 24
2.4 Journal Publication on Storm Drainage System Design
2.4.1 Introduction…………………………………………………….. 29
2.4.2 Rational Method for Peak Flow Estimation……………………. 29
2.4.3 Street Stormwater Storage Capacity…………………………… 29

ii
 2.4.4 Hydraulic Performance of Highway Storm Sewer
Inlet………………….………………………………………… 30
2.4.5 Improvements in Curb Opening and Grate Inlet
Efficiency……………………………………….…….……….. 30
2.4.6 Stormwater Flow on Curb TxDOT Type Concrete
Roadway………………………………………….…………… 30
2.4.7 Design of Curb Opening Inlet Structure………………………. 31
2.4.8 Storm Sewer Design Sensitivity Analysis Using
ILSD-2 Model………………………………………………… 34
2.4.8.1 Effect of Time Distribution of Rainfall……………….. 35
2.4.8.2 Effect of Overland Flow Hydrograph
Generation Method…………………………………… 36
2.4.8.3 Effect of Technique for Routing Flow
Through Sewers………………………………………. 37

3. COMPUTER MODELS FOR STORMWATER SYSTEM DESIGN

3.1 Introduction…………………………………………………………… 40

3.2 WinStorm……………………………………………………………… 40
3.3 StormCAD…………………………………………………….………. 41
3.4 Hydraflow……………………………………………………………... 41
3.5 SWMM………………………………………………………………... 42
3.6 FHWA Stormwater System Model, HYDRA………………………… 43
3.7 MIDUSS………………………………………………………………. 43
3.8 HydroCAD…………………………………………………….……… 44

4. DRAINAGE SYSTEM FOR CASE STUDY

4.1 Introduction……………………………………………………………. 45
4.2 Hypothetical Drainage System for Case Study………………………... 45

5. RESULTS OF CASE STUDY

5.1 Introduction……………………………………………………………. 47
5.2 Case Study by Using WinStorm and StormCAD……………………… 47
5.3 Case Study by Using Hydraflow
5.3.1 Simulation Result With Analysis w/Design Option…………… 51
5.3.2 Simulation With Enhanced Modeling System Option………… 51
5.4 Conceptual Disconnect in Storm Drain Design……………………….. 53
5.5 Case Study by Using SPLIT Program…………………………………. 54

iii
 5.6 Case Study by Using SWMM…………………………………………. 55
5.6.1 Results From Runoff Layer……………………………………. 55
5.6.2 Results From Runoff and Hydraulic Layer……………………. 57
5.6.3 Results From SCS Method in Runoff Layer…………………… 58
5.6.4 Comparison Between WinStorm and SWMM………………… 58

6. IMPLEMENTATION OF MODEL TESTING FOR HIGHWAY 77/83

6.1 Description of Study Site and TxDOT Design Policy…………………. 60
6.2 Simulation Results by Using StormCAD and WinStorm……………… 61

7. FEASIBILITY OF INLINE WATER QUALITY TREATMENT

7.1 Concepts of Inline Water Quality Treatment………………………….. 56
7.2 Flow Simulation in Larger Pipes by Using SWMM…………………… 71

8. SUMMARY AND CONCLUSIONS………………………………………… 77

REFERENCES……………………………………………………………………….. 81

APPENDIX A CALCULATION OF WINSTORM AND STORMCAD

5. Flow From Watershed to Inlet by Rational Method…………………………… 85
6. Carryover Flow Computation for Each Inlet………………………………….. 86
7. Carryover Flow From Inlet Upstream…………………………………………. 89
8. Flow Computation in Pipe…………………………………………………….. 92

APPENDIX B HYDRAFLOW SETUP………………………………………. 95

APPENDIX C SPLIT PROGRAM

C.1 Introduction to SPLIT Program……………………………………………….. 98

C.2 Modification of SPLIT Program………………………………………………. 99
C.3 Discussion on SPLIT Program Output……………………………………….... 99

APPENDIX D CASE STUDY USING SWMM

D.1 Setup of Storm Drainage Network…………………………………………….. 107

D.2 Precipitation Data……………………………………………………………… 107
D.3 Sub-catchment Data and Hydrograph Generation……………………………... 109
D.4 Simulation Control Data……………………………………………………….. 113

iv
D.5 Inlet Input in Extran (Hydraulic) Layer………………………………………... 115
D.6 Simulation Result by Using Runoff Layer for Both Nodes and Pipes………… 117
D.7 Simulation Result by Using Runoff and Hydraulic Layers Without Inlet
Restriction……………………………………………………………………… 119
D.8 Simulation Result by Using Inlet Restriction
D.8.1 Maximum Capacity of 3.0 cfs to Inlet Without Gutter………………… 124
D.8.2 Maximum Capacity of 3.0 cfs to Inlet With Gutter…………………… 126
D.8.3 Using Rating Curve With/Without Gutter Section……………………. 128
D.9 Simulation Using SCS Unit Hydrograph Method…………….……………….. 130

v
 List of Tables

Chapter 2 Page No.

Table 2.1 Values of runoff coefficient for Rational Formula

(ASCE, 1992)……………………………………………………… 8

Table 2.2 Resistance Factor For Overland Flow…………………………………. 9

Table 2.3 Suggested Minimum Design Frequency and

Spread (Brown et. al, 1996)………………………………………... 16

Table 2.4 Manning’s n for Street and Gutter Pavement

Gutters (FHWA, HDS-3)…………………………………………... 18

Table 2.5 Recommended Pavement Cross Slopes………………………………... 19

Table 2.6 Types of Grates for which Design Procedures

are developed (Brown. et.al, 1996)………………………………… 26

Table 2.7 Values of Coefficient K in Equation (2.32)

(Fiuzat, et. al, 2000)………………………………………………... 32

Chapter 4 Page No.

Table 4.1 Configuration data for four catchments………………………………... 46

Table 4.2 Configuration data for grade inlets…………………………………….. 46

Table 4.3 Configuration data for underground conduits………………………….. 46

Chapter 5 Page No.

Table 5.1 Carryover calculation at inlets by StormCAD…………………………. 49

Table 5.2 Carryover calculation at inlets by WinStorm………………………….. 49

Table 5.3 Pipe flow calculation using StormCAD……………………………….. 50

Table 5.4 Flow calculation in pipes using WinStorm……………………………. 50

vi
Table 5.5 Calculated flow parameters at inlets by Hydraflow with Analysis
w/Design option…………………………………………………… 52
Table 5.6 Calculated flow parameters in “Lines” by Hydraflow with Analysis
w/Design option…………………………………………………… 52

Table 5.7 Calculated flow parameters in “Lines” by Hydraflow

with EMS option…………………………………………………… 53

Table 5.8 Results from WinStorm and modified SPLIT program……………….. 54

Table 5.9 Peak flow (cfs) in the nodes and conduits simulated by using Rational
method and SCS method under Runoff layer……………………… 56

Table 5.10 Peak discharge in the pipes resulted from SWMM under different inlet
settings using Rational method……………………………………. 58

Table 5.11 Peak discharge in the pipes resulted from SWMM under
different inlet settings using SCS hydrology method……………… 59

Chapter 6 Page No.

Table 6.1 Characteristics of inlets tested by WinStorm and

StormCAD………………………………………………………… 63

Table 6.2 Simulated intercepted and bypassed flow from

WinStorm and StormCAD………………………………………… 63

Table 6.3 Pipe parameters and simulated pipe flow using

WinStorm and StormCAD………………………………………… 63

Table 6.4 Discharge obtained from WinStorm for different return

periods for pipes of U.S 77/83 Highway………………………….. 65

vii
 List of Figures

Chapter 1 Page No.

Figure 1.1 The urban drainage system (from Proctor

and Redfern, 1976)………………………………………………… 2

Figure 1.2 Combined urban drainage system (from

Metcalf and Eddy et al., 1971)…………………………………….. 3

Chapter 2 Page No.

Figure 2.1 Intensity-Duration-Frequency (IDF) curves…………………………… 11

Figure 2.2 (a) Hydrograph when duration of rainfall is greater than tc……………….. 14

Figure 2.2 (b) Hydrograph when duration of rainfall is equal to tc…………………… 14

Figure 2.2 (c) Hydrograph when duration of rainfall is less than tc…………………... 15

Figure 2.3 Typical gutter sections (Brown et. al, 2001)…………………………… 17

Figure 2.4 Types of storm drain inlets (Brown et. al, 2001)………………………. 21

Figure 2.5 Depressed curb opening inlet (Brown et. al, 2001)……………………. 23

Figure 2.6 Curb opening inlet with different throats

(Mays, 2001)……………………………………………………….. 25

Figure 2.7 Grate inlet frontal flow interception efficiency

(Brown et. al, 1996)………………………………………………... 28

Figure 2.8 Characteristics spread TA and TB corresponding

to inlet sizes LA and LB (after Fiuzat et.al, 2000)………………….. 33

Figure 2.9 Sewer district (Nouh, 1987)……………………………………………. 35

Figure 2.10 Effect of time distribution of rainfall on least-cost

sewer system design (Nouh, 1987)………………………………… 36

Figure 2.11 Effect of overland flow hydrograph generation method

on least-cost Sewer system design (Nouh, 1987)………………….. 37

Figure 2.12 Effect of technique for routing flow through sewers on

viii
 least-cost sewer system design (Nouh, 1987)……………………… 39

Chapter 4 Page No.

Figure 4.1 Hypothetical storm drainage system for case study…………………… 45

Chapter 5 Page No.

Figure 5.1 Color-coded simulation results using StormCAD…………………….. 48

Chapter 6 Page No.

Figure 6.1 Topographic location of project site and U.S. 77/83

Expressway………………………………………………………… 61

Figure 6.2 Set up of trunk line A of U.S. 77/83 under StormCAD………………... 62

Chapter 7 Page No.

Figure 7.1 Typical Layout of city with Stormceptor installed (from

http://www.stormceptor.com/applications.php)…………………… 68

Figure 7.2 A typical Stormceptor (from

http://www.stormceptor.com/single_inline_unit.php)……………... 68

Figure 7.3a Available Stormwater treatment systems (CE News, 2003)………….. 69

Figure 7.3b Available Stormwater treatment systems (CE News, 2003)………….. 70

Figure 7.3c Available Stormwater treatment systems (CE News, 2003)…………… 71

Figure 7.4 Layout of dynamic section view before simulation……………………. ..72

Figure 7.5 Storage capacity available and the head developed

at time 8:25:20……………………………………………………... 73

Figure 7.6 Fluctuation of flow with time on pipe conduit L3

using SWMM………………………………………………………. ..74

Figure 7.7 SWMM output of time distributed flow in pipe conduit L4…………… ..74

Figure 7.8 SWMM output of time distributed flow in pipe conduit L2…………… ..75

ix
Figure 7.9 Dynamic section view of flooding occurred in Node 3 due
to a smaller pipe size……………………………………………….. ..76

Figure 7.10 Dynamic section view at the end of simulation due to

a smaller pipe size……………………………………………………… 76

Appendix A Page No.

Figure A.1 Rainfall data input as table form in StormCAD……………………….. 85

Figure A.2 Watershed (catchment) setting and simulation results

for Inlet I-1…………………………………………………………. 86

Figure A.3 Inlet setting and simulation results for Inlet I-1………………………... 87

Figure A.4 Flow simulation results for Inlet I-1…………………………………… 88

Figure A.5 Watershed (catchment) setting and simulation results

for Inlet I-2…………………………………………………………. 89

Figure A.6 Inlet setting and simulation results for Inlet I-2………………………... 90

Figure A.7 Flow simulation results for Inlet I-2…………………………………… 91

Figure A.8 Pipe settings and flow simulation results in pipe P-1………………….. 92

Figure A.9 Pipe settings and flow simulation results in pipe P-2………………….. 93

Figure A.10 Rainfall data input using equations in StormCAD ……………………. 94

Appendix B Page No.

Figure B.1 Network setup of the case study in Hydraflow………………………… 95

Figure B.2 Rainfall IDF Curve generated in Hydraflow…………………………... 96

Figure B.3 Design constraints and initial condition setup in

Hydraflow………………………………………………………….. . 97

Figure B.4 Calculation option dialog in Hydraflow……………………………….. 97

Appendix C Page No.

Figure C.1 Simulated surface flow components at the inlet I-1

x
 (Node 1) by using modified SPLIT program………………………. 101

Figure C.2 Simulated subsurface flow components at the inlet

I-1 (Node 1) by using modified SPLIT program…………………... 102

Figure C.3 Simulated surface flow components at the inlet I-2

(Node 2) by using modified SPLIT program………………………. 103

Figure C.4 Simulated subsurface flow components at the inlet

I-2 (Node 2) by using modified SPLIT program…………………... 104

Figure C.5 Simulated surface flow components at the inlet I-3

(Node 3) by using modified SPLIT program………………………. 105

Figure C.6 Simulated surface flow components at the inlet I-2 (Node 2)
by using modified SPLIT program (Tc= 5 min.)…………………... 106

Appendix D Page No.

Figure D.1 Network setup of case study in CAiCE SWMM………………………. 107

Figure D.2 Infiltration Dialog box in SWMM……………………………………... 108

Figure D.3 Input parameters for Horton Equation in SWMM……………………... 108

Figure D.4 Rainfall data as constant time interval for duration

of 10 minutes………………………………………………………. 109

Figure D.5 Sub-catchment input parameter dialog box……………………………. 110

Figure D.6a Different sub-catchment shapes to illustrate effect

of sub-catchment width (Huber, 1988.)…………….……………… 111

Figure D.6b Sub-catchment hydrographs for different shapes of

Figure D.6a (Huber, 1988)……….…………………………….…... 111

Figure D.6c Sub-catchment hydrographs developed by Kinematic

Wave Method with different basin widths…………………………. 112

Figure D.7 Unit hydrograph methods available in SWMM………………………... 113

Figure D.8a Runoff interface files selection dialog box…………………………….. 114

Figure D.8b Interface file selection dialog box in Extran mode…………………….. 114

Figure D.9 Node data input dialog box in Extran layer……………………………. 115

xi
Figure D.10a Rating curve for the curb inlet of L = 10 ft…………………………….. 116

Figure D.10b Graphical view of rating curve in Figure D.10a……………………….. 116

Figure D.11 Gutter section defined in case study to carry non

captured flow downstream…………………………………………. 117

Figure D.12 Hydrograph output of SWMM for Node N1…………………………... 118

Figure D.13 Time series plot of flows and velocity by SWMM

for conduit L1……………………………………………………… 118

Figure D.14 Time series flow output by SWMM for Node-N2……………………... 119

Figure D.15 Time series plot of flows and velocity by SWMM

for conduit L4……………………………………………………… 119

Figure D.16 SWMM output for Node-N2 solved in Runoff layer

only (Tc = 10 minutes)……………………………………………... 120

Figure D.17 SWMM output for Node-N3 solved in Runoff layer only…………….. 121

Figure D.18 SWMM output for Node-N4 solved in Runoff layer only…………….. 121

Figure D.19 SWMM output for Node-N2 solved in Runoff layer only
(Tc = 5 minutes)……………………………………………………. 122

Figure D.20 Time series output of flow, velocity and elevation for
conduit L-1 resulted from Extran layer of SWMM………………... 123

Figure D.21 Time series output of flow, velocity and elevation for
conduit L-2 resulted from Extran layer of SWMM………………... 123

Figure D.22 Time series output of flow, velocity and elevation for
conduit L-3 resulted from Extran layer of SWMM………………... 124

Figure D.23a Time series output of flow, velocity and elevation for conduit L-1 resulted
from Extran layer of SWMM for a Maximum capacity of 3.0 cfs for
inlet (Time step for simulation 5 seconds)………………………… 125

Figure D.23b Time series output of flow, velocity and elevation for conduit L-1 resulted
from Extran layer of SWMM for a Maximum capacity of 3.0 cfs for
inlet (Time step for simulation 60 seconds)……………………….. 125

xii
Figure D.24 Time series output of flow, velocity and elevation for conduit L2 resulted
from Extran layer of SWMM for a maximum interception capacity of 3.0
cfs for inlet………….…………………………………………………. 126

Figure D.25 Variation in flow and elevation with respect to time in

the surface conduit Gutter 1………………………………………... 127

Figure D.26 Variation in flow and elevation with respect to time in the
surface conduit Gutter 2……………………………………………. 127

Figure D.27 Time series output of flow, velocity and elevation for
conduit L-1 resulted from Extran layer of SWMM
using rating curve as inlet restriction………………………………. 128

Figure D.28 Time series output of flow, velocity and elevation for
conduit L-1 resulted from Extran layer of SWMM
using rating curve as inlet restriction………………………………. 129

Figure D.29 Variation in flow and elevation with respect to time

in the surface conduit Gutter 1 using rating curve…………………. 129

Figure D.30 Variation in flow and elevation with respect to time

in the surface conduit Gutter 2 using rating curve…………………. 130

Figure D.31a Simulated hydrographs using SCS Method in

CAiCE SWMM……………………………………………………..131

Figure D.31b Simulated hydrograph using SCS Method in

XP SWMM………………………………………………………… 132

Figure D.32 Time series output from SWMM for conduit L-1……………………... 133

Figure D.33 Time series output from SWMM for conduit Gutter1…………………. 133

Figure D.34 Time series output from SWMM for conduit L-2……………………... 134

Figure D.35 Time series output from SWMM for conduit Gutter 2………………… 134

Figure D.36 Time series output from SWMM for conduit L-3……………………... 135

Figure D.37 Time series output from SWMM for conduit Gutter 3………………… 135

Figure D.38 Time series output from SWMM for conduit L-4……………………... 136

xiii
 DISCLAIMER

The contents of this report reflect the views of the authors, who are responsible
for the facts and accuracy of the data presented herein. The contents do not necessarily
reflect the official view or policies of the Texas Department of Transportation (TxDOT).
This report does not constitute a standard, specification, or regulation. The United States
government and the State of Texas do not endorse products or manufacturers. Trade or
manufacturer’s names appear herein solely because they are considered essential to the
object of this report. The researcher in charge of this project was David Thompson.
There was no invention or discovery conceived or first actually reduced to
practice in the course of or under this contract, including any art, method, process,
machine, manufacture, design, or composition of matter, or any new useful improvement
thereof, or any variety of plant, which is or may be patentable under the patent laws of
the United States of America or any foreign country.

xiv
 ACKNOWLEDGEMENT

The authors would like to thank the project director, James J. Mercier, who has
provided needed guidance and assistants. Special thanks go to the program coordinator,
David Stolpa, and the project advisors, George (Rudy) Herrmann and Michael Stan,
whose recommendations have been very helpful. Numerous other TxDOT personnel also
took time to provide insights and information that assisted the project, and we sincerely
appreciate their help. Finally, we wish to express our appreciation to TxDOT for its
financial sponsorship of the project.

xv
 CHAPTER ONE
INTRODUCTION

1.1 Hydrological Perspectives of Storm Drainage Systems

The study of flood prevention and mitigation is a focus in both hydrology and
hydraulics. The storm-induced flood is the most severe and frequent natural flood
disaster in the world. A rainstorm may generate a large rate of surface runoff in the short
period of time in response to high-intensity rainfall. The resulting runoff cannot be
drained quickly and leads to water accumulation and flooding in streets, roads and
residential areas. Storm drain systems are typically designed to carry flow from a rainfall
event away from areas where it is unwanted (such as parking lots and roadways).
Flooding occurs when either a heavy storm that exceeds the design criteria of the
structure or inadequate capacity to drain flood flows exists. Thus stormwater drainage
(storm sewers) design is an important part of civil engineering. Appropriate drainage
design should maintain compatibility and minimize interference with existing drainage
patterns; control flooding of property, structures and roadways for design flood events;
and minimize potential environmental impacts in stormwater runoff. Stormwater
collection systems must be designed to provide adequate surface drainage while at the
same time meeting other stormwater management goals such as water quality, stream
bank channel protection, habitat protection and groundwater recharge.
The development of the Rational formula and the Manning formula in the late
1880’s represents two major advances in modern hydrology. Gradually, hydrologists
replaced empiricism with Rational analysis and observed data to solve practical
hydrological problems. Green and Ampt (1911) developed a physically based model for
infiltration; Sherman (1932) devised the unitgraph (unit hydrograph) method to transform
effective rainfall to the direct runoff hydrograph; Horton developed infiltration theory
(1933) and a description of drainage basin form (1945); and Gumbel (1941) proposed the
extreme value law for hydrologic studies.
Storm sewers, in general, can be divided into two categories according to the
functional classification as “separate system” that carries only the stormwater from the
roadways and adjacent areas (Figure 1.1) and “combined sewers”, which carry waste
water from residences, offices, industrial complexes, etc (Figure 1.2) during dry periods
and which also convey stormwater during rainfall events. This study mainly focuses on
the “separate system” or the stormwater drainage system. The principal or the major
hydraulic components in stormwater drainage system include (Figures 1.1 and 1.2):
• Inlets, structures that pass runoff from the surface of the land into the closed
conduit system,
• Conduits, structures that convey water received from the inlets from one
location to another,
• Junctions, structures that connect adjacent conduits,

xvi
 • Manholes, structures that provide ventilation and access for inspection and
maintenance, and
• Outfalls, structures that release the runoff into a surface drainage system such
as a manmade channel or natural stream.
All of above appurtenances are essential components of a storm sewer system, and each
exerts some influence over the system performance. The features that this study most
closely investigated are the inlets and the conduits. Junctions and manholes typically
involve little control of flow, and outfalls exert control only when partially or fully
submerged by water in the receiving stream.

Figure 1.1 The urban drainage system (from Proctor and Redfern, 1976).

2
Figure 1.2 Combined urban drainage system (from Metcalf and Eddy et al., 1971).

The design of stormwater drainage is generally accomplished by rainfall

modeling, runoff modeling, followed by the design of conduits or pipes and
appurtenances. Before designing or evaluating any stormwater drainage system, the
engineer must first determine the acceptable level of risk of failure (in the hydrologic
sense), which is expressed in the return interval, or the average length of time (in years,
taken over a long period of time) between subsequent hydrologic failures. From the
return interval, and from an analysis of the contributing watershed, storm duration and
depth are determined, from which are computed the discharges that the system must
convey.
Rainfall information can be obtained from a variety of sources, most likely
intensity-duration-frequency relations or from rainfall atlases such as HYDRO-35
(Frederick et. al. 1977) or TP-40 (Hershfield, 1961). Once the hyetograph for a design
has been developed, it is used to compute runoff rates to be used in the drainage system
design. Currently, there are numerous hydrological methods available for computing
peak flows, developing runoff hydrographs, and routing hydrographs. Some of them
include Rational method, Modified Rational method, SCS/NRCS method, Clark method,
Snyder method, Kinematic Wave method, EPA Runoff method, Nash method, and SBUH

3
(Santa Barbara Urban Hydrograph) Method. The choice and selection of method
depends on geographic location, whether a hydrograph is required or only a peak
discharge is needed, available data, and available resource.
Flow of water through a conduit is said to be closed conduit flow or open channel
flow based on whether or not the surface of water is at atmospheric pressure. In closed
conduit flow, the cross-sectional area of the flow is the same as that of the inside of the
conduit and the hydraulic grade line is above the grade of the crown of the pipe.
However, in open channel flow, the cross-sectional area of the flow, and hence other flow
parameters such as the velocity, depend not only on the size and shape of the conduit, but
also on the depth of flow. As a consequence, open channel flow is more difficult to treat
from an analytical point of view than is pipe flow. Flow can be further classified as
steady and unsteady flow depending on change of flow parameters (e.g., depth) with
time. Although most collection system design is based on steady state flow hydraulics,
flow in storm sewers is inherently unsteady because of the nature of precipitation and
runoff transformation. A variety of methods exist for evaluating unsteady flow in
conveyance systems, ranging from the most sophisticated numerical solutions of the
Saint-Venant to simple approximation methods.
Today expenditure for urban drainage works and pollution control facilities are
among the largest items in the budget of most municipalities, and represent a significant
percentage of federal funding of public works. Widespread access to computers and the
commencement of sampling have led to the development of urban runoff models that
have been calibrated and validated by comparisons with measured data. The need for
comprehensive approaches for the simulation of flow quantities and the limitations of the
Rational method was recognized in the late 1950’s, with the development and application
of these models, even though hydrograph methods had been introduced much earlier.
The first uses of hydrologic models for urban flow simulation followed the development
of the Road Research Laboratory Model (RRL) in the United Kingdom, and the Chicago
Model in the U.S. (Watkins, 1962; Kiefer et al., 1970). Many models are used in the
U.S., such as the EPA’s Storm Water Management Model (SWMM), the Water
Resources Engineers (WRE) model, the University of Cincinnati model, Illinois Urban
Drainage Area Simulator (ILLUDAS), and HYDROCOMP (Brandstetter, 1977).
Computer models are important to engineers because they can be used to execute
engineering computations, either resulting in a savings of time and money, or by allowing
more complex analyses or more alternatives to be examined for the same cost in
resources. During the last 25 years there has been a proliferation of computer models
that can be used for various aspects of the design of stormwater collection, storage and
conveyance structures. Computer modeling became an integral part of hydrologic and
hydraulic design and analysis in the early to mid 1970’s when several federal agencies
began the development of software. Some private civil engineering software companies
also developed good computer models. Many of these computer models were developed
by adding more graphical user interface features to the existing governmental computer
models.

1.2 Background and Scope of the Study

4
 The Texas Department of Transportation (TxDOT) currently uses the Rational
method for development of design peak flow rates for its highway storm drainage design.
Watersheds for which it is used have drainage areas less than 200 acres. The Rational
method is an "instantaneous" peak discharge method that is popular due to its simplicity.
The Rational method assumes a linear relation between rainfall rate for the time of
concentration of the watershed and peak instantaneous discharge. The drawback of the
Rational method, however, is that the time distribution and accumulation of flow cannot
be precisely accounted for through each node (inlet) and run (conduit) of the system.
Instead, the accumulated effects of all contributing sub-basins and branches are assumed
to be “lumped” into a single equivalent basin when designing or analyzing each
successive node or run. Use of peak discharge values also limits the hydraulic design and
analysis of the system to the assumption of simple steady-state flow conditions. While
this may result in a simple design process, the inability to consider unsteady flow and the
inherent storage available in these systems may result in the missed opportunity to
develop more cost-effective designs. Simple steady-state flow assumptions may also be
inadequate to address the complex hydraulics that could be associated with the need to
include non-traditional hydraulic features, such as in-line water quality basins.
Therefore, the proposed study is intended to be the first phase in evaluating TxDOT
procedures for storm drain design, not only in terms of the adequacy of current TxDOT
practice relative to new directions in the field, but also in anticipation of the need to
evaluate more complex features that might be required by changes in water quality
regulations.
The study is accomplished by completing two tasks: (1) a literature review to
synthesize both the technical approach (Rational method versus other hydrological
methods) and modeling efforts of drainage networks with various computer software
packages, and (2) use of modeling tests on simple cases to examine storm drainage
design. The study performed is summarized in this report as follows: Chapter Two
provides not only a thorough literature review on Rational method, Modified Rational
method, and inlet design, but also a review of recent journal papers dealing with storm
drainage design. Computer software packages to analyze the stormwater system are
presented in the Chapter Three. Chapter Four provides the basic setup of the case study.
Results of the case study are summarized in the Chapter Five, which presents model
results of the simple drainage system by using different software packages like
WinStorm, StormCAD, Hdyraflow and SWMM. Detailed information on model setup
and some intermediate results of the case study are presented separately in the Appendix
A to Appendix D. This case study allowed the researchers to examine technical
approaches implemented in these computer models, and developed useful results and
conclusions on storm drainage design procedures. For example, most of the simple
models for storm drainage design, for example, WinStorm and Hydraflow, exhibit a
conceptual disconnect1 between the sizing of inlets and the design of the pipeline
network. The study of unsteady processes from rainfall and runoff was examined by

1
The conceptual disconnect referred to here occurs the inlet design approach allows some of the incoming
flow to bypass or carryover from one inlet to the next. This avoids the requirement of very long inlet
lengths. However, in the conduit design procedure, all flow from the subwatershed is assumed to enter the
system through the inlet. Therefore, one set of flows is used to design the inlets and a second, and greater,
set of flows is used to design the conduit.

5
performing SWMM simulation on the simple system. Furthermore, part of the storm
drainage system designed for U.S. Highway 77/83 was tested under different return
periods and are presented in the Chapter Six, and this led us to conclude conservative
design does exist in many TxDOT storm drainage systems. Feasibility of in-line water
quality treatment by using extra capacity of over-designed conduits is also explored and
discussed on the Chapter Seven. Chapter Eight addresses summary and conclusions of
the study.

6
 CHAPTER TWO
LITERATURE REVIEW

2.1 Rational Method for Storm Drain Design

2.1.1 Introduction
Early stormwater or catchment runoff estimation throughout the world was based
on designer’s experience and judgment. Current practice is that the watershed that is to
be drained by a proposed storm sewer system will be generally divided into one or more
sub-catchments or sub-watersheds that are of reasonable size and are approximately
homogeneous in nature. These watersheds may include residential, commercial or
industrial areas, but usually have larger proportions of pavement and the streets and roads
which are the principal surface drainage conveyance, have short time of concentration,
and have well-defined flow paths, typically through gutters, ditches and medians of
streets and roads. Mr. Emil Kuichling, City Engineer of Rochester, New York,
developed a method (Rational formula) based on his measurements on five sub-basins in
Rochester, ranging in size from 25 to 357 acres (Kuichling, 1889). Based on his
measurements, he concluded the following:
1. Runoff volume is proportional to imperviousness.
2. Maximum discharge occurs when the rainfall lasts long enough for the entire
watershed area to contribute flow.
3. Peak discharge is proportional to intensity of rainfall.
4. Antecedent moisture levels are likely to have a significant effect on peak flow.
Now known as the Rational method, the technique developed by Kuichling is
used extensively in the United States and has encountered little change since its original
development.

2.1.2 Assumptions of Rational Method

The following assumptions (Mays, 2001) are generally made when one applies
the Rational formula:
• The rainfall intensity is constant with respect to time.
• The rainfall intensity is constant with respect to space over the watershed drainage
area.

• The frequency distributions of the event rainfall and the peak runoff rate differ in
mean value but have the same variance (are parallel if plotted in probability
space).
• The time of concentration of a basin is constant and is easily determined.

7
 • Despite the natural temporal and special variability of abstractions from rainfall,
the percentage of event rainfall that is converted to runoff can be estimated
reliably.
• The runoff coefficient is invariant, regardless of season of the year or depth or
intensity of rainfall.

2.1.3 Rational Formula

The Rational method is the most frequently used urban hydrology method. It is
used to estimate the peak instantaneous discharge from the watershed, and it is assumed
that the peak runoff rate is proportional to the peak intensity of rainfall multiplied by the
contributing area. The constant of proportionality is called a “runoff coefficient”, always
lesser than unity.
Mathematically Rational formula is represented as
Q = Cd Cr i A (2.1)
where, Q = maximum runoff rate (ft3/s in English units, m3/s in SI units),
Cr = the runoff coefficient (dimensionless),
Cd = dimensional correction factor (1.008 in English units, 1/360 = 0.00278 in SI
units),
i = average rainfall intensity (inches/hour in English units, mm/hour in SI units),
A = contributing watershed area (acres in English units, hectares in SI units).

2.1.3.1 Runoff Coefficient

The runoff coefficient Cr is the variable of the Rational method least susceptible
to precise determination and requires judgment and understanding on the part of the
design engineer. While engineering judgment will always be required in the selection of
runoff coefficients, typical coefficients represent the integrated effects of many drainage
basin parameters. Recommended runoff coefficients for the Rational method are
presented on Table 2.1.
Where watershed is not homogeneous but is characterized by dispersed areas that
can be characterized by different runoff coefficients, a weighted runoff coefficient should
be determined. The weighting is based on the area of each land use and is computed
using the following equation:
n

∑C
j =1
j Aj
Cw = n
(2.2)
∑A j =1
j

where Aj is the area for land cover j,

8
 Cj is the runoff coefficient for area j,
n is the number of distinct land covers within the watershed, and
Cw is the weighted runoff coefficient.

2.1.3.2 Rainfall Intensity, Rainfall Duration and Time of Concentration

Historic rainfall data are compiled and analyzed to predict storm characteristics.
Based on previous statistical analysis, as the duration of rainfall increases, the average
intensity of that rainfall tends to decrease. This means that a short burst of rainfall, while
it might not result in a greater depth of rainfall, will, in general, have a greater intensity
than a longer burst. Rainfall-runoff models typically require determination of the time of
concentration or a similar timing parameter. In the Rational method, design rainfall
intensity is a direct function of the time of concentration. Time of Concentration is
defined as the length of time it takes for water to travel from the hydraulically most
remote point in a basin, sub-watershed, or watershed to the outlet. There is no practical
way of measuring the time of concentration (Mays, 2001). From elementary
considerations of free-surface flow (that is. velocity of flow increases with increasing
depth of flow), we know that for any given storm duration, greater rainfall depths will
induce greater depths of flow in the drainage network, and travel times through the basin
that will be less than those that will occur during smaller, more frequent rainfall events.
Time of concentration is the sum of the flow time for overland or sheet flow, which
occurs in headwater areas; the time for shallow concentrated flow (swales, natural
channels), which occurs immediately downstream of overland flow; and the flow time for
open channel or sewer flow, which tends to occur in the lower reaches of a tributary area
(US Department of Agriculture, 1986). However, sometimes only one or two of these
components exists. Therefore, in general, it can be said that estimation of time of
concentration requires significant engineering judgment.
Table 2.1 Values of runoff coefficient for Rational Formula (ASCE, 1992).
Land Use C Land Use C
Business: Lawns:
Downtown areas 0.70 – 0.95 Sandy soil, flat, 2% 0.05 - 0.10
Neighborhood areas 0.50 - 0.70 Sandy soil, avg., 2-7% 0.10 - 0.15
Sandy soil, steep, 7% 0.15 - 0.20
Residential: Heavy soil, flat, 2% 0.13 - 0.17
Single-family areas 0.30 – 0.50 Heavy soil, avg., 2-7% 0.18 - 0.22
Multi units, detached 0.40 – 0.60 Heavy soil, steep, 7% 0.25 - 0.35
Multi units, attached 0.60 – 0.75
Suburban 0.25 - 0.40 Streets:
Asphalt 0.70 - 0.95
Industrial: Concrete 0.80 - 0.95

9
 Light areas 0.50 – 0.80 Brick 0.70 - 0.85
Heavy areas 0.60 – 0.90
Unimproved areas 0.10 - 0.30
Parks, cemeteries 0.10 – 0.25 Drives and walks 0.75 - 0.85
Playgrounds 0.20 – 0.35 Roofs 0.75 - 0.95
Railroad yard areas 0.20 – 0.40

There are several different methods for estimating the time of concentration.
Because available procedures are based on a wide variety of hydrologic and hydraulic
conditions, the selection of a procedure or procedures for a given sub-basin should
include comparison of the hydrologic-hydraulic characteristics of the sub-basin to the
hydrologic-hydraulic characteristics of the sub-basins used to develop the time of
concentration. Generally, the disparity between estimates of time of concentration by the
various methods decreases as basin area decreases (Mays, 2001).
One commonly used method for estimation of time of concentration for overland
flow, concentrated flow, and conduit flow respectively is discussed here. The overland
flow is usually estimated using Kerby/Hathaway equation. The Kerby/Hathaway
equation is an empirical relation developed by Kerby (1959) on the basis of published
research on airport drainage done by Hathaway (1945).
0.467
⎡ 0.67 N L0 ⎤
t0 = ⎢ ⎥ (2.3)
⎢⎣ S 0 ⎥⎦

where t0 = travel time of overland flow, minutes,

N = overland flow resistance factor, dimensionless,
L0 = length of the overland flow segment, feet, and
S 0 = slope of overland flow segment, feet vertical/feet horizontal.
The distance L0 is recommended to be less than 300 feet (SCS, 1986; Mays, 2001)
or up to 525 ft (TxDOT, 2002). The variable, N, is analogous to Manning’s coefficient of
friction. The usual values of Manning’s n that are considered for various surfaces and
channel linings for channelized flow should not be used for overland flow computations.
Table 2.2 was excerpted from Table 3.5, HEC-1 Flood Hydrograph Package, Users
Manual.

Table 2.2. Resistance Factor for Overland Flow.

Surface n value
Asphalt/concrete 0.05-0.15
Bare packed soil, free of stone 0.10

10
Poor grass cover on moderately rough round 0.30
Light turf 0.20
Average grass cover 0.40
Dense turf 0.17-0.80
Dense grass 0.17-0.30
Bermuda grass 0.30-0.48

One of the more common methods for estimating the travel time of concentrated
flow is by Kirpich (1945) equation. The equation is a power function and is given by
−0.385
⎛ ΔH ⎞
t ch = 0.0078 L 0.77
⎜ ⎟ (2.4)
⎝ L ⎠
where tch = travel time of channelized flow, minutes,
L = length of channelized flow reach, feet, and
ΔH = the difference in elevation (feet) between the upper and lower ends of the
channelized reach of length L (slope).

Flow time in channels, gutters, and closed conduits can be computed by using
Manning’s equation to compute flow velocity V (ft/s).
L Ln
t lc = = (2.5)
60 V 60 × 1.49 R 2 / 3 S 1 / 2

where tlc = time of flow traveling in the lined conduit/channel, minutes,

n = Manning’s roughness coefficient, dimensionless,
R = hydraulic radius of the lined conduit/channel, feet, and
S = longitudinal slope of the lined conduit, dimensionless.
In the selection of a time of concentration for any individual component, the
engineer should probably use two or three familiar methods and for which the necessary
independent variables and parameters exist or can be economically determined. A
representative value can then be selected from the estimates.
Rainfall data are available from a variety of resources, including governmental
organizations and agencies. These can be presented in various formats like Intensity-
Duration-Frequency curves (IDF curves) or depth-duration frequency curves (DDF
curves), cumulative rainfall depths, and rainfall hyetographs. The Intensity-Duration-

11
Frequency curves (IDF curves) or depth-duration frequency curves (DDF curves) are
plots of rainfall intensity (or depth) versus duration of event rainfall. Usually, there are
several curves on a single graph, one for each of several different rainfall frequencies
(return periods). These curves are hyperbolic or exponential decay type curves, which
vary by geographical location (e.g. by county), and for many counties such relationships
have been developed for the use of designers. The general shape of an Intensity-
Duration-Frequency (IDF) curve is shown on Figure 2.1, and illustrates the average
rainfall intensities corresponding to a particular storm recurrence interval for various
storm durations.

Figure 2.1 Intensity-Duration-Frequency (IDF) curves.

Mathematically these curves can be represented in different forms as follows

a
i= (2.6)
(b + D) n

a ( RP ) n
i= (2.7)
(b + D) n

i = a + b(ln D) + c(ln D) 2 + d (ln D) 3 (2.8)

where i is rainfall intensity, in/hr or mm/hr,
D is duration of rainfall in minutes,

12
 RP is the Return Period (yr), and
The parameters, a, b ,c ,d, m, n define the shape and appropriate units, and are
determined for curve fitting to IDF data. For example, TxDOT uses the Equation (2.6) to
compute rainfall intensity for its hydrologic design, coefficients a, b, and n are available
for each county in Texas and for return periods of 2-, 5-, 10-, 25-, 50-, and 100-years,
respectively. This is also implemented in TxDOT’s design software WinStorm 3.0.

2.1.4 Limitation of Rational Method

Though simple to use, the Rational method does have limitations. First is the
runoff coefficient. The runoff coefficient incorporates a large number of variables into a
single index that ranges between 0 and 1. In addition, the parameter can take on a wide
range of values even for the same land use characteristics. As a result, there appears to
be an amount of subjectivity in selecting a runoff coefficient.
Secondly, the Rational method is simplistic in its accounting of runoff and loss
processes, so must be limited to small, relatively homogenous and simple watersheds,
usually less than 200 acres (80 ha.), which typically have times of concentration of less
than 20 minutes (ASCE, 1992; Wanielista et al., 1997).
A final limitation of the Rational method is that it only provides a single value on
the discharge hydrograph, the peak discharge. If the objective is to determine the size for
an inlet or a pipe, then this single point on the discharge hydrograph is adequate.
However, to design a detention basin one must have the direct runoff hydrograph, that is,
a time history of the runoff. Therefore, the Rational method does not give any time-
distributed information in any sense.

2.2 Hydrograph Generation Methods

2.2.1 Hydrograph Development for Inlets
The Rational method only provides designer a peak discharge, not a hydrograph.
Use of the peak discharge limits the hydraulic design and analysis of storm drain systems
to the assumption of steady-state flow conditions. In order to consider unsteady flow and
inherent storage available in storm drain systems for cost effective design, hydrographs at
inlets and inside pipes are required. A hydrograph is a time series of instantaneous
discharge versus time at a particular location within a watershed. Hydrographic analysis
is performed when flow routing is important, such as in the design of stormwater
detention, water quality facility and pump stations. It can also be used to evaluate flow
routing through large storm drainage systems to more precisely reflect flow peaking
conditions in each segment of the system.
Most approaches to hydrograph estimation are based on the concept of the unit
hydrograph, first introduced by Sherman (1932), which is a hydrograph produced by a
unit depth of runoff distributed uniformly over a basin for defined period of time. The
basic theory rests on the assumption that the runoff response of a drainage basin to an
effective rainfall input is linear; that is, it may be described by a linear differential
equation or the concepts of proportionality and superposition can be applied. Unit

13
hydrographs can be developed from rainfall and runoff data of a watershed, and can
typically be applied to the same watershed for other rainfall events. For un-gaged basins,
synthetic unit hydrographs can be developed from theoretical or empirical formulas
relating hydrograph peak flow and time characteristics of the basin to watershed or
rainfall characteristics, or can be transposed from nearby, hydrologically similar
watersheds, for which exist rainfall-runoff data. However, the synthetic unitgraphs have
certain limitation and the engineer or the hydrologist should apply them with caution to
new areas. Extensive literature exists on various methods (e.g., Clark, Snyder, Nash,
SCS.) to develop synthetic unit hydrographs. A brief discussion on two popular methods
of generating synthetic hydrograph is given here.
Snyder’s Method: Snyder (1938) was the first to develop a synthetic unit
hydrograph based on a study of watersheds in the Appalachian Highlands. It allows
computation of lag time, time base, unit hydrograph duration, peak discharge, and
hydrograph time widths at 50 and 75 percent of peak flow. By using these seven points,
a sketch of the unit hydrograph is obtained and checked to see if it contains 1 unit of
direct runoff.
Dimensionless SCS Unit Hydrograph: The method developed by Soil
Conservation Service (SCS) is a dimensionless unit hydrograph developed by Victor
Mockus (NRCS, 1972), derived from a large number of unit hydrographs ranging in size
and geographic location. The hydrograph is represented as a simple triangle with rainfall
duration, time of rise, time of fall and peak flow. This method requires only the
determination of the time to peak and the peak discharge.
Methods for developing synthetic unit hydrographs discussed in hydrology
textbooks are typically for relative large watersheds. For example, the relation of
Snyder’s method is considered applicable to drainage areas ranging in size from 10 to
10,000 mile2. Some of the stormwater simulation models, for example, XP or Visual
SWMM (which will be introduced later), also use Snyder’s method, SCS method, and
other unit hydrograph methods to develop hydrographs from local catchments (few acres
or more) for inlets of a storm drain system. Designer and engineers should pay attention
while applying those methods. In the next section, modified Rational method is
discussed in detail, which has been widely used for developing hydrograph for inlets.

2.2.2 Modified Rational Method

The modified Rational method (MRM) is an extension of the traditional Rational
method. The MRM produces a runoff hydrographs, and hence the runoff volume while
the original Rational method is meant to produce only the peak design discharge. The
MRM, which has found widespread use in the engineering practice in recent years, is
used to size detention/retention facilities for a specified recurrence interval and
concurrent release rate.
The MRM is based on the same assumptions as the conventional Rational
method. An important additional assumption is that the rainfall intensity averaging time
used in the modified Rational method equals the storm duration (“Urban Surface Water
Management” by Walesh, 1989). This assumption means that the rainfall, and the runoff

14
generated by that rainfall, occurring before or after the rainfall averaging period is not
accounted for.
In the MRM, it is also assumed that an urban stormwater runoff hydrograph under
the design storm can be approximated as being either triangular or trapezoidal in shape.
The rising and the falling limbs follow a linear time-area relationship trend for the sub-
basin, that is, their contributions are linear.
Three different possible types of hydrograph can be developed for the given sub-
basin using the MRM. Hydrograph type is a function of the storm duration or the length
of rainfall, d, with respect to the time of concentration, tc. The following three types are
possible (“Urban Surface Water Management” by Walesh, 1989):
(a) If the storm duration (d) is greater than the watershed time of
concentration (tc), the resulting hydrograph is a trapezoidal in shape with
uniform maximum discharge as determined from the conventional
Rational method Q = C d C r i A for the difference between tc and duration
of storm. The linear rising and falling limbs each has duration of tc as
shown in Figure 2.2(a).
(b) If the storm duration (d) is equal to time of concentration (tc), there is a
rise to full contribution (peak), followed by a recession over tc back to
zero. The resulting hydrograph is triangular in shape as shown in Figure
2.2 (b) with a peak discharge of Q = C d C r i A .
Discharge

Time

Figure 2.2(a) Hydrograph when duration of rainfall is greater than tc.

15
 Discharge

Time

Figure 2.2(b) Hydrograph when duration of rainfall is equal to tc.

(c) If the storm duration (d) is less than the time of concentration (tc),
then the resulting hydrograph is trapezoidal in shape with a
maximum uniform discharge of Q p' = C d C r i A (d / t c ) from the end
of the storm (d) to the time of concentration tc. The linear rising
and falling portions of the hydrograph each has a duration of d as
show in Figure 2.2(c).

Thus, the MRM can obtain the time-distributed discharge, which is useful to
predict downstream flooding or to determine the size of the detention basin. The
designers should limit use of the MRM for sizing detention basins for watershed drainage
areas not exceeding 20 to 30 acres (Mays, 2001).

16
 Discharge

Time

Figure 2.2(c) Hydrograph when duration of rainfall is less than tc.

2.3 Design of Stormwater System (Inlets)

2.3.1 Introduction
When rain falls on a sloped pavement surface, it forms a thin layer of water that
increases in thickness as it flows to the sides of the roadway. This accumulation of water
can disrupt traffic flow, reduce vehicular skid resistance, increase potential for motorist
hydroplaning, and contribute to pavement deterioration. The objective in highway
drainage design is to minimize such problems by collecting runoff in gutters and
intercepting runoff using stormwater inlets that direct flow to subsurface conveyance
systems, culverts, or ditches. Proper design of drainage facilities is therefore essential to
maintaining safe vehicular travel conditions and ensures that highway service levels will
avoid disruption.
This section of the report provides the guidelines for evaluating roadway features
and design criteria as they relate to gutter and inlet hydraulics and storm drain design.
Procedures for performing gutter flow calculations are based on a modification of
Manning's equation. For detailed discussion on inlet capacity calculations, the reader is
referred to Federal Highway Administration guidance documents, including Hydraulic
Engineering Circular (HEC) No. 12 and HEC No. 22. Storm drain design is based on the
use of the Rational formula.

2.3.2 Design Frequency and Spread

Two of the major design variables considered in sizing and locating highway
drainage structures are the frequency and the allowable spread of water on the pavement.

17
Spread and design frequency are interrelated, because the implication of allowable spread
can be significantly different for storms of different recurrence intervals (Brown et al.,
1996). Thus, the main objective is to collect runoff in the gutter and convey it to inlets in
a way that provides safety for traffic during the design storm event at a reasonable cost.
Selection of recurrence interval and spread for the design are dependent on the acceptable
risks and the budgetary limitation for the drainage system. The factors to be considered in
selecting design frequency and spread include highway classification, design speed,
traffic volumes, rainfall intensity and the capital cost. Moreover, it is the responsibility of
the designer to select a design frequency and spread that meets the needs of a particular
project. Suggested minimum design frequencies and spread based highway classification
and design speed are presented on Table 2.3.

Table 2.3. Suggested Minimum Design Frequency and Spread (Brown et al., 1996).
Road Classification Design Frequency Design Spread
<70 km/hr (45 mph) 10-year Shoulder + 1 m (3 ft)
High volume or divided
>70 km/hr (45 mph) 10-year Shoulder
or bi-directional
Sag point 50-year Shoulder + 1 m (3 ft)
<70 km/hr (45 mph) 10-year ½ Driving Lane
Collector >70 km/hr (45 mph) 10-year Shoulder
Sag point 10-year ½ Driving Lane
Low ADT (Average
5-year ½ Driving Lane
Daily Traffic)
Local Street
High ADT 10-year ½ Driving Lane
Sag Point 10-year ½ Driving Lane

2.3.3 Curbs and Gutters

A curb serves as the outside edge of pavements and performs multiple functions,
such as the following
• Act as a boundary between the roadway and the adjacent properties
• Provide pavement delineation
• Prevent erosion
A gutter is a section of pavement adjacent to the curb that is designed to convey
water to curb inlets during a runoff event. The gutter may include a portion or all of a
traffic lane. Gutter cross slopes may be the same as that of the pavement or may be
designed with a steeper cross slope, usually 80 mm per meter (1 inch per foot). Gutter
sections can be categorized as conventional or shallow swale type, as shown in
Figure 2.3. Conventional curb and gutter sections usually have a triangular shape with
the curb forming the near-vertical leg of the triangle. Conventional gutters may have a
straight cross slope, a composite cross slope where the gutter slope varies from the
pavement cross slope or a parabolic section. Shallow swale gutters typically have V-

18
shaped or circular sections, as shown in Figure 2.3, and are often used in paved median
areas on roadways with inverted crowns.

Figure 2.3. Typical gutter sections (Brown et al., 2001)

2.3.4 Flow in Gutters with Uniform Sections
Flow in gutter varies with the gutter geometry and the definition of the wetted
perimeter. For a uniform section (flow in triangular channel), where the wetted perimeter
is assumed to be equal to the spread width of the water flowing in the gutter, gutter flow
can be derived from horizontal integration of the Manning’s velocity equation for an
increment of cross sectional width (Izzard, 1946). Assuming resistance due to the curb
face is negligible, a reasonable assumption for uniform cross slopes less than 10 percent,
the integration yields
K u 1.67 0.25 2.67
Q= Sx SL T (2.9a)
n
or in terms of spread T
0.375
⎛ Qn ⎞
T = ⎜⎜ 1.67
⎟
0.5 ⎟
(2.9b)
⎝ Ku Sx SL ⎠

19
 where
Ku = 0.376 (0.56 in English units),
n = Manning’s coefficient (Table 2.4),
Q = gutter flow rate m3/s (or cfs),
T = spread of water onto the pavement in m (or ft), or top width of flow,
Sx = gutter cross slope in m/m (or ft/ft), and
SL = longitudinal slope, or grade, of the highway in m/m (ft/ft).

Table 2.4. Manning’s n for Street and Pavement Gutters (FHWA, HDS-3).
Type of Gutter or Pavement Manning’s n
Concrete gutter, trowel finish 0.012
Asphalt pavements:
Smooth 0.013
Rough 0.016
Concrete gutter-asphalt pavement:
Smooth 0.013
Rough 0.015
Concrete pavement:
Float finish 0.014
Broom finish 0.16
For gutters with small slope, where sediment may accumulate, 0.02
increase above values of “n” by

Equation (2.9) includes an adjustment factor because the hydraulic radius is

incapable of fully describing a shallow, gutter cross section, particularly when the spread
can exceed 40 times the flow depth (ASCE, 1992). Subsequently, spread can be related to
flow depth at the curb, d, by
d = T Sx , (2.10)
where d is the depth of flow.
From Equation (2.10), we observe that the effects of cross slope on gutter
capacity can be relatively large. Therefore, the designer should balance the need for
steeper cross slopes for effective drainage and the need for flatter slopes for driver
comfort and safety. An acceptable range of pavement cross slopes for drainage practice,
as provided by AASHTO (1990), is shown on Table 2.5. Spread on the pavement and
flow depth at the curb are often used as criteria for spacing drainage inlets.

Table 2.5. Recommended Pavement Cross Slopes.

20
Surface type Range of cross slope
High-type surface
2- Lanes 0.015-0.020
3 or more lanes, each direction 0.015 minimum; increase 0.005-0.010 per lane;
0.040 maximum.
Intermediate surface 0.015-0.030
Low-type surface 0.020-0.060
Shoulders
Bituminous or concrete with curbs 0.020-0.060

2.3.5 Flow in Gutters with Composite sections

Design computations for composite gutter sections require additional
consideration of flow in the depressed section. The depression serves to capture more
flow from the gutter and thus increasing gutter capacity and inlet efficiency. Thus, the
total flow incorporating the depressed section flow is given by
Q = Qw + Qs (2.11)
where
Q = total gutter flow rate in m3/s (or cfs),
Qw = flow rate in the depressed section of the gutter m3/s (or cfs), and
Qs = flow capacity of the gutter section above the depressed section m3/s (or cfs).
Qs can be evaluated using Equation (2.9) if T is taken as only the spread over the
un-depressed portion of the gutter, Ts (Fig. 2.3). Equation (2.11) can be used in
conjunction with the following expressions for computing flow in a composite cross
section (Brown et al., 1996).
⎧ ⎫
⎪ ⎪
⎪ Sw / Sx ⎪
E 0 = 1 / ⎨1 + 2.67 ⎬ (2.12)
⎪ ⎡ Sw / Sx ⎤ ⎪
⎪ ⎢1 + (T / W ) − 1⎥ − 1⎪
⎩ ⎣ ⎦ ⎭
and
Qs
Q= (2.13)
(1 − E0 )
where
E0 = ratio of flow in a chosen width to total gutter flow (Qw/Q), and
Sw = cross slope of the depressed portion of the gutter in m/m (or ft/ft), and Sw is
expressed as

21
 a
Sw = Sx + , (2.14)
W
where a = depth of gutter depression in m (ft), and
W = width of the depressed section m (or ft) given in Fig. 2.3.

2.3.6 Drainage Inlet Design

The primary purpose of the storm drain inlet is to intercept all or a portion of the
flow as flow accumulates in gutters and spread encroaches upon pre-specified design
values, and to discharge it into an underground storm drainage conveyance system. The
design characteristics of inlets eventually control the rate at which runoff is removed
from the gutter and enters the storm drainage system. Subsequently, inadequate inlet
capacity or poorly located inlets can cause hazardous flooding to the traffic or the
property. Therefore, the responsibility of the designer is to determine the type, size, and
spacing of inlets to intercept a sufficient portion of the design gutter flow, while
preserving attention to cost. In addition, the designer should ensure that inlets do not
project significantly above a pavement surface or pose as an obstacle to oncoming traffic.
Inlets commonly used in practice for the drainage of highway surfaces include
(Figure 2.4)
1) Curb-opening inlets
2) Grate inlets
3) Slotted drain inlets
4) Combination inlets
Inlets can be further classified as being on a “continuous grade” or in “sump”.
The term “continuous grade” refers to an inlet so located that the grade of the street
(road) has a continuous slope past the inlet and therefore ponding does not occur at the
inlet. The sump condition exists whenever water is restricted to the inlet area because the
inlet is located at a low point. A sump condition also called as “inlets on sag”, can occur
at a change in grade of the street (road) from positive to negative or at an intersection due
to the crown slope of a cross street.

22
 Figure 2.4. Types of storm drain inlets (Brown et al., 2001).

2.3.6.1 Interception Capacity and Efficiency on Continuous Grade Inlet

Inlet capacity, Qi, is the amount of gutter flow intercepted by an inlet under a
given set of conditions, which is conveyed to the stormwater pipe. The efficiency of an
inlet, E, is the percent of total runoff that the inlet will convey to the underground pipe
for those conditions. The efficiency of an inlet depends on cross slope, longitudinal slope,
total gutter flow, inlet geometry, and, to lesser extent, pavement roughness. Whereas
interception capacity of all inlets increases with increasing gutter flow rates, efficiency
generally decreases with increasing gutter flow (Brown, et al., 1996). Mathematically,
efficiency E is defined by the following equation
Qi
E= (2.15)
Q
where
E = inlet efficiency,
Q = total gutter flow in m3/s (or cfs), and

23
 Qi = intercepted capacity in m3/s (or cfs).
Any flow that is not intercepted by an inlet is termed carryover flow, or bypass
flow and is defined as follows:
Qb = Q – Qi (2.16)
where
Qb = bypass flow in m3/s (or cfs),
Q = total gutter flow in m3/s (or cfs), and
Qi = interception capacity in m3/s (or cfs).

2.3.6.2 Curb-opening Inlets

Curb-opening inlets are vertical opening in the curb covered by a top slab as
shown in Figure 2.4 (b). They are most effective on flatter slopes (less than 3%) and in
sags, and are less susceptible to clogging by debris. The primary factor affecting curb-
opening capacity and inlet efficiency are the depth of water next to the curb and length of
curb opening.
For uniform cross slopes, the length of curb-opening inlet on grade required to
intercept 100 percent of gutter flow can be expressed as
0 .6
⎛ 1 ⎞
Lt = K u Q 0.42
S ⎜⎜
0 .3
L
⎟⎟ (2.17)
⎝ nS x ⎠
where
Lt = curb opening length in m (or ft) required to intercept all of the gutter flow,
Ku = empirical constant equal to 0.817 (0.6 in English units),
Q = design discharge reaching gutter m3/s (or cfs),
SL = longitudinal slope, or grade, of the highway in m/m (or ft/ft),
Sx = gutter cross slope in m/m (or ft/ft), and
n = Manning’s roughness coefficient.
When actual length (L) of the curb is shorter than the length required for total

interception, the efficiency of the curb-opening inlets, is given by:

1.8
⎛ L⎞
E = 1 − ⎜⎜1 − ⎟⎟ . (2.18)
⎝ Lt ⎠
Increasing the cross slope tends to reduce the required length of curb opening for
total interception as it can be viewed from Equation (2.17). Moreover, the cross slope
can be increased using locally or continuously depressed gutter sections, as shown in

24
Figure 2.5. Therefore, the length of inlet required for 100 percent interception can be
computed by use of an equivalent cross slope, Se, in Equation (2.17) in place of Sx. The
term Se can be determined by
S e = S x + S w' E 0 , (2.19)
where
Eo = ratio of flow in the depressed section to total gutter flow determined by the
gutter configuration upstream of the inlet, defined in Equation (2.12), and
S w' = cross slope of the depressed section measured form the cross slope of the
pavement, m/m (or ft/ft) and can be expressed as
a
S w' = (Figures. 2.3 and 2.5). (2.20)
W

Figure 2.5. Depressed curb opening inlet (Brown et al., 2001).

Thus, for curb-opening inlets with less than 100 percent interception, depressed
sections can significantly increase the interception capacity and efficiency. Equation
(2.18), for calculating efficiency, is applicable for both uniform and composite cross
slopes.
Curb-opening inlets on Sump (Sag) can operate either as a weir or as an orifice.
They act as a weirs for ponding depth at the curb less than or equal to the height of the
curb opening (Brown et al., 1996). In this case, the equation for the interception capacity
of a curb-opening inlet is given by
Qi = C w L d 3 / 2 (2.21)
where
Cw = weir discharge coefficient 1.60 (3.0 in English units),
L = length of the curb-opening in m (or ft), and

25
 d = depth at curb measured form the normal cross slope in m (or ft).
For depressed curb-opening inlet (Fig. 2.5), the capacity is computed by
Qi = C w (L + 1.8W ) d 3 / 2 , (2.22)
where
W = lateral width of depression in m (or ft), and
Cw = weir discharge coefficient 1.25 (2.3 in English units).
The application of Equation (2.22) is limited to depths at the curb less than or
equal to the height of the opening plus the depth of depression. Curb-opening inlets
operate as orifices at depths greater than approximately 1.4 times the opening height.
The interception capacity for depressed or undepressed curb opening inlet operating as
horizontal orifice throat, as shown in Figure 2.6 (a), is computed by the following
equation
Qi = C 0 h L(2 g d 0 )
0.5
(2.23)
For other throat configurations as shown in Figure 2.6(b) and 2.6(c), this expression is
generalized as
1/ 2
⎡ ⎛ h ⎞⎤
Qi =C 0 Ag ⎢2 g ⎜ d i − ⎟⎥ , (2.24)
⎣ ⎝ 2 ⎠⎦
where
C0 = orifice discharge coefficient (0.67),
Ag = effective area of the curb opening, m2 (or ft2),
g = gravitational acceleration,
di = depth at lip of curb-opening, m (or ft),
h = height of curb-opening orifice, m (or ft), and
d0 = effective head on the center of the orifice throat, m (or ft).

2.3.6.3 Grate Inlets

Grate inlets consist of an opening in the gutter or ditch covered by one or more,
flush-mounted grates placed parallel to the flow, as shown in Figure 2.4 (a). The main
advantage of grate inlets is that they can be installed in the direct path of runoff. The
highly susceptible of grate inlet to debris clogging is however, the principal disadvantage.
Consideration should be given where the bicycle or pedestrian traffic occurs. The grates
for which design procedures have been developed are listed in Table 2.6 (Brown et al.,
1996).

26
 Figure 2.6. Curb opening inlets with different throats (Mays, 2001).

When the velocity approaching the grate is less than the “splash-over’ velocity,
the grate will intercept essentially all of the frontal flow and conversely, when the gutter
flow velocity exceeds the “splash-over” velocity for the grate, only part of the flow will
be intercepted (Brown et al., 1996). A part of the flow along the side of the grate will be
intercepted, dependent on the cross slope of the pavement, the length of the grate, and
flow velocity.

27
 Table 2.6. Types of grates for which design procedures is developed (Brown et. al.
1996).
P-50 Parallel bar grate with bar spacing 48 mm (1-7/8 in) on center.
P-50x100 Parallel bar grate with bar spacing 48 mm (1-7/8 in) on center and 10
mm (3/8 in) diameter lateral rods spaced at 102 mm (4 in) on center
P-30 Parallel bar grate with 29 mm (1-1/8 in) on center bar spacing
Curved Vane Curved vane grate with 83 mm (3-1/4 in) longitudinal bar and 108 mm
(4-1/4 in) transverse bar spacing on center
45°- 60 Tilt Bar45° tilt-bar grate with 57 mm (2-1/4 in) longitudinal bar and 102 mm (4 in)
transverse bar spacing on center
45°- 85 Tilt Bar45° tilt-bar grate with 83 mm (3-1/4 in) longitudinal bar and 102 mm (4 in)
transverse bar spacing on center
30°- 85 Tilt Bar30° tilt-bar grate with 83 mm (3-1/4 in) longitudinal bar and 102 mm (4 in)
transverse bar spacing on center
Reticuline "Honeycomb" pattern of lateral bars and longitudinal bearing bars

The ratio of frontal flow to total gutter flow, E0, for uniform cross slope can be
expressed by Equation (2.25):
8/3
Q ⎛ W⎞
E 0 = w = 1 − ⎜1 − ⎟ , (2.25)
Q ⎝ T ⎠
where
Q = total gutter flow, m3/s (or cfs),
Qw = flow in width W, m3/s (or cfs),
W = width of depressed gutter or grate, m (or ft), and
T = total spread, m (or ft).
Similarly, the ratio of side to gutter flow is expressed as:
Qs ⎛Q ⎞
= 1 − ⎜⎜ w ⎟⎟ = 1 − E 0 , (2.26)
Q ⎝ Q ⎠
where
Qs = side flow, m3/s (cfs).
The ratio of intercepted flow to total frontal flow, or frontal flow efficiency, Rf, is
expressed by:
R f = 1 − K f (V − V0 ), (2.27)

where
Kf = 0.0295 (0.09 in English units),

28
 V = velocity of flow in gutter, m/s (or ft/s), and
V0 = gutter velocity where splash over first occurs, also called as splash-over
velocity, m/s (or ft/s).
The frontal flow efficiency can be determined graphically using the curves in Figure 2.7,
which accounts for grate length, bar configuration, and gutter velocity at which splash
over occurs. The ratio of side flow intercepted to total side flow is expressed as:
1
Rs = , (2.28)
⎛ K s .V 1.8 ⎞
⎜⎜1 + ⎟
⎝ S x .L2.3 ⎟⎠
where
Ks = 0.0828 (0.15 in English units), and
L = Length of gutter section, m (or ft).
The overall efficiency, E, of the grate can be evaluated as a function of the frontal
and side flow efficiencies by using
E = R f E 0 + Rs (1 − E 0 ). (2.29)
Combination inlets (Figure 2.4 (c)) and the slotted inlets (Figure 2.4 (d)) are less
commonly used and are not discussed in detail here. The interested reader is suggested to
refer Chapter 4 of FHWA HEC-22 Manual for detailed discussion.

29
Figure 2.7. Grate inlet frontal flow interception efficiency (Brown et al., 1996).

30
2.4 Journal Publications on Storm Drainage System Design
2.4.1 Introduction
In the 1970s, one of the active water resources research areas was urban
stormwater drainage. It evolved from concern of urban flood mitigation, primarily with
respect to water quantity, to the concept of stormwater quality and quantity management.
As a partial response to the need, the first international conference on urban storm
drainage was held April 11-15, 1978, at the University of Southampton in England. The
second and third international conferences on urban storm drainage were held at Urbana,
Illinois, USA, on June 15-19, 1981, and in Göteberg, Sweden in June 1984. Dr. Yen
(1981a, 1981b) developed two volumes of proceedings for the second conference: the
first is “Urban Stormwater Hydraulics and Hydrology” containing 50 papers, and the
second is “Urban Stormwater Quality, Management and Planning” containing 60 papers.
Unfortunately, papers dealing with inlet design or inlet efficiencies are few. In the
following sections, several papers from the literature are presented.
Although some studies were performed and reported in the literature, most of
computer software packages for inlet design discussed in the Chapter Four and Five are
basically automated versions of a method developed for use by hand, that is the classical
Rational method to compute peak discharge and to size stormwater pipe system
(Herrmann, 2002).

2.4.2 Rational Method for Peak Flow Rate Estimation

The Rational method continues to be the most widely used approach for
estimating T-year return frequency peak flow rates for small catchments of about one
square mile or less in area (Hromadka et al., 1987). The balanced design storm (U.S.
Army Crops of Engineers, 1990) unit hydrograph method is perhaps the second most
widely used technique for estimating peak flow rates (and is the most widely used
method for developing runoff hydrographs) but is generally considered to be more
accurate than the Rational method. Hromadka and Whitley (1994, 1996) reported that
both of these techniques for estimating peak flow rates are mathematically comparable.
They concluded that the Rational method can be significantly improved by including an
additional multiplicative constant that corresponds to the S-Hydrograph or unit
hydrograph type (e.g., SCS, Mountain, Desert, Valley, etc.). They extended the Rational
Equation to Qp = ε CIA for the developed valley in Orange County, California. The
adjustment factor, ε, has a reported value of about 1.0 (mean of 0.98 with a range of 0.97
to 1.02), but for undeveloped portions of Orange County, it has a value of 0.86, ranging
from 0.83 to 0.89. The similarity between these values may explain why the Rational
method continues to be widely used even though other, more computationally
sophisticated techniques, are readily available.

31
2.4.3 Street Stormwater Storage Capacity
The primary function of a street is to maintain the movement of traffic. Under the
assumption that street drainage will be designed to collect stormwater as fast as possible,
the street stormwater capacity has been defined as its hydraulic conveyance, estimated by
Manning's formula. This practice has resulted in a prevailing experience that street
intersections are often flooded. Guo (2000) presented an investigation on street hydraulic
capacity. It was found that the street stormwater capacity at a sump is in fact dictated by
the storage capacity rather than the conveyance capacity. A new design methodology was
developed in Guo’s study to consider the street depression storage as a criterion when
sizing a sump inlet. Design parameters required by this method include the local
intensity-duration-frequency information, catchment area, runoff coefficient, street
transverse slope, and the configuration of the sump area as a fraction of a circle (Guo,
2000).

2.4.4 Hydraulic Performance of Highway Storm Sewer Inlets

Laboratory experiments (Hotchkiss et al., 1991) were performed with highway
stormwater curb inlets to 1) reduce the oblique standing wave that extends into the
highway from the downstream side of the inlet and 2) determine the effect of highway
resurfacing on inlet efficiency. The work was performed at the University of Nebraska
Hydraulic Modeling Basin on a section of full-scale single lane highway with a
longitudinal slope of three percent and a transverse slope of two percent. All measured
flows were supercritical. Four alternatives to reduce the oblique standing wave were
tested but produced only minimally better conditions. Careless highway resurfacing that
covers inlet transitions drastically reduces inlet efficiency. Efficiency for this case can be
predicted with previously developed equations and is less than one-half that achieved
with standard design transitions (Hotchkiss et al., 1991).

2.4.5 Improvements in Curb-Opening and Grate Inlet Efficiency

Draining stormwater quickly and efficiently from highways is an important part
of every highway design. Laboratory experiments (Hotchkiss, 1994) were conducted to
develop curb-opening and grate inlet efficiency curves for the Nebraska standard inlet
(single and in series), the city of Lincoln canted inlet, a new grate inlet (single and in
series), and an inlet affected by resurfacing. Experiments were performed for the on-
grade inlets on a full-scale roadway surface that was treated with sand-imbedded paint to
produce an average Manning's n of 0.016. The constant longitudinal and cross slopes
were 3 and 2 percent, respectively. Supercritical flow prevailed over the flow range of 0.5
to 5 ft3/sec. Based on results from Hotchkiss (1994), the Nebraska standard inlet
provides about 20 percent greater efficiency than the equivalent AASHTO-type inlet.
Canted inlet performance was only marginally better than that of the Nebraska standard
inlet. The new grate inlet performance was very similar to that of curb-opening inlets.
Inlets in series increased efficiencies by almost 20 percent over the efficiencies of single
inlets. Finally, roadway resurfacing that covers inlet transitions reduces efficiency by
about 50 percent (Hotchkiss, 1994).

32
2.4.6 Stormwater Flow on a Curbed TxDOT Type Concrete Roadway
In 1946, C.F. Izzard introduced the gutter flow equation derived via integration of
Manning’s velocity equation across a gutter section. Literature reviews (Ickert and
Crosby, 2003) indicated that Izzard suggested increasing Manning’s n when using the
integrated equation. Although the integrated equation is used extensively, the suggested
increase in Manning’s n appears to not have been implemented. There is a discrepancy
between the geometric equation (the product of Manning’s velocity equation and flow
area) and the integrated equation that results in overestimation of a gutter’s flow capacity
when using the integrated equation and traditional values of Manning’s n. With intent to
improve driving conditions with more accurate gutter flow computation methods, the
derived geometric and integrated equations are evaluated for three definitions of the
wetted perimeter of a curbed section. Also, experimental data were evaluated to
determine Manning’s n for a curbed TxDOT (Texas Department of Transportation) type
concrete roadway at various longitudinal and transverse slopes.

2.4.7 Design of Curb Opening Inlet Structure

The South Carolina Department of Transportation (SCDOT) has experienced
hydraulic and structural problems with two of its curb-opening structures, called Type 5
and Type 6, primarily used on city streets. Hydraulically, the structure has a tendency to
be plugged by debris with the accompanying loss of capacity. Furthermore, the top slab
of the structures frequently failed structurally under the load of vehicle wheels (usually
large heavy trucks) that jumped the curb. Replacing broken slabs burdened SCDOT
with substantial budget and personnel costs. A new structure was developed in 1989 at
Clemson University. The study (Fiuzat et al., 2000) was conducted mainly to investigate
the hydraulic efficiency of curb opening inlet structures adopted by the SCDOT. They
discussed the design equations and methodology for the utilization of the new structure.
The primary features of the structure they studied are listed below:
• Throat opening of 9.9 cm to avoid clogging
• Gutter depression of 5.1 cm from its normal level to allow for the large vertical
opening. Further depression of the gutter would be unsafe for pedestrians and
bicycles.
• Top slab of 20.3 cm thick (capable to withstand 10 tons of wheel load)
• The smallest structure is 122 cm (4 ft) along the road, and empties into a standard
catch basin of size 122 cm x 122 cm.
The hydraulic efficiency E of the curb inlet structure is defined as E = 100 Qi / Q , and
where Q is the flow rate on the road gutter, and Qi is the flow rate intercepted by the inlet.
The parameters affecting the efficiency are roadway geometry (longitudinal and cross
slope of road), flow properties (road/gutter flow rate, flow spread) and the inlet geometry
(length, gutter depression, transition length, sharpness of the edge of curb opening). The
effect of these factors was studied by Bowman (1988) and Soares (1991) and, based on

33
experimental studies, a gutter depression of 5.1 cm, a transition length of 0.9 m, and
sharp structure corners were suggested.
The development and use of an empirical equation for the efficiency of the curb
inlet as a tool for spacing and sizing the structure was developed by Soares (1991).
Factors that were constant in the development of the equation were normal gutter cross
slope 17:1, normal gutter width of 46 cm and gutter depression of 5.1 cm. Fiuzat and his
coauthors (2000) suggested that Soares’ equation should to be calibrated for other
geometries to improve its range of applicability.
Wasley (1961) investigated the hydrodynamics of flow into curb opening inlets in
a simple road section with no gutter depression. He found efficiency to be a linear
function of a dimensionless inlet length as:
⎛L ⎞
E = 100⎜⎜ i ⎟⎟,
⎝ L1 ⎠
(2.30)
where Li = the length of curb inlet and L1 = length of inlet required to achieve 100%
efficiency. Izzard (1977) presented a method for the design of curb inlets based on data
gathered by Bauer and Woo (1964). In his results, efficiency was a linear function of Li
up to a certain length, beyond which efficiency follows an exponential function:
0.4
⎛L ⎞
E = 100⎜⎜ i ⎟⎟ ,
⎝ L3 ⎠
(2.31)
where, L3 = the length of curb inlet for 100 percent efficiency over the exponential
portion of the curve. As a first step in the development of the design equation, the
measured value of efficiency (E) and flow spread on the road (T) were plotted for
different cross slopes (Sx), longitudinal slopes (S), and inlet length (L, 1.22m, 2.44m,
3.66m, and 4.88m). The values of flow spread for 100 percent efficiency (E = 1.0), T1,
called characteristic spread, was determined by linear regression, which resulted in the
relationship (Fiuzat et al., 2000),
2
T1 ⎡ S⎤
= K ⎢ln ⎥ .
L ⎣ 10 ⎦
(2.32)
The coefficient K varies with the inlet and the cross slope, as shown in Table 2.7, and has
to be determined empirically.

Table 2.7. Values of the coefficient K in Equation (2.32) (Fiuzat et al., 2000).
Cross Slope Coefficient K for inlet lengths in (m)
H:V 1.22 2.44 3.66 4.88
48:1 0.037 0.021 0.016 0.013

34
 36:1 0.031 0.019 0.014 0.011
48:1 0.027 0.016 0.012 0.010
36:1 0.018 0.012 0.010 0.008

The efficiency E, in percent, is reported to be related with the dimensionless flow

spread T1/T as follows:
0.8
⎛T ⎞
E = 100⎜ 1 ⎟ forT ≥ T1 , and
⎝T ⎠
(2.33)
E = 100 forT < T1 .
(2.34)
A plot of calculated efficiencies using Equation (2.33) was compared with a plot
of measured values, and there was reasonable agreement between the two approaches. As
a final step, Equations (2.32) and (2.33) were combined to obtain:
0.8
⎧⎪⎛ KL ⎞ ⎡ ⎛ S ⎞ 2 ⎤ ⎫⎪
E = 100⎨⎜ ⎟ ⎢ln⎜ ⎟ ⎥ ⎬ .
⎪⎩⎝ T ⎠ ⎢⎣ ⎝ 10 ⎠ ⎥⎦ ⎪⎭
(2.35)
Equation (2.35) provides the efficiency of the curb inlet of length (L) for given values of
flow spread (T, design spread), cross slope (Sx), and longitudinal slope (S).
Wasley (1961) and Izzard (1977) previously found that the efficiency of curb
inlets as a function of the dimensionless inlet length (L/L1). The dimensionless spread
T1/T was related to the dimensionless length L/L1 (Fiuzat et. al., 2000). In Figure 2.8, LB
is a fraction of LA, and TA and TB are the characteristic spreads T1 of the respective inlets.
The efficiency of the inlet LB, given the flow spread TA is given by Wasley (1961) as
LB
EB = .
LA
(2.36)

Flow

TA
TB
Inlet
LB

LA

35
Figure 2.8 Characteristic spread TA and TB corresponding to inlet sizes LA and LB
(after Fiuzat et al., 2000)

From Equation (2.33), the efficiency of the inlet shown on Figure 2.8 is
0.8
⎛T ⎞
E B = 100⎜⎜ B ⎟⎟ .
⎝ TA ⎠
(2.37)
Combining Equations (2.36) and (2.37), and using Equation (2.32) gives
0.8
⎡⎛ K ⎞⎛ L ⎞⎤
E B = ⎢⎜⎜ B ⎟⎟⎜⎜ B ⎟⎟⎥ .
⎣⎝ K A ⎠⎝ L A ⎠⎦
(2.38)
The efficiency estimated by Equation (2.38) is greater than that from Equation (2.37).
This is because Wasley (1961) studied curb openings without a gutter depression. By
adjusting Equation (2.38), Wasley showed that as the length of inlet LB increases, the gain
in efficiency becomes smaller. This effect can be used for gaining substantial economic
advantage. Izzard (1977) obtained the same conclusion.
The efficiency obtained form Equation (2.35) was compared to that of HEC-12
for L=1.22m (Fiuzat et al., 2000). Fiuzat, et al. concluded that HEC-12 underestimates
the capacity of smaller curb inlets somewhat. For a longer structure, Fiuzat et al.
reported reasonable agreement between Equation (2.35) and the efficiency generated
using the HEC-12 procedure.
The results obtained with the efficiency equation agreed with the experimental
results within 10% error for most of the situations. Though the efficiency equation was
developed for specific geometry adopted by SCDOT, it could be generally used for other
conditions, but the values of the coefficient K should be experimentally determined for
good accuracy (Fiuzat et al., 2000).

2.4.8 Storm Sewer Design Sensitivity Analysis Using ILSD-2 Model

Illinois Least-Cost Sewer System Design (ILSD-2) model (Wenzel et al., 1979)
basically considers conjunctively the concept of flow routing through sewers, and risks
and uncertainties associated with the design, which is optimized by using the discrete
differential dynamic programming technique. The risk in a sewer design is considered as
the probability of having a flow imposed on a sewer which exceeds the capacity of the
sewer (Nouh, 1987). This condition may be due to hydrologic and hydraulic
uncertainties, uncertainties due to construction and materials, and uncertainties regarding
the cost function (Yen et al., 1976). The main objective of the study by Nouh (1987) was
to perform a comparative evaluation for variations in the generated designs, which might
occur as a result of application of different methodologies to construct the design

36
hyetograph, to generate the overland flow hydrographs, and /or to route the flow through
the sewers. Computer subroutines were developed for constructing ten different shapes of
hyetographs, for generating overland flow hydrographs by three different methods, and
for routing flow hydrographs by different techniques (Nouh, 1987).
Nouh (1987) applied ILSD-2 model to a simple case study, but produced some
results that are useful for other complex systems. For the testing of the model, a district
of east Riyadh City, Saudi Arabia was selected by Nouh (1987): watershed area = 1.278 x
106 square meters (315 acres) consisting of 18% industrial sites, 8% retail stores, 70%
residential sections, and 4% grassy land. The area (Figure 2.9) was divided into 32 sub-
catchments, ranging in size from 2x104 to 8.8x104 m2, and in imperviousness, from 45 to
80%. Time of concentration was about 20 minutes for normal antecedent conditions in
the pervious area, the initial infiltration capacity, the constant infiltration and decay rate
of infiltration selected were 76 mm/hr, 7.4 mm/hr and 0.0012/s, respectively, for
Horton’s infiltration equation. The sewer system layout, manhole locations and data on
the cost of the sewers are the inputs into the ILSD-2 model to generate the least-cost
design.

Figure 2.9. Sewer district (Nouh, 1987)

2.4.8.1 Effect of Time Distribution of Rainfall

The hyetograph can have significant effect on the peak flow and the volume for
the least-cost design of storm sewer systems. Ten different shapes of hyetograph,
including the Uniform, Hershfield, Sifalda, SCS-6 hr, Keifer and Chu, Composite, Huff
(1st quartile), Pilgrim and Cordery, Triangular, and Trapezoidal distributions were
investigated by ILSD-2 model (Nouh, 1987). Infiltration losses corresponding to each
shapes of hyetograph were determined using the mathematical model developed by Akan

37
and Yen (1981 and 1984). These losses were used to determine the aerially averaged
infiltration parameters required for generation of overland flow hydrographs.
Each hyetograph was constructed for a 10-yr design rainfall and used together
with the ILLUDAS model to generate the overland flow hydrographs to the
inlets/manholes of the case study sewer system which was designed by the ILSD-2
Model. The risk of failure was evaluated for a 2-year service period. The baseline for
comparison was the design developed for a uniform (constant rate) hyetograph. The
expressed relative values for a particular design hyetograph reflect use of that hyetograph
instead of the uniform hyetograph in the least cost design of the storm sewer system.
From Figure 2.10, the shape of the design hyetograph resulted in the reduction in the total
sewer cost (design and construction cost) and in an increase in the risk of failure. In
general, the more the reduction in the total sewer cost, the greater the risk of failure.
The greatest amount of cost reduction results from using the trapezoidal hyetograph,
followed by the triangular, Pilgrim and Cordery, and Huff hyetographs.

38
Figure 2.10 Effect of time distribution of rainfall on least-cost sewer system design
(Nouh, 1987)
2.4.8.2 Effect of Overland Flow Hydrograph Generation Method
Several methods were developed to generate overland flow hydrographs for the
purpose of designing urban facilities (Nouh, 1987). To investigate the value of using a
more sophisticated hydrology and hydraulics method for runoff hydrograph generation on
the least cost design of storm sewer system, three methods with differing degrees of
accuracy were considered. These methods, ILLUDAS (Illinois Urban Drainage Area
Simulator, Terstriep and Stall, 1974), UCUR (University Cincinnati Urban Runoff
Model, Papadakis and Preul, 1972), and SWMM (Metcalf and Eddy, 1971), were used
independently with each of the design hyetographs to generate overland flow
hydrographs to the inlets of the sewer system designed using the ILSD-2 model. The

39
results from different methods were compared and expressed relative to those using the
Rational method for peak flow estimation at the inlets of the system. Generally, the use
of SWMM, followed by the UCUR and the ILLUDAS methods resulted in the least
design cost and highest risk of failure. On the other hand, the use of the Rational method
resulted in the greatest design cost and lowest risk. From Figure 2.11, the total sewer
cost associated with the trapezoidal hyetograph together with the SWMM method was
about 67 percent of the cost for the same system designed using the traditional Rational
method. Such significant saving in the total sewer cost may encourage the use of
methods, which are more accurate than the Rational method for the design of storm sewer
system (Nouh, 1987).

Figure 2.11 Effect of overland flow hydrograph generation method on least-cost sewer
system design (Nouh, 1987)

40
2.4.8.3 Effect of Techniques for Routing Flow through Sewers
For hydraulic routing of hydrographs through storm sewers, the most common
techniques are steady-uniform flow (traditional Rational method approach), the
ILLUDAS, nonlinear kinematics wave, Muskingum–Cunge, and SWMM (unsteady and
hydrodynamic methods). The routing method in a particular model represents the
unsteady gravity flow in sewers. This part of sensitivity analysis was to give a
comparative evaluation for the effect of using these techniques on the least cost design of
storm sewer systems.
Based on results presented in Figure 2.12, if the routing mechanism better
approximates the physics associated with flow movement, then the reduction in the
design cost is increased along with the risk of hydraulic failure. This is true for all of the
investigated hyetographs. The ILLUDAS technique is more accurate than the steady flow
technique because it considers the effect of water storage in the sewers. After comparing
all of the approaches, it is clear that SWMM is the most accurate among those used
because it considers, at least partially, the backwater effects. Designs developed using
SWMM cost 83 percent of those developed using the Rational method, but the risk of
failure was 2.3 times greater than that of the system designed using the Rational method.
These findings encourage use the flow routing techniques that better represent the physics
of flow in the design of storm sewers (Nouh, 1987).
Nouh (1987) made the following conclusions: The hyetograph, which resulted the
least total sewer cost, is trapezoidal. The more sophisticated the methods used for runoff
generation and flow routing through the storm sewers, the lower the resulting total sewer
cost of design, but the greater the risk of hydraulic failure. The SWMM method for
overland flow hydrograph generation provides no significant improvements in the design
over the UCUR method but it does over the ILLUDAS and the Rational methods. Thus, it
is recommended for the design of storm sewer systems. The ILLUDAS method may be
used for small sewer systems. However, it is not advisable to use the Rational method
especially when the storm sewer system is large because it results in designs that are
more costly than required. Finally, SWMM is recommended for large storm sewer
systems, while the rest of the techniques, including the steady flow technique may be use
for small storm sewer systems.

41
Figure 2.12 Effect of technique for routing flow through sewers on least-cost sewer
system design (Nouh, 1987)

42
 CHAPTER THREE
COMPUTER MODELS FOR STORMWATER SYSTEM

3.1 Introduction
Computer models are important tools for engineers because they can help
engineers perform engineering task in a faster and better way. Numerous computer
models exist for stormwater system analysis and design. Some of the commercial
software packages that will be discussed in the following sections are listed below
1. WinStorm (storm drain analysis/design, TxDOT)
2. StormCAD (design and analysis of gravity flow pipe networks, Haestad Methods)
3. Hydraflow (hydrology and hydraulic analysis of storm sewer network,
InteliSOLVE)
4. SWMM (urban runoff quantity and quality modeling, USEPA)
5. HYDRA (Storm Sewer Design Model, FHWA)
6. MIDUSS by Alan A. Smith Inc.
7. HydroCAD by Applied Microcomputer Systems

3.2 WinStorm
WinStorm Version 3.0 (http://www.dot.state .tx.us/isd/software/software.htm),
developed by TxDOT, is public domain software and can be downloaded from TxDOT’s
web site. It is used to model storm drainage systems using a drainage network comprised
of three basic drainage components, namely drainage areas, node and links. The user
describes the components of the system by proceeding through a series of dialog
windows defining each portion of the drainage component, drainage area, nodes and
links. WinStorm is capable of designing and analyzing a system simultaneously when
sizes of features are specified. Additionally, two storms of different frequency can be run
simultaneously in order to evaluate the performance of a system during different events,
or design a system based on one event and analyze the design under a different event.
The current or most recent version includes Graphical User Interface (GUI), which was
an upgrade from WinStorm 2.0.
WinStorm computes the peak discharge associated with the drainage area and has
the capability of designing/analyzing seven different types of storm drain inlets. It allows
six conveyance elements to be used to connect the inlets, which includes pipes, box
culverts, arch-pipes, elliptical pipes, semicircular pipes, and ditches. WinStorm can
optionally compute the junction loss at the nodes and provides graphical visualization of
the hydraulic grade line for a selected reach. SCS TR20 and the Rational method are the
two runoff computational procedures involved in WinStorm.
Because WinStorm was particularly developed for the state of Texas, the user has
the option of only choosing the county in the state of Texas for which the IDF curve is

43
developed and described as Equation (2.6). Pumps in the system are not supported in
WinStorm. While multiple incoming links to a node is allowed, only one outgoing link
from a node is supported. Diversion and loop within a system must be modeled as
separate networks and the network must terminate at a single point (outfall).

3.3 StormCAD
StormCAD (version 5) is a computer program for the design and analysis of
gravity-flow pipe networks. The program can be run either within the AutoCAD
environment or in stand-alone mode using its own graphical interface. To use the
program, a graphical representation of the pipe network, containing all information such
as pipe data, inlet characteristics, watershed areas and rainfall information is constructed
(WinStorm doesn’t has the capability to do that). Rainfall information is calculated using
rainfall tables, Equations (2.6, 2.7, 2.8), or the National Weather Service’s HYDRO-35
data. It is one of the commercial software products of Haestad Method Inc., which can
be purchased at the URL: http://www.haestad.com/software/ stmcstandalone/default.asp.
Network flows are computed in StormCAD using built-in numerical models,
which use both the direct step and standard step gradually-varied flow method. Flow
from inlets is calculated using the Rational method, and both intercepted flow and
carryover flow are computed. Flow calculations are valid for both pressure pipe flow and
varied open-channel flow situations, including hydraulic jumps, backwater, and drawn-
down curves. Its capabilities include analyzing various storm sewer design scenarios and
presenting the results both in report format and as a graphical plot. The entire network or
a portion of the network can be designed based on a set of user-defined constraints. These
design constraints include minimum/maximum velocity, slope and cover; choice of pipe
invert and crown matching at structures; inlet efficiency; and gutter spread and depth.
The invert elevations and diameters of pipes, as well as the size of a drainage inlet
necessary to maintain a given spread (for inlets in sag) or capture efficiency (for inlets on
grade) can be computed. Profiles of the network can be generated, which are useful for
viewing the hydraulic grade line.

3.4 Hydraflow
Hydraflow is also a commercial computer program used for storm drain network
design and analysis, and is developed by InteliSOLVE. Hydraflow can be purchased at the
URL: http://www.intelisolve.com/stormsew.html. The current version of Hydraflow
(version 2003) allows the user to design and analyze a storm sewer system consisting up
to 250 pipelines and inlets. The data input is by spreadsheet. Conduits are specified one at
a time, beginning at the downstream end. The lines (conduits) are automatically
numbered in the order in which they are input.
Once the system data is input, Hydraflow offers four different ways to compute
results: analysis with design, enhanced modeling system, full design, and capacity only.
The method depends on the level of accuracy needed and whether the user is modeling an
existing system or designing a new one. In either case, the results are the hydraulic grade

44
line, the full flow capacity of the system, flows at inlets including carryover, captured and
bypassed flows, and total cost of the system.
Hydraflow allows specifying certain design constraints and initial conditions. It
has the option of plotting or generating IDF curves from existing data, third degree
polynomial equation, NWS Hydro-35 data, or from a NOAA (National Oceanic and
Atmospheric Administration, Atlas 2 published by NWS in 1973). In addition, it also has
the feature of interactive design where the user can make changes during the solution
process. It displays the results in Summary, DOT style, Inlet, FL-DOT (Florida DOT)
style, calculation and cost reports along with custom report.

3.5 SWMM
SWMM, developed by the U.S. Environmental Protection Agency (USEPA), is a
comprehensive computer model for mathematical simulation of urban runoff quantity and
quality in storm sewers, combined sewers (storm and sanitary), and natural drainage
systems on the basis of rainfall (hyetograph) and other meteorological inputs and system
characterization (catchment, conveyance, storage/treatment) (Huber and Dickinson,
1992). Both single-event and continuous-event (long-term) processes can be simulated by
the SWMM. Rainfall, snowmelt, surface and sub-surface runoff, and flow routing can be
simulated. The watershed is broken into a finite number of sub-catchments that can be
readily described by their hydraulic and geometric properties. SWMM can be used for
screening, planning, design, and operation purposes for stormwater management.
SWMM consists of three principal computational blocks and five service blocks.
The computational blocks are Runoff, Transport, and Extran blocks. The service blocks
include Combine, Statics, Rain, Temp and Graph. The runoff block produces
hydrographs and pollutographs at inlet locations with an option of snowmelt simulation,
and simulates flow in pipes by non-linear reservoir routing. The transport block routes
the incoming hydrographs and pollutographs through the sewer system based on
kinematic wave routing, and uses the first order decay, scour and deposition to generate
dry-weather flow and quality. The Extran block is intended to solve the St. Venant
equations (one-dimensional, unsteady partial differential equations describing
conservation of mass and conservation of momentum). In the Extran bloc, dynamic wave
simulation is used to route flows through an open or closed conduit stormwater system
(Roesner et al., 1988).
A node in SWMM represents the junction of hydraulic elements (links) and acts
as a location for input of flow and pollutants into the drainage system. A node can also
represent a storage device such as a pond or lake, a point junction representing a point of
change in channel or conduit geometry, a boundary condition in the model, a sewage
treatment plant in the transport layer, or a watershed in runoff layer. Links represent the
hydraulic elements for flow and constituent transport through the system (for example
pipe, channel, pump, weir, orifices regulator, real-time control device, etc.). There are
more than 30 different types of conduits included in SWMM.
Several companies have modified the original version of SWMM. Some of latest
versions of SWMM include:

45
 • PCSWMM 2002 (developed by Computational Hydraulic Inc.)
• Mike SWMM 2002 (developed by Danish Hydraulic Institute)
• Visual SWMM 2000 (developed by CAiCE Software Corporation)
• XP-SWMM 2000 (developed by XP software Pty. Ltd.)

3.6 FHWA Storm Sewer Design Mode, HYDRA

HYDRA is a part of a package of integrated design computer programs called
HYDRAIN developed by the U.S. Federal Highway Administration. HYDRA is used as
storm drain design model by federal and other engineers. Like all FHWA software
packages, the model HYDRA is distributed under contract with the FHWA through
McTrans Software Center at the Civil Engineering Department of the University of
Florida at Gainesville. Commercial vendors have linked HYDRA to an integrated
CADD/GIS system for interactive design.
The program’s primary use is in analyzing the adequacy of existing storm drains
or designing new storm drains and inlets by the Rational method as described previously
(Section 2.1) or by the modified Rational method (Section 2.2); the latter represents the
hydrograph as a trapezoid having a volume equal to the calculated net rain.
In addition to the modified Rational method, commercial versions of HYDRA
allow design by SCS methods or the Santa Barbara hydrograph method. HYDRA has an
uncommon but useful feature of allowing the hydraulic grade lines to be checked through
underground storm sewers. Backwater calculations determine the total system energy
losses and hydraulic grade line elevation, allowing an indication of whether inlets,
manholes, or junction boxes will overflow, which is implementation of steady state
hydraulics. Another useful feature is that street and gutter flows that exceed the inlet
capacity of the storm sewers are routed by HYDRA to the next downstream location and
added to the hydrograph at that point.

3.7 MIDUSS Software

MIDUSS is a stormwater modeling program intended for designers of stormwater
systems with emphasis on the detailed design of devices for centralized or on-site BMPs.
The program is positioned midway between detailed design programs with no simulation
capability and hydrological simulators with no design features. BMPs include wet and
dry ponds and MIDUSS has many tools for the design of outflow controls for a variety of
storage facilities including rooftop and parking lot storage. It includes an exfiltration
trench design. One can design a drainage network consisting of an unlimited number of
pipes, channels, ponds and diversion structures.
The hydrology in MIDUSS is versatile with many built-in models:
• 5 storm types,
• 3 infiltration models, and
• 4 overland flow methods plus a lag and route tool for modeling large catchments.

46
 Another feature of MIDUSS is the ability to run in Automatic mode where a

previously created output file serves as the basis for the input database. This is when

testing a design under a more severe storm or completing a design in two or more

sessions.

The design options in MIDUSS include:

• Pipe sizing (in which hydraulic gradient is reported if the pipe is surcharged).
• Open channels of either a generalized trapezoidal shape or a more complex
cross-section defined graphically and modified with up to 50 co-ordinate pairs.
• Hydrograph flood routing in part-full pipes or open channels.
• Detention ponds including a variety of tools for computing depth-discharge and
depth-storage curves for a variety of outflow control devices and pond
geometries.
• Ex-filtration trenches with multiple perforated and non-perforated pipes.
• Diversion structures for separation of hydrograph components (e.g. major and
minor).
The above detailed design tools are available at all points in the development of
the drainage network. MIDUSS is developed by Alan A. Smith Inc., a software
development company established in 1978.

3.8 HydroCAD
HydroCAD is a Computer Aided Design tool for use by civil engineers for
modeling stormwater runoff. It is based largely on the hydrology techniques developed
by the Soil Conservation Service (SCS/NRCS), combined with other hydrology and
hydraulics calculations. The general features include Rational method, SCS method, and
SBUH (Santa Barbara Urban Hydrograph) as runoff hydrograph generation method,
reach and pond routing plus built-in hydraulics, graphics, and on-screen routing diagram.
HydroCAD was developed by Applied Microcomputer Systems in 1986 and is
being continuously updated. HydroCAD can be used for drainage projects ranging from
small runoff studies to detention pond designs. The capabilities of HydroCAD also
include
• Time of concentration calculation using sheet flow method, shallow
concentrated flow, channel flow, curve number method,
• Rainfall management,
• Unit hydrographs method (SCS, custom unit hydrograph),
• Exporting and importing hydrographs, and

47
• Twelve types of pond outlet hydraulics and pond storage capabilities.

48
 CHAPTER FOUR
2.1.1 Drainage System for Case Study

4.1 Introduction

In this chapter, a simple drainage system developed by Mr. George (Rudy)

Herrmann is presented. This is a hypothetical system (Herrmann, 2002) and is to provide
a test bed for the comparison of different design approaches implemented in the several
computer software packages. In addition to model testing on hypothetical drainage
system, a select subset of the software packages was tested on a real drainage system, and
will be presented in the Chapter 6.

4.2 Hypothetical Drainage System for Case Study

A simple hypothetical drainage system (analogous model) was used to test the
differences in results obtained from application of different computer software packages
for storm drainage design. The model system comprises of four drainage areas, and is
shown in Figure 4.1. The system consisted of four inlets (I-1 to I-4 in Figure 4.1)
interconnected by underground conduits (P-1 to P-4 in Figure 4.1) and terminating with a
free outlet (O-1 in Figure 4.1). For simplicity, the inlets were curb-opening inlet, because
this type of inlet has fewer parameters. Rainfall intensity is one of input variables for
applying the Rational method and is computed as a function of the time of concentration
of the subwatershed. Rainfall intensity for Lubbock County (Texas) was used for the
case study. The 2-year design storm was used to develop the basic design of the network,
and it was analyzed using a 5-year design storm.
The catchments, inlet, and conduit configuration parameters used in the case
study are given in Tables 4.1 to 4.3. Because abstractions in the Rational method are
characterized by the runoff coefficient, C, no other forms of direct losses were
incorporated into the model. For the simple model, the runoff coefficient was set to a
value of 1.0 (implying an impervious area) for all catchments. Results and conclusions
developed through the case study are independent of values of runoff coefficients used.

I-1 P-1 I-2 P-2 I-3 P-3 I-4 P-4 O-1

Figure 4.1. Hypothetical storm drainage system for case study.

49
 Table 4.1 Configuration data for four catchments.
Inlet ID C Value Area (acre) TC (minutes)
I-1 1.0 1.00 10
I-2 1.0 1.00 5
I-3 1.0 1.00 15
I-4 1.0 1.00 12

Table 4.2 Configuration data for grade inlets.

Inlet Inlet Slopes Gutter Bypass to

Inlet ID
type Length (ft) Long. Trans. n Depression Inlet ID
(%) (%) (ft)
I-1 Curb 10.00 0.10 2.50 0.014 0.04 I-2
I-2 Curb 10.00 0.10 2.50 0.014 0.04 I-3
I-3 Curb 10.00 0.10 2.50 0.014 0.04 I-4
I-4 Curb 10.00 0.10 2.50 0.014 0.04 O-1

Table 4.3 Configuration data for underground conduits.

Flow line
elevation 2.1.1 Shape Span Rise Length Slope n ID
Run
U/S D/S (ft) (ft) (ft) (%) values D/S
(ft) (ft)
P-1 88.00 86.00 Box 2.00 2.00 600.00 0.33 0.013 I-2
P-2 86.00 83.00 Box 2.00 2.00 1000.00 0.30 0.013 I-3
P-3 81.00 80.50 Box 3.00 3.00 400.00 0.13 0.013 I-4
P-4 77.00 74.00 Box 3.00 3.00 300.00 1.00 0.013 O-1

50
 CHAPTER FIVE
RESULTS OF CASE STUDY

5.1 Introduction
The results obtained for the case study from different software packages are
discussed in this chapter. Detailed calculation on the results and the basic setup required
to operate the software packages are presented in the Appendix A to D. The results
acquired from the software packages are presented in tabular format.
Typically in storm drain design, inlet flow quantity calculations are done
separately from pipe system flows. In other words, most designers assume that all
of the flows computed by the Rational Method (using maximum travel time as time
of concentration) will enter the pipes, which is implemented in Hydraflow.
WinStorm and StormCAD use the same methodology to handle the flow entering
the pipe system even they have different methods to compute cumulative CA values
before the Rational Method is applied. The following sections further explain
results developed for the case study by using steady models: WinStorm, StormCAD
and Hydraflow, and by using unsteady models: SPLIT program and SWMM. The
steady models only compute peak discharge for the storm drain design, while the
unsteady models develop hydrographs at inlets and route through the pipe system.

5.2 Case Study by Using WinStorm and StormCAD

The hypothetical drainage system presented in the Chapter Four was first
tested by using WinStorm and StormCAD. The Appendix A gives detailed
calculations on flows from watershed to inlet, carryover flow at inlet, carryover flow
from upstream inlet, and pipe flow. Results of computations using StormCAD are
graphically shown in Figure 5.1. For example, intercepted flow at the inlet I-1
(0.839 CA) is indicated in red. This flow directly enters the pipe P-1. Carryover CA
from inlet I-1 (0.161 CA) plus local CA of inlet I-2 is indicated in green (total 1.161
CA). Intercepted flow from I-2 combined with pipe flow in P-1 is indicated by the
yellow pipe, which is the flow in the pipe P-2. The same process is then repeated for
remaining part of the system, as depicted in Figure 5.1 (tp is travel time in pipe).
Carryover calculation results for four inlets developed by StormCAD and
WinStorm, respectively, are given in Tables 5.1 and 5.2. When StormCAD was used
(Table 5.1), for example, total CA for the inlet I-3 is equal to carryover CA from the
inlet I-2 (0.228, blue square in catchment 2 of Figure 5.1) plus local CA (1.0).
Therefore StormCAD actually uses local rainfall intensity and carryover CA to
recalculate carryover flow. WinStorm does calculate the carryover simply by using
equation (2.16), and WinStorm directly adds carryover flow from upstream inlet
and the local flow together (Table 5.2). Total flow to each inlet simulated using
StormCAD and WinStorm are highlighted in both Tables 5.1 and 5.2, and the
difference between simulated discharges is relative small (<0.2 cfs in the case study).

51
 Flow computation results for four pipes developed by StormCAD and
WinStorm, respectively, are given in Tables 5.3 and 5.4. It was found that
StormCAD incorporates the carryover flow in sizing the conduits whereas
WinStorm calculates the carryover but does not take into account in sizing the
conduits. Since procedures used to calculate flow in pipes by using WinStorm and
StormCAD are different as discussed in Appendix A, total pipe flows calculated by
using WinStorm are always greater than ones by StormCAD. In the case study, the
drainage area for each catchment is only 1 acre; the peak discharge in the pipe
produced by WinStorm is about 1 cfs greater than that produced in StormCAD, as
shown in Tables 5.3 and 5.4. For real watersheds, the difference in computed pipe
flow could be much greater because drainage areas would commonly be larger than
those used in this simple test case.

Tc=1 Tc=1 Tc=1 Tc=1

0 0 5 2
0.839 CA 1.161 CA 1.228 CA 1.205 CA

E=83.9% E=80.3% E=83.3% E=81.3%

tp=3.56 tp=4.21 tp=2.17

Figure 5.1. Color-coded simulation results using StormCAD.

52
 Table 5.1. Carryover calculation at inlets by StormCAD.
Catchment Total CA/flow for Intercepted
parameter Upstream inlet CA/flow
Intensity Inlet Carryover
Node carryover
(in/hr) Flow Efficiency Flow CA
CA Tc CA CA CA
(cfs) (cfs)

I-1 1 10 5.11 0 1 5.149 83.9 0.839 4.32 0.161

I-2 1 5(10) 5.11 0.161 1.161 5.980 80.3 0.933 4.80 0.228

I-3 1 15 4.26 0.228 1.228 5.271 83.3 1.023 4.39 0.205

I-4 1 12 4.73 0.205 1.205 5.740 81.3 0.980 4.67 0.225

Table 5.2. Carryover calculation at inlets by WinStorm.

Catchment Upstream Local inflow Total flow

parameter Intensity Intercepted Carryover
Node carryover to inlet
(in/hr) Q = CIA (cfs) flow (cfs) flow (cfs)
flow (cfs) (cfs)
CA Tc

I-1 1 10 5.11 0 5.11 5.11 4.336 0.774

I-2 1 5(10) 5.11 0.774 5.11 5.88 4.795 1.085

I-3 1 15 4.26 1.085 4.26 5.34 4.478 0.862

I-4 1 12 4.73 0.862 4.73 5.59 4.625 0.965

Table 5.3. Pipe flow calculation using StormCAD.

53
 Travel Full
Overall Intensity Intercepted Total CA Flow in
Pipe Length Velocity time in
Tc (in/hr) CA into pipe Pipe
Capacity
pipe (cfs)

P-1 600 2.81 0 10.0 5.11 0.839 0.839 4.32 16.63

P-2 1000 3.96 3.56 13.56 4.47 0.933 1.772 7.98 15.78

P-3 400 3.08 4.21 17.77 3.91 0.695 2.795 11.02 30.02

P-4 300 6.88 2.17 19.94 3.67 0.732 3.775 13.96 84.92

Table 5.4. Flow calculation in pipes using WinStorm.

Full
Travel time Total CA Flow in
Pipe Length Velocity Overall Tc Intensity Capacity
in Pipe into Pipe Pipe
(cfs)

P-1 600 3.63 0 10.00 5.11 1 5.11 16.63

P-2 1000 4.09 2.75 12.75 4.60 2 9.20 15.78

P-3 400 3.06 4.07 16.82 4.02 3 12.06 30.02

P-4 300 7.03 2.18 19.00 3.77 4 15.08 84.93

54
5.3 Case Study by Using Hydraflow
The input for Hydraflow for the case study is given in Appendix B. Hydraflow
has four options for computing/generating results. For the case study the Return Period
chosen was 5 year and the calculation options investigated were on two different options
available namely “Analysis w/ Design” and “Enhanced Modeling System” to study the
difference in the process of simulating results. The “Full Design” and “Capacity Only”
options, which can be used to design the pipe size, are of less concern in the study.
Furthermore, the calculations of the former two options are processed and their results are
compared and tabulated separately.

5.3.1 Simulation with Analysis w/ Design Option

Table 5.5 below gives the total flow from catchment; the captured and the
bypassed flows resulted from Hydraflow for the inlet configuration while flow
computation in the pipes is shown in Table 5.6. Calculation procedure used to develop
results in Table 5.5 was based on FHWA guidelines as discussed in Chapter 2. The
results obtained (Table 5.5) are similar to those of WinStorm outputs (Table 5.4) except
the precision.
Under this calculation option, the time of concentration (18.80 minutes) for Line
(pipe) L-2 in Table 5.6 is the summation of the upstream Tc (10 minutes) and the travel
time in the conduit by assuming that pipe is flowing full and not considering pipe slope:
8.8 minutes = [(600ft)/{(4.53cfs)/(4ft2)}]/(60s/min), where 4 ft2 is the full-flow cross
section area, and 4.53 cfs is intercepted flow. This travel time in pipe L-2 is much greater
than the travel time (3.5 seconds) computed by using Manning’s equation, e.g., in
WinStorm (Table 5.4) and StormCAD (Table 5.3). This treatment on computing travel
time in pipes may not produce reliable results.
Rainfall intensity and then Total Runoff by the Rational method in Table 5.6 are
calculated and based on the time of concentration calculated by the above method. While
the total flow in Table 5.6 is simply the summation of captured flows from the inlets
(Table 5.5) in the upstream direction (9.44cfs = 4.53cfs + 4.91cfs). The result so
obtained for the total flow was due to using the option “Use inlet captured flows in
system” under Flow Options of Design Codes (Fig. B.3), otherwise the total flow would
be the same as the total runoff (system flows) as illustrated in Table 5.7.

5.3.2 Simulation With Enhanced Modeling System (EMS) Option

This option is used when one has to analyze existing systems and where the
hydraulic analysis is critical and maximum accuracy is of importance (Hydraflow, Storm
Sewers 2003 User’s Guide). While solving the model with this option on, Hydraflow
undergoes three system iterations to achieve a practical balance between accuracy and the
time required to produce results. The model was solved without using the Flow Option
“Use inlet captured flows in system” under Design Codes (Fig. B.3), therefore the total
runoff is the same as the total flow (Table 5.7) for each pipe. The model resulted the
same values of flow parameters for the catchment and the inlet as those developed in

98
previous case (Table 5.5). The results of the flow data in conduits only by using EMS
option are summarized in Table 5.7.

Table 5.5. Calculated flow parameters at inlets by Hydraflow with Analysis w/Design
option.

Inlet Inlet time Intensity Runoff Q = CIA Q Carryover Q Captured Q Bypassed

ID (minutes) (in/hr) Coefficient (cfs) (cfs) (cfs) (cfs)

I-1 10 5.12 1.00 5.12 0.00 4.53 0.60

I-2 5 5.12 1.00 5.12 0.60 4.91 0.81
I-3 15 4.27 1.00 4.27 0.81 4.50 0.58
I-4 12 4.74 1.00 4.74 0.58 4.66 0.66

2.1.2 Table 5.6. Calculated flow parameters in “Lines” by Hydraflow with Analysis
w/Design option

Rainfall Total
Line Total Time of Total Capacity
Line intensity Runoff
Length CxA Conc. (min) Flow (cfs) Full (cfs)
(in/hr) (cfs)
L-1 600 1.00 10.00 5.10 5.12 4.53 16.63
L-2 1000 2.00 18.80 3.80 7.60 9.44 15.77
L-3 400 3.00 25.90 3.20 9.53 13.93 30.02
L-4 300 4.00 30.20 2.90 11.58 18.59 84.91

Total CA in Tables 5.6 and 5.7 is simply the addition of CA in the upstream
direction, which is handled in the same way as WinStorm does. Travel time in the
conduit (L-1) to reach from I-1 to I-2 (3.10 minutes) is calculated by dividing the length
(600 ft.) by the average velocity (3.21 ft/s.), obtained from Manning’s equation.
Therefore, the time of concentration for the next line L-2 (13.10 minutes) is calculated by
summing upstream time of concentration and the travel time in the conduit to reach the
next inlet [13.10 min.=10min. (upstream) + 3.10min. (travel time)]. Likewise, the time
of concentration for the last line is calculated and found that a total flow of 14.99 cfs
flows through the line L-4 to the outlet (Table 5.7).

Table 5.7. Calculated flow parameters in “Lines” by Hydraflow with EMS option.

Line Total Time of Rainfall Total Total Capacity Velocity

Line

99
 Length CxA Conc. intensity Runoff Flow Full (ft/s)
(min) (in/hr) (cfs) (cfs) (cfs)

L-1 600 1.00 10.00 5.10 5.12 5.12 16.63 3.21

L-2 1000 2.00 13.10 4.60 9.11 9.11 15.77 4.07

L-3 400 3.00 17.20 4.00 11.96 11.96 30.02 3.12

L-4 300 4.00 19.30 3.70 14.99 14.99 84.91 5.44

5.4 Conceptual Disconnect in Storm Drain Design

Total flows computed by Hydraflow with EMS option is essential same as ones
developed by using WinStorm (Table 5.4), therefore Hydraflow as well as WinStorm has
the conceptual disconnect between inlet design and pipe design. The conceptual
disconnect occurs when the inlet design approach allows some of the incoming flow to
bypass or carryover from one inlet to the next (This avoids the requirement of very long
inlet lengths), however, in the conduit design procedure, all flow from the subwatershed
is assumed to enter the system through the inlet. Therefore, one set of flows is used to
design the inlets and a second, and greater, set of flows is used to design the conduit.
Even the intercepted flow is calculated by WinStorm and Hydraflow, but is not
used in calculating the travel time in the pipe; instead, the peak discharge from the sub-
catchment is used. Second, the CA product is simply added algebraically (lumped) from
upstream catchments (Table 5.4, 5.6, and 5.7), even though not all of the flow enters the
pipe through the inlet. The lumped CA and the travel time are the only two variables
required for calculation of flow in pipe when using the Rational method.
In contrast, StormCAD overcomes the conceptual disconnection between two
parts of network design by considering intercepted and carryover CA. Basically
StormCAD doesn’t simply add (lump) CA from upstream catchments as discussed in
Appendix A and the section 5.2. Also StormCAD does use the intercepted flow to
calculate the travel time of flow in pipes. StormCAD like WinStorm and Hydraflow
recalculates the discharge entering the pipe by applying the Rational method and using
different rainfall intensities for different inlets based on the maximum travel time.
Therefore, the peak discharges in pipes (Table 5.3) (final result for determining pipe size)
obtained using StormCAD are always less than those obtained using WinStorm (Table
5.4) and Hydraflow (Table 5.7). However, the differences are only 7% to 16% percent
less than discharges computed by WinStorm, which are significant, but not large. These
three models still could result the same (or similar) pipe size for the drainage system,
depending on the policy used to select the pipe size from calculated flow. All the three
software packages deal only with the instantaneous peak obtained from the watershed
using Rational Method with steady state process in pipe, which is not the actual case.
Although some experimental and analytical studies on storm drainage system
design were performed and reported in the literature as reviewed and presented in
Chapter Two, most of computer software packages for inlet design explored/discussed in

100
previous sections are basically automated versions of a method developed for use by
hand, that is the classical Rational method to compute peak discharge and to size
stormwater pipe system. This is the same finding developed by Herrmann (2002).
Despite known limitations of the Rational method and computer tools based on
the Rational Method, most engineers employ these tools. Because of the limitations in
inherent in the development of the design discharge, and the problems with inlet inflow
and bypass, pipes sized/designed/adopted using these methods are usually larger than
necessary (over-designed).
Apart from conservative approach (Rational method), the selection of over-
capacity pipe is also governed availability of pipe in particular diameter and by design
code, which will be discussed in the Chapter Six. Designers may think that over-
designed or conservative pipe sizes in storm drainage system are capable of passing
discharges from unexpected heavy rainfalls that are exceed the design event (perhaps
resulting from increased urbanization), but because inflows to the system are limited by
inlet capacity, that extra pipe capacity may never be used.

5.5 Case Study by Using SPLIT Program

The hypothetical drainage system for the case study was then analyzed using the
“modified SPLIT” program. The system configuration data for which the simulation
results are discussed in this section are given in Tables 4.1 to 4.3. Table 5.8 below shows
the computed results of discharges obtained from WinStorm and the “modified SPLIT”
program for comparison. The design return period for simulation was chosen to be 5
years. The simulations were performed for different durations of storm and intensities of
rainfall.

Table 5.8. Results from WinStorm and modified SPLIT program.

Discharge (cfs) by Modified SPLIT program

WinStorm
Pipe
Discharge
Run D =10 min D =12 min D =15 min D =20 min
(cfs)
i = 5.1 in/hr i = 4.7 in/hr i = 4.3 in/hr i = 3.7 in/hr
P-1 5.11 4.30 4.05 3.80 3.39
P-2 9.19 7.84 7.93 7.93 7.00
P-3 12.05 11.28 11.58 11.61 10.67
P-4 15.07 14.11 14.43 14.82 14.36

Although results computed using the SPLIT program are less than those
computed using WinStorm, the differences are only 5% to 16% percent of discharges
developed by WinStorm, significant, but not large. The time distributed sequence of flow
in the pipe and the carryover flow in the gutter can only be visualized or interpreted
through the Modified SPLIT program. Basic information on SPLIT program and further
discussion of output (hydrographs) generated are presented in the Appendix C.

101
5.6 Case Study by Using SWMM
The hypothetical drainage system was extensively analyzed using Visual and XP
SWMM. The background information on SWMM, model setup, and many intermediate
results developed by SWMM are given in Appendix D. SWMM consists of three
principal computational blocks (layers or modes): Runoff, Transport (Sanitary), and
Extran (Hydraulic) blocks. There are two options in SWMM to connect conduits with
four inlet nodes and the outfall node for the case study. If the conduits connecting nodes
are created under the Runoff layer, non-linear reservoir approach is used to simulate
runoff passing through conduits. If the conduits connecting nodes are created under the
Hydraulic (Extran) layer, the St. Venant equations (one-dimensional, unsteady state
continuity and momentum partial differential equations) for dynamic wave simulation
(Roesner et al., 1988) on an open or closed conduit wastewater or stormwater system are
solved to generate hydrographs.

5.6.1 Results From Runoff Layer

First, both inlets and pipes (links) were set up under runoff layer. SWMM
simulates runoff hydrograph at inlets, and all flow collected/generated at inlets gets into
underground pipe (no inlet restriction under Runoff layer) and routes through pipe by
non-linear reservoir method. For all inlets except the most upstream one, routed
hydrograph from upstream pipe combines with local inflow hydrograph generated
through catchment modeling. Figure 5.2 shows hydrograph generated at the inlet N-1
and by using Rational formula for unit hydrograph method. Peak discharge is 4.6 cfs at
8:13AM (rainfall starts at 8:00AM and lasts 10 minutes with 5.11”/hr intensity, time of
concentration is 10 minutes for catchment at the inlet N-1). Figure 5.3 shows hydrograph
and flow velocity after pipe routing, and peak discharge is slight smaller (4.1 cfs) and at a
later time (8:15AM). These are two examples of hydrograph for illustration only, and
other simulated hydrographs by SWMM are presented in the Appendix D.
The peak discharges at all the nodes and the conduits obtained after solving the
model in the SWMM Runoff layer using the Rational method and the SCS method as the
hydrograph generation technique, respectively, are summarized in Table 5.9. A typical
value of 484 for the shape factor, as determined by Soil Conservation Service for most
watersheds, though actual value may range from 300 for flat swampy country to 600 for
steep terrain, was utilized while simulating under SCS method. Because of the fact that
Rational method is more conservative than SCS method, the peak flows simulated, in
nodes and eventually in conduits, obtained from the Rational method are found to be
greater than one obtained from the SCS method. Peak discharges in pipes are always less
than one at corresponding nodes (Table 5.9) because they are peak values after non-linear
reservoir routing in the pipes.

Table 5.9 Peak flow (cfs) in the nodes and conduits simulated by using Rational
method and SCS method under Runoff layer.

102
 Rational Method SCS Method
Node
Peak Discharge at Peak Discharge in Peak Discharge at Peak Discharge in
Node Pipe Node Pipe
N-1 4.61 4.14 3.50 3.11
N-2 8.52 7.05 6.42 5.15
N-3 10.34 9.96 7.76 7.32
N-4 13.21 13.17 9.64 9.62

Figure 5.2. Hydrograph output of SWMM for Node N1.

103
 Figure 5.3. Time series plot of flows and velocity by SWMM for conduit L1.

5.6.2 Results from Runoff and Hydraulic Layers

The SWMM model was solved for both the Runoff and the Hydraulic layers with
and without considering inlet restriction. At first, the model was solved in the Runoff
layer only to generate inlet hydrograph. Because of conduits not connected in the Runoff
layer, the maximum flow at a node (inlet) represents the flow from individual catchment
alone. These hydrographs are the basis of flow input for the Extran layer.
The results of solving the model in the Hydraulic layer are presented in Appendix
D. As expected, the maximum flow in the conduit L1 (4.5 cfs at 8:14AM) is slightly less
than the maximum inflow from the catchment node N1 (4.6 cfs). The maximum flow
from the conduit L1 is then added to the inflow from the catchment at node N2 (4.6 cfs)
and again routed in the downstream conduit L2. This process of dynamic routing is
continued to the downstream conduits in a similar manner until an outfall is encountered.
Table 5.10 summarizes the output results of peak flows simulated from SWMM by using
the Rational method under different inlet settings.
The model was tested again with an inlet restriction at all nodes with an allowable
maximum interception capacity of 3.0 cfs at an inlet and a rating curve with and without
a gutter along side of the road to carry the flow downstream. Results are also given in
Table 5.10. A considerable amount of flow, with a maximum rate of 0.73 cfs (Table
5.10) and peaking at 8.16 AM, was found to travel in the gutter section from first node
N1 towards the downstream node N2. It was found that simulated flows in underground
conduits were reported to be the same with or without the gutter layer, this led us to
believe that SWMM model actually does not combine flow from local catchment with
flow from upstream surface gutter due to carryover.

104
Table 5.10 Peak Discharge in the pipes resulted from SWMM under different inlet
settings using Rational method.

Peak flow (cfs)

Rational Method
With inlet Restriction
Rational Hydraulics
Pipe Method at layer Constant discharge = 3 cfs Rating curve
Runoff without
Layer With gutter Without With gutter Without
inlet
restriction gutter gutter
In
In Pipe Gutter In Pipe In Pipe In Gutter In Pipe

L-1 4.61 4.52 3.34 0.73 3.34 3.89 0.29 3.89

L-2 4.61 6.56 5.11 0.35 5.11 5.79 0.15 5.79
L-3 3.40 9.76 7.99 0.17 7.99 8.73 0.09 8.73
L-4 4.10 13.18 10.75 10.75 11.48 11.48

5.6.3 Results from SCS Method In Runoff Layer

Like the Rational method, the SCS method was also tested under different
combinations of flow, rating curve and with and without considering the gutter section on
the surface. Table 5.11 summarizes the peak discharges in the four pipes under different
conditions with/without inlet restrictions. Under same setting and condition, the peak
discharges in all the cases acquired from SCS method were found to be lesser than those
obtained by Rational method (Table 5.10).

5.6.4 Comparison between WinStorm and SWMM

WinStorm and SWMM are completely different models for storm drain design
and analysis. WinStorm uses the Rational method to compute the peak discharge at inlets
and carryover flow by the HEC-22 method. For pipe flow WinStorm always recalculates
rainfall intensity from IDF curves or equations using the maximum time of concentration,
and then recalculate pipe discharge using cumulative CA. Therefore different pipe
discharges at a drainage system are associated with different design storm and
contributing watershed area. SWMM is used to design and analyze a drainage system
under specified rainfall input (hyetograph) over specified duration. In this study, constant
rainfall intensity of 5.11 in/hr with duration of 10 minutes was used for SWMM analysis,
therefore all watersheds, inlets, and pipes are subjected to this rainfall input. This rainfall
intensity was determined from IDF curves of Lubbock County, Texas, for the return
period of 5 years and duration of 10 minutes. In spite of fundamental differences
between WinStorm and SWMM, the peak discharges of pipes developed by WinStorm

105
(Table 5.4) and SWMM (Tables 5.10 and 5.11) are compared. The peak flows in the
conduits produced by SWMM varied to some extent, depending on methods for
hydrograph generation and flow routing in the pipe. Peak flows simulated in pipes using
SWMM are smaller than those obtained by WinStorm (Table 5.4), the differences are up
to 24% (Rational method in Table 5.10) to 41% (SCS method in Table 5.11) percent of
discharges developed by WinStorm, significant, but not large. The differences between
peak discharges developed using SWMM and WinStorm could be greater for actual
application to design problems, instead of the simple rainfall settings and storm drain
network tested here.

Table 5.11. Peak Discharge in the pipes resulted from SWMM under different inlet
settings using SCS hydrology method.
Peak flow (cfs)
SCS Method
With inlet Restriction
SCS Hydraulics
Pipe Method Constant discharge = 3 cfs Rating curve
layer
Runoff without With gutter Without With gutter Without
Layer inlet gutter gutter
In
restriction In Pipe In Pipe In Gutter
Gutter In Pipe In Pipe
L-1 3.50 3.77 3.35 0.10 3.35 3.43 0.09 3.43
L-2 3.50 4.75 4.46 0.04 4.46 4.45 0.04 4.45
L-3 2.64 7.07 6.77 0.02 6.77 6.75 0.02 6.75
L-4 3.11 9.25 8.91 8.91 8.89 8.89

CHAPTER SIX

IMPLEMENTATION OF MODEL TESTING FOR

HIGHWAY 77/83
6.1 Description of Study Site and TxDOT Design Policy
For comparison with the previous example, data were taken from an existing
drainage system on the U.S. 77/83 expressway for comparison using WinStorm and
StormCAD. Furthermore, some of the current stormwater sewer design policies and
regulation that TxDOT are currently pursuing for highway stormwater design are also
discussed.
Texas Department of Transportation (TxDOT) with the Federal Aid, proposed
state highway improvement on U.S. 77/83 expressway. The expressway, located on the
Pharr District (southern part of Texas) of Cameroon County, was only covered for a total

106
length of 3.086 miles as shown by dark line in the Figure 6.1. The main aim of the
project was to upgrade the existing stormwater drainage system. The stormwater
conveyance line was, for convenience, divided into 5 trunk lines A, B, C, D and E
respectively, based on discharging outlet. Trunk line A is comprised of 34 junctions with
24 inlets, 8 manholes and an outlet while Trunk line B consisted of 31 junctions among
which 20 were inlets, 10 manholes and an outfall, and Trunk line C, D and E took
account of 33, 27, 14 junctions including the outlet, respectively. TxDOT Engineers
utilized their in-house developed computer software called WinStorm for all of TxDOT
drainage design and analysis projects, so did they for the U.S. 77/83 project too.
The process of computing a hydrograph begins with selection of a design storm,
the first step of which is to select a design frequency. Often, the local approving
authority (city, county, drainage district, etc.) will specify the level of design to be used
for any particular type of structure. Likewise, TxDOT has its policy of using either 2 or 5
year, preferably 5 year, as the storm frequency for roadways of functional classification,
while design frequency for interstate and limited access highways is 10 years. In order to
alleviate or eliminate some common mistakes in storm drain design that result in
operational problems, TxDOT also has certain regulation in selection of conveyance
(pipe) sizes. The minimum sizes of conduit that TxDOT utilizes is 18 inches in order to
facilitate cleaning and debris clearing, but occasionally designer may use 12 inches for
short laterals form an inlet to a junction box with one joint or so (Herrmann, 2003). The
standard sizes that TxDOT generally employ include 18, 21, 24, 30, 36, 42, 48, 54, 60
and 72 inches of diameter. Even pipes on 3-inch increment are available on market,
however, TxDOT designers experienced that the bid prices for 3-inch increment pipes
(e.g., 15, 21, 27, 33 inches) are always more than for the next even 6-inch increment
pipes (e.g., 18, 24, 30, 36 inches) (Herrmann, 2003). The conduit of the storm drain
system shall be placed at not less than the grade that maintains a minimum of 2 feet per
second velocity with a maximum of 12 feet per second.

107
Figure 6.1 Topographic location of project site and U.S. 77/83 Expressway.

6.2 Simulation Results by Using StormCAD and WinStorm

StormCAD was used to calculate the discharge and compare to that obtained
using WinStorm. Only a part of Trunk line A (Figure 6.2), though it consists of different
inlet types, was tested using StormCAD. Table 6.1 shows characteristics of inlets
including inlet type and length tested under WinStorm and StormCAD. Table 6.2 shows
results of simulated flow from local catchment, intercepted and bypassed flow at inlets.
WinStorm gave slightly smaller flow from local catchment since unit conversion factor is
treated as 1.0 in WinStorm (StormCAD uses 1.008) when rational method is applied. As

108
seen from Table 6.2, the intercepted and bypassed flows computed by either computer
software package do not vary substantially (do not show appreciable difference). Table
6.3 reports pipe parameters used (real data from U.S. 77/83 project) and simulated pipe
flows under WinStorm and StormCAD. Peak discharges in pipes simulated by both are
essential the same with only very small difference.

Figure 6.2. Layout of trunk line A of U.S. 77/83 under StormCAD.

109
Table 6.1 Characteristics of inlets tested by WinStorm and StormCAD.

Inlet C A (Acre) Tc (min) Inlet type Inlet length (ft)

A2 0.90 2.89 46.66 Grate on Sag 3.0
A4 0.42 2.12 18.25 Grate on Sag 3.0
A6 0.45 1.97 17.23 Curb on grade 20.0
A10 0.56 1.19 64.44 Grate on grade 7.0
A8 0.77 2.22 37.78 Curb on grade 15.0
A14 0.59 2.36 66.66 Grate on grade
A12 0.74 2.31 44.05 Curb on grade 15.0

Table 6.2 Simulated intercepted and bypassed flow from WinStorm and StormCAD.

Runoff from catchment (cfs) Intercepted Bypassed Intercepted Bypassed

Inlet
WinStorm StormCAD WinStorm StormCAD
A2 8.64 8.71 8.639 0.00 8.71 0.00
A4 5.22 5.26 5.221 0.00 5.26 0.00
A6 5.51 5.55 5.507 0.00 5.55 0.00
A10 1.80 1.82 1.739 0.06 1.73 0.08
A8 6.51 6.56 6.507 0.00 6.56 0.00
A14 3.63 3.66 3.479 0.21 3.48 0.26
A12 5.89 5.94 5.889 0.00 5.94 0.00

Table 6.3 Pipe parameters and simulated pipe flow using WinStorm and StormCAD.

Pipe Upstream/ Full Flow in pipe

Diameter
Pipe Length Downstream Capacity WinStorm StormCAD
(ft)
(ft) Elevations (cfs) (cfs) (cfs)
10 100 24.52/23.4 2.0 25.94 5.51 5.55
25 48 25.6/25.38 2.0 18.65 8.64 8.71
20 145 24.88/23.47 2.5 43.82 8.64 8.69
30 490 23.38/22.47 3.0 31.14 14.55 14.67
40 79 23.75/23.66 2.0 8.27 6.51 6.56
50 588 22.45/21.60 3.0 27.48 18.13 18.2
60 49 23.38/23.07 1.5 8.91 5.89 5.94
70 98 21.24/21.01 3.5 52.81 25.75 25.71

110
 As noted in Table 6.3, full capacity available in the pipes is found to be much
greater than simulated flows by WinStorm and StormCAD. Hence one can conclude that
the conduits are over-designed for the specified return period, which was 5 year (TxDOT
output was labeled as 10 year return period, but all flows reported were based on 5 year
return period). Therefore, in order to analyze the optimum return period for the full
capacity available, the storm drainage system for U.S. 77/83 was modeled several times
for different return periods without changing any of its physical characteristics of the
system. Table 6.4 below summarizes the discharge on pipe obtained from WinStorm by
altering the return period for each case. Allowing a tolerance discharge of 5 cfs as
compared with the full capacity of pipe (Qf), they are classified as over-designed (Q < Qf
–5), under-designed (Q > Qf) and appropriately designed pipes (Qf –5 < Q < Qf). In the
Table 6.4, the discharges shown in normal font indicate that pipes are appropriately
designed, while the discharges in underlined and bold fonts indicate that pipes are under-
designed and over-designed, for corresponding return period, respectively.
If U.S. 77/83 was designed for 5-year return period, 20 out of 32 pipes were over-
designed. If U.S. 77/83 was designed for 10-year return period, 15 out of 32 pipes were
still over-designed, while there were six pipes under-designed. One can also conclude
from Table 6.4 that some of the pipes (nine pipes) were over-designed even up to return
period of 100-year. Minimum pipe size used for U.S. 77/83 was 18 inches (1.5 ft) in
order to facilitate cleaning and debris clearing. This modeling test clearly indicates that
there are many pipes, designed by using WinStorm and TxDOT policy in selecting pipe
size, may have large capacity present that will never be utilized. These extra capacities
could be utilized for other purpose, e.g., in-line water quality treatment, which will be
further discussed in the next section.
As noted from Table 6.4, pipe 70 is over designed up to a return period of 100
year while the inlet A14 linking the pipe 70 is under designed as it does not capture all
flow reaching it (Table 6.2). Due to conservatism of Rational Method, which results over-
designed pipes, it is recommended that TxDOT engineers should design inlet as big as
possible or distribute inlets with a shorter distance to capture all incoming flow. There
were two sag inlets (A2 and A4) analyzed (Table 6.1) both WinStorm and StormCAD
always consider 100% incoming flow to be intercepted, which may not be true in the
reality. Sag inlet design in conjunction to geometry design is needed be investigated
further.

111
Table 6.4. Discharge obtained from WinStorm for different return periods for pipes of
U.S 77/83 Highway.

Discharge (cfs) Full

Run Diameter
Capacity
No. (ft) 5 yrs 10 yrs 25 yrs 50 yrs 100 yrs (cfs)
10 2.0 5.51 6.46 7.30 7.96 8.55 25.94
20 2.5 8.64 10.17 11.68 13.05 14.07 43.82
30 3.0 14.55 17.14 19.69 22.02 23.73 31.14
40 2.0 6.51 7.65 8.76 9.73 10.49 8.27
50 3.0 18.13 21.38 24.69 27.85 30.04 27.48
60 1.5 5.89 6.93 7.95 8.87 9.56 8.91
70 3.5 25.75 30.40 35.14 39.66 42.79 52.81
80 3.5 26.14 30.86 35.68 40.27 43.46 30.53
90 3.5 26.47 31.25 36.15 40.83 44.08 50.58
100 3.5 26.85 31.72 36.70 41.47 44.78 46.72
110 1.5 3.51 4.14 4.78 5.39 5.81 7.26
120 3.5 32.24 38.09 44.10 49.84 53.82 42.20
130 1.5 3.95 4.65 5.34 5.96 6.42 10.80
140 3.5 34.62 40.93 47.41 53.65 57.99 41.55
150 1.5 4.16 4.89 5.61 6.25 6.74 22.15
160 3.5 37.42 44.24 51.25 58.03 62.75 177.48
170 1.5 0.35 0.41 0.46 0.50 0.53 9.13
180 3.5 37.52 44.36 51.38 58.18 62.91 40.98
190 1.5 2.94 3.44 3.86 4.17 4.46 5.69
200 3.5 38.34 45.33 52.53 59.50 64.35 59.92
210 3.5 40.84 48.29 55.97 63.42 68.57 52.98
220 1.5 2.38 2.79 3.13 3.37 3.61 4.76
230 2.0 3.15 3.70 4.15 4.48 4.80 57.53
240 3.5 41.43 48.99 56.81 64.42 69.70 52.39
250 1.5 5.77 6.76 7.62 8.28 8.88 6.02
260 1.5 8.31 9.75 11.05 12.09 12.99 17.43
270 3.5 46.97 55.55 64.43 73.15 79.19 55.00
280 1.5 4.77 5.61 6.40 7.07 7.61 5.58
290 3.5 49.60 58.69 68.08 77.38 83.81 71.76
300 3.5 49.60 58.69 68.08 77.38 83.81 49.81
25 2.0 8.64 10.17 11.68 13.05 14.07 18.65
205 1.5 6.00 7.04 7.97 8.72 9.37 7.18

112
 CHAPTER SEVEN
FEASIBILITY OF INLINE STORMWATER QUALITY
TREATMENT

7.1 Concepts of Inline Water Quality Treatment

Urban drainage systems are vital infrastructure assets, which protect our town and
cities from flooding and the transmission of waterborne diseases. They are usually
constructed as a network of buried pipelines that can be either “combined sewers system”
or “separate system”. Initially the separate system used to transport the stormwater away
from one area and directly disposed to the nearest river. It is now realized that the
scheme, whilst removing a potential threat from one area, often simply passes large
quantities of water forward so that it becomes someone else’s problem downstream.
Uncontrolled, rapid urban runoff presents not only an increase in the risk of downstream
flooding, but also has an adverse effect on river habitat due to changes in channel
morphology through man’s actions in the name of flood defense. The modern drainage
engineer is therefore faced with some interesting challenges in maintaining the levels of
flood protection demanded by society without any cause of damages to the natural
environment.
An urban stormwater Best Management Practice (BMP) is a “technique, measure
or structural control that is used for a given set of conditions to manage the quantity and
improve the quality of stormwater runoff in the most cost-effective manner”. BMP can
generally be divided into structural and nonstructural categories. Structural BMPs are
techniques that can be used to address flow control and pollution removal in stormwater
runoff and include infiltration systems, detention systems, retention systems, filtration
systems, and wetlands. Nonstructural BMPs on the other hand are practices designed to
prevent pollutants from entering stormwater runoff or to reduce the volume of runoff that
may include public education, minimizing pollutants disposal, good housekeeping etc.
In-line storage refers to the practices designed to use the unused volume
temporarily available within the stormwater system to store stormwater runoff. While
these practices can reduce storm peak flows, they are unable to improve water quality or
protect downstream channels, as the intent will be to make the system self-cleaning to
reduce maintaining requirements. Storage is achieved by placing devices in the storm
drain system to restrict the rate of flow. Controls to restrict flow can either be fixed or
adjustable. Fixed systems will probably be cheaper and require less maintenance as
compared to that of adjustable systems. Hence, if storage is combined with an end-of-
pipe treatment, the flow attenuation will help equalize the load to the treatment process
and, hence, optimize the treatment plant.
An example of BMP is Stormceptor, an in-line treatment structure. A layout of
typical urban area with Stormceptor installed is shown in Figure 7.1. A Stormceptor
(Figure 7.2) is a pre-fabricated concrete structure designed for remove free oil (i.e.,
hydrocarbons) and suspended solids (i.e., sediment) from stormwater runoff. Two

113
working condition applies: one is the Normal operating conditions and the other is the
By-pass operating condition. Under normal operating conditions (more than 90% of all
storm events), stormwater flows into the upper chamber and is diverted by a u-shaped
weir, into the separation-holding chamber (Figure 7.2). Right angle outlets direct flow
around the circular walls of the chamber. Fine and coarse sediments settle to the floor of
the chamber, while the petroleum products rise and become trapped beneath the
fiberglass insert. During infrequent, high flow events (less than 10% of all storm events),
peak stormwater flows pass over the diverting weir and continue into the downstream
storm sewer system. This by-pass activity creates pressure equalization across the by-
pass chamber, preventing scouring and re-suspension of previously trapped pollutants.
Based on the loading condition, location and the operational condition they can be
positioned in series, in the inlet junction and submerged condition respectively.
There are numerous stormwater treatment systems available today. Their
mechanism or operational procedure that a particular treatment pursues, the various
treatment processes that a system can undergo, the possible pollutants removed, vendor
address are summarized in Figure 7.3a, 7.3b and 7.3c, obtained from CE News (ASCE,
2003). As discussed and concluded through model testing by using U.S. 77/83 data, there
are many pipes having extra capacity which never will be utilized. Can these extra
capacities be used for in-line water quality treatment? For example, pipes with extra
capacity can function as a storage reservoir just before Stormceptor. SWMM will be
used to study feasibility of large pipe as storage reservoir for in-line water quality
treatment and to examine dynamics of flow in the reservoir in the next section.

114
Figure 7.1 Typical layout of city with Stormceptor installed (from
http://www.stormceptor.com/applications.php).

115
Figure 7.2 A typical Stormceptor (from
http://www.stormceptor.com/single_inline_unit.php).

Figure 7.3a Available stormwater treatment systems (ASCE, 2003).

116
Figure 7.3b Available stormwater treatment systems (ASCE, 2003).

117
Figure 7.3c Available stormwater treatment systems (ASCE, 2003).

7.2 Flow Simulation in Larger Pipes by Using SWMM

The overall treatment effectiveness is a function of its pollutant removal rate and
the volume of run-off treated. A high flow by-pass is generally designed into treatment
measures for protection from large flood flows that could damage the device or scour and
transport previously collected pollutants downstream (e.g., Stormceptor). The primary
problems for cost effective stormwater quality treatment facility in the urban area are
space and head. Some treatments require large amounts of hydraulic head for operation.
These are obviously not suitable for use in low-lying areas with mild drain slopes. Also
there are numerous proprietary technologies most of which operate on centrifugal
principles requiring significant head (e.g., hydro cyclones) with specialized maintenance
equipment. Other research shows that this type of technology is not the most efficient
technology available for stormwater treatment. Preliminary investigations done in
association with TxDOT project 0-1837 and 0-4273 suggest that it is possible to combine
some simple detention strategies for water management and off the shelf reinforced
concrete pipe and boxes to make very effective non-proprietary small footprint BMPs for
urbanized areas. If large pipe with extra capacity links with a Stormceptor, the pipe
functions as detention pond or storage reservoir and could be used for in-line stormwater
treatment.
A simple SWMM model was developed to illustrate how the storage capacity
available in the pipe can be used to develop certain amount of head that is required by
some treatment for its normal operation. The simple, hypothetical storm drainage system
used for the case study earlier (Figure 4.1), modeled under WinStorm, StormCAD and
SWMM, was modified to demonstrate the utilization of storage available in the pipe to
generate head. The first three conduits were unchanged, and the last pipe was modified
as a smaller pipe (0.75’ by 0.75’) as shown in Figure 7.4. Upstream and downstream
elevations for last two pipes are also modified but still maintaining the pipe slope. It is
assumed that the last pipe hydraulically functions as a Stormceptor, which only allows

118
limited flow pass through or say it is equivalent to the treatment facility hydraulically.
The model with the same rainfall pattern was simulated using Runoff layer (SCS
Method) and the Hydraulic layer under XP SWMM for a simulation period of 1 hour 30
minutes. Figure 7.5 shows the basic layout in dynamic profile view after solving the
model in Hydraulic layer of SWMM. Figure 7.5 shows clearly the storage capacity and
the head developed in the pipe at time 8:25:20 that can be used (needed) in the
stormwater treatment facility as the rainfall was started as 8:00 for only ten minutes.

Figure 7.4 Layout of dynamic section view before simulation started.

119
Figure 7.5 Storage capacity available and the head developed at time 8:25:20.
Figure 7.6 shows fluctuation of flow with time in conduit L3. It should mention
that pipe size of L3 is 3 by 3 ft, and upstream and downstream invert elevations for L3
are 83 ft and 81 ft respectively. Therefore, when simulated water surface elevation at the
downstream end of pipe L3 is above 84 ft (Figure 7.4), pipe L3 was found to be flowing
full at the downstream end, but not at the upstream end since water surface elevation was
less than 86 ft. This is also depicted in Figure 7.5. Figure 7.7 and Figure 7.8 illustrate
the corresponding time distributed flow in pipe conduit L4 and conduit L2 respectively.
One can see the fluctuation in both discharge and velocity in conduit L3 (Figure 7.6),
which is due to dynamic routing by applying unsteady St. Venant equation. The last pipe
with smaller size, which is hypothetically assumed as Stormceptor or treatment facility,
only allows maximum of 3.07 cfs passing through (Figure 7.7).

120
Figure 7.6 Fluctuation of flow with time on pipe conduit L3 using SWMM.

Figure 7.7 SWMM output of time distributed flow in pipe conduit L4.

121
Figure 7.8 SWMM output of time distributed flow in pipe conduit L2.
Moreover care should be taken in selecting flow capacity of the treatment facility
so that there occur no flooding while generating the head needed. Figure 7.9 shows the
case of flooding in the node and the conduit L3 by making the last pipe size further
smaller (0.25 by 0.25 ft). This resulted maximum flow in L4 to be only 0.2 cfs. The
maximum hydraulic grade line (purple color) shows flooding at the downstream end of
conduit L3 on 8:31:50. Current hydraulic grade line is given as blue color. Figure 7.10
shows water surface elevations in pipes at the end of simulation (9:30). Even maximum
flows through smaller pipe for both model testing may not real flow capacity for any
Stormceptor or other treatment facility, SWMM simulation does give us idea that extra
pipe capacity is feasible as storage reservoir to build necessary head for treatment facility,
and size of treatment facility is also important in order to avoid street flooding and any
further damage to public and private properties.

122
Figure 7.9 Dynamic section view of flooding occurred in Node 3 due to a smaller
pipe size.

Figure 7.10 Dynamic section view at the end of simulation due to a smaller pipe size.

123
 CHAPTER EIGHT
SUMMARY AND CONCLUSION

Storm drain systems are typically designed with one purpose: to carry flow from a
rainfall event away from areas where it is unwanted (such as parking lots and roadways).
Because of this primary functionality, most storm drainage systems are designed to
adequately convey a peak flow rate, based on the characteristics of the watershed and the
rainfall event. Design of storm drainage systems in most of the agencies is generally
accomplished by application of the rational method. The rational method assumes a
linear relation between rainfall rate for the time of concentration of the watershed and
peak instantaneous discharge. Actually the process by which rainfall translates into
runoff is a very complex and unsteady process affected/controlled by many variables.
Rational method is probably adequate for design of simple culverts; the impact of
hydraulic routing in more complex drainage networks calls this procedure into question.
Storm drains are dendritic or "branched" hydraulic systems consisting of contributing sub
basins and their respective inlets, laterals, junctions, trunklines, manholes, and outfalls.
The hydraulics at any point throughout the system can change significantly. These
changes can be due to both the variability of the hydrologic characteristics of the
contributing sub basins and the associated hydraulic features of each branch or feature in
the storm drain system. This study performed is intended to evaluate TxDOT procedures
for storm drain design, not only in terms of the adequacy of current TxDOT practice
relative to new directions in the field, but also in anticipation of the need to evaluate more
complex features that might be required by changes in water quality regulations.

An in-depth literature review was performed to synthesize both the technical

approach (Rational method versus other hydrological methods) and modeling efforts of
drainage networks with various computer software packages, which were summarized in
Chapter Two. Rational Method for storm drainage design was reviewed, including basic
assumptions, how to quantify runoff coefficient, rainfall intensity, time of concentration,
and limitations of the method. Modified rational method was reviewed as one of the
procedures used to develop runoff hydrographs for inlets. Many other methods are
available to develop hydrographs for watersheds to account for unsteady processes from
rainfall to runoff, they are not reviewed in detail (briefly discussed) in this study and are
illustrated in many standard textbooks of hydrology. The design of inlets (especially curb
inlet in grade and sag), flow in uniform and composite gutter, interception flow, inlet
efficiency, concept of carry over flow and bypassed flow were reviewed in detail in
Chapter Two. Moreover, several different computer models or software packages, which
are often used for the storm drainage design, were also included in the literature review.
Several journal papers were reviewed and summarized to explore recent developments in
the field of storm drainage design.

As a simple testing case study, a hypothetical storm drainage system was

examined in detail, as reported in Chapter Four, using several software packages,
including WinStorm, StormCAD, Hydraflow, the custom-developed SPLIT program, and
SWMM. This hypothetical system, comprising four inlets, four conduits and one outfall,
was initially developed and used by a TxDOT engineer, Mr. George (Rudy) Herrmann

124
during his graduate program. The first three models use only the Rational Method and
the steady flow approach for the storm drainage analysis and design. The fourth model,
the SPLIT program, considers a simple time distribution of surface and the sub-surface
flow. The last model, Visual or XP SWMM, developed from the original EPA SWMM,
not only has the option of estimating runoff hydrographs at inlets using different methods
(for example, unit hydrograph, NRCS/SCS method, and kinematic wave) but also
includes the complete unsteady hydraulic open channel pipe flow with a graphical
interface. Results developed using these computer models were summarized, analyzed
and compared. Model results were presented in graphical and tabular format.
Furthermore, results obtained were also compared to manual calculations wherever
applicable. Model testing was then implemented on a real world project, the storm
drainage system for U.S. Highway 77/83 on the Pharr District (southern part of Texas) of
Cameroon County. Model testing was conducted for different return periods. Finally, the
concept and feasibility of in-line water quality treatment was discussed with the aid of
SWMM simulation results.

Through literature review, reviewing the results of the case study and examining
the calculation procedures of various computer models (packages), the following
conclusions were made:

(1). Some computer models, for example, WinStorm and Hydraflow, have a
conceptual disconnect between the inlet and pipe design process. Though the carryover
flow is calculated in the inlet design process (typically based on FHWA HEC-22), it is
ignored in the computation of flow in pipes.

First, the intercepted flow is not used in calculating the travel time in the pipe;
instead, the peak discharge from the sub-catchment is used. Second, the CA product is
simply added algebraically (lumped) from upstream catchments, even though not all of
the flow enters the pipe through the inlet. The lumped CA and the travel time are the
only two variables required for calculation of flow in pipe when using the Rational
method.

(2). In contrast, StormCAD overcomes the conceptual disconnection between two

parts of network design by considering intercepted and carryover CA. Therefore, the
peak discharges in pipes (Table 5.3) (final result for determining pipe size) obtained
using StormCAD are less than those obtained using WinStorm (Table 5.4) and
Hydraflow (Table 5.7). However, the peak discharges are still in the same order of
magnitude. Therefore, these three models could result the same (or similar) pipe size for
the drainage system, depending on the policy used to select the pipe size from calculated
flow. All the three software packages deal only with the instantaneous peak obtained
from the watershed using Rational Method with steady state process in pipe, which is not
the actual case.

(3). Despite known limitations of the Rational method and computer tools based
on the Rational Method, most engineers employ these tools. Because of the limitations in
inherent in the development of the design discharge, and the problems with inlet inflow

125
and bypass, pipes sized/designed /adopted using these methods are usually larger than
necessary (over-designed).

(4). Results from the SPLIT or the modified SPLIT program include simple time
distributed hydrographs of inlet from the watershed and in the pipes, while those from
SWMM include inlet hydrographs and the complete dynamic solution of pipe flows.
SWMM allows the user to examine unsteady flow process in a storm drainage system
subjected to a specified rainfall event. The peak flows in the conduits produced by these
two models varied to some extent, depending on methods for hydrograph generation and
flow routing in the pipe. Peak flows simulated in pipes using SWMM (Table 5.10 and
5.11) and SPLIT (Table 5.8) are smaller than those obtained by WinStorm (Table 5.4),
however the peak flows are still in the same order of magnitude. The differences
between peak discharges developed using SWMM and WinStorm could be greater for
actual application to design problems, instead of the simple rainfall settings and storm
drain network tested here.

(5). Based on the literature review (especially, “Storm Sewer Design Sensitivity
Analysis Using ILSD-2 Model, Nouh 1987), use of the Rational method results in the
highest construction cost and lowest failure risk (conservative), while the SWMM
technique is the most accurate (by considering unsteady and hydrodynamic processes)
among those studied, it produced cost and risk values of 0.83 and 2.30 times the steady
flow values (Rational method). Overall designs developed using SWMM can result in
less construction cost and greater risk of hydraulic failure in comparison to the other
design approaches, if the agency is willing to reduce the amount of hidden conservatism
in the resulting designs. Finally, SWMM is recommended for large drainage systems,
while the rest of the techniques, including the steady flow technique may be used for
small drainage systems.

(6). Two commercial versions of SWMM, namely CAiCE Visual SWMM and
XP-SWMM 2000, were tested in the study. Though both were developed from the EPA
SWMM module, they produced somewhat different results. A simple case was presented
in Figure 3.55a and Figure 3.55b. Therefore, a question arises about which product is
more correct. Little research work of the nature represented by the current study is
present in the professional literature.

(7). Numerous model applications to an actual system for a variety of return

periods were conducted. Some pipes were larger than required (over-designed) for return
periods of up to 100 years. Apart from conservative approach (Rational method), the
selection of over-capacity pipe is also governed availability of pipe in particular diameter
and by design code. Designers may think that over-designed or conservative pipe sizes in
storm drainage system are capable of passing discharges from unexpected heavy rainfalls
that are exceed the design event (perhaps resulting from increased urbanization), but
because inflows to the system are limited by inlet capacity, that extra pipe capacity may
never be used. However, the unused volume (storage) available in the pipe can be
possibly used for in-line treatment of storm discharges to improve the water quality with/
without flow control device (if head is required) as discussed in the Chapter Seven.

126
 (8). If pipe capacity is sufficient to manage flows up to a return period of 100
years (Table 6.4), if the inlet does not capture the incoming flow (Table 6.2), the capacity
is “wasted.” Therefore, it is recommended that TxDOT engineers modify the design
guidelines either to include additional inlet capacity commensurate with pipeline
capacity, or that pipeline capacity can be reduced. Furthermore, WinStorm and
StormCAD assume that sag inlets intercept all incoming flow instantaneously, which may
not be true. During heavy storms, sag inlets may be subjected to a long-standing pool of
water, and none of the models is currently treating the dynamic nature of flow at sag
inlets. Further investigation of the interaction of hydrographs and inlet capacity at sag
inlets is suggested.

Overall, it is concluded that further research into development or application of

sophisticated modeling technology is not currently warranted for design and analysis of
TxDOT drainage network designs. The SWMM model is useful for analyzing flow
dynamics of in-line treatment of storm discharges. Furthermore, TxDOT engineers
should realize the disconnect between inlet design and conduit design using current tools.
In general, the result is that conduits are larger than required for the flows they actually
convey under design conditions. Therefore, there is implicit in current design practice
additional conservatism beyond that implied by the design criteria. This is important
because of the expectation of the number of failures based on anecdotal observations.
Finally, further research into sag inlet design is warranted.

127
 REFERENCES

Akan, A. O. and B. C. Yen. 1981. Mathematical Model of Shallow Water Flow over

Porous Media. Journal of the Hydraulics Division, ASCE: 107 (HY4): 479-494.

Akan, A. O. and B. C. Yen, 1984. Effect of Time Distribution of Rainfall on Overland

Runoff. Proceedings of the 3rd International Conference on Urban Storm

Drainage. Goteborg. Sweden. 193-202.

American Association of State Highway and Transportation Officials (AASHTO). 1990.

A Policy on Geometric Design of Highways and Streets, Washington, DC.

American Society of Civil Engineers (ASCE), 1992. Design and Construction of Urban

Stormwater Management Systems. Manuals and Reports of Engineering Practice

No. 77, New York, NY: WEF Manual of Practice RD-20.

American Society of Civil Engineers (ASCE), 2003. CE News Magazine.

Bauer, W. J. and D. C. Woo. 1964. Hydraulic design of depressed curb opening inlets.

Research Report No. 58, Highway Research Board.

Bedient, P. B. and W. C. Huber. 1988. Hydrology and Floodplain Analysis, Addison-

Wesley Publishing Company.
Bowman, K. 1988. Hydraulic analysis of alternative South Carolina curb inlet
structures. Clemson, SC: Department of Civil Engineering, Clemson University.
Brandstetter, A. B. 1977. “Assessment of Mathematical Models for Storm and Combined
Sewer Management.” EPA-600/22-76-175a, (NTIS PB-259597), Environmental
protection Agency. Cincinnati.
Brown, S. A., S. M. Stein and J. C Warner. 1996. Urban Drainage Design Manual.
Hydraulic Engineering Circular No. 22. FHWA-SA-96-278. Washington, D.C.
U.S. Department of Transportation, Federal Highway Administration.
Brown, S. A., Stein, S. M. and J. C. Warner, 2001. Urban Drainage Design Manual.
Hydraulic Engineering Circular No. 22. Washington, DC: U.S. Department of
Transportation, Federal Highway Administration.

128
Federal Highway Administration (FHWA), 1977. Design Charts for Open-Channel Flow.

Hydraulic Design Series No. 3. Washington DC: U.S. Department of

Transportation.

Federick, R.H., V.A. Myers, and E.P. Auciello. 1977. Five-to 60-Minutes Precipitation

Frequency for the Eastern and Central United States. NOAA Technical

Memorandum NWS HYDRO-35. Silver Spring, Maryland: National Weather

Service, Office of Hydrology.

Fiuzat A. A., C. E. Soares and B. L Sill. 2000. Design of Curb Opening Inlet Structure.

Iranian journal of science and technology 24 (1): 11-21.

Green, W.H. and G. A. Ampt. 1911. Studies on Soil Physics. J. Agric. Sci. 4:1. 1-24.

Gumbel, E. J. 1941. The Return Period of Flood Flows. Ann. Math. Stat. 12:2. 163-190.

Guo, J. C.Y. 2000. Street stormwater storage capacity. Water Environment Research 72:
(5) 626-630.
Hathaway, G. A. 1945. Design of Drainage Facilities. Transactions of the American

Society of Civil Engineers vol. 110.

Herrmann, G. R. 2002. Individual studies on: Time Distributed Storm Sewer

Performance, Considering the Effect of Inlet Restriction.

Herrmann, G. R. 2003. Personal communication through e-mail.

Hershfield, D. M. 1961. Rainfall Frequency Atlas of the United States for Durations from
30 Minutes to 24 Hours and Return Period from 1 to 100 years. Technical Paper
No. 40. Washington, D.C.: Weather Bureau, U.S. Dept. of Commerce.
Horton, R. E. 1933. The role of Infiltration in the Hydrologic Cycle. 14 Trans. Am.

Geophys. Union. 446-460.

Horton, R. E.1945. Erosional Development of Streams and Their Drainage Basins:

129
 Hydrophysical Approach to Quantitative Morphology. 56 Bull. Geol. Soc. Am,

275-370.

Hotchkiss, R. H. and D. Bohac 1991. Efficiency of Highway Storm Water Inlets,

Nebraska Department of Roads. Report RES1 (0099) p 440.
Hotchkiss, R. H., 1994. Improvements in curb-opening and grate inlet efficiency.
Transportation Research Record (1471): 24-30.
Hromadka II, T.C. and R. J. Whitley. 1994. The Rational Method for Peak Flow Rate
Estimation. Water Resources Bulletin 3 (6). American water Resources
Association.
Hromadka II, T.V., r. H. McCuen, and C. C. Yen. 1987. Computational Hydrology in
Flood control Design and Planning. Lighthouse Publications. Mission Viejo.
California.
Huber, W.C. and R. E. Dickinson. 1988. Stormwater Management Model (SWMM)

User’s Manual, Version 4.0. U.S. Environmental Protection Agency. Program

Developed by Environmental Research Laboratory Office of Research and

Development.

Huber, W.C. and R. E Dickinson. 1992. Stormwater Management Model (SWMM)

User’s Manual, Version 4.0. U.S. Environmental Protection Agency. Program

Developed by Environmental Research Laboratory Office of Research and

Development.

Ickert, R. A. and C. E. Crosby. 2003. Storm Water Flow on a Curbed TxDOT Type

Concrete Roadway.

Izzard, C. F. 1946. Hydraulics of Runoff from Developed Surfaces, Proc. Highway

Research Board 26: 129-150, Washington, DC.

Izzard, C. F. 1964. Tentative results on capacity of curb opening inlets. Highway

Research Board. Research Report No. 11-13.

130
Izzard, C. F. 1977. Simplified method for design of curb opening inlets. Transportation

Research Record No. 631.

Kerby, W. S. 1959. Time of Concentration for Overland Flow. Civil Engineering 29(3).

Kiefer, C. J., J. P. Harrison and T. Hixson. 1970. Unpublished Preliminary Report.

Chicago Hydrograph Method.

Kirpich, Z. P. 1940. Time of Concentration of Small Agricultural Watersheds. Civil

Engineering 10(6).

Kuichling, E. 1889. The Relation between Rainfall and the Discharge in Sewers in
Populous Districts. Trans. ASCE No 20.
Mays, L. M. 2001. Stormwater collection systems design handbook. McGraw-Hill

Companies, Inc.

McCuen, R. H. 1998. Hydrologic Analysis and Design. Second Edition, Prentice-Hall,

Inc. Upper Saddle River, NJ.

Metcalf and Eddy Inc. 1971. Storm Water Management Model, Final Report. U.S.

Environmental Protection Agency. Washington, D. C. Water Quality office.

Nouh, M. 1987. Storm Sewer Design Sensitivity Analysis using ILSD-2 Model. Journal

of Water Resources Planning and Management113 (1).

NRCS (National Resources Conservation Service). 1972. National Engineering

Handbook, Section 4, Hydrology. U.S. Department of Agriculture. Washington

DC.

Papadakis, C. N. and H. C. Preul. 1972. University of Cincinnati Urban Runoff Model.

Journal of the Hydraulics Division, ASCE 98 (HY10): 1789-1804.

131
Papadakis, C. N. and H. C. Preul. 1973. Testing of Methods of Determination of Urban

Runoff. Journal of the Hydraulics Division, ASCE 99 (HY10): 1319-1335.

Proctor and Redfern Ltd. and James F. MacLaren Ltd. 1976. Stormwater Management

Model Study, Vol. I, Final Report. Ministry of the Environment. Research Report

No. 47. Ontario, Canada: Environment Canada.

Roesner, L. A., J. A. Aldrich and R. E. Dickinson. 1988. Stormwater Management Model

User’s Manual, Version 4, Addendum I, EXTRAN. Environmental Protection

Agency. 203, Athens.

Sherman, L. K. 1932. Streamflow from Rainfall by the Unitgraph Method. Engineering

News Record 108: 501-505.

Soarres, C. E. 1991. Analysis of the efficiency of an alternative South Carolina storm

water drainage structure. Clemson: Department of Civil Engineering, Clemson

University.

Soares, C. E. and B. L. Sill. Winter, 2000. Design of curb opening inlet structure. Iranian
Journal of Science & Technology 24 (1): 11-21.
Soil Conservation Service (SCS). 1986. Urban Hydrology for Small Watersheds.

Technical Release 55. Washington D.C.

Terstriep, M. L. and J. B. Stall. 1974. The Illinois Urban Drainage Area simulator –

ILLUDAS. Bulletin No. 58. Illinois State Water Survey, Urbana III.

TxDOT (Texas Department of Transportation). 2002. Hydraulic Design Manual.

U.S. Army Corps of Engineers. 1990. HEC-1, Flood Hydrograph Package. Hydraulic

Engineering Center.

U.S. Federal Highway Administration (USFHWA). 1999. User’s Manual for HYDRAIN

132
 Integrated Drainage Design Computer System: Version 6.1. 400 Seventh Street,

SW, Washington, DC 20590. Office of Engineering, Office of Technology

Applications,

Uyumaz, A. 1992. Discharge capacity for curb-opening inlets. Journal of Hydraulic

Engineering 118 (7): 1048-1051.
Viessman, W and G. L. Lewis. 2003. Introduction to Hydrology. Fifth Edition, Pearson
Education Inc., Upper Saddle River, NJ.
Walesh, S.G. 1989. Urban Water Management., New York: Wiley, c1989.
Wanielista, M. P. 1979. Stormwater Management Quantity and Quality. Ann Arbor
Science Publishers Inc. Michigan.
Wanielista, M. P., R. Kersten, and R. Eaglin. 1997. Hydrology: Water Quantity and
Control. 2nd Edition, Wiley. New York.
Wasley, R. J. 1961. Hydrodynamics of flow into curb opening inlets. Journal of the
Engineering Mechanics Division. ASCE 87, No.EM4, 1-18.
Watkins, L. H. 1962. The Design of Urban Sewer Systems. Road Research Technical
Paper No. 55. Department of Scientific and Industrial Res., London, H. M.
Stationary Office.
Wenzel, H. C., B. C. Yen and W. H. Tang. 1979. Advanced Methodology for
Storm Sewer Design – Phase II. Water Resources Center Report No. 140. Urbana,
University of Illinois.
Yen, B. C., H. G. Wenzel, L. W. Mays and W. H. Tang. 1976. Advanced Methodology
for Design of Storm Sewer Systems. Water Resources Center Report No. 112.
Urbana, University of Illinois.
Yen, B. C. 1981. Urban Stormwater Hydraulics and Hydrology. Proceedings of the
second international conference on Urban Storm Drainage Held at Urbana.
Illinois, USA.
Young, G. K., S. E.Walker and F. Chang. 1993. Design of Bridge Deck Drainage,

Hydraulic Engineering Circular No. 21. FHWA-SA-92-010. Washington DC:

U.S. Department of Transportation, Federal Highway Administration.

133
 APPENDIX A

CALCULATION OF WINSTORM AND STORMCAD

A.1 Flow from Watershed to Inlet by Rational Method

Both StormCAD and WinStorm use Equation (2.1) to calculate the peak runoff
rate Q. The dimensional correction factor (Cd = 1.008) is a constant that should be used
if values of intensity and area are in US customary units. The StormCAD incorporates
this factor, whereas the WinStorm neglects it (treated 1.008 as 1.0). In order to illustrate
further differences between WinStorm and StormCAD, Figures A.1 to A.10 are screen
captures of windows showing input data and output results from StormCAD.
The intensity of rainfall for a particular watershed is calculated provided the time
of concentration (tc) in both StormCAD and WinStorm. While in StormCAD, one can use
any of three different sets of general formulations, Equations (2.6) to (2.8), (three
empirical factors are required) or a table for calculating intensity i for a given duration D
and return period, WinStorm uses only Equation (2.6) in conjunction with a database
containing three parameters for each Texas counties. In WinStorm 3.0, the user only
needs to select the return period for design or analysis and county from the menu options.

Figure A.1 Rainfall data input using Equations in StormCAD.

134
 Figure A.2 Rainfall data input as table form in StormCAD.

A.2 Carryover Flow Computation for Each Inlet

Both in StormCAD and WinStorm, after determining the total flow reaching the
gutter and which includes flow from the contributing watershed area and/or the carryover
flow from upstream inlets, compute the total bypassed flow and intercepted flow in the
inlet based on the efficiency (E) of the inlet. The efficiency and the theoretical curb
length, Lt, of the curb were computed using Equation (2.15) and Equation (2.17)
respectively. Furthermore, the efficiency, when multiplied by the discharge from
Equation (2.15), yields the quantity of water taken in by an inlet (intercepted flow) into
the storm pipe and the remainder, Qb, is allowed to continuously flow in the gutter (over
pavement) to the next downstream inlet for inlets on grade, Equation (2.16). In other
words, the quantity of water bypassed is calculated from the total runoff less the
intercepted capacity to the inlet. This is the method that WinStorm quantifies the
quantity of water bypassed (carryover) as described in HEC 22.
On the other hand, StormCAD manipulates the carryover for each inlet on the
basis of non-contributing area (total inlet CA less intercepted CA) of the inlet under
consideration. The intercepted CA is equal to total inlet CA times efficiency (CAt*E or
CAt*Qj/Q), and therefore the carryover CA equals CAt*(1-E) or CAt*Qb/Q. The total
inlet CA (CAt) is the summation of CA from local catchment and the carryover CA from
upstream inlets.
In Figure A.3, local intensity of 5.11 in/hr is calculated based on return period,
time of concentration (Tc = 10 minutes) and coefficients a, b, and n for Lubbock County
in conjunction with Equation (2.6). Local rational flow of 5.149 cfs is the flow from the
contributing watershed, and is calculated on the basis of Q = 1.008 C I A, Equation (2.1).

135
Total flow to inlet I-1 is also 5.149 cfs because there is no carryover from upstream for
the first inlet. Figure A.3 also shows that local inlet CA is 1.0, carryover CA from
upstream is 0.0, and total inlet CA is 1.0.
Figure A.4 shows the inlet computation data. Total intercepted flow is
4.317 cfs from Equation (2.15), and the total bypassed flow is computed to be 0.831 cfs,
from Equation (2.16). The capture efficiency calculated is equal to 83.9%, from Equation
(2.18). The gutter spread (T=16.89ft) is calculated using the Equation (2.9b).

Figure A.3 Watershed (catchment) setting and simulation results for Inlet I-1.

Figure A.5 shows the flow result reaching the Inlet I-1. Intercepted CA is
calculated as 0.839; total CA (=1.0) times efficiency (83.9%). System CA, system
intensity, and system flow time are same as that of inlet CA, intensity and flow time
because of the absence of any node upstream. Since no additional flow is incorporated,
the system rational flow is same as the total system flow in all the cases of the first inlet.
System flow is the discharge getting into the underground storm drainage pipe, therefore
flow into pipe P-1 (Figure A.5) is 4.317 cfs from inlet I-1.

136
Figure A.4 Inlet setting and simulation results for Inlet I-1.

137
 Figure A.5 Flow simulation results for Inlet I-1.

Because not all water flowing in the gutter can enter the inlet (due to its
restriction), carryover flow has to be considered while designing. StormCAD
incorporates the carryover flow in sizing the conduits whereas WinStorm calculates
the carryover but does not take into account in sizing the conduits2. This important
observation will be further explained in details through the following sections.

A.3 Carryover Flow From Inlet Upstream

StormCAD calculates the carryover flow for each inlet (Qbup in Equation 3.1)
based on the non-contributing area of the upstream inlet (carryover CA) and the local
intensity (ilocal) of the inlet under consideration rather than directly considering the
quantity of flow bypassed (not using Qb given in equation 2.16).

Qbup = (Carryover CA)up x ilocal (3.1)

2
And it is important to notice that WinStorm is the principal design tool that TxDOT designers use for
storm drain networks.

138
 As discussed earlier, WinStorm does calculate the carryover simply by using
equation (2.16). As shown in Figure A.6, for Inlet I-2, the carryover CA equals 0.161
acres, which is calculated from the total CA (1) less the intercepted CA (0.839) for the
upstream Watershed I-1 (Figures A.4 and A.5). Carryover rational flow was computed to
be 0.831 cfs; and was calculated by the rational method (1.008 times Carryover CA times
local intensity). Thus, the total flow to the Inlet I-2 (5.980 cfs) is the summation of local
flow (5.149 cfs) and carryover rational flow (0.839 cfs) or equal to total inlet CA (1.161)
times local intensity. The total inlet CA is equal to local inlet CA (1.0) plus upstream
inlet carryover CA (0.161) as present in Figure A.6. In summary, StormCAD uses local
rainfall intensity and carryover CA to recalculate carryover flow, while WinStorm
directly uses carryover discharge from upstream inlet. The local intensity is a
function of time of concentration at local inlet, therefore rainfall intensity for different
inlets can be different. In order to compare results between WinStorm and StormCAD,
the time of concentration for the catchment of Inlets I-1 and I-2 was set to 10 minutes, but
StormCAD allows use of a minimum of 5 minutes for storm drainage design.

Figure A.6 Watershed (catchment) setting and simulation results for Inlet I-2.

139
 The results of computations displayed on Figure A.7, i.e., inlet setting and

simulation results for Inlet I-2, is similar as the previous inlet I-1 (Figure A.4). The total

intercepted flow (4.803 cfs) and total bypass flow (1.177 cfs) were calculated based on

the total flow to inlet (5.980 cfs in Figure A.6), and capture efficiency is 80.3% also

based on the total flow to inlet (4.803/5.980).

Figure A.7 Inlet setting and simulation results for Inlet I-2.

In both WinStorm and StormCAD, the maximum time of concentration is used to

calculate the intensity and, therefore, the peak discharge of the watershed into
underground pipes. The maximum time of concentration is the maximum value of either
the sub-watershed time of concentration or the summation of pipe travel time and the
time of concentration for upstream watershed(s), whichever is greater. The flow
simulation results are shown on Figure A.8 for Inlet I-2. The upstream tc equals 13.56
minutes, which is the time of concentration of the Inlet I-1 plus the travel time in the pipe
[3.56 minutes = 600ft/(2.81ft/s)/(60s/min)].

140
 Figure A.8 Flow simulation results for Inlet I-2.

Intercepted CA for the Inlet I-2 equals 0.933 (Figure A.8), obtained by the total
CA contributing this inlet (I-2, which is 1.161 not only local CA =1.0) times the
efficiency (0.803) of the same inlet. System CA is estimated as 1.711, which is obtained
by the summation of carry over CA from upstream inlet I-1 (0.839) (related to flow into
pipe P-1 from I-1) and intercepted CA (0.933) (related to flow into pipe P-2 from I-2).
This system CA in conjunction with intensity calculated from the maximum time of
concentration will be used to compute flow into the pipe by applying Rational method as
discussed below. Computation for other inlets at downstream and outlet will be the same
as discussed above and is not presented here.

A.4 Flow Computation in Pipes

The flow that enters the inlet is conveyed to the downstream through the conduit.
Knowing the discharge, slope, and the cross section of pipe, the depth of flow in pipe is
calculated by an iterative process using steady normal flow equation. The velocity of
flow is then interpreted by the continuity equation (Q = AV). The time of flow traveling
in the conduit, which is an important factor in determining the total time of concentration
on inlets other than in the first one, is reckoned once the velocity and the length of pipe is
known. Both WinStorm and StormCAD apply the above-mentioned procedure to

141
manipulate the velocity, time of flow traveling in the pipe and the depth of flow, while
flow entering the pipe may be different.
For example, the pipe shape, material, pipe size, upstream and downstream invert
level and length of the conduit are the inputted or assumed parameters (Table 3.3) for this
case study. The average velocity is manipulated after calculating the flow depth in the
conduit as system flow from Inlet I-1 was calculated as 4.317 cfs (Figures A.4 and A.5).
The computed values of full capacity, excess design capacity, and excess full capacity are
as shown in Figure A.9.

2.1.2 Figure A.9 Pipe settings and flow simulation results in pipe P-1.

Pipe setting and simulation results for pipe P-2 using StormCAD are shown on
Figure A.10. The total system flow used is 7.977 cfs, which is the same as system flow at
inlet I-2 (Figure A.8). Since the system flow at the inlet does consider carryover flow
from upstream inlets as explained previous, this clearly indicates that StormCAD
considers carryover flow from upstream inlets for its pipe flow calculation, therefore
for sizing the pipe in the storm drainage design. While StormCAD does not just and
simple add flow from upstream pipe (e.g., 4.317 cfs at pipe P-1) and intercepted flow

142
from current Inlet I-2 (e.g., 4.803 cfs in Figure A.7), because those peak flows most
likely occur at different times.
If WinStorm is used, it simply adds the CA from upstream catchments and the
local CA to get the total CA for flow computation, therefore the total CA for flow
entering pipes P-1 and P-2 (Figure 3.1) in WinStorm is 1 and 2, respectively, in the case
study (CA = 1 for both I-1 and I-2 catchment). Intercepted and carryover flows are
calculated at each inlet in WinStorm and carryover flow is also combined with local flow
in computation of spread and gutter depth for each downstream inlet. Intercepted and
carryover flows are neither used for pipe flow computations nor to size storm drainage
pipes. The procedure implemented in WinStorm results in designed pipe sizes that are
greater than required by actual design flows.

Figure A.10 Pipe settings and flow simulation results in pipe P-2.

143
 APPENDIX B

HYDRAFLOW SETUP
The primary network setup of the case study in Hydraflow is shown in Figure B.1.
The network in Hydraflow is basically made up of “Lines”. A line has a length of
stormwater pipe with a junction at the upstream end. Junctions can be manholes, inlets, j-
boxes (junction), or other structures where losses or gains of flow occur. The graphical
part of the network of storm drainage system in Hydraflow is portrayed in the Plan view.
It automatically numbers the lines in the order that they are input. However, the first line
starts from the outfall or the most downstream end. Therefore, the first line that is input
will be line number 1 and the second line input will be number 2, and so on. The
remaining data for the links are completed by switching to the “Lines” view and filling
the blanks needed. For example, the flow data and the physical data such as invert
elevations, pipe sizes, etc. is filled in the spreadsheet display table provided. The
junction data (inlet data) are completed by switching to “Inlet” view. The data in the
“Inlet” view, for the study is acquired from the Table 4.1 and Table 4.2.

Figure B.1. Network setup of the case study in Hydraflow.

144
 During calculation, Hydraflow automatically computes the rainfall intensity from
its own IDF curves for use in the Rational Method. Rainfall data in Hydraflow is entered
either in the form of coefficient defined in Equation 2.6 (shown in Figure B.2) or in the
polynomial mode as characterized in Equation 2.8. Moreover, it also has the option of
entering the intensity values for different time durations and different return periods
respectively, in generating the IDF curve. Additional option of creating the IDF curve is
from Map Data (NWS precipitation data). This option is of less concern to the study and
not discussed further.

Figure B.2. Rainfall IDF Curve generated in Hydraflow

Hydraflow allows specification of certain design constraints and initial conditions

before calculating the results. The values, as shown in Figure B.3 are editable at any time.
The data for the design codes are divided into eight categories plus an additional section
for inlet default values. A minimum time of concentration of 10 minutes was the only
option altered in Design Codes while the rest were allowed to be the default values.
Checking the Flow Option “Use inlet captured flows in system” will ignore the system
flows for pipes calculated through the system intensity but only incorporate the flow
captured by the inlets. Hydraflow has four options for computing/generating results as
shown in the Figure B.4.

145
Figure B.3. Design constraints and initial condition setup in Hydraflow.

Figure B.4. Calculation option dialog in Hydraflow.

146
 APPENDIX C
SPLIT PROGRAM

C.1 Introduction to SPLIT Program

George R. Herrmann, P.E., executed a study titled “Time Distributed Storm Sewer
Performance, Considering the Effect of Inlet Restriction.” In his report, he stated that
“no extra security is gained in the system by the conservative design procedure, due to
the limiting factor in the system, which is the restriction of the inlet. The performance of
a storm sewer system can be only as good as the performance of its most restrictive
component…” (Herrmann, 2002). The central premise of his report was based on three
hypotheses, that are as follows:
1. The traditional method of sizing conduits is overly conservative, resulting in
conduits that are overly large for their purpose,
2. The system cannot perform at a level any greater than its most restrictive
features, which are the inlets, and
3. Hypothesis (1) is compounded by hypothesis (2), and therefore any excess
capacity built into the system is valueless, as it can never come into use, because of the
restriction of the inlets.
In order to verify the hypothesis and support his statement, he developed program
coded in FORTRAN 77 called “SPLIT”. The SPLIT program considers the following
conditions to exist:
• The actual area is multiplied by the runoff coefficient, giving an “effective area”,
• The area is essentially rectangular, such as a stretch of roadway, and that the rise and
recession of contribution are both linear.
• If time of concentration tc for a watershed is less than the storm duration, there is a
rise to full contribution, constant flow for the difference between tc and duration, then
recession to zero over tc after cessation of rainfall.
• If tc is equal to the storm duration, there is a rise to full contribution, followed by a
recession over tc back to zero.
• If tc is greater than the storm duration, there is a rise at the same rate of rise as above
until cessation of rainfall, then a recession at the same rate to zero.
• Rainfall duration should be integer minutes.
SPLIT takes into account only the traditional rational method for analysis. It is a simple
program to assess the hypotheses presented above and not intended as a design or
analysis tool. The following items are not addressed in the analysis:
1. It was developed only for curb inlets without depression. No other types of
inlets and gutter with depression can be tested or analyzed.

147
 2. Backwater effects within a system
3. Kinematic, diffusive, or dynamic hydraulic effects
4. Realistic rainfall distributions.

C.2 Modification of SPLIT Program

After reviewing the FORTRAN program - SPLIT, the program was modified in
manipulation of efficiency of the inlets. A gutter depression of 0.04 ft was used in
WinStorm for calculating the discharge and the carryover flow to the downstream inlet.
The program was therefore modified to incorporate the gutter depression for its
calculation and also to update hydrograph generation method - the modified Rational
method when the rainfall duration is less than the time of concentration.

C.3 Discussion on SPLIT Program Outputs

TxDOT design guidelines recommend that time of concentration be greater than
or equal to ten minutes. Therefore, that value was used for all watersheds when using
WinStorm for storm drainage design. To compare results from WinStorm with those
from the SPLIT program, the time of concentration for the second node was chosen to be
ten minutes despite the fact that the local time of concentration was estimated to be five
minutes (Table 4.1). For a ten-minute storm duration, total time period for simulation
was 35 minutes (at a computational interval of one minute), which was sufficient to trace
the full flow path, i.e. for all flow to drain from the gutter to the inlet.
For the first node, runoff from the catchment and the flow in the gutter coincide
due to the fact that all of the flow from the catchment has to be conveyed to the gutter
provided there is no loss of any form. Peak flow occurs when the hydrological remote
point of the watershed contributes the outlet, i.e. the time of concentration, which in this
case is 10 minutes. The peak discharge was found to be 5.1 cfs, which can be obtained
from the rational formula equation. The carryover flow to the subsequent inlet starts
when the length of local inlet is not sufficient to capture the inflow reaching it and equals
to the inlet efficiency times inflow reaching the node 1. As a result, the carryover flow
starts at 4 minutes after rainfall starts (Figure C.1) and tends to reach a peak value (0.8
cfs) at 10 minutes and then decreases to zero after certain period of time (16 minutes in
Figure C.1).
After entering the inlet, the flow is conveyed to the downstream through the
underground conduit. The flow entering the conduit P-1 (subsurface flow of inlet I-1) is
shown in the Figure C.2 with a peak of 4.3 cfs. The total flow in the pipe is the
summation of flow intercepted through the inlet and any flow from upstream node. The
total flow and the intercepted flow are same for pipe P-1, because of no flow is added
from the upstream to Node 1.
Similarly for node 2, total flow in the gutter (peak flow = 5.4 cfs occurs at t = 10
minutes in Figure C.3) is the addition of flow from the local catchment (5.1 cfs at t =10
minutes) plus the carryover flow from the previous node (0.3 cfs at t = 10 minutes). As
seen in Figure C.3 the time position of the carryover flow from node 1 (Figure C.1) has a

148
lag of 3 minutes in node 2, which is due to the gutter travel time (3 minutes). The
carryover at node I-2 to the subsequent inlet (I-3) begins only after the efficiency of the
inlet start falling below 100%, and has a peak of 0.9 cfs at t = 10 minutes with turning
point of 0.5 cfs at t = 13 minutes (Figure C.3). The total subsurface flow for the inlet I-2
(flow in conduit P-2 shown in Figure C.4) is the summation of the flow reaching the node
2 from the upstream pipe (P-1) and the flow intercepted by the inlet I-2. The flow rate in
the pipe P-2 reaches 7.4 cfs at t = 10 minutes, has a peak of 7.8 cfs at t = 13 minutes, and
reduces to zero at t = 24 minutes. This hydrograph includes integrated effects from
catchment runoff, intercepted flow and carryover flow of both inlets, and upstream pipe
flow. This process of interception and carryover is continued (due to the inlet restriction)
until the flow reaches the downstream outlet.
Figure C.5 gives the hydrograph at the inlet I-3, where rainfall duration (10
minutes) is shorter than the catchment time of concentration (15 minutes), by using
modified SPLIT program. The response of flow from the catchment to the inlet tend to
rise linearly from zero until the duration of storm (10 minutes) is reached and remains
constant for a duration equal to time of concentration (15 minutes) and then descend back
to zero at the same rate of increase. Therefore, the inlet is likely to receive runoff from
the watershed for a total duration of 25 minutes (Figure C.5) even though the storm last
for only 10 minutes. The peak discharge between 10 to 15 minutes is 3.4 cfs, which is
only a portion of peak discharge computed by the rational method: 10/15 CIA = 10/15 x
5.1cfs = 3.4 cfs. Figure C.5 also shows other surface flow components at the inlet I-3.
Figure C.6 gives an example of the surface flow hydrograph at the inlet 2 if actual
time of concentration of 5 minutes (Table 4.1) is used for catchment linking to the inlet.
This results that the storm duration (10 minutes) is greater than the watershed time of
concentration (tc). From Figure C.6, it is observed that though the time of concentration
of the watershed is only 5 minutes, the peak discharge of 5.1 cfs (Rational method) starts
at 5 minutes, remains constant until the end f storm, and then falls linearly to zero within
5 minutes (tc). this is a trapezoidal hydrograph based on modified rational method as
discussed in details in the section 2.2.2 of Chapter two.

149
 NODE 1 (surface flow)

Runoff from catchment Carryover flow from U/S Total flow in gutter carryover flow to D/S

4
Flow (cfs)

0
0 5 10 15 20
Time (min)

Figure C.1. Simulated surface flow components at the inlet I-1 (node 1) by using modified SPLIT program.

150
 NODE 1 (Subsurface flow)

Intercepted flow Flow from U/S Node Total flow through Pipe 1

4
Flow (cfs)

0
0 5 10 15 20
Time (min)

Figure C.2. Simulated subsurface flow components at the inlet I-1 (node 1) by using modified SPLIT program.

151
 NODE 2 (surface flow)

Runoff from catchment Carryover flow from U/S Total Gutter Flow Carryover flow to D/S

4
Flow (cfs)

0
0 5 10 15 20
Time (min)

Figure C.3. Simulated surface flow components at the inlet I-2 (node 2) by using modified SPLIT program.

152
 NODE 2 (Sub surface flow)

Intercepted flow Flow from U/S Node Total flow

6
Flow (cfs)

0
0 5 10 15 20
Time (min)

Figure C.4 Simulated subsurface flow components at the inlet I-2 (node 2) by using
modified SPLIT program.

153
 NO DE 3 (Surface Flow)

Runoff from catchment Carryover flow from U/S Total Gutter flow carry over

3
Flow (cfs)

0
0 5 10 15 20 25
Tim e (m in)

Figure C.5. Simulated surface flow components at the inlet I-3 (node 3) by using
modified SPLIT program.

154
 NODE 2 (surface flow)

Runoff from catchment Carryover flow from U/S Total Gutter Flow Carryov

4
Flow (cfs)

0
0 5 10 15
Time (min)

Figure C.6. Simulated surface flow components at the inlet I-2 (node 2) by using
modified SPLIT program (Tc=5 min.).

155
 APPENDIX D
Case Study Using SWMM

D.1 Setup of Storm Drainage Network

SWMM consists of three principal computational blocks (layers or modes):
Runoff, Transport (Sanitary), and Extran (Hydraulic) blocks. The basic setup of the case
study model in SWMM is shown in Figure D.1. Four inlet nodes (N1 to N4 in Figure
D.1) were built in Runoff layer while the fifth outfall node (N5 in Figure D.1) was
created in Hydraulic Layer. There are two options to connect conduits with four inlet
nodes and the outfall node. If the conduits connecting nodes are created under the
Runoff layer, non-linear reservoir approach is used to simulate runoff passing through
conduits. If the conduits connecting nodes are created under the Hydraulic (Extran)
layer, the St. Venant equations (one-dimensional, unsteady state continuity and
momentum partial differential equations) for dynamic wave simulation (Roesner et al.,
1988) on an open or closed conduit wastewater or stormwater system are solved to
generate hydrographs.

Figure D.1. Network setup of case study in CAiCE SWMM.

D.2 Precipitation and the Infiltration Data

The precipitation data and the infiltration data are the minimum required data in
SWMM that are to be inputted in the Global database. The unavailability of the
infiltration data made us to select the smallest possible value using Horton infiltration

156
equation, so that the effect is negligible, as shown in the Figure D.3 below. The record
name for infiltration used was TxDOT.inf (Figure D.2).

Figure D.2. Infiltration Dialog box in SWMM.

Figure D.3. Input parameters for Horton Equation in SWMM.

Rainfall data are the most important hydrological input required by SWMM. A
hyetograph of rainfall intensities versus time is required for the period of the simulation.

157
For single event simulation, the data is usually entered for a gauge directly as a synthetic
design storm. For continuous simulation, a historical rainfall sequence is normally used.
Each sub-catchment in SWMM references a single rain gauge. The data for each rain
gauge may be input manually or defined to come from an external interface file. It is
allowable to use a combination of manually entered rain gauges and some to be read in
from an external interface file. Current case study used constant time interval of 1
minute with intensity of 5.11 in/hr with duration of 10 minutes (TxDOT rainfall) as input
for rainfall data (Figure D.4). This rainfall intensity was used for the case study under
WinStorm and StormCAD and was based on IDF curve at Lubbock County, Texas. One
should keep in mind that WinStorm and StormCAD always recalculate rainfall intensity
based on time of concentration for each inlet and pipe before the Rational method is
applied to compute the peak discharge, while SWMM applies given rainfall hyetograph
to all inlets and pipes and to simulate their hydrological responses. SWMM does
allow user to link different rainfall patterns to different catchments, which is not used in
our case study.

Figure D.4. Rainfall data as constant time interval for duration of 10 minutes.

D.3 Sub-catchment Data and Hydrograph Generation

SWMM has the capability of analyzing multiple sub-catchments for each inlet
and allow the user to input up to five sub-catchments for a single node as shown in the
Figure D.5 when hydrograph generation technique used is original SWMM runoff non-
linear reservoir method. Required data for each sub-catchment are area, percent
impervious, width, and longitudinal slope. Sub-catchments are modeled as idealized
rectangular areas with the slope of the catchment perpendicular to the width. In this case
study, the catchment area is 1 acre square with width and length of 208 feet, and the slope

158
of the catchment was assumed to be 1%. For a real application, catchment slope and
area should be estimated from topographic map or determined from field surveying;
while the width of idealized rectangular can be still a calibration parameter.

Figure D.5. Sub-catchment input parameter dialog box.

How does hydrograph generated from sub-catchment depend on watershed

parameters? Wayne C. Huber (1988) tested “runoff non-linear reservoir” hydrograph
generation technique using five hypothetical sub-catchments (Figure D.6a), and results
are given in Figure D.6b. These outflow hydrographs were generated for continuous
rainfall with duration of 20 minutes and by using a time step of 5 minutes. Clearly, as the
sub-catchment width is narrowed (i.e., the outlet is constricted), the time to equilibrium
increases. Thus, the equilibrium discharge is achieved quite rapidly for cases A and B
and more slowly for cases C, D and E. A shape effect is also evident. Theoretically, all
the hydrographs reach a peak simultaneously (at the cessation of rainfall). However, a
large width with shorter time of concentration will cause equilibrium outflow to be
achieved rapidly, producing a flat-topped hydrograph for the remainder of the rainfall
(similar to modified Rational method discussed in Chapter two). Thus, for a catchment
schematized with several sub-catchments and subject to variable rainfall, increasing the
widths tends to cause peak flows to occur sooner (Huber, 1988).
A storage effect is very noticeable, especially when comparing hydrographs A
and E for duration of 20 minutes. The sub-catchment thus behaves in the familiar
manner of a reservoir. For case E, the outflow is constricted; hence, for the same
amount of inflow (rainfall) more water is stored in the catchment and less released before
the cessation of rainfall. For case A, on the other hand, water is released rapidly and little
is stored in the catchment. Thus case A has both the fastest rising and recession limbs of
the hydrographs. Figure D.6c gives sub-catchment hydrographs developed by Kinematic
Wave Method with different basin widths from SWMM simulation for comparison.

159
Figure D.6a. Different sub-catchment shapes to illustrate effect of sub-catchment width
(Huber, 1988.)

Figure D.6b. Sub-catchment hydrographs for different shapes of Figure D.6a (Huber,
1988).

160
 Ki nemat ic W ave M et ho d

4 Width = 208 ft
Discharge (cfs)

Width = 50 ft
3 Width = 100 ft
Width = 300 ft
2

0
8.00 8.10 8.20 8.30 8.40 8.50 8.60 8.70
Tim e (hr)

Figure D.6c. Sub-catchment hydrographs developed by Kinematic Wave Method with

different basin widths.

For each sub-catchment, SWMM requires user to specify a routing method in

order to convert rainfall into a hydrograph, and this is so called hydrograph generation
techniques for catchment. SWMM can route the runoff from the rainfall data and
generate the hydrograph using five groups of techniques: SWMM Runoff Non-linear
Reservoir Method, Kinematic Wave Method, Laurenson Non-linear Methods, SCS Unit
Hydrograph Method, and Unit Hydrograph methods. There are six available methods to
specify unit hydrographs: Nash, Snyder (Alameda), Snyder, Rational Formula,
Time/area, and Santa Barbara Urban Hydrograph (Figure D.7). In spite of different
available techniques, the “Rational Formula” is used as the primary routing method in
this chapter to route the rainfall data to each inlet, and is compared with SCS unit
hydrograph method later. For Rational formula, it requires to input runoff coefficient
and time of concentration to generate unit hydrograph. Except SWMM “Runoff Non-
linear Reservoir” method, other routing methods only use catchment area and do not use
other catchment geometry parameters (Figure D.5) to generate input hydrograph for inlet.

161
 Figure D.7. Unit hydrograph methods available in SWMM.

D.4 Simulation Control Data

The runoff simulation is controlled by Runoff Job Control dialog box under
“Special” pull-down menu. Evaporation, Time control, and Print control are the three
essential control data to be inputted while the rest are optional. A default value of 0.1
mm/inch is selected as the evaporation rate which otherwise it pops a warning message
stating that the model during the simulation will use the default evaporation rate. Time
series of simulated flow parameters at individual node and conduit will be printed in the
OUT file if selected in the print control (default option). Finally, time control
information consisting of starting and ending time of the simulation has to be entered in
the time control dialog box. The runoff layer has three time steps: wet which is used
when the precipitation is occurring, transition which is used when the watershed has
surface storage but the precipitation has stopped, and dry which is used for the inter-event
times of the simulation. The same date as in the rainfall starting date should be used for
the runoff layer Job control.
After Job Control is set up appropriately, simulation results, i.e., runoff
hydrograph at each inlet and pipe, are stored on binary interface files when “solve”
button is clicked. Each of the SWMM layers, when are solved, can read and write
certain types of interface files. The most common set up for stormwater system
modeling uses rainfall interface files for runoff layer, runoff layer saves the flows
(hydrograph) for all active nodes to an output SWMM interface files, and hydraulics
layer reads the existing runoff layer interface file (i.e. hydrograph at each inlet) to
perform dynamic routing in storm drainage pipes. The interface file name should have an
extension .INT as convention. The Runoff module is also capable of creating and reading
Rainfall interface files plus reading a Temperature interface file created by the utilities
module. An interface file resulting from runoff layer is created for later use as an input in
the Extran or Hydraulics layer (Figure D.8 a &b). Not only the interface file name must
be the same, but also file location or directory, e.g. C:\CAiCE\samples\Final\” in Figure
D.8a &b, must be specified the exact same, otherwise hydrographs generated in Runoff
layer can not be passed into the Extran (Hydraulics) layer.

162
 The Transport or Sanitary layer determines the quality and quantity of dry
weather flow, calculates the system infiltration, and calculates the water quality of the
flows in the system. Since the quality was not our desired determination the Sanitary
layer is not tested in our case study. Only Runoff and the Hydraulics layer were used for
the study.

Figure D.8a. Runoff interface files selection dialog box.

Figure D.8b. Interface file selection dialog box in Extran mode.

163
2.1.3 D.5 Inlet Input in Extran (Hydraulics) Layer
If nodes have already set up at runoff layer, nodes should add into hydraulic layer
by using “+” button (add object to layer). In this case, flow (hydrograph) at nodes (inlets)
is simulated from runoff layer, and route hydrograph through pipes by dynamic waves.
In an alterative way, one can add both nodes and pipes into hydraulics layer directly, and
one has to specify time series of flow as either user flow or gauged flow at the node.
In SWMM, a node can be a simple inlet accepting storm runoff, or a junction or
manhole linking to several conduits. The hydraulics layer will use the nodal depth at
junction/manhole to divide the flow in downstream conduits based on the nodal
continuity equation and the depth of water in the conduits. A hydraulics node requires a
ground or spill crest elevation and an invert elevation (Figure D.9). Outfall, inflow and
storage information is not mandatory to simulate a node in the hydraulics layer. In the
case study, ground elevation for node N1 (Figure D.9) was set to 94, an invert elevation
of 88 feet (Table 4.2), and no inflow information is given since an interface file from
runoff layer is used (Figure 3.32D.8a &b).
Since the case study has a stipulated inlet size, inlet restriction has to be
considered, and SWMM allows the users to accurately simulate the restriction of flows
through checking “Inlet Capacity” option (Figure D.9). Though inlet types and inlet flow
computation based on HEC–22 are yet to be included in future SWMM model, street
flooding due to capture inefficiency of inlets can be specified either as maximum
capacity or as the rating curve. The rating curve for each inlet in this case study was
developed using WinStorm or StormCAD by altering the discharge and observing the
gutter and intercepted flow. The tabular and the graphical rating curve used for all the
inlets are is shown in Figure 3.34aFigure D.10a and Figure 3.34bFigure D.10b.

Figure D.9. Node data input dialog box in Extran layer

164
 Figure D.10a. Rating curve for the curb inlet of L = 10 ft.

FigureD.10b. Graphical view of rating curve in Figure D.10a.

Conduits or links are added connecting the nodes. The physical parameters of the
conduits are shown in the Table4.3 for the case study. Moreover due to the inlet
restriction, all the flow approaching the node (inlet) may or may not enter the
underground pipe (link). Flow that doesn’t enter the inlet is carried over to the next
subsequent inlet as an overland flow. By default, SWMM does not route the flow on the
road surface, which keeps above the ground due to inlet restriction or spills out from a

165
manhole/inlet due to limited capacity of underground pipes. These flows above the
ground are accounted by modeling the link to act as a dual drainage in SWMM. This is
accomplished by converting the existing link (pipe) to a multiple conduit. In this study,
the overland flow (carry over flow) is carried to the next inlet through gutter. The gutter
section (Figure D.11) has been defined with a maximum depth of 0.5 ft from the curb,
taking road width as 50 ft, and transverse slope of 2.5%. Similar section was defined to
all the following links. The last node (N5) was defined as a free outlet with minimum of
critical or the normal depth as the outfall controlling characteristics.

Figure D.11 Gutter section defined in case study to carry non captured flow
downstream.

2.1.4 D.6 Simulation Results Using Runoff Layer For Both Nodes And Pipes
First, both inlets and pipes (links) were set up under runoff layer. SWMM
simulates runoff hydrograph at inlets, and all flow collected/generated at inlets gets into
underground pipe (no inlet restriction under Runoff layer) and routes through pipe by
non-linear reservoir method. For all inlets except most upstream one, routed hydrograph
from upstream pipe combines with local inflow hydrograph generated through catchment
modeling. Figure D.12 shows hydrograph generated at the inlet N-1 and by using
Rational formula for unit hydrograph method. Peak discharge is 4.6 cfs at 8:13AM
(rainfall starts at 8:00AM and lasts 10 minutes, time of concentration is 10 minutes for
catchment at the inlet N-1). Figure D.13 shows hydrograph and flow velocity after pipe
routing, and peak discharge is slight smaller (4.1 cfs) and at a later time (8:15AM).
Figure D.14 shows hydrograph after combining hydrographs from previous pipe and
local inlet (peak discharge of 8.5 cfs and at 8:14AM). Figure D.15 shows hydrograph at
the last pipe with peak discharge of 13.2 cfs at 8:18AM. The peak discharges at all the
nodes and the conduits obtained after solving the model in the SWMM Runoff layer
using the Rational method and the SCS method as the hydrograph generation technique,
respectively, were summarized in Table 5.9. Results were further explained in the
Chapter Five.

166
 Figure D.12. Hydrograph output of SWMM for Node N1.

Figure D.13. Time series plot of flows and velocity by SWMM for conduit L1.

167
 Figure D.14. Time series flow output by SWMM for Node-N2.

Figure D.15. Time series plot of flows and velocity by SWMM for conduit L4.

D.7 Simulation Results Using Runoff and Hydraulic Layers without Inlet
Restriction
SWMM was used to solve for both the Runoff and the Hydraulic layers without
considering inlet restriction. At first, the model was solved in the Runoff layer only to

168
generate inlet hydrograph. Because of conduits not connected in the Runoff layer, the
maximum flow at a node (inlet) represents the flow from individual catchment alone (not
combined as presented in Table 5.9 and explained in the Chapter Five). Individual
catchment hydrographs so generated in the Runoff layer are shown in the Figures D.16 to
D.18. These hydrographs are the basis of flow input, which are saved as an interface file
for later use in the Extran layer. Secondly, with the aid of the interface file generated in
Runoff layer as the input, the model is now solved in Hydraulics layer only, though
SWMM facilitates to solve concurrently, both or all layer at once.
The lesser maximum flow as noted in Figures D.17 and D.18 for Node-N3 (3.4
cfs at 8:19AM, tc = 15 minutes) and Node-N4 (4.1 cfs at 8:16AM, tc = 12 minutes)
respectively, compared to that of Node-N2 (Figure D.16, 4.6 cfs at 8:13AM, tc = 10
minutes), is due to the evident fact of time of concentration as discussed in Figure D.6b.
Under constant rainfall intensity, the equilibrium discharge (Q = CiA) for a node with less
the time of concentration is achieved faster than one with longer the time of
concentration, hence the peak flow is greater for the node with less the time of
concentration. For a comparison, Figure D.19 represents hydrograph for the Node-N2
under the same arrangement (Max. flow = 5.2 cfs) but simulated by only varying the time
of concentration to 5 minutes (Table4.1). A significant change of discharge is clearly
envisaged.

Figure D.16. SWMM output for Node-N2 solved in Runoff layer only (Tc = 10 minutes).

169
Figure D.17. SWMM output for Node-N3 solved in Runoff layer only.

Figure D.18. SWMM output for Node-N4 solved in Runoff layer only.

170
Figure D.19. SWMM output for Node-N2 solved in Runoff layer only (Tc = 5 minutes).

Hydraulic layer performs a dynamic routing of stormwater and/or sanitary flows

throughout the major storm drainage system to the points of outfall or to the receiving
water system. It uses a link-node description of the sewer system which facilitates the
discrete representation of the physical prototype and the mathematical solution of the
gradually-varied unsteady flow (St. Venant) equation which forms the mathematical basis
of the model. It receives hydrograph input(s) at locations by interface file, which is
transferred from an upstream mode (e.g. the Runoff Block).
The result of solving the model in the Hydraulic layer is revealed in Figures D.20
to D.22. As expected, the maximum flow in the conduit L1 (4.5 cfs at 8:14AM) is
slightly lesser than the maximum inflow from the catchment node N1 (4.6 cfs). The
maximum flow from the conduit L1 is then added to the inflow from the catchment at
node N2 (4.6 cfs) and again routed in the downstream conduit L2. The result after
routing the combined flow is represented in Figure D.21, which shows a maximum flow
of 6.6 cfs and maximum velocity of 3.84 ft/sec. This process of dynamic routing is
continued to the downstream conduits (e.g., Figure D.22 for L3) in a similar manner until
an outfall is encountered.

171
Figure D.20 Time series output of flow, velocity and elevation for conduit L-1 resulted
from Extran layer of SWMM.

Figure D.21 Time series output of flow, velocity and elevation for conduit L-2 resulted
from Extran layer of SWMM.

172
Figure D.22 Time series output of flow, velocity and elevation for conduit L-3 resulted
from Extran layer of SWMM.

D.8 Simulation Results Using Inlet Restrictions

D.8.1 Maximum Capacity of 3.0 cfs to Inlet without Gutter
The model was reworked with an inlet restriction at all nodes using an allowable
maximum interception capacity of 3.0 cfs at an inlet but without a gutter along side of the
road to carry the flow downstream. The modeled was solved for both Runoff and
Hydraulic layers of SWMM using different time step of simulation to test its sensitivity.
Figure D.23a shows the time series output from SWMM with a simulation time step of 5
seconds (Maximum flow = 3.34 cfs) while Figure D.23b illustrates the same model
solved with a time step of 60 seconds (Maximum flow = 3.37 cfs). As a result, very
small fluctuation in the peak discharge was observed by varying the time step of
simulation. The peak discharge in all the conduits seemed to decrease (Table 5.10 in the
Chapter Five), due to the constraining maximum flow through the inlet, as compared to
the peak discharge in the conduit without an inlet restriction. The captured pipe flow will
be limited to or around 3.0 cfs (Figure D.23a). It is possible for the peak flow to be
momentarily a bit higher (for example 3.34 cfs for L1 in Figures D.23a and D.23b) in a
surcharge condition due to volume changes in conduits and downstream effect. Figure
D.24 shows simulated hydrograph and flow velocity at conduit L2. Graphic results
developed by SWMM Hydraulics Layer (Figures D.20 to D.24) also give water surface
elevations at upstream and downstream ends of conduits. SWMM also allows user to
visualize changes of water surfaces at all conduits over time as video images, which will
be given later as example.

173
Figure D.23a Time series output of flow, velocity and elevation for conduit L-1 resulted
from Extran layer of SWMM for a Maximum capacity of 3.0 cfs for inlet
(Time step for simulation 5 seconds).

Figure D.23b Time series output of flow, velocity and elevation for conduit L-1 resulted
from Extran layer of SWMM for a Maximum capacity of 3.0 cfs for inlet
(Time step for simulation 60 seconds).

174
Figure D.24 Time series output of flow, velocity and elevation for conduit L2 resulted
from Extran layer of SWMM for a maximum interception capacity of 3.0
cfs for inlet.

D.8.2 Maximum Capacity of 3.0 cfs to Inlet with Gutter

A diversion or a multiple conduit was designed in the Hydraulics layer, which is a
special type of link object shown in a network as a dashed line. A diversion is used to
divert sanitary sewage out of the storm drainage system or to relieve the storm load on
sanitary interceptors. In the Hydraulics layer, all diversions are assumed to take place at
a node and are handled as inter-nodal transfers.
In order to take into account the lost flow in the nodes due to inlet restriction, a
gutter of section as shown in Figure D.11 was designed to bypass the carryover or non-
captured flow to downstream, so that no flow is lost in the system or node. The flows in
the sub-surface (underground) conduits were noted to be the same as without a gutter
section as summarized in Table 5.10 (Chapter Five), while the quantity of flow along the
road side in the gutter was now viewed. Figure D.25 and D.26 show the quantity of flow
rate in the gutter section with respect to time in the surface conduit Gutter 1 and Gutter 2
respectively. A considerable amount of flow, with a maximum rate of 0.73 cfs and
peaking at 8.16 AM, was found to travel in the gutter section from first node N1 towards
the downstream node N2. One may consider that the approach flow will not only be the
inflow to N2 from the Runoff interface file (from local catchment) but also the flow
arriving in the channel from the conduit “Gutter 1”. Since simulated flows in
underground conduits were reported to be the same with or without the gutter layer, this
led us to believe that SWMM model actually does not combine flow from local
catchment with flow from upstream surface gutter due to carryover.

175
Figure D.25 Variation in flow and elevation with respect to time in the surface conduit
Gutter 1.

Figure D.26 Variation in flow and elevation with respect to time in the surface conduit
Gutter 2.

176
D.8.3 Using Rating Curve With/Without Gutter Section
The case study was investigated again using a rating curve as the restriction for
the inlet. The rating curve (Figure D.10a & b) for each inlet in this study was developed
using WinStorm or StormCAD, by altering the discharge and observing the gutter flow
and intercepted flow. As discussed in the previous section, the conduits resulted the same
flow amount either considering with or without the gutter section (Table 5.10 in the
Chapter Five). The only difference being the flow is reflected graphically if gutter
section is considered. As seen from Figure D.27 and D.28 and also in Table 5.10, the
intercepted flow in the conduit using rating curve is greater than that of restricting the
inlet with a maximum of 3.0 cfs (Figure D.23a and D.24). Moreover, the time to reach
the peak flow is also shifted to a greater value of 8:14AM for L1 for this case (Figure
D.27) than with the similar one previously dealt in section (8.12 AM for L1 in Figure
D.23a).
A lesser amount of flow in gutter, 0.29 cfs in gutter 1 and 0.15 cfs for gutter 2
(Figure D.29 and D.30, Table 5.10) respectively, using the rating curve as compared to
constant discharge of 3.0 cfs, was simulated, and is due to more interception capacity for
the inlet with rating curve than one restricting to constant 3 cfs.

Figure D.27 Time series output of flow, velocity and elevation for conduit L-1 resulted
from Extran layer of SWMM using rating curve as inlet restriction.

177
Figure D.28 Time series output of flow, velocity and elevation for conduit L-1 resulted
from Extran layer of SWMM using rating curve as inlet restriction.

Figure D.29 Variation in flow and elevation with respect to time in the surface conduit
Gutter 1 using rating curve.

178
Figure D.30 Variation in flow and elevation with respect to time in the surface conduit
Gutter 2 using rating curve.

D.9 Simulation Using SCS Unit Hydrograph Method

As an alternative to the Rational method of generating hydrograph by Runoff
layer, hydrographs may optionally be generated by the SCS method also. The difference
between the two methods is that instead of specifying runoff coefficient in Rational
method, pervious area curve number along with the shape factor or the peak rate factor is
employed in the SCS method. The pervious area curve number CN is a dimensionless
number depending on hydrologic soil group, cover type, treatment, hydrological
condition, and antecedent moisture conditions. This number has a valid range from 0 to
100 with typical values from 60 to 90 and 98 for impervious surfaces. The pervious
curve number in our case study was selected as 100, which is an ideal condition of
impervious surface. The typical value for the shape factor K as suggested by Soil
Conservation Service is 484 for a hydrograph where the volume under the rising side of
the triangular unit hydrograph is equal to the volume under the rising limb of the
curvilinear unit hydrograph. Actual values may vary from 300 in very flat swampy
country to 600 in steep terrain. This shape factor depends not only on the size of the
watershed but also on the geographic location.
The SCS method was tested on two different computer software packages of
SWMM, i.e., CAiCE Visual SWMM and XP-SWMM, under the same rainfall pattern as
in the case study. The hydrographs obtained as a result of solving in the Runoff layer for
the first catchment by CAiCE SWMM and XP-SWMM are illustrated in Figures D.31a
and D.31b for different shape factor K values, respectively. Comparing the Figures one
can clearly envisage that the hydrograph (flow vs. time) generated by CAiCE SWMM
and XP-SWMM are inconsistent, since the peak discharge computed by former (Q=3.5

179
cfs) is lesser than computed by the latter (Q=3.9 cfs). The catchment was tested for its
sensitivity to the shape factor value (K=300, K=400, K=484, K=500 and K=600), while
the other input parameters remained unchanged. It is observed that by increasing the
value of shape factor the peak flow is increased while the time base of the hydrograph is
reduced.
Like the Rational method, the SCS method was also tested under different
combinations of flow, rating curve and with and without considering the gutter section on
the surface. Table 5.11 given in the Chapter Five summarizes the peak discharges in the
four pipes under different conditions of with/without inlet restrictions. Under same
setting and condition, the peak discharges in all the cases acquired from SCS method
were found to be less than those obtained by Rational method (Table5.10 in the Chapter
Five).

Com parison of Sim ulated Hydrographs in CAiCE SWMM

(SCS Method)

5.0

4.0
K=300

K=400
Discharge (cfs)

3.0 K=484

K=500
2.0
K = 600

1.0

0.0
0 10 20 30 40
Tim e (m in.)

Figure D.31a. Simulated hydrographs using SCS Method in CAiCE SWMM.

180
 Com parison of Sim ulated Hydrographs in XP SWMM
(SCS Method)

5.0

K = 300
4.0
K = 400
Dishcarge (cfs)

3.0 K = 484
K = 500
2.0
K = 600

1.0

0.0
0 10 20 30 40
Tim e (m in.)

FigureD.31b. Simulated hydrograph using SCS Method in XP SWMM.

The case study was tested again using SCS methodology at the Runoff layer for
nodes, and multiple links (pipes) with gutter were set up at hydraulics layer. Inlets at
hydraulics layer use rating curve (FigureD.10a &D.10b) for inlet capacity. A standard
shape factor of 484 was utilized while simulating the model, as suggested by Soil
Conservation Service for most watersheds. The first watershed contributes a peak flow
of 3.5 cfs (Table5.11) to the node N-1, a part of which enters the node while the rest is
carried over to the next inlet N-2 through the gutter. Figure D.32 shows the flow vs. time
(hydrograph) in the conduit L-1 where the maximum flow reads 3.43 cfs at 8.13 AM,
while Figure D.33 gives for the gutter flow (very small). Figure D.34 to D.38 show
simulated hydrographs in pipes L-2 to L-4 and in their associated gutters.

181
 Figure D.32. Time series output from SWMM for conduit L-1.

Figure D.33. Time series output from SWMM for conduit Gutter1.

182
 Figure D.34. Time series output from SWMM for conduit L-2.

Figure D.35. Time series output from SWMM for conduit Gutter 2.

183
 Figure D.36. Time series output from SWMM for conduit L-3.

Figure D.37. Time series output from SWMM for conduit Gutter 3.

184
Figure D.38. Time series output from SWMM for conduit L-4.

185