INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION

IN BUILDING AND CONSTRUCTION

T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

A comparison of the various methods used worldwide

for the stormwater systems design
G. Badea (1), C. Bacoţiu (2)
(1) gheorghe.badea@insta.utcluj.ro
(2) ciprian.bacotiu@insta.utcluj.ro
Department of Building Services, Technical University of Cluj-Napoca, Romania

Abstract
The objective of this paper is to introduce the Romanian drainage engineers to
different methods for estimating those components of the hydrologic cycle which affect
design decisions (from precipitation to runoff) and to compare these methods in order to
see what has to be done in Romania in order to safely use one model/method or another.
Special attention was given to design parameters such as the flow (runoff) coefficient,
the time of concentration and the rainfall intensity-duration-frequency curves. Finally,
critical observations concerning the present Romanian Standard STAS 1846-90 were
made.

Keywords
Urban hydrology, runoff, sewer systems, hydrologic computer models

1 Introduction
The urban population is increasing rapidly and will continue to do so, as expected
over the following years. Urban development, mainly the construction of buildings and
the extension of roads, is making the soil impermeable and therefore modifies
stormwater flow conditions by generating more surface runoff instead of soil
infiltration. The effect is a constant need to resize, to reinforce and to improve the
efficiency of underground stormwater networks.
Urban hydrologists have focused mainly on sizing these stormwater networks in
order to minimize the risk of flooding. Too often in the past a simplistic approach to the
design of storm sewers was taken, with the design and sizing of conduits derived from
nomographs or basic hydraulic flow equations. As a result of this, excessive surcharging
has been experienced in many instances due to improper design of the hydraulic
structures. Therefore, a thorough hydraulic analysis should be performed to assure that
the system operates efficiently. Urban hydrologists developed a set of methods and

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-1
 INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION
IN BUILDING AND CONSTRUCTION
T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

models capable of simulating the reaction of urban catchments to high-intensity rainfall

events. During the last 30 years, many computer models have been created for the
simulation of the rainfall/runoff process: SCS TR-55, SCS TR-20, HEC-1, HSPF,
ILLUDAS, MIDUSS, ORAGE, SWMM, RERAM, STORM etc.
However, there is sometimes a “black box” impression when using such models and
interpreting the results, especially when their characteristics and work hypothesis are
not fully emphasized or even understood.
In Romania we are facing during recent years an interesting challenge: which design
method or computer model is acceptable to be used according to the present national
design specifications and standards? Or, eventually these standards and design
prescriptions need to be changed? How?

2 Some common stormwater drainage design methods

2.1 The Rational Method
The Rational Method is an empirical equation for the discharge from a catchment
and is applied since 1889 to urban drainage systems. It is widely used around the world,
with different modifications and adaptations. In 1975, a survey among 37 canadian
municipalities revealed that more than 97% of them used the Rational Method for the
design of their stormwater systems.
Mainly, it is an intensity based rainfall prediction method, so it can be used to
estimate peak flows, according to the watershed characteristics and the rainfall event.
The Rational Method relies on several assumptions:
• the drainage area must be small;
• peak flow only occurs when the entire watershed is contributing;
• the rainfall intensity is uniform for a period of time greater then or equal to the time
of concentration;
• the rational coefficient C is independent of the rainfall intensity
The equation used to calculate the peak flow is :
Q = 0.00278 ⋅ C ⋅ i ⋅ A (1)
where Q – the estimated peak flow (m3/s)
C – the weighted rational coefficient for the drainage basin
i – the rainfall intensity (mm/h)
A – the drainage area (ha)
The coefficient of runoff C (rational coefficient) is a unitless number between 0 and
1 and expresses the ratio between the rate of discharge from a watershed and the rate of
rainfall over that watershed. C is closer to 1 when the drainage area is covered and
impervious (e.g. asphalt). The choice of C coefficient shall be based on ground slope,
type of ground or surface cover, size of drainage area and the expected ultimate land use
of the properties within the drainage area. The choice of the coefficient shall be guided
by the following ranges:
- Industrial areas : 0.80 to 1.00
- Commercial business areas, multi-family : 0.65 to 0.90
- Single family residential & low density multi-family : 0.50 to 0.80
- Rural areas, parks, golf courses : 0.25 to 0.55

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-2
 INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION
IN BUILDING AND CONSTRUCTION
T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

Most watersheds contain more than one type of land cover. Therefore, the composite
C will be calculated as a weighted average of individual, elementary C values.
The intensity of a storm is closely related to the duration and frequency (return
period) of that storm. In order to predict the storm characteristics for a certain region, it
is necessary to have measurements over long periods of time. These historical storm
data are analysed, systematised and usually presented in the form of Intensity-Duration-
Frequency (IDF) curves. Graphical rainfall curves may be good for quick estimates and
hand calculations, but they are not well suited for computer analysis. Software programs
prefer to use an equation form. North American manuals [4] recommend one of the
following equations:
a
i= (2a)
(b + t ) n

a ⋅ ( P) m
i= (2b)
(b + t ) n

i = a + b ⋅ (ln(t )) + c ⋅ (ln(t )) 2 + d ⋅ (ln(t )) 3 (2c)

where i – the rainfall intensity (mm/h)
t – the rainfall duration (h)
P – the return period=inverse of frequency (years)
a,b,c,d,m,n - coefficients
The time of concentration is another important parameter which raised many
discussions over time. It is defined as the longest time of flow in the upstream area and
is made up of a time of entry te and a time of flow tf. There are many methods for
calculating the time of entry: Kerby, Kirpich, Schaake, Federal Aviation Agency etc.
These methods are different, but in essence all are based on the type of ground cover,
the slope of the terrain and the distance along the flow path. In most cases, a minimum
time of concentration (5 or 10 minutes) is recommended for small watersheds.
When calculating time of concentration, two common errors should be avoided.
First, in some cases runoff from a portion of the drainage area which is highly
impervious may result in a greater peak discharge than would occur if the entire area
were considered. In these cases, adjustments can be made to the drainage area by
disregarding those areas where flow time is too slow to add to the peak discharge.
Second, when designing a drainage system, the overland flow path is not necessarily the
same before and after development and grading operations have been completed.
Limitations of the Rational Method:
-gives only the peak flow and there is no information about the shape of the runoff
hydrograph;
-is limited to small watersheds (max. 2 km2);
-cannot deal with surcharged or flooded systems;
-storage tanks and overflows cannot be taken into account;
-it is assumed that the shape of the watershed is somehow rectangular, otherwise one
may obtain erroneous results. In particular, if the contributing area does not increase
proportionally with time, the maximum runoff flow may be produced by a design storm
whose duration is less than the time of concentration;
-the runoff coefficient is assumed to be uniform in time (which in reality is not true)

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-3
 INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION
IN BUILDING AND CONSTRUCTION
T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

2.2 Wallingford Modified Rational Method

The peak discharge formula is:
Q = 2.78 ⋅ CV ⋅ C R ⋅ i ⋅ A (3)
where Q – the peak flow (l/s)
CV – the volumetric runoff coefficient for impermeable area (ha)
CR – the routing coefficient (a recommended value is 1.3)
i – the rainfall intensity (mm/h)
A – total upstream connected impermeable area (ha)
The volume of effective rain is calculated using a percentage runoff method PR
CV = 0.01 ⋅ PR ⋅ (Total _ area / A) (4)
The Wallingford PR equation is:
PR = 0.829 ⋅ PIMP + 25 ⋅ SOIL − 0.078 ⋅ UCWI − 20.7 (5)
where PR – the percentage runoff from the total area
PIMP – the percentage of impermeable surface
SOIL – a soil index obtained from a map of Water Rainfall Acceptance
Potential
UCWI – the Urban Catchment Wetness Index, related to the Standard
Average Annual Rainfall depth
The rainfall intensity is obtained from methods derived by the Meteorological Office
for the Flood Studies Report (1975).

2.3 The Superficial Method (Caquot)

The peak flow formula is:
1 v 1 w
Q( F ) = k u ⋅ I u ⋅ C u ⋅ A u (6)
3
where Q – the peak flow (m /s)
F – the rainfall frequency (years-1)
I – the average slope of the watershed (m/m)
C – the runoff coefficient
A – the watershed area (ha)
k,u,v,w – coefficients based on parameters a(F) şi b(F) which are defining
the rainfall intensity, as follows:
0.5 b ( F ) ⋅ a ( F )
k= (7)
6.6
u = 1 + 0.287 ⋅ b( F ) (8)
v = −0.41 ⋅ b( F ) (9)
w = 0.95 + 0.507 ⋅ b( F ) (10)
The rainfall intensity depends on its duration and frequency, as follows:
i (t , F ) = a ( F ) ⋅ t b ( F ) (11)
The peak flow formula is only valid for watersheds characterised by a so-called
“average elongation” M=2, but in any other case it must be corrected with a special
coefficient m, extracted from nomographs.
The watershed elongation is defined as a ratio between the longest hydraulic path L
and the square root of the area of that basin:

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-4
 INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION
IN BUILDING AND CONSTRUCTION
T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

L
M= ≥ 0.8 (12)
A
The evaluation of the equivalent parameters for groupings of j homogeneous
watersheds characterised by Aj, Cj, Ij, Lj is done according to the following table:

Table 1 – Equivalent parameters for a heterogeneous watershed (after [1])

Equivalent
Aech Cech Iech Mech
Parameters
2
 
 
Serial ∑A j ∑C ⋅ A
j j  ∑ Lj  ∑L j

grouping ∑A j

∑
Lj 

∑A j

 I j 

Parallel ∑A j ∑C ⋅ A
j j ∑ I j ⋅ Q pj L ⋅ Q pjMAX
grouping ∑A j ∑Q pj ∑A j

Validity of the Caquot formula:

a) This method gives good results for small urban watersheds (under 200 ha), with
the slope between 0.2% and 5%, and the average (weighted) runoff coefficient between
0.2 and 1.
b) If the urban watershed is larger than 200 ha, it is recommended to use computer
models which can simulate the actual “mechanisms” involved in the network behaviour:
- a hydrologic model for determining the runoff hydrograph at the exit point of each
sub-basin;
- a hydrologic model capable to perform the propagation, composition or
amortisation (for stockage reservoirs) of the runoff hydrographs along the stormsewer
network.

2.4 Computer models used in stormwater drainage design

Models are generally used for studies of quantity and quality associated with urban
runoff in which four major objectives may be identified: screening, planning, design
and operation. Each objective typically produces models with somewhat special
characteristics, and the different models overlap to a certain degree.
Screening models are preliminary, first-cut desktop procedures that require no
computer. They are intended to provide a first estimate of the magnitude of urban runoff
quantity and quality problems, prior to an investment of time and resources into more
complex computer based models. Latter models should be used only if the screening
model indicates the necessity. As an example of screening model is the american
SWMM (Storm Water Management Model) Level I procedure.
Planning models are used for an overall assessment of the urban runoff problem.
They may also be used for first-cut analyses of the rainfall-runoff process and illustrate
trade-offs among various control options. They are typified by relatively large time
steps (hours) and long simulation times (months and years), i.e., continuous simulation.
Data requirements are kept to a minimum and their mathematical complexity is low.

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-5
 INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION
IN BUILDING AND CONSTRUCTION
T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

Various continuous simulation models are on the market. SWMM has had this
capability since 1976, following the earliest work of the Stanford Watershed Model and
the latter widely-used Corps of Engineers STORM model.
A planning model may also be run to identify hydrologic events that may be of
special interest for design or other purposes. These storm events may then be analyzed
in detail using a more sophisticated design model. SWMM can be used in both the
planning and design mode. Planning or long-term models may also be used to generate
initial conditions (i.e., antecedent conditions) for input to design models.
Design models are oriented toward the detailed simulation of a single storm event.
They provide a complete description of flow and pollutant routing from the point of
rainfall through the entire urban runoff system and often into the receiving waters as
well. Such models may be used for predictions of flows and concentrations anywhere in
the rainfall/runoff system and can illustrate the detailed and exact manner in which
abatement procedures or design options affect them. Design models are generally used
for simulation of a single storm event and are typified by short time steps (minutes) and
short simulation times (hours). Data requirements may be moderate to very extensive
depending upon the particular model employed.
In its original form, SWMM was strictly a design model. However, it may now be
used in both a planning and design mode. In addition, it has acquired additional design
potential through inclusion of the Extended Transport Model, Extran. Extran is probably
the most sophisticated program available in the public domain for detailed hydraulic
analysis of sewer systems.
Operational models are used to produce actual control decisions during a storm
event. Rainfall is entered from telemetered stations and the model is used to predict
system responses a short time into the future. Various control options may then be
employed, e.g., in-system storage, diversions, regulator settings.
These models are frequently developed from sophisticated design models and
applied to a particular system. Examples are operational models designed for
Minneapolis-St. Paul and Seattle.

Table 2 – Hydrologic Computer Models (adapted from [8])

Models
QUALHYMO
OTTHYMO

Model
SCS-TR 55
ILLUDAS

characteristics
MIDUSS

STORM

SWMM
HSPF

Single Event Type x x x x x

Continuous Type x x x x
Infiltration x x x x x x x x
Evapotranspiration x x x
Snowmelt x x x
Surface runoff x x x x x x x x
Subsurface flow x x
Reservoir routing x x x x x x
Channel routing x x x x x x x
Water quality x x x x
High ease of use x x x x x x
Low ease of use x x

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-6
 INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION
IN BUILDING AND CONSTRUCTION
T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

SWMM is a large, relatively sophisticated hydrologic, hydraulic and water quality

simulation program. It is not appropriate for all applications or for all personnel (see
Table 2). For instance, hydrologic routing (e.g. prediction of runoff from rainfall) may
be performed simply using standard techniques (e.g. unit hydrographs, linear reservoirs)
described in hydrology texts and suitable for microcomputers. In addition, many other,
smaller Fortran programs are available for urban hydrologic simulation that may be
entirely suitable for a given problem and much easier to implement on a given computer
system. Notable among the hundreds of such program are the Corps of Engineers,
Hydrologic Engineering Center program STORM for continuous simulation and the
Illinois State Water Survey program ILLUDAS for single-event simulation and pipe
sizing. Both have good documentation and user support and have been extensively
tested and utilized by engineers other than the model developers. HSPF is another
alternative for catchments that are primarily nonurban or that require more sophisticated
simulation of pollutant interaction.
In Europe, among other models, the HVM model created by Dorsch firm (Munchen,
Germany) is used in many countries. In North America, models provided by State
Agencies are preferred (ILLUDAS, STORM, SWMM etc.). Here, local design
regulations (usually in the form of a Stormwater Design Manual) are explicitly
reccomending one or another model to be used for a certain operation.
How these computer models work ?
They use simplified algorithms type Muskingum or the linear reservoir model.
For example the HYDROG model (France) is based on the following principles:
a) First step consists in performing the transformation of the storm hyetograph into a
hydrograph at the exit point of each sub-basin of the watershed, with a Muskingum type
relationship, associated with the continuity equation, as follows:
( I − Q)dt = K m ⋅ [ xdI + (1 − x)dQ] (13)
where I – the neat rainfall intensity at the t moment
x – the Muskingum coefficient
Q – the flow at the exit point from the sub-basin at the t moment
Km – a parameter connecting the accumulated volume to a combination
between the input flow and the output flow:
dV = K m ⋅ [ xdI + (1 − x)dQ] (14)
The Km value (which has time units) must be correctly established after a calibration
process. In general, Km is given by empirical functions depending on both the longest
hydraulic path within the sub-basin and the average terrain slope on that path.
b) The second step consists in the routing and addition of the obtained hydrographs
along the tree-shaped sewer network (see Figure 1). Using the Muskingum model with
x=0, combined with the continuity equation, we obtain:
(Q1 − Q2 )dt = K p ⋅ dQ2 (15)
where Q1 – the flow upstream the pipe at the t moment
Q2 – the flow downstream the pipe at the t moment
Kp – the routing parameter of the hydrograph for the current pipe
The Kp value, as well as Km, is determined as a result of several adjustments made
during the calibration of the model. The initial value is given by some empirical
functions which are based on the pipe length, on the friction coefficient and also on the
hydraulic slope and radius.

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-7
 INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION
IN BUILDING AND CONSTRUCTION
T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

Figure 1 – Generation of the hydrographs along the network

3 Conclusions
In Romania, stormwater systems design engineers use a variant of the Rational
Method, imposed by the standard STAS 1846-90. Unfortunately, this standard is almost
unchanged for more than 35 years and needs some amendments. First of all, there is not
enough information regarding the limitations of the method. For several years, in some
cases, design engineers found strange results by applying this method. Questions need
answers, so we decided to take a closer look to methods used in other countries. Also,
another dangerous phenomenon arised a few years ago, with the arrival of specialised
computer software to our market. Many firms and bureaus involved in sewer design
started to use these powerful tools, without having the theoretical knowledge of the
models running behind, and also designing outside the present Romanian regulations.
Or, the design storm has very specific characteristics, varying strongly from country to
country (even from county to county). Therefore, a blind run of an imported model will
definitely produce even worse results than the present standard.
For example, in Romania, IDF curves are provided in STAS 9470-73. The country is
divided into 19 hydrologic regions, each region having its own IDF nomograph. In
order to use these curves inside a computer program, one must obtain the regression
equation which best fits the data.

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-8
 INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION
IN BUILDING AND CONSTRUCTION
T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

It is interesting to observe that for the Cluj-Napoca region, none of the equations 2a-
2c was satisfactory in providing the best fit. For a frequency f=1/20, a Weibull model
y=a-b*exp(-c*x^d) provided a very good correlation coefficient r=0.999862, where
a=812.11168, b=807.48609, c=3.74652, d=-0.85067 and the standard error S=2.207.
The conclusion is a that a serious research is needed in order to bring to the romanian
engineer the latest methods available worldwide, but safely, according to an up-to-date
standard, with respect to the rainfall data available in our country. The work in under
way, but it will require a lot of effort and money, maybe under an international wider
programme, for harmonizing engineering design prescriptions in the EU.

4 References
1. Berthier E., Auzizeau J., Fasquel M., Flahaut B., Rouaud J.-M., Andrieu H. - Le
suivi hydrologique de bassins versants expérimentaux en milieu urbanisé, Bulletin
des laboratoires des Ponts et Chaussées, 1998, no. 218, pg. 59-75
2. Bourrier R.- Les réseaux d’assainissement : calculs, applications, perspectives,
3ème éd., Editions TEC&DOC, Paris, 1991
3. Brière F.G. - Distribution et Collecte des Eaux, Presses Internationales
Polytechnique, Montréal, Canada, 2000
4. Haestad Press - Computer Applications in Hydraulic Engineering, Haestad Press,
Waterbury CT, USA, 1997
5. Hall M.J. - Urban Hydrology, Elsevier Applied Science Publishers, London, 1984
6. Huber W.C., Heaney J.P., Nix S.J., Dickinson R.E., Polmann D.J. - Storm Water
Management Model (SWMM), Version III, Users Manual, Municipal Environmental
Research Laboratory, U.S. Environmental Protection Agency, Cincinnati, Ohio,
1982.
7. Hydrologic Engineering Center - HEC-1 Flood Hydrograph Package-Users
Manual, U.S. Army Corps of Engineers, Davis, California, 1973.
8. Hydrologic Engineering Center - Storage, Treatment, Overflow Model (STORM) -
Users Manual, U.S. Army Corps of Engineers, Davis, California, 1977.
9. Johanson R.D., Imhoff J.C., Davis H.H. - Hydrological Simulation Program
Fortran (HSPF), Users Manual, U.S. Environmental Protection Agency,
Environmental Research Laboratory, Athens, Georgia, 1980.
10. Lazaro T.R.- Urban Hydrology - A Multidisciplinary Perspective, Technomic
Publishing Co, Lancaster, USA, 1990
11. Linsley R.K., Kohler M.A., Paulhus J.L.H - Hydrology for Engineers, 3rd edition,
McGraw-Hill, New-York, 1982
12. National Corrugated Steel Pipe Association - NCSPA Blue Book – Modern Sewer
Design, Captured in 06.06.2002. Internet page. On line.
http://www.ncspa.org/bluebook.pdf
13. Natural Resources Conservation Service (NRCS) - Urban Hydrology for Small
Watersheds. Technical Release No. 55, USDA, Washington D.C., 1986
14. Pederson J.T., Peters. J.C., Helweg D.J. - Hydrology by Single Linear Reservoir
Model, Proceedings ASCE, Journal of Hydraulics Division, 106 (HY5), pp. 837-
842,1980.
15. Rowney A.C., Wisner P.E. - QUALHYMO Users Manual, Release 1.0, Department
of Civil Engineering, University of Ottawa, Ottawa, Ontario, 1984.

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-9
 INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION
IN BUILDING AND CONSTRUCTION
T E C H N IC A L U N IV E R S IT Y “G H .A S A C H I ” IA ŞI, R O M A N IA

16. Smith A.A. - Microcomputer Interaction Design of Urban Stormwater Systems

(MIDUSS), Users Manual, Version 4.2, Dundas, Ontario, 1987.
17. Terstriep M.L., Stall J.B.- Illinois Urban Drainage Area Simulator (ILLUDAS),
Illinois State Water Survey, Bulletin 58, Urbana, Illinois, 1974.

5 Main autor presentation

Gheorghe BADEA is Professor of Sanitary Engineering at the
Technical University of Cluj-Napoca, Department of Building
Services.
Prof. Badea is the President of the “Association of Romanian
Building Services Engineers” (AAIR) - Transylvania subsidiary.
Also, he is member of many other professional associations:
ASHRAE, “Association des Energéticiens de Bordeaux” etc.
He published more than 300 scientific articles, 6 books and he
supervised more than 100 research contracts.
Competences: sanitary engineering, water supply, sewerage, wastewater treatment,
HVAC installations.

2002 CIB W62 INTERNATIONAL SYMPOSIUM WATER SUPPLY AND DRAINAGE FOR BUILDINGS C4-10