MULTIMEDIA UNIVERSITY

URBAN DRAINAGE AND FLOOD

PROTECTION

Lecture Notes

Page 1
Objectives
At the end of the course the student should be able to:
1. propose solutions to the problems created by floods from the catchment area outside the urban area;
2. design facilities for surface and sub-surface drainage of the urban area;
3. carry out cost-benefit analysis of urban drainage and flood protection works;
4. explain how to control the possible negative effects of urban drainage on other areas.

Course content
Flood protection: run-off from the catchment outside the urban area. Land use, rainfall, discharge measurements data.
Determination of normative floods by direct or indirect methods. Coping with floods of short and long duration. Flood
attenuation measures such as upstream erosion control measures, retention reservoirs, spillways. Flood warning systems.
Design of floodways and floodway structures such as through-ways, bypasses and mobile weirs. Sediment transport and
sedimentation, energy principle. Cost-benefit analysis of flood protection works. Urban drainage: sub-surface drainage.
Groundwater control. Infiltration and retention of rainwater. Land subsidence. Classification of surface drainage systems;
primary, secondary and tertiary systems. Determination of normative discharges by rainfall data, rational method and
other methods such as unsteady flow methods and hydrograph method. Flow routing for steady and unsteady flow
conditions. Calculation and design of retention reservoirs. Design of surface drainage watercourses. Design of pumping
stations. Design of automatic and non-automatic level and discharge control structures. Water quality control of surface
drainage watercourses and retention reservoirs. Flushing of water courses; methods, facilities and structures. Maintenance
of watercourses and structures. Weed cutting, removal and disposal of aquatic plants, floating debris, sediments and solid
waste. Mitigative measures against negative effects of drainage works in catchments outside the urban area.

Evaluation
End of Semester Examination 70%
Continuous Assessment 30%

Reference
1. Butler, D and J.W. Davies. 2011. Urban Drainage. 3rd Edition. Taylor and Francis.
2. Chow, Ven Te., D.R. Maidment and L.W. Mays 1988. Applied hydrology
3. Ghosh, S.N. 2014. Flood Control and Drainage Engineering. 4th Edition. Taylor and Francis.
4. Gupta B.L. and A. Gupta2010. Water Resources Systems and Management.
5. Mays, L.W. 2011. Water Resources Engineering. 2nd Edition. Hamilton.
6. Novak, P. A.I.B. Moffat and C. Nalluri and R. Narayanan. 2007. Hydraulic Structures. 4th Edition.
7. Raghunath. H.M. 2006. Hydrology:-principles, Analysis, Design. 2nd Edition.
8. Subramanya, K. 2008. Engineering Hydrology. 3rd Edition. Tata McGraw-Hill.
9. Viessmann W and G.L. Lewis. Introduction to Hydrlogy. 4th Ediion.
10. Any other relevant text.

Page 2
 1
INTRODUCTION
1.0 Introduction
A Flood can be defined as a covering by water of land not normally covered by water resulting in
severe detrimental effects. Flood control implies all measures taken to reduce the damaging effects of
flood.

There exist several types of flooding such as:

• River Flooding: This is the major cause of flooding extensive areas as a result of heavy rains
in the catchment areas as well as local areas thereby increasing the river levels.
• Flash Floods: These results due to heavy rains which cause water levels to rise to dangerous
level within a short period of time say 6 to 12 hours. Heavy and continuous rains in local
areas can also cause flash floods.
• Urban Flooding: Local heavy rains over the city and larger towns can cause damaging and
disruptive flooding due to poor or chocked drainage and rapid runoff.
• Strom Surge or Tidal Flooding: This results mostly due to tropical disturbances, developing
to cyclones and crossing surrounding coastlines.
• Floods Arising due to Failure of Dam: Due to poor maintenance and due to exceptionally
high precipitation a severe flood may occur from failure of the dam. This causes a surging
water front travelling with high velocity causing destruction of properties and loss of life.

This course will mainly focus on flooding related to urban areas

1.1 Flood Control Strategies

The objective of flood control is to reduce or to alleviate the negative consequences of flooding. The
methods for flood control can be broadly grouped into:
− Structural measures
− Non-structural measures

1.1.1 Structural measures

Structural measures, which consist of construction of flood-control facilities, are aimed at mitigating
flood damage by regulating the movement of flood water. The structural measures include the
following:
• Flood-control reservoir
They are used to store flood waters for release after the flood event, reducing the magnitude
of the peak discharge.

• Diversion
They are used to reroute or bypass flood flows from damage centres in order to reduce the
peak flows at the damage centres.

• Levees and floodwalls

They are used to protect flood flows from floodplain areas where damage can occur.

Page 3
 • Channel modifications
Channel modification is performed to improve the conveyance characteristics of a stream
channel. They increase the carrying capacity thereby lowering the stage.

1.1.2 Non-structural measures

They are used to reduce potential damages. The main idea is to keep the general civil and industrial
activities undiminished during flood. They include:
• Flood forecasting and warning- for action before flood stages increase. They help to keep
alert concerned people in advance and to take timely action for evacuation in case of
impending danger.
• Flood proofing- the measures include re-arrangement of structural working space, water-
proofing of exterior walls and elevating structures. Flood proofing is most applicable to new
facilities.
• Land-use control alternatives- refer to administrative and other actions in order to modify
floodplain landuse so that uses are compatible with the potential flood hazard.

1.2 Urban Drainage

Development of an urban area, involving covering the ground with artificial surfaces, has a
significant effect on hydrological processes. The artificial surfaces increase the amount of surface
runoff in relation to infiltration, and therefore increase the total volume of water reaching the river
during or soon after the rain. Surface runoff travels quicker over hard surfaces and through sewers
than it does over natural surfaces and along natural streams. The effects of urbanization on the flood
hydrograph include increased total runoff volumes and peak flow rates (Fig. 1.1). In addition,
reduced infiltration means poorer recharge of groundwater reserves. This obviously increases the
danger of sudden flooding of the river. It also has strong implications for water quality.

Fig. 1.2.1: Effect of urbanisation on peak flow

Drainage systems are needed in developed urban areas because of the interaction between human
activity and the natural water cycle. Stormwater, if not drained properly, would cause inconvenience,
damage, flooding and further health risks. It contains some pollutants, originating from rain, the air or
Page 4
the catchment surface. Urban drainage systems aim at minimising the problems caused to human life
and the environment due to flooding.

Page 5
 2
PEAK FLOW ESTIMATION

2.0 Introduction
The methods for estimating runoff can be grouped into two broad categories:
 Direct and
 Indirect methods.

Direct methods consist of measurements while indirect methods consist of use of equations to derive
runoff. This section will focus on indirect methods for estimation of peak runoff (sometimes referred
to as flood flow).

The hydrograph of extreme runoff corresponding to peak flow provide valuable data for purposes of
hydrologic design. Methods for estimating the magnitude of peak flow include :
1. Rational formula
2. Empirical methods
3. SCS curve number method
4. Unit hydrograph technique

These methods are briefly described in the following sections.

2.1 Rational Method

The Rational Method is probably the oldest runoff estimation method and remains very popular in
urban drainage design. It is given as:

1
Qp = Cic A 2.1.1
3.6

where,
Q p is the peak runoff (m3/s)
C is the runoff coefficient (based on land cover)
ic is the rainfall intensity at the time of concentration (mm/hr) and
A is the watershed area (km2).

2.1.1 Intensity of rainfall

The intensity of rainfall employed in the rational method is equal to the design intensity or critical
intensity of rainfall, ic , corresponding to the time of concentration, tc , for the catchment for a given
recurrence interval, T. The design intensity of rainfall, ic , can be found from the intensity-duration-
frequency curves, for the catchment corresponding to tc and T. If the intensity-duration-frequency
(IDF) curves, are not available for the catchment the critical rainfall intensity can be determined from
total precipitation, P, and its duration, t R .

Consider a rainfall P falling lasting for a period, t R . The rainfall intensity is calculated as:

Page 6
 P
I= 2.1.1.1
tR
where,
P is the total rainfall (mm)
t R is the duration of rainfall (hrs)
I is the mean intensity of rainfall (mm/h)
If i (cm/h) is the rainfall intensity taken for a small interval, t , it has been found that the
intensity of rainfall is an inverse function of its duration as:

i tR + C 2.1.1.2
=
I t +C
where,
C is a constant.
The value of C is equal to 1 except for extreme cases. Thus:
 t + 1 2.1.1.3
i=I R 
 t + 1 

Substituting for the value of I yields:

P  tR + 1 2.1.1.4
i=  
tR  t + 1 

Thus, if IDF curve is not available, then the critical rainfall intensity, ic , can be computed for a time
t = tc . Hence:
P  t R + 1
ic =   2.1.1.5
t R  tc + 1 

When the time of concentration, tc , is not known, ic , can be approximated as:

P
i = 2.1.1.6
c
tR

2.1.2 Time of concentration

The design rainfall that is used in the rational method is the one with duration equal to the
watershed’s time of concentration, tc . The time of concentration is the time required for water to
travel from the hydrologically most remote point of the basin to the outlet or point of interest. It is
therefore the amount of time from the beginning if a rainfall event until the entire catchment area is
contributing to flow at the outlet.

The time of concentration is sum of overland flow and channel flow times. Thus:

tc = tov + tch 2.1.2.1

where,
Page 7
 tc is the time of concentration (hr)
tov is the overland travel time (hr)
tch is the channel travel time (hr).

The overland and channel flow times are calculated as follows:

Lslp 0.6 .n 0.6

t =
ov 2.1.2.2
ov
18s0.3

0.62 L n 0.75
t ch = ch 2.1.2.3
0.125 0.375
A s ch
where,
Lslp is the slope length of the catchment (m)
nov is the Manning’s roughness coefficient for overland flow
s is the average slope in the catchment (mm-1).
nch is the Manning’s roughness coefficient for channel flow
s ch is the channel slope (mm-1).
L is the channel length from the most distant point to the outlet of the catchment (km)

When the travel pattern cannot be differentiated as overland and channel flow or in the case of
insufficient data, empirical equations can be employed to estimate the time of concentration. There
are a number of empirical equations for estimation of the time of concentration. The Kirpich equation
is popularly used empirical formula relating the time of concentration to the length of travel and slope
of the catchment as:

t c = 0.01947 L0.77 s −0.385 2.1.2.4

where,
tc is the time of concentration (min)
L is the maximum length of travel of water (m)
s is the slope of the catchment

2.1.3 Runoff coefficient

The runoff coefficient, C, is a function of ground cover. Some tables of C provide for variation due to
slope, soil, and the return period of the design discharge. Should the basin contain varying amounts of
different covers, a weighted runoff coefficient for the entire basin can be determined as:

Weighted C = C A i i
2.1.3.1
A
where,
Ci is the runoff coefficient for cover type i that covers area Ai and
A is the total area.

Table 2.1.3.1 gives the runoff coefficients for rational method.

Page 8
Substituting the weighted C into the rational formula yields:

Qp =
1
. C A i i
.iA
1
3.6 A
= .i. CA 2.1.2

3.6  i i

Taking the case of two contributing subbasins, the longest time of concentration of the two subbasins
is used to determine the time of concentration for the combined area and the rainfall intensity
corresponding to this time of concentration is applied to the rational equation. For the combined area
and runoff coefficient, the peak discharge is calculated as:
Q =
1
(C A + C A )i 2.1.3
p 1 1 2 2
3.6

where,
C1 is the runoff coefficient for subbasin 1
A1 is the area for subbasin 1
C2 is the runoff coefficient for subbasin 2
A2 is the area for subbasin 2.

Page 9
Table 2.1.3.1: Runoff coefficients for the rational formula by hydrologic soil group and slope

The assumptions in the rational formula are as follows:

1) A storm that has duration equal to tc produces the highest peak discharge for this frequency.
2) The rainfall is uniformly distributed over the watershed.
3) The rainfall intensity remains constant during the time period equal to the time of
concentration, tc .
4) The return period of the peak discharge rate is the same as the return period of the rainfall
intensity. For instance, a 10-year rainfall intensity, i, is assumed to produce the 10-year peak
discharge.

Page 10
The rational method has the following limitations:
1. When basins become complex, and where sub-basins combine, the rational formula tend to
overestimate the peak flow.
2. The method assumes that the rainfall intensity is uniform over the entire watershed. This
assumption is true only for small watersheds and time periods, thus limiting the use of the
formula to small watersheds.

Example 2.1.1
The catchment area of a river is 1,200 hectares. The hydrological length of the catchment is 24 km.
The fall in level from the critical point to the outlet is 168 m. A severest storm yielded 20 cm of rain
in 5 hours. If the average runoff coefficient for the catchment is 0.4 and Kirpich equation applies, use
the rational method to estimate peak runoff of the catchment.

Solution
Peak runoff, Q p , can be estimated as:
1
Qp = Cic A
3.6
A = 1200 hectares = 12 km 2

The critical rainfall intensity is calculated as:

P  tr + 1
ic =  
t r  tc + 1 

P = 20 cm = 200 mm ; tr = 5 hrs

The time of concentration (min) can be estimated using Kirpich equation as:

t c = 0.01947 L0.77 s −0.385

where,
L = 24 1000 = 24,000 m
168
s= = 0.007
24000
Therefore,
t c = 0.01947  240000.770.007 −0.385
= 310.289 min
= 5.171 hrs
Thus:

ic = 200  5 + 1 
5  5.171 + 1 
= 38.892 mm / hr

Therefore, the peak runoff is computed as:

Page 11
 1
Qp =  0.4  38.892 12
3.6
= 51.856 m3 / s

Example 2.1.2
An urban catchment of area 300 hectares consists of different surface cover as shown in the table
below:
Type of cover Runoff coefficient Percentage
Roofs 0.9 15
Pavements and yards 0.8 15
Lawns and gardens 0.15 25
Roads 0.4 20
Open ground 0.1 15
Single family dwelling 0.5 10

The intensity of rainfall is 30 mm/h for rain with duration equal to the time of concentration.
Determine the peak runoff.

Solution
Overall runoff coefficient is calculated as:

15  0.9 + 15  0.8 + 25  0.15 + 20  0.4 + 15  0.1 + 10  0.5

C=
15 + 15 + 25 + 20 + 15 + 10
= 0.438

The catchment area, A = 300 ha = 3 km 2 .

The peak runoff is thus calculated as:

1
Qp = CiA
3.6
1
=  0.438  30  3
3.6
= 10.95 m3 / s

Example 2.1.3
An urban catchment has an area of 85ha. The slope of the catchment is 0.006 and the maximum
length of travel of water is 950m. The maximum depth of rainfall with a 25-year return period is as
shown in the table below:

Duration (mm) 5 10 20 30 40 60
Depth of rainfall (mm) 17 26 40 50 57 62

If a culvert for drainage at the outlet of this area is to be designed for a return period of 25 years, use
the rational method to estimate the required peak-flow rate, assuming a runoff coefficient of 0.3. Use
Kirpch equation where necessary.

Page 12
Solution
The ration method is stated as:
1
Qp = Cic A
3.6
The rainfall intensity is derived from the rainfall depth at the time of concentration. The time of
concentration is estimated using Kirpich formula:
t c = 0.01947L0.77 s −0.385
= 0.01947  9500.77  0.006−0.385
= 27.392 min
(27.392 − 20) (
The rainfall depth corresponding to this time = 40 +  50 − 40) = 47.392 mm
(30 − 20)
27.392 47.392
The rainfall intensity, ic = 47.392  =  60 = 103.808 mm / h
60 27.392
Thus the required peak-flow rate is:

1  0.3 103.808  85
Qp =
3.6 100
= 7.353m / s
3

2.2 SCS Peak Discharge Method

2.2.1 Runoff Depth Estimation

In developing the SCS rainfall-runoff relationship, the total rainfall is separated into three
components: direct runoff (Q), actual retention (F), and the initial abstraction (Ia). The retention F is
assumed to be a function of the depths of rainfall and runoff and the initial abstraction. The
development of the equation yields:

Qsurf
=
(P − I a )2 2.2.1.1
( P − Ia ) + S
where,
P is depth of precipitation, mm
Ia is initial abstraction, mm
S is maximum potential retention and
Qsurf is depth of direct runoff, mm.

The initial abstractions are determined empirically as:

I a = 0.2S 2.2.1.2

Substituting Equation 2.2.2 into Equation 2.2.1 yields:2

(P − 0.2S )
Qsurf = 2.2.1.3
P + 0.8S

Page 13
The retension parameter, S, is estimated as:  1000 
S = 25.4 − 10 2.2.1.4
 
 CN 
where,
CN is curve number.
when a catchment has more than one landuse, the weighted CN is calculated as:

CNw =
 A CN
i i
2.2.1.4a
A i

where,
CN w is the weighted CN
CNi is the CN for landuse i
Ai is the area for landuse i
The curve number depends on the hydrologic soil group and a land use. SCS developed a soil
classification system that consists of four groups, which are identified by the letters A, B, C, and D.
Soil characteristics that are associated with each group are as follows:
Group A: they have low runoff potential and high infiltration rate when wetted. They include
deep sand, deep loess; aggregated silts
Group B: shallow loess; sandy loam
Group C: clay loams; shallow sandy loam; soils low in organic content; soils usually high in
clay
Group D: they have high runoff potentisl. They include soils that swell significantly when
wet; heavy plastic clays; certain saline soils.

CN Values for Urban Land Uses is based on a specific percentage of imperviousness. For example,
the CN values for commercial and business land use are based on an imperviousness of 85 percent.
Curve numbers for other percentages of imperviousness can be computed using a CN of 98 used for
the impervious areas and the CN for open space (good condition) used for the pervious portion of the
area. Thus CN values of 39, 61, 74, and 80 are used for hydrologic soil groups A, B, C, and D,
respectively. These are the same CN values for pasture in good condition. Thus the following
equation can be used to compute a CN for a specific percentage of imperviousness:

CN u = CN (1 − f ) + f (98)
p 2.2.1.5
where,
CNu is the curve number for urban landuse
CN p is the curve number for pervious portion of the area and
f is the fraction of imperviousness.

For instance, the CN values for commercial and business areas with 85 percent imperviousness can
be calculated as:

A soil : 39 (1 − 0.85) + 98(0.85) = 89

B soil : 61(1 − 0.85) + 98(0.85) = 92
C soil : 74 (1 − 0.85) + 98(0.85) = 94
D soil : 80 (1 − 0.85) + 98(0.85) = 95

Page 14
Table 2.2.1.1 shows the curve numbers for a number of urban land cover.

Table 2.2.1.1: SCS curve number for urban areas

Page 15
2.2.2 Peak Discharge Estimation
Once the runoff depth has been determined, the peak discharge may be estimated using or the rational
method or the SCS method. If the rational method is employed, the rainfall intensity is calculated
from the excess rainfall depth, Qsurf , and time of concentration, tc . Thus:

1
Qp = Ci A
3.6 c
Qsurf .A 2.2.2.1
=
3.6 tc
where,
Q p is peak runoff rate (m3/s)
Qsurf is surface runoff depth (mm)
A is the catchment area (km2) and
tc is the time of concentration (hr).

Computational details of the rational method are given in the previous section.

The peak discharge can be computed with the SCS method as:

Q p = qu AQsurf Fp 2.2.2.2

where,
Q p is the peak discharge (m3/s)
q u is the unit peak discharge (m3/s/km2/mm)
A is the drainage area (km2)
Qsurf is the depth of runoff (mm)
Fp is the pond factor.
The pond factor accounts for pond and swamp storage within the watershed. The factor is selected
from Table 2.2.2.1 based on percentage of ponds and swamps.

Table 2.2.2.1: Adjustment factor (Fp) for pond and swamp Areas

Page 16
The unit peak discharge is obtained from the time of concentration (tc) in hours and the initial
abstraction/rainfall (Ia/P) ratio as:

qu = 10C0 +C1 log tc +C2 (log tc )

2
2.2.2.3

where,
Co, C1, and C2 are regression coefficients
α is the unit conversion constant equal to 0.000431 in SI units and
tc is the time of concentration (hrs)

The regression coefficients Co, C1, and C2 are determined from Table 2.2.2.2 for various rainfall
types and Ia/P ratios. This is done by first determining the SCS type that best describes the maximum
precipitation event in the desired basin. To represent various regions of the United States, NRCS
developed four synthetic 24-hour rainfall distributions (I, IA, II, and III). Type IA is the least intense
and type II the most intense short duration rainfall. Types I and IA represent the Pacific maritime
climate with wet winters and dry summers. Type III represents Gulf of Mexico and Atlantic coastal
areas where tropical storms bring large 24-hour rainfall amounts. Type II represents the rest of the
US. For regions outside the US, the rainfall type that best suits the region is selected.

When (I a P)  0.1 , values of Co, C1, and C2 corresponding to (I a P) = 0.1 should be used, and if
(I a P)  0.5 , values of values of Co, C1, and C2 corresponding to (I a P) = 0.5 should be used.

Page 17
Table 2.2.2.2: Coefficients for SCS Peak Discharge Method

Example 2.2.1
An urban watershed with a drainage area of 5.18 km2 has a 6.0-hr time of concentration; CN of 75
and 5 percent of the area is ponded. The 25-year, 24-hr rain is 152.4 mm. Find the 25-year peak
discharge. Assume type II rainfall distribution.

Solution
Excess rainfall depth is calculated as:
(P − 0.2S )2
Qsurf =
P + 0.8S

The retension parameter, S, is estimated as:

Page 18
  1000 
S = 25.4 − 10
 
 1000
CN 
= 25.4 − 10
 
 75 
= 84.667
Therefore:

Qsurf =
(152.4 − 0.2  84.667)2
152.4 + 0.8  84.667
= 83.364 mm

The peak discharge is given as:

Q p = qu AQsurf Fp

The unit peak discharge is given as:

qu = 10C0 +C1 log tc +C2 (log tc )

2

The initial abstractions are calculated as:

I a = 0.2 S
= 0.2  84.667
= 16.933
Therefore:
(Ia P) = 16.933
152.4
= 0.111
From Table 2.2.2, the factors for unit peak discharge are found by interpolation. Thus:

C0 = 2.54839; C1 = −0.61553; C2 = −0.16142

Therefore, the unit peak discharge is given as:

qu = 0.000431102.54839+(−0.61553log 6+[−0.16142(log 6) ]
2

= 0.04 (m3/s)/km2/mm

From Table 2.2.2, the pond factor for 5 percent pondind is:

F p = 0.72

Therefore, the peak discharge is :

Qp = 0.04  5.18  83.364  0.72

= 12.437 m3 / s

Page 19
2.3 Emprical Formulae
Empirical formulae are essentially regional formulae based on statistical correlation of the observed
peak flow and important catchment properties. One major limitation of the empirical formual is that
they are applicable only in the region from which they were developed. When applied to other areas,
they can at best give approximate values.

The simplest formula relates the peak flow to the catchment area as:

Q p = f (A) 2.3.1

This takes a general form as:

Q p = CAn 2.3.1a
where,
Q p is the peak flow
C is an empirical constant
n is an exponent (between 0 and 1)
A is the catchment area

A number of these methods are summarised below.

1. Dicken’s formuala (1865):

Q p = C D A3 4 2.3.2
where,
Q p is the peak discharge (m3/s)
CD is the Dicken’s constant
A is the area of the catchment (km2)

This formula is applicable to Northern and Central India

2. Ryves formula (1884):

Q p = C R A2 3 2.3.3
where,
Q p is the peak discharge (m3/s)
CR is the Ryves coefficient
A is the area of the catchment (km2)

This formula is applicable India.

3. Inglis formula (1930):

124 A
Qp = 2.3.4
A + 10.4
where,

Page 20
 Q p is the peak discharge (m3/s)
A is the area of the catchment (km2)

This formula is also applicable to Indian catchments.

4. Fuller’s formula:

QTp = C f A0.8 (1 + 0.8 log T ) 2.3.5

where,
QTp is the maximum 24-h flood (m3/s)
T is the frequency (years)
A is the area of the catchment (km2)
C f is a constant (0.18 to 1.88)
Fuller’s formula was derived for catchment sin the USA.

5. Baird and Mcllwraith (1951):

3025A
Qp = 2.3.6
(278 + A)0.78
where,
Q p is the peak discharge (m3/s)
A is the area of the catchment (km2)

This equation was developed based on the maximum recorded floods throughout the world.

6. Envelope curves
In this method, the available flood peak data from a large number of catchments which do not
significantly differ from each other in terms of meteorological and topographical
characteristics are collected. The data are then plotted on a log-log paper as flood peak vs
catchment area. Envelop curves thus obtained are very useful in getting quick rough
estimation of peak values. A typical envelope curve is shown in Figure 2.3.1.

Page 21
 Fig. 2.3.1: Envelope curves for Indian rivers

2.4 Synthetic Unit Hydrograph

The unit hydrograph developed from rainfall and streamflow data on a watershed applies only for that
watershed and for the point on the stream where the streamflow data were measured. Synthetic unit
hydrograph procedures are used to develop unit hydrographs for other locations on the stream in the
same watershed or for nearby watersheds of a similar character. Synthetic unit hydrographs are used
to estimate peak flow based on watershed characteristics. The following sub-sections describe
synthetic hydrographs for estimation of peak flow.

2.4.1 Snyder's Synthetic Unit Hydrograph

This type of synthetic unit hydrograph was developed by Snyder after a study of watersheds located
mainly in the Appalachian highlands of the United States, and varying in size from about 30 to
30,000 km2. The method proposed by Snyder for unit hydrograph synthesis is derived from
relationships between the characteristics of a standard unit hydrograph and descriptors of basin
morphology. It relates the time from the centroid of the excess rainfall to the peak of the unit
hydrograph (i.e., lag time) to geometric characteristics of the basin in order to derive critical points
for interpolating the unit hydrograph. Figure 2.4.1.1 summarises the key components of Snyder’s
hydrograph.

Page 22
 Figure 2.4.1.1: Snyder’s synthetic unit hydrograph

The basin lag time is evaluated by

t = 0.75C (LL )0.3 2.4.1.1
L t CA
where,
t p is the basin lag (hrs)
L is the length of the main stream from the outlet to the upstream divide (km)
LCA is the is the length of stream channel from a point nearest the centre of the
basin to the outlet (km) and
Ct is an empirical watershed storage coefficient. It is derived from gaged watersheds in
the same region.

The standard duration, t d , of excess rainfall is computed empirically by:

tL
t = 2.4.1.2
d
5.5

Adjusted values of lag time, t La , for other durations of rainfall excess can be obtained by:

t La = t L + 0.25 (tda − t d ) 2.4.1.3

where,
tda is the desired duration of the unit hydrograph.

Time to peak can be computed as a function of lag time and duration of excess rainfall,
expressed as:

t p = t La + 0.5tda 2.4.1.4

The corresponding peak discharge per unit rainfall depth, q p , is:

Page 23
 0.275Cp A
qp = 2.4.1.5
tLa
where,
q p is peak discharge ((m3/s)/mm)
Cp is an empirical constant derived from gaged watersheds in the same region.
A is watershed area (km2)

The corresponding peak discharge is:

0.275C p AQsurf
Qp = 2.4.1.6
t La
where,
Q p is peak discharge (m3/s)
Qsurf is depth of rainfall excess (mm)

Example 2.4.1
A watershed has an area of 25.9 km2. The hydraulic flow length from the catchment boundary to the
outlet is 4.02 km. Using Snyder’s method; determine the peak discharge for a 1-hour synthetic unit
hydrograph for the basin. Assume that LCA = 1.61 km , Ct = 1.9 , and C p = 0.6 .

Solution
Lag time is computed as:

t = 0.75C (LL )0.3

L t CA

= 0.75 1.9  (4.02 1.61)0.3

= 2.495hrs

The duration of rainfall excess is:

tL
td =
5.5
2.495
=
5.5
= 0.454 hrs

Since duration of excess rainfall is not equal to the desired duration, tda , i.e. 1 hour, so the lag time
should be adjusted as:

tLa = t L + 0.25 (t − td )
da
= 2.495 + 0.25 (1 − 0.454)
= 2.632 hrs

The corresponding peak discharge, q p , per unit rainfall depth is:

Page 24
 0. .275C p A
qp =
t La
0.275  0.6  25.9
=
2.632
= 1.624 m 3 / s / mm

2.4.2 SCS Unit Hydrograph

The SCS unit hydrograph is shown in Figure 2.4.2.1. The unit hydrograph used by the SCS is based
upon an analysis of a large number of natural unit hydrographs from a broad cross-section of
geographic locations and hydrologic regions.

Figure 2.4.2.1: SCS dimensionless unit hydrograph and its equivalent triangular hydrograph

The time of concentration is calculated through the lag time method. The lag time is the time in hours
from the center of mass the rainfall excess to the peak discharge. The lag time method relates the lag
time, ( t L ) of the excess rainfall to the slope, the hydraulic length, and the potential maximum
retention as:

Page 25
  1000  0.7
2.587 L0.8 −9
 
tL =  CN  2.4.2.1
1900 s 0.5
where,
t L is the lag time (hrs)
L is the hydraulic watershed length (m)
CN is the hydrologic area-weighted curve number.
s is the average watershed slope (%).
The SCS uses the following relationship between the time of concentration and the lag time:

5
t = t 2.4.2.2
c
3 L
where,
tc is the time of concentration (hrs)

Thus:
 1000  0.7
2.587 L0.8 −9
 
tc =  CN  2.4.2.3
1140 s 0.5

The time to peak, t p , is the time from the beginning of the rainfall to the time of the peak discharge.
Thus:
tR
t = + tL 2.4.2.4
p
2

where,
t p is the time to peak (hrs)
t R is the duration of the rainfall excess (hrs)
SCS recommends that t R be 0.133 of the time of concentration of the watershed. Thus:

t R = 0.133tc 2.4.2.5

Substituting t L = 0.6tc and t R = 0.133tc , the time to peak can now be expressed as:

0.133tc
tp= + 0.6tc 2.4.2.6
2
Thus:

t p = 0.67tc 2.4.2.6a

The area under the unit hydrograph equals the depth of excess runoff Q, which is 1 mm (1 in) for a
unit hydrograph. Based on geometry of the triangular unit hydrograph:

Page 26
 Q=
1
q
p
(t
p
+ tr ) 2.4.2.7
2
where,
t p is the time to peak (hrs)
t r is the recession time (hrs)
q p is the peak discharge for unit excess runoff
Q is the depth of direct runoff, which is 1 mm (1 in) for a unit hydrograph

Solving for q p and noting that Q = 1 for a unit hydrograph yields:

(1)  
2
qp =   2.4.2.8
t p 1 + tr t p 
 2 
Letting K =   , then
1 + tr t p 
K (1)
qp = 2.4.2.9
tp
This equation was developed using imperial units. Therefore the equation can be modified to express
q in ft3/s, t in hours and Q in inches:
p p

A (1)
q p = 645.33K 2.4.2.10
tp
where,
A is the watershed area (miles2).

The factor 645.33 is the rate necessary to discharge 1 inch of runoff from 1 mi2 in 1 hr. Using
tr = 1.67t p gives K = 2 (1 + 1.67)= 0.75 . The equation then becomes:

A (1)
q p = 484 2.4.2.11
tp
For SI units:

A(1)
q p = 0.208 2.4.2.12
tp
where,
q p is the peak discharge (m3/s/mm)

Thus the peak discharge can be given as:

AQsurf
Q p = 0.208 2.4.2.12
tp
where,
Q p is the peak discharge (m3/s)
t p is the time to peak (hrs)

Page 27
 Qsurf is the depth of direct runoff (mm).
A is the watershed area (km2).

Example 2.4.2.1
An commercial and business urban zone with a drainage area for a 1.2 km2 is 85% impervious. The
flow length is 1982 m, the slope is 1.3 percent, and the soil is of group B. Determine the peak runoff
per unit rainfall depth, from the zone, using the SCS unit hydrograph procedure.

Solution
The peak runoff per unit rainfall depth is given as:
A(1)
q p = 0.208
tp
A = 1.2 km2 ; Qsurf =1 mm
The time to peak:
0.133tc
tp = + 0.6tc
2
The time of concentration:
 1000  0.7
2.587 L0.8 −9
 
tc =  CN 
0.5
1140 s

The curve number CN = 92 (from the table of urban curve numbers, for soil class B and commercial
and business landcover 85% impervious).

Thus:
 1000 
2.587  19820.8 − 9 0.7
 
t =  92 
c
1140 1.30.5

= 1.339 hrs

Hence:
0.133 1.339
tp = + 0.6 1.339
2
= 0.892 hrs

Therefore:
1.2 1
q p = 0.208
0.892
= 0.28 m / s / mm
3

Page 28
 3
ESTIMATION OF DESIGN FLOOD
3.0 Introduction
The design flood is a flood adopted for the design of hydraulic structures. Estimation of design flood
is one of the important components of planning, design and operation of water resources projects.
Information on flood magnitudes and their frequencies is needed for design of hydraulic structures.
The design flood may be taken as probable maximum flood (PMF) or the standard project flood
(SPF).

PMF is an extreme flood that is physically possible in a region as a result of severemost

combinations, including rare combinations of meteorological and hydrological factors such as
minimum water losses and probable maximum precipitation. PMF is used in situations where the
failure of the structure would result in loss of life and catastrophic damage and as such complete
security from potential floods is sought. Storage structures such as high dams located on the upstream
of large cities must be designed for very severe floods. The failure of such dams may cause great loss
of life and property.

SPF is the flood would result from a severe combination of the meteorological and hydrological
conditions that are reasonably characteristic of the drainage basin considered. It excludes extremely
rare conditions. Typically, SPF is about 40% to 60% of PMF for the same drainage basin. SPF is
often used where the failure of a structure would cause less severe damages. In remote areas, small
structures such as culverts can be designed for less severe floods as the damage due to failure may
not be of much significance.

For a particular hydraulic structure, the design flood is selected after careful consideration of a
number of factors:
a) Type of structure,
b) Importance of structure,
c) Economy of the structure, and
d) Development of the area surrounding the project.

There are various methods by which the estimate of design flood can be made. These include:
1. Rational method
2. Empirical method
3. SCS curve number method
4. Unit hydrograph procedure
5. Observation of the highest flood level or maximum historical flood,
6. Flood frequency studies

The design flood can be estimated from design storm of a given frequency (and/or watershed
characteristics) by applying any of the following methods: rational, empirical, SCS curse number or
the unit hydrograph method. These methods have been discussed in the previous chapter. In the
absence of meteorological data, an idea of the magnitude of the maximum historical flood flow that
might have occurred may be taken as the design flood. This chapter focuses on estimation of design
flood through frequency studies. The basic ideas and computation procedures behind the method are
briefly described in the sections that follow.

Page 29
3.1 Probability Concepts
The following is a summary of some of important probability concepts relevant to hydrologic design:

1. The probability of exceedence of a peak flow, QA , is approximated by the relative number of

exceedences of QA after a long series of observations, i.e.,

Pr (QA ) =
No. of exceedence s n
= e 3.1.1
Total number of observations n

where,
Pr (QA ) is the probability of exceedence (relative frequency)
ne is the number of exceedences (frequency)
n is the total number of observations

2. The probability of a nonexceedence of an event such as peak flow, QA , is given by:

( )
Pr QA =
n − ne = 1 − ne
n
= 1 − Pr (QA ) 3.1.2
n
where,
( )
Pr Q A is the probability of nonexceedence.

3. The sum of probabilities of exceedence and nonexceedence equals 1.

( )
Pr (Q A ) + Pr QA = 1 3.1.3

4. The return period, T, is related to the probability of exceedence by:

1
T = 3.1.4
Pr (QA )
where,
T is the return period.
The return period gives the probability of exceedence of an event over a given period. For
example, a 100-year storm has an annual probability of occurrence of 1 in 100 years i.e.,
Pr = 1 100 = 0.01 . Therefore, it is synonymous to speak of a 1 percent storm as a 100-year
storm.

5. The probability that QA will be equalled or exceeded in any year is:

Pr (QA ) =
1
3.1.5
T

6. The probability that QA will not be equalled or exceeded in any year is:
( )
P r QA = 1 − Pr (QA ) = 1 −
1
3.1.7
T

7. The probability that QA will not be equalled or exceeded in any of n successive years is:

Page 30
 P (Q ) P (Q )...  P (Q )= P (Q ) n  1 n
=1 −  3.1.6
r1 A r2 A rn A r A
 T

8. The probability, R , called risk, that QA will be equalled or exceeded at least once in n
successive years is:
R = 1 − P (Q ) = 1 − 1 −
n  1n
3.1.7
r A  
 T

where n is referred to as the expected life of the structure.

It can be seen that the return period for which a structure should be designed depends upon
the acceptable level of risk.

9. The reliability of a system is defined as the probability that a system will perform its required
function for a specified period of time under stated conditions. Reliability is the complement
of risk, or the probability that the loading will not exceed the capacity. Thus, reliability, Re , is
given as:
 1 n
R =1− R = 1− 3.1.8
e  
 T

Although natural hydrologic uncertainty can be accounted for using risk concept as previously
mentioned, other kinds of uncertainty are difficult to calculate. These are often treated using a safety
factor, SF, or a safety margin, SM. Letting the hydrologic design value be L and the actual capacity
adopted for the project be C, the factor of safety is:

C
SF = 3.1.9
L
and the safety margin is:

SM = C − L 3.1.10

The actual capacity is larger than the hydrologic design value because it has to allow for other kinds
of uncertainty: technological (hydraulic, structural, construction, operation, etc.), socio-economic,
political, and environmental.

Example 3.1.1
What return period must a highway engineer use in designing an underpass drain to accept only a 10
percent risk that flooding will occur in the next 5 years?

Solution
Risk is given as:  1 n
R =1− 1−
 
 T

Page 31
Thus:
10  15
= 1 − 1 − 
100  T

 T = 47.958 years

Example 3.1.2
A bridge has an expected life of 25 years and is designed for a flood magnitude of return period 100
years.
a) What is the risk of this hydrologic design?
b) If a 10% risk is acceptable, what return period will have to be adopted?

Solution
a) The risk is calculated as:
 1 n
R =1− 1−
 
 T
 1  25
=1− 1−
 
 100 
= 0.222

b) Risk is given as:

 1 n
R =1− 1−
 
 T
 1  25
0.1 = 1 − 1 −
 
 T
 T = 237.781, say T = 240 years

3.1.1 Probability of exceedence and plotting positions

The first step in determining the probability of exceedence is the ranking of the hydrological data in
descending order, with the highest value first and the lowest value last. After the data are ranked, a
serial rank number (m), ranging from 1 to n (number of observations), is assigned. Subsequently the
probability of exceedence is calculated based on the rank number and the total number of
observations. There are a number of formulae, known as ‘plotting positions’ for calculating the
probability of exceedence. Thus if n is the total number of values and m is the rank of a value in a list
ordered by descending magnitude, the exceedence probability of the mth largest value, xm (for large n)
is given by different plotting formulae as:

Pr (X  x m ) =
m
1. California: 3.1.1.1
n

m − 0.5
2. Hazen: Pr (X  x m ) = 3.1.1.2
n

Page 32
 Pr (X  x m ) =
m
3. Weibull: 3.1.1.3
n +1
3m −1
4. Tukey: Pr (X  x m ) = 3.1.1.4
3n +1
m − 0.3
5. Chegodayev: Pr (X  x m ) = 3.1.1.5
n + 0.4

6. Blom: P (X  x )= m − 3 8 3.1.1.6
n +1 4
r m

Most plotting position formulas follow the general form:

m−b
Pr (X  x m ) = 3.1.1.7
n + 1 − 2b

where b is a parameter. For example, b=0.5 for Hazen's formula; b=0.3 for Chegodayev's; b=0 for
Weibull's; b = 3/8 for Blom's formula and b=1/3 for Tukey's formula,

The California's formula produces a probability of 100 percent for m = n , which may not be easily
plotted on a probability scale. The Weibull formula is a compromise with more statistical
justification. Thus it is considered by many as the most efficient and most commonly used formula
for computing plotting positions for unspecified distributions, for most sample data.

Example 3.1.1.1
The following table shows the annual maximum discharge of a river for 45 years. Determine the
probability of exceedence using Weibull formula and hence the corresponding return period.

Year 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964
Data 804 1090 1580 487 719 140 1583 1642 1586 218 623 507 1303 197 583

Year 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
Data 377 348 804 328 245 140 49 1651 716 286 671 3069 306 116 162

Year 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Data 425 1982 277 1254 430 260 276 1657 937 714 855 399 1543 360 348

Solution
The first step involves arranging the data in descending order as shown in column 3 of the table
below. The ranked data are then assigned rank numbers, m, with m=1 for the highest value (column
4). The probability that a value will be equalled or exceeded (column 5) is then calculated using
Weibull formula as:

Pr (X  x m ) =
m
; where n = 45 for the given data.
n +1

The return period, shown in column 6, is calculated as the reciprocal of the probabilities. Thus:
Page 33
 1
T=
Pr (X  xm )

Annual Arranged Pr (X  xm ) Return

Year maxima data Rank (m) (Weibull) period, T
1950 804 3069 1 0.02174 46.00
1951 1090 1982 2 0.04348 23.00
1952 1580 1657 3 0.06522 15.33
1953 487 1651 4 0.08696 11.50
1954 719 1642 5 0.10870 9.20
1955 140 1586 6 0.13043 7.67
1956 1583 1583 7 0.15217 6.57
1957 1642 1580 8 0.17391 5.75
1958 1586 1543 9 0.19565 5.11
1959 218 1303 10 0.21739 4.60
1960 623 1254 11 0.23913 4.18
1961 507 1090 12 0.26087 3.83
1962 1303 937 13 0.28261 3.54
1963 197 855 14 0.30435 3.29
1964 583 804 15 0.32609 3.07
1965 377 804 16 0.34783 2.88
1966 348 719 17 0.36957 2.71
1967 804 716 18 0.39130 2.56
1968 328 714 19 0.41304 2.42
1969 245 671 20 0.43478 2.30
1970 140 623 21 0.45652 2.19
1971 49 583 22 0.47826 2.09
1972 1651 507 23 0.50000 2.00
1973 716 487 24 0.52174 1.92
1974 286 430 25 0.54348 1.84
1975 671 425 26 0.56522 1.77
1976 3069 399 27 0.58696 1.70
1977 306 377 28 0.60870 1.64
1978 116 360 29 0.63043 1.59
1979 162 348 30 0.65217 1.53
1980 425 348 31 0.67391 1.48
1981 1982 328 32 0.69565 1.44
1982 277 306 33 0.71739 1.39
1983 1254 286 34 0.73913 1.35
1984 430 277 35 0.76087 1.31
1985 260 276 36 0.78261 1.28
1986 276 260 37 0.80435 1.24
1987 1657 245 38 0.82609 1.21
1988 937 218 39 0.84783 1.18
1989 714 197 40 0.86957 1.15
1990 855 162 41 0.89130 1.12
1991 399 140 42 0.91304 1.10

Page 34
 1992 1543 140 43 0.93478 1.07
1993 360 116 44 0.95652 1.05
1994 348 49 45 0.97826 1.02

3.2 Probability Distributions

The design of hydrologic systems is influenced by extreme events such as floods. The magnitude of
an extreme event is inversely proportional to its probability of occurrence. Thus, more severe events
occur less frequently. The concept of probability distribution is used to analyse hydrologic data so as
to relate the magnitude of extreme events to their frequency of occurrence. Hydrologic data to be
analysed is assumed to be independent and identically distributed and the hydrologic system is
assumed to be stochastic, space-independent and time independent. The data should be properly
selected so that the assumptions of independence and identical distributions are satisfied. The
assumption of identical distribution or the homogeneity is achieved by selecting the observations
from same population (i.e., no changes in the watershed and recording gauges are made). The
assumption of independence is achieved by selecting the annual maximum of the variable being
analysed as the successive observations from year to year will be independent.

The probability distribution is a function representing the frequency of occurrence of the value of a
random variable. Consider an 85-year record of annual streamflow as shown in Table 3.2.1. The table
shows streamflow values (row 1) and the corresponding frequencies (number of occurrences) in row
2.

Table 3.2.1: Illustrational streamflow data

Streamflow (cumecs) 500 1500 2500 3500 4500 5500 6500 7500 8500
No. of occurrences 5 8 15 18 13 8 10 5 3
Relative frequency 0.059 0.094 0.176 0.212 0.153 0.094 0.118 0.059 0.035
Cumulative Freq. 0.059 0.153 0.329 0.541 0.694 0.788 0.906 0.965 1.000

The total number of observations n = 85 for the given data. The associated probabilities (row 3) are
calculated as :
no
P (X = x ) = 3.2.1
r i
n
where
Pr ( X = x) is the probability (relative frequency) that streamflow equals a value, x
no is the frequency of a given streamflow value
n is the total number of observations.

The cumulative frequency (probability) is given in row 4 of the table. Figure 3.2.1 (a) shows the
histogram while Figure 3.2.1 (b) and Figure 3.2.1 (c) show the probability distributions of the
streamflow data. This is a discrete type of probability distribution because the number of occurrences
is exact, in the record from which the frequencies were taken.

Page 35
 Fig. 3.2.1: Probability distribution of the streamflow data

Observe that each of the streamflow events has a finite probability and the sum is 1; that is:

 P (X = x )
i
r i = 1 3.2.2

Another important property of random variables is the cumulative distribution function (CDF)
defined as the probability that any outcome in X is less than or equal to a stated, limiting value x. The
cumulative distribution function is denoted F ( x ) . Thus:

F(x) = Pr ( X  x) 3.2.3

The function F (x ) , for these data, increases monotonically from a lower limit of zero to an upper
bound of unity. The CDF (figure 3.2.1 (c)) is derived from figure 3.2.1 (b) by taking cumulative
probabilities.

Figure 3.2.1 (b) thus gives the probability, Pr ( X = x) , while figure 3.2.1 (c) shows the probability,
Pr ( X  x) . For instance, the probability Pr ( X = 5500) = 0.094 while Pr ( X  5500) = 0.788 .
Page 36
Continuous variables present a slightly different picture. As the number of observations increase and
the size of the intervals is reduced, continuous distribution will be developed. The continuous forms
of figures 3.2.1 (b) and 3.2.1 (c) are known as probability density function (PDF) and cumulative
distribution function respectively. These continuous probability distributions are shown in figure
3.2.2.

Fig. 3.2.2: Continuous probability distributions: (a) probability density function and (b) cumulative
distribution function.

The probability density function is defined as:

F (x) dF(x)
f (x) = lim = 3.2.4
x→0 x dx

Also, the function has the property that:


 f (x)dx = 1
−
3.2.5

which again is the requirement that the probabilities of all outcomes sum to 1.

Furthermore, the probability that x will fall between the limits, a and b , is written as:
P (a  X  b) = f (x)dx
b
3.2.6
r a

Page 37
Note that finite probabilities are defined only as areas under the probability density function (PDF).
The CDF can now be defined in terms of the PDF as:

P (−   X  ) = P (X  x) = F (x) = f (u)du 3.2.7
r r  −

where u is used as a dummy variable to avoid confusion with the limit of integration. The area under
the CDF has no meaning, only the ordinates, or the difference in ordinate. For example,
P (x  X  x ) can be evaluated as F (x ) − F (x ) .
r 1 2 1 2

The complementary cumulative probability, G(x) is given as:

G(x) = 1 − F (x) = Pr (X  x) 3.2.8

A number of probability distribution functions are used in analysis of hydrologic data. The following
sections illustrate on the application of some probability functions relevant to flood management.

3.2.1 Normal distribution

The normal distribution has a symmetrical, bell-shaped probability density function. It describes
many processes that are subject to random and independent variations. Although it often does not
perfectly fit sequences of hydrologic data, it has wide application, for example, in dealing with
transformed data that do follow the normal distribution.

The normal distribution has two parameters, the mean , and the standard deviation , for which x
and s , derived from sample data, are substituted. The probability density function for a normal
distribution is given as:

1  1  x −  2 
f (x) = exp−    for − x
 2  2    
3.2.1.1

By a simple transformation, the distribution can be written as a single-parameter function only, the
standardised variate, z , defined as:

x−
z= 3.2.1.2


in which z has a zero mean and unit variance. The PDF of a normal distribution can now be re-
written as:
f (z) =
1
e−z
2
2
3.2.1.3
2

The cumulative probability function (CDF) thus becomes:

 x−
F (z) = = P (Z  z)
1
e
z 2
−u 2
du =  3.2.1.4
Pr  Z 

r
2 −  

Page 38
These cumulative probabilities have been tabulated for various values of z . A table of the CDF for a
normal distribution is shown in Table 3.2.1.1.

One disadvantage of the normal distribution is that it is unbounded in the negative direction whereas
most hydrologic variables are bounded and can never be less than zero. For this reason and the fact
that many hydrologic variables exhibit a pronounced skew, the normal distribution usually has
limited applications. However, these problems can sometimes be overcome by performing a log
transform on the data. Often the logarithms of hydrologic variables are normally distributed.

Page 39
Table 3.2.1.1: Cumulative probability of the standard normal distribution

Page 40
Example 3.2.1.1
The table below shows the annual daily maximum rainfall for a given catchment. Assuming that the
rainfall data is normally distributed, determine the 100-year rainfall.

Year 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Max. daily rainfall (mm) 43 44 38 31 47 49 52 31 51 40 41 43

Year 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Max. daily rainfall (mm) 37 36 34 38 36 37 43 34 53 49 47

Solution
For a normal distribution:  x−
P (X  x) = F (z) = = P (Z  z)
1 z
e
2
−u 2
du = 
Pr  Z 

r
2
r
−  

Two parameters, mean and standard deviation, of the normal distribution are calculated first from the
given data. From the data given:
Mean = 41.478 mm
S tan dard deviation = 6.673 mm

The exceedence probability is calculated as:

P (X  x) =
1
r
T
1
=
100
= 0.01
Therefore the non-exceedence (cumulative) probability is:

Pr (X  x) = 1 − 0.01
= 0.99
From the normal distribution table, the z value corresponding to this cumulative probability is:

z = 2.326

The 100-year rainfall is thus calculated from the relation:

x− x − 41.478
z  2.326 
 6.673

Thus the 100-year rainfall is:

x = 57 mm

Example 3.2.1.2
The mean and standard deviation of annual daily rainfall at a station are 14.9 mm and 5.9 mm
respectively. Determine the 10-year rainfall, assuming that the rainfall is normally distributed.
Page 41
Solution
Let Pr (X  x) be the probability, corresponding to the 10-year return period. Thus:
P (X  x) =
1
r
T
1
=
10
= 0.1

Therefore:
P (X  x) = 1 − P (X  x)
r r
= 1 − 0.1
= 0.9

The CDF for normal distribution is:

 x−
P (X  x) =  = P (Z  z )
r Pr  Z 
  
r

Substituting the given data into the equation gives:

0.9 = Pr (Z  z )

From the normal distribution the value of z corresponding to this probability is:

z = 1.282
But:
x−
z=

Thus:
x − 14.9
1.282 =  x = 22.5
5.9

Therefore the 10-year rainfall is 22.5 mm.

3.2.2 Log-normal distribution

The log-normal distribution has the same characteristics as the normal distribution except that the
dependent variable, X, is replaced with the variable, Y = ln X . The log-normal distribution is bounded
on the left by zero and it has a pronounced positive skew. These are both characteristics of many of
the frequency distributions that result from an analysis of hydrologic data.

To analyse the log-normal distribution, the following steps may be followed:

i. Find the logarithms of the data given i.e Y = ln X .
ii. Analyse the log-transformed data like a normal distribution i.e. determine the mean and
standard deviation of the variable Y and use the normal tables for analysis
iii. Find the log inverse of Y.
Page 42
The mean and standard deviation of the log-transformed data ( Y = ln X ) can be determined directly
from those of the X, without transforming the data, as:

1  2 
 y = ln  x  3.2.2.1
2 1 + x 2 

(
 y2 = ln 1 +  x2 ) 3.2.2.2

x
 = 3.2.2.3
x
x

The PDF of the log-normal random variable is:


1 1  y − y 2
 
f ( y) = exp−   for 0 y
x y 2  2   
  y   3.2.2.4
where,
y = ln x 3.2.2.5
and  y and  y are the mean and standard deviations of the log-transformed data.

If a logarithmic transformation is performed and the resulting logarithmic distribution is normally

distributed, the z values tabulated for a standard normal distribution are used in the log-normal
frequency analysis. The standardised variate, z , is then calculated as:
y − y
z= 3.2.2.6
y

Examples of variables that have been known to follow a log-normal distribution include:
• Annual series of peak flow rates.
• Daily precipitation depths and streamflow volumes (also monthly, seasonal, and annual).
• Daily peak discharge rates.
• Annual precipitation and runoff

Example 3.2.2.1
The following table shows the annual peak floods for a river for the period 1940-1982. Assuming that
the data are log-normally distributed, determine 100-year flood.

Year 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951
Peak flood (m3/s) 71.9 195.1 495.6 342.7 56.6 100.3 900.6 41.6 58.1 492.8 160.3 60.9

Year 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963
Peak flood (m3/s) 22.7 140.5 24.5 34 49.6 146.7 261.1 94.9 90.6 86.4 112.1 25.2

Year 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975

Page 43
Peak flood (m3/s) 60.6 153.8 61.2 155.2 371 77.3 95.2 83.5 180.1 903.4 274.1 117

Year 1976 1977 1978 1979 1980 1981 1982

Peak flood (m3/s) 212.7 130.8 267.3 134.5 56.1 410.6 231.1

Solution
The logarithms of the data are shown in row 3 of the table below:

Year 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951
Peak flood (X) 71.9 195.1 495.6 342.7 56.6 100.3 900.6 41.6 58.1 492.8 160.3 60.9
Y=ln X 4.275 5.274 6.206 5.837 4.036 4.608 6.803 3.728 4.062 6.200 5.077 4.109

Year 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963
Peak flood (X) 22.7 140.5 24.5 34 49.6 146.7 261.1 94.9 90.6 86.4 112.1 25.2
Y=ln X 3.122 4.945 3.199 3.526 3.904 4.988 5.565 4.553 4.506 4.459 4.719 3.227

Year 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975
Peak flood (X) 60.6 153.8 61.2 155.2 371 77.3 95.2 83.5 180.1 903.4 274.1 117
Y=ln X 4.104 5.036 4.114 5.045 5.916 4.348 4.556 4.425 5.194 6.806 5.613 4.762

Year 1976 1977 1978 1979 1980 1981 1982

Peak flood (X) 212.7 130.8 267.3 134.5 56.1 410.6 231.1
Y=ln X 5.360 4.874 5.588 4.902 4.027 6.018 5.443

Thus the mean and standard deviation of the log-transformed data are 4.815 and 0.907 respectively.

Let Pr (Y  y100 ) be the probability of the 100-year flood of the log-transformed data. Thus:
P (Y  y
r 100 )= 1
T
1
=
100
= 0.01

Therefore:
P (Y  y ) = 1 − P (Y  y )
r 100 r 100

= 1 − 0.01
= 0.99

The CDF for normal distribution is:

 y100 −  
P (Y  y )=  = P (Z  z )
Pr  Z 

r 100 r
 

Substituting the given data into the equation gives:

0.99 = Pr (Z  z )

Page 44
From the normal distribution the value of z corresponding to this probability is:

z = 2.326
But:
y100 −  y
z=
y

Thus:
y100 − 4.815
2.326 = y = 6.925
100
0.907

But y100 = ln x100

Thus,
x100 = e
y100

= e6.925
= 1017.1
Therefore the 100-year flood, x100 , is given as 1,017.1 m3/s.

Example 3.2.2.2
The annual maximum series of flood magnitudes has a mean and standard deviation of 6000 m3/s and
4000 m3/s. respectively. If the data is assumed to follow a log-normal distribution:
a) What is the probability in each year that a flood magnitude would exceed 7000 m3/s?
b) Determine the flood magnitude with a return period of 100 years.

Solution
a) Let Pr (Y  7000) be the probability that the flood magnitude will exceed 7000 m3/s. This
probability is expressed in terms of the CDF as:
P (Y  7000) = 1 − P (Y  7000)
r r

For a log-normal distribution:

P (Y  y) = F (z) =
1
P Y  y − y  = P (Z  z)
z −u2 2 
e du =
r
2 
−
r
  y  r

The mean and standard deviation of the log-transformed data are estimated directly from
those of the original data using the following formulae:

1  
2
 y = ln   x 
2
2 1 +  x 

(
 y 2 = ln 1 +  x 2 )
Page 45
 x
 =
x
x

4000
Therefore:  = = 0.667
x
6000
1
 y = ln  60002 2 
2 1 + 0.667 = 8.515
 

 y2 = ln (1 + 0.6672 ) = 0.368   y = 0.607

The standardised variate is:

y − y
z=
y
But y = ln x = ln 7000 = 8.854

Thus:
8.854 − 8.515
z= = 0.558
0.607

Therefore:
P (Y  7000) = P (Z  z ) = P (Z  0.558)
r r r

From the table of normal distribution, the cumulative probability for Z  0.558 is 0.712.

Therefore:
P (X  7000) = 1 − P (X  7000)
r r
= 1 − 0.712
= 0.288

Hence probability in each year that a flood magnitude would exceed 7000 m3/s is 0.288.

b) The probability, Pr (X  x) , of flood magnitude with a return period of 100 years is calculated
as:
P (X  x) =
1
r
T
1
=
100
= 0.01
Therefore:

Pr (X  x) = 1 − 0.01
= 0.99

Page 46
The CDF for normal distribution is:
P (X  x) = P (Z  z )
r r

Substituting the given data into the equation gives:

0.99 = Pr (Z  z )

From the normal distribution the value of z corresponding to this probability is:

z = 2.326
But:
y − y
z= 
y

Thus:
y − 8.515
2.326 =  y = 9.927
0.607

But y = ln x

Thus,
x = ey
= e9.927
= 20475.8

Therefore the flood magnitude with a return period of 100 years is 20,476 m3/s

3.2.3 Gumbel’s (Extreme value type I) distribution

This extreme value distribution was introduced by Gumbel and is therefore commonly known as
Gumbel’s distribution. It is one of the most widely used probability distribution functions for extreme
values in hydrologic ad meteorological studies for prediction of flood peaks, maximum rainfall,
maximum wind speed etc. Gumbel’s probability density function is given as:

f (x) =
1 −( x −  ) 
e −(x− )  −e for − x 3.2.3.1


and thus its cumulative distribution function is given as:

P(X  x) = F (x) = exp− exp− (x −  )   ; − x 3.2.3.2

where the parameters are estimated as:

x 6
= 3.2.3.3


Page 47
  =  x − 0.5772 3.2.3.4

The inverse of Gumbel’s CDF is given as:

x =  −  ln − ln F (x) 3.2.3.5

Example 3.2.3.1
A rainfall series has a mean of 1.09 mm and a standard deviation of 0.343. Determine the 100-year
flood using the Gumbel distribution.

Solution
The parameters,  and  , of the Gumbel distribution are determined as:
x 6
=

0.343 
=
3.142
= 0.267

 =  x − 0.5772
= 1.09 − 0.5772  0.267
= 0.936
The probability of the 100-year rainfall is calculated as:

P(X  x) =
1
T
1
=
100
= 0.01
Therefore:

P(X  x) = F (x) = 1 − 0.01 = 0.99

Substituting the information into the CDF:

F (x) = exp− exp− (x − 0.936) 0.267= 0.99

The inverse of the CDF is thus:

x = 0.936 − 0.267  ln − ln 0.99

= 2.164 mm

Therefore the 100-year rainfall is 2.164 mm

Page 48
3.2.4 Frequency factors
Computation of magnitudes of extreme events such as flood flows requires that the cumulative
probability function be invertible; that is, given a value of the return period T, the corresponding
value xT be determined. The magnitude, xT , of a hydrologic event may generally be represented as a
function of the mean and the standard deviation as:

xT = x + KT sx 3.2.4.1
where,
xT is a magnitude corresponding to return period, T
x is the mean
s x is the standard deviation and
K T is a frequency factor dependent on the type of distribution.

The following illustrates the use of frequency factors for some probability distributions:

1. Normal distribution
For a normal distribution, the magnitude xT corresponding to a return period,T, may be determined
from:
x−
z= 3.2.4.2

as:

xT = x + KT sx 3.2.4.3

where,
x is the mean (that approximates )
s x is the standard deviation (that approximates )
KT = z is the frequency factor for a normal distribution.

The values x and s x are determined from the data. The value of z corresponding to an exceedence
probability of p = 1 T determined from the normal tables or estimated as follows:

2.515517 + 0.802853w + 0.010328w2   1 1 2

z=w− ; w = ln  for 0  p  0.5
1 + 1.432788w + 0.189269w2 + 0.001308w3   p 2 

  
3.2.4.4
and
 2.515517 + 0.802853w + 0.010328w2    1  1 2
z = − w −  = ln 
2  p  0.5
  (1 − p) 
; w for
 1 + 1.432788w + 0.189269w2 + 0.001308w3 
  
3.2.4.5
Example 3.2.4.1
The mean and standard deviation of annual daily rainfall at a station are 14.9 mm and 5.9 mm
respectively. Determine, using frequency factors, the 10-year rainfall, assuming that the rainfall is
normally distributed.

Page 49
Solution
The frequency factor equation is given as:
xT = x + KT sx
where,
K T = z for a normal distribution.

The probability of exceedence for a 10-year rainfall is given as:

1
p=
T
1
=
10
= 0.1
Thus:
  12
w = ln 1 
  p 
2
  
  1 1 2
= ln 
2 
  0.1 
= 2.146
and

2.515517 + 0.802853  2.146 + 0.010328  2.1462

z = 2.146 −
1 + 1.432788  2.146 + 0.189269  2.1462 + 0.001308  2.1463
= 1.282

Hence the 10-year rainfall is given as:

xT = x + KT sx
= 14.9 + 1.282  5.9
= 22.5 mm

2. Log-normal distribution
The log-normal distribution is determined as the normal distribution using the variable yT = ln xT .
Thus:
yT = y + KT s y 3.2.4.6

where,
y is the mean of the log-transformed data
s y is the standard deviation of the log-transformed data
K T = z is the frequency factor for a log-normal distribution.

The value of z is determined the same way as for normal distribution. The value of xT is then
determined as:

CVS 547E: LECTURE NOTES Page 50

 xT = e yT 3.2.4.7

Example 3.2.4.2
The annual maximum series of flood magnitudes has a mean and standard deviation of 6000 m3/s and
4000 m3/s. respectively. If the data is assumed to follow a log-normal distribution, determine the
flood magnitude with a return period of 100 years using frequency factors.

Solution
The frequency factor equation for a log-normal distribution is given as:

yT = y + KT s y
where,
K T = z for a log-normal distribution.

The mean and standard deviation of the log-transformed data are estimated directly from those of the
original data using the following formulae:

1  2 
y = ln  x 
2 1 + x 2 

(
s y2 = ln 1 +  x 2 )
sx
 =
x
x

4000
Therefore:  = = 0.667
x
6000

1  60002 
y = ln   = 8.515
2 1 + 0.6672 

( )
s y2 = ln 1 + 0.6672 = 0.368  s y = 0.607

The probability of exceedence for a 100-year flood is given as:

1
p=
T
1
=
100
= 0.01

CVS 547E: LECTURE NOTES Page 51

Thus:
  12
w = ln 1 
  p 
2
  
  1 12
= ln 
2 
  0.01 
= 3.035
and

2.515517 + 0.802853  3.035 + 0.010328  3.0352

z = 3.035 −
1 + 1.432788  3.035 + 0.189269  3.0352 + 0.001308  3.0353
= 2.327

Therefore:
yT = y + KT s y
= 8.515 + 2.327  0.607
= 9.927

The 100-year flood is thus given as:

xT = e yT
= e9.927
= 20,475 m3 / s

3. Gumbel distribution
For the Gumbel distribution, the frequency factor can be derived by substituting the frequency
equation xT = x + KT s x into the cumulative distribution function F (x )T= exp− exp− (x T−  )  
so that:

F (xT ) = exp− exp− (x T −  )  = 1 − 1 T 3.2.4.8

Substituting the values for  and  ans solving for KT gives:

 6    T 
KT = −  0.5772 + ln ln  
     T − 1 
 
3.2.4.9
This equation is valid when the sample size is large ( n  100 ). When the sample is small, the
following table is used to estimate K T .

The return period can also be expressed as:

−1
   K 
T = 1 − exp− exp− 0.5772 − T
 3.2.4.10
   6 

CVS 547E: LECTURE NOTES Page 52

Table 3.2.4.1: Gumbel’s frequency factors, K T , various return period and sample sizes.

Example 3.2.4.3
A rainfall series has a mean of 1.09 mm and a standard deviation of 0.343. If the rainfall series
follows the Gumbel’s distribution, determine the 100-year flood using frequency factor. Assume the
sample size is large.

Solution
The frequency factor equation is given as:

xT = x + K T sx
The sample size is large, thus:
 6    T 
KT = −  0.5772 + ln ln   for Gumbel distribution.
     − 
  T 1

For T = 100 :


KT = −  0.5772 + ln ln  100 
6 

3.142    100 − 1
 
= 3.136

Therefore the 100-year rainfall is given as:

xT = x + KT sx
= 1.09 + 3.136  0.343
= 2.166 mm

CVS 547E: LECTURE NOTES Page 53

Example 3.2.4.4
The mean of annual maximum discharges at a streamflow site with 25 years of record is 1000 m 3/s.
the standard deviation is 400 m3/s. estimate the magnitude of the 50-year flood for a Gumbel
extreme-value distribution.

Solution
The frequency factor equation is given as:

xT = x + KT sx
The sample size is small, thus the frequency factor, K T , is obtained from the table of frequency
factors for Gumbel distribution. For a sample size of 25 and return period of 50:

KT = 3.088

Therefore the 50-year dischrge is given as:

xT = x + KT sx
= 1000 + 3.088  400
= 2235 m3 / s

3.3 Hydrologic Design for Flood Management

Hydrologic design is the process of assessing the impact of hydrologic events on a water resource
system and choosing values for the key variables of the system so that it will perform adequately.
Hydrologic design may be used to develop plans for a new structure or to develop management
programs for better control of an existing system. Besides hydrology, other factors to consider when
designing a water resource systems include public welfare and safety, economics, aesthetics, legal
issues, and engineering factors such as geotechnical and structural design.

The hydrologic design scale is the range in magnitude of the design within which a value must be
selected. The most important factors in selecting the design value are cost and safety. The optimal
magnitude for design is one that balances the conflicting considerations of cost and safety.

Although the true upper limit is usually unknown, for practical purposes an estimated upper limit
may be determined. This estimated limiting value (ELV) is defined as the largest magnitude possible
for a hydrologic event at a given location, based on the best available hydrologic information. The
range of uncertainty for the ELV depends on the reliability of information, technical knowledge, and
accuracy of analysis.

Design of hydrologic structures may be based on a percentage of the ELV or on a design return
period. Based on past experience and judgment, some generalized design criteria for water-control
structures have been developed as summarized in Table 3.3.1.

CVS 547E: LECTURE NOTES Page 54

Table 3.3.1: Generalized design criteria for water-control structures

CVS 547E: LECTURE NOTES Page 55

 4
STORMWATER CONVEYANCE AND DETENTION
4.0 Introduction
Storm water management is knowledge used to understand, control, and utilize waters in their
different forms within the hydrologic cycle. The overall key component of stormwater management
is the drainage system. The following vital functions of a drainage system:
1. It removes stormwater from the streets and permits the transportation arteries to function
during bad weather; thus the life expectancy of street pavement is extended.
2. It controls the rate and velocity of runoff along gutters and other surfaces in a manner that
reduces the hazard to local residents and the potential for damage to pavement.
3. It conveys runoff to natural or manmade major drainage ways.
4. It can be designed to control the mass of pollutants arriving at receiving waters.
5. Major open drainage ways and detention facilities offer opportunities for multiple use such as
recreation, parks, and wildlife preserves.

Storm drainage criteria are the foundation for developing stormwater control. The following provides
a checklist for developing storm drainage criteria:
• Governing legislation and statements of policy and procedure.
• Initial and major drainage system provisions.
• Data required for design.
• Details of major system components.
• Detention requirements.
• Water quality criteria.
• Special considerations.

4.1 Stormwater Sewer Systems

4.1.1 Information needs and design criteria
To begin the design process of a storm sewer system, one must collect a considerable amount of
information such as:
• Local drainage criteria and design standards.
• Topographical maps.
• All existing and proposed utilities.
• Layout of design area.
• Soil characteristics and properties.
• Seasonal water table levels.
• Local intensity-duration-frequency storm data.
• Available pipe materials and sizes.

The design criteria for storm sewers may vary from country to country or from city to city. The
assumptions and constraints used in the design of storm sewers include:
a) The sewer system is designed for “gravity flow” so that pumping stations and pressurized
sewers are not considered.
b) The sewers consist of commercially available circular pipes.
c) The design diameter is the smallest commercially available pipe that has flow capacity equal
to or greater than the design discharge and satisfies all the appropriate constraints.

CVS 547E: LECTURE NOTES Page 56

 d) Storm sewers must be placed at a depth that will not be susceptible to frost, will drain
basements, and will allow sufficient cushioning to prevent breakage due to ground surface
loading.
e) The sewers are joined at junctions such that the crown elevation of the upstream sewer is no
lower than that of the downstream sewer.
f) To prevent or reduce excessive deposition of solid material in the sewers, a minimum
permissible flow velocity at design discharge or at barely full-pipe gravity flow is specified.
g) To prevent the occurrence of scour and other undesirable effects of high-velocity flow, a
maximum permissible flow velocity is also specified.
h) At any junction or manhole, the downstream sewer cannot be smaller than any of the
upstream sewers at that junction.
i) The sewer system is a dendritic network converging towards downstream without closed
loops.

Typical technical considerations and limitations to consider in the design of stormwater sewers are
listed in Table 4.1.1.1.

Table 4.1.1.1: Technical considerations and limitations in stormwater sewer design

Item Permissible value
Velocity:
Minimum design velocity 0.6 – 0.9 m/s
Maximum design velocity
Rigid pipe 4.6 – 6.4 m/s
Flexible pipe 3.0 – 4.6 m/s
Maximum manhole spacing 122 – 183 m
Minimum size of pipe 0.3 – 0.6 m
Minimum depth of soil cover 0.3 – 0.6 m

4.1.2 Design method and steps

A storm sewer system is a network of pipes used to convey storm runoff in a city. It conveys
stormwater from a surface inlet such as a manhole to an outfall. The rational method is commonly
used method used for the design of stormwater sewers. The design assumptions of the method are
given in the previous chapters. The design process of a stormwater sewer can be grouped into two
main parts: runoff prediction and pipe sizing. The steps involved in the design are:
1. Identify the contributing subbasin(s) to a pipe under consideration.
2. Determine the total area
3. Determine the equivalent runoff coeffient for the subbasin(s).
4. Determine the time of concentration for the subbasin(s).
5. Using information on storm IDF and time of concentration, determine the rainfall intensity.
6. Calculate the peak runoff using the rational method.
7. Determine the required pipe diameter.
8. Select the minimum commercially available pipe with a diameter greater than that computed
in step 7.
9. Check for velocity requirements and general losses.

CVS 547E: LECTURE NOTES Page 57

4.1.3 Important formulae
Manning’s equation (Equation 4.1.3.1) can be applied in the sizing of pipes.
1 2 1
Q = 3 2
4.1.3.1
p AR S o
n
where,
Qp - is the peak runoff (computed using ration method)
n - is the Manning’s coefficient
A - is the area of the pipe
R - is the hydraulic radius of the pipe section
S o - is the slope (dependent on topography).

For circular pipes the area is computed as:

D 2
A= 4.1.3.2
4
where,
D - is the diameter of the pipe

The wetted perimeter is calculated as:

P = D 4.1.3.3
where,
P - is the wetted perimeter of the circular pipe

The hydraulic radius is then calculated as:

A D
R= = 4.1.3.4
P 4
where,
D - is the diameter of the pipe

Substituting for area and hydraulic radius into Manning’s equation yields:
2
1  D 2  D  3 1
Qp =    So 2 4.1.3.5
n  4  4 

The required diameter of the storm sewer can then be calculated as:

38
 
D =  3.21Qn 4.1.3.6
 
 So 
The time of concentration to any point in a storm drainage system is the sum of the inlet time and the
flow time in the upstream sewers connected to the subbasin:

tc = to + t f 4.1.3.7
where,
tc - is the time of concentration
to - is the inlet time (time for overland flow)

CVS 547E: LECTURE NOTES Page 58

 t f - is the time of flow in upstream sewers connected to subbbasin.
The flow time is computed as:
L
t = 4.1.3.8
f
v
where,
L - is the length of the pipe and
v - is the flow velocity in the pipe
.
In stormwater sewer design, each sewer is designed individually and independently (except for the
computation of sewer flow time). The corresponding rainfall intensity is computed repeatedly for the
area drained by the sewer. Thus, as the design progresses towards the downstream sewers, the
drainage area increases and usually the time of concentration increases accordingly. This in turn gives
a decreasing rainfall intensity.

4.1.4 Hydraulic losses in storm sewer systems

To analyse the hydraulic effectiveness of storm sewer design, it is necessary to check the hydraulic
gradient losses. The hydraulic-gradient loss analysis can be used to determine if design flows can be
accommodated without causing flooding at various locations or causing flows to exit the system at
locations where this is not acceptable. Some of the losses related to sewer design include transition,
manhole and bend losses.

Transition losses occurs as a result of pipe size transitions through manholes. Contraction losses can
be expressed as:

H = 0.1 v2 − v1 
2 2
for v v 4.1.4.1
c  2g 2g  2 1
 
where,
H c - contraction losses
v1 , v2 - upstream and downstream velocities respectively
g - gravitational constant

Expansion losses can be expressed as:

H = 0.2  v1 − v2 
2 2
for v v 4.1.4.2
e  2g 2g  1 2
 
where,
H e - expansion losses

Manhole losses are related to pipe configurations at storm sewer junctions (manholes). For a straight-
through manhole with no change in pipe sizes, manhole losses can be expressed as:
v2
H m = 0.05 4.1.4.3
2g
where,
H m - manhole losses

CVS 547E: LECTURE NOTES Page 59

Losses at terminal manholes can be estimated as:
v2
Hm = 4.1.4.4
2g

Bend losses in storm sewers can be estimated as:

v2
H b = Kb 4.1.4.5
2g
where H b are bend losses and the factor, K b , is a sewer bend loss coefficient dependent on the angle
of deflection of sewer segments.

Example 4.1.1
A hypothetical drainage basin of a city is shown in Fig. 4.1.1E. The basin characteristics are given in
Table 4.1.1E, while the pipe layout data are given in Table 4.1.2E. The intensity-duration-frequency
relation for storms in the city is given as:

3048T 0.175
i= ;
(t + 27)
where T - is the return period in years and t - is the time of concentration in minutes.

Design the storm sewer for a 10-year return period. The Manning’s coefficient for all sewers is 0.012.
The diameters (in mm) of the commercially available pipes are: 150, 225, 300, 375, 450, 525, 600,
760, 825, 900, 1050.

Fig. 4.1.1E: Hypothetical drainage basin of a city

CVS 547E: LECTURE NOTES Page 60

Table 4.1.1E: Characteristics of the drainage basin
Catchment Area (km2) Runoff coefficient Inlet time (min)
I 0.00809 0.7 5
II 0.0121 0.7 7
III 0.0162 0.6 10
IV 0.0162 0.6 10
V 0.0202 0.5 15
VI 0.0182 0.6 15
VII 0.0182 0.8 15

Table 4.1.2E: Pipe Layout characteristics

Pipe Length (m) Slope
EB 137.16 0.0064
AB 167.64 0.0081
BC 121.92 0.0064
CD 137.16 0.0064

Solution
Pipe EB:
Contributing subbasins: III
Total contributing area:
A = 0.0162 km2
Composite runoff coefficient:
Ce = 0.6
Time of concentration= Max10= 10 min
Rainfall intensity:

3048T 0.175 3048  5

0.175
i= = = 109.177 mm / h
(t + 27) (10 + 27)
Discharge:
1 1
Q= ciA =  0.6 109.177  0.0162 = 0.295 m3 / s
3.6 3.6

Required pipe diameter:

 3.21Qn  38  3.21 0.295  0.015  38

D=  =  = 0.523 m = 523 mm
 s0   0.0064 
 

Therefore try 525 mm diameter pipe

CVS 547E: LECTURE NOTES Page 61

Flow velocity in the pipe:
Q
v= =
A    0.5252  = 1.363 m / s ; OK
0.295
 
 4 

ADOPT 525 mm diameter pipe for sewer EB

Sewer flow time:

L 137.16
t = = = 100.6 s = 1.68 min
f
v 1.363

Pipe AB:
Contributing subbasins: I and II
Total contributing area:
A = 0.0202 km2
Composite runoff coefficient:
0.7  0.00809 + 0.7  0.0121
Ce = = 0.7
0.0202

Time of concentration= Max5,7= 7 min

Rainfall intensity:

3048T 0.175 3048  5

0.175
i= = = 118.811 mm / h
(t + 27) (7 + 27)

Discharge:
1 1
Q= ciA =  0.7 118.811 0.0202 = 0.467 m3 / s
3.6 3.6

Required pipe diameter:

 3.21Qn  38  3.21 0.467  0.015  38

D=  =  = 0.594 m = 594 mm
 s   0.0081 
 0 

Therefore try 600 mm diameter pipe

Flow velocity in the pipe:

Q
v= = 0.467
A    0.62  = 1.652 m / s ; OK
 
 4 

CVS 547E: LECTURE NOTES Page 62

ADOPT 600 mm diameter pipe for sewer AB

Sewer flow time:

L 167.64
t = = = 101.5 s = 1.69 min
f
v 1.652

Pipe BC:
Contributing subbasins: I -V
Total contributing area:
A = 0.0728 km2
Composite runoff coefficient:
0.7  0.00809 + 0.7  0.0121+ 0.6  0.0162 + 0.6  0.0162 + 0.5  0.0202
Ce = = 0.6
0.0728

Time of concentration= Max(7 + 1.69), (10 + 1.68),10,15= 15 min

Rainfall intensity:

3048T 0.175 3048  5

0.175
i= = = 96.18 mm / h
(t + 27) (15 + 27)

Discharge:
1 1
Q= ciA =  0.6  96.18  0.0728 = 1.167 m3 / s
3.6 3.6

Required pipe diameter:

 3.21Qn  38  3.211.167  0.015  38

D=  =  = 0.876 m = 876 mm
 s   0.0064 
 0 

Therefore try 900 mm diameter pipe

Flow velocity in the pipe:

Q
v= = 1.167
A    0.92  = 1.834 m / s ; OK
 
 4 

ADOPT 900 mm diameter pipe for sewer BC

Sewer flow time:

L 121.92
t = = = 66.5 s = 1.11 min
f
v 1.834

CVS 547E: LECTURE NOTES Page 63

Pipe CD:
Contributing subbasins: I -VII
Total contributing area:
A = 0.109 km2
Composite runoff coefficient:
0.7  0.00809 + 0.7  0.0121 + 0.6  0.0162 + 0.6  0.0162 + 0.5  0.0202
Ce =
0.109
0.5  0.0182 + 0.5  0.0182
+
0.109
= 0.568

Time of concentration= Max(15 + 1.11),15,15= 16.11 min

Rainfall intensity:

3048T 0.175 3048  5

0.175
i= = = 93.704 mm / h
(t + 27) (16.11 + 27)

Discharge:
1 1
Q= ciA =  0.568  93.704  0.109 = 1.612 m3 / s
3.6 3.6

Required pipe diameter:

 3.21Qn  38  3.211.612  0.015  38

D=  =  = 0.989 m = 989 mm
 s0   0.0064 
 

Therefore try 1050 mm diameter pipe

Flow velocity in the pipe:

Q
v= =
A   1.052  = 1.862 m / s ; OK
1.612
 
 4 

ADOPT 1050 mm diameter pipe for sewer CD

Sewer flow time:

L 137.16
t = = = 73.7 s = 1.23 min
f
v 1.862

The computations are summarised in Table 4.1.3E:

CVS 547E: LECTURE NOTES Page 64

Table 4.1.3E: Summary of computations
Sewer Length Slope Total Runoff Tc Rainfall Design Computed Pipe Flow Flow
pipe (m) (so) area coef. (min) intensity discharge, sewer size velocity time
drained (C) (mm/h) Q (m3/s) diameter used (m/s) (min)
(km2) (mm) (mm)
EB 137.16 0.0064 0.0162 0.6 10 109.177 0.295 523 525 1.363 1.68
AB 167.64 0.0081 0.0202 0.7 7 118.811 0.467 594 600 1.652 1.69
BC 121.92 0.0064 0.0728 0.6 15 96.18 1.167 876 900 1.834 1.11
CD 137.16 0.0064 0.109 0.568 16.11 93.704 1.612 989 1050 1.862 1.23

4.2 Stormwater Drainage Channels

Stormwater drainage channels (or flood-control channels) must behave in a stable, predictable
manner to ensure that a known capacity will be available for a design storm event. In most cases, the
design goal is a non-eroding channel boundary, although, in recent cases, a dynamic channel is
desired. Two types are usually designed: rigid-lined and flexible-lined channels.

4.2.1 Rigid-lined channels

Rigid-lined channels have non-erodible channel sides typically lined with concrete, grouted riprap,
stone masonry, or asphalt. The general steps in the design of such a channel include:
1. Select the Manning roughness coefficient n (Table 4.2.1.1), the channel side slope z and the
channel bottom slope So. The bottom slope is based upon the topography and other
considerations such as alignment.
2. Compute the uniform flow section as:

2 Qn
AR 3
= 4.2.1.1
So
in which A is the cross-sectional area of flow (m2), R is the hydraulic radius (m) and Q is the
design discharge (m3/s).
3. Determine the channel dimensions and flow depth for the uniform flow section factor
computed in step 2. Choose the expression for the uniform flow section factor as a function
of depth. For a trapezoidal channel:

1
  3
 (B + zy)5 y 5 2

 w
 = AR 3 4.2.1.2
2
 B w + 2 y 1 + z 2  
  
where Bw is the bottom width (m) and y is the depth of flow (m). By assuming several values
of Bw and z, a number of combinations of section dimensions can be obtained. Final
dimensions should be based upon hydraulic efficiency and practicality. For the best
hydraulically efficient trapezoidal channel (Table 4.2.1.2):

13
 y8  2
3  = AR 3
4.2.1.3
 4 
4. Check the minimum velocity requirement.
5. Add an appropriate freeboard to the depth of the channel section.

CVS 547E: LECTURE NOTES Page 65

Table 4.2.1.1: Manning roughness coefficient for channel design

Table 4.2.1.2: Best hydraulically efficient sections without freeboard

Example 4.2.1.1
Design a non-erodible trapezoidal channel to carry a discharge of 11.33 m3/s. Assume a Manning
coefficient of 0.025 and channel slope of 0.0016

CVS 547E: LECTURE NOTES Page 66

Solution
1. Manning roughness coefficient n=0.025 and the channel bottom slope So=0.0016.
2. The uniform flow section is given as:

2 Qn 11.33  0.025
AR 3
= = = 7.081
So 0.0016

3. For the best hydraulic trapezoidal section:

1
 y8  3 2
3  = AR 3 = 7.081
 4 
 y = 2.016 m

The best hydraulic trapezoidal section is half a hexagon, thus:

z = tan 30o = 0.577 .
A = y2 = 2.0162  = 7.039 m 2
Therefore:
(Bw + zy )y = 7.039
(Bw + 0.577  2.016)2.016 = 7.039
 Bw = 2.328 m

4. The velocity is given as:

Q 11.33
v= = = 1.61m / s
A 7.039
This is greater than the minimum permissible velocity for silt deposition.

5. An appropriate freeboard e.g 1 m can be added to the depth of the channel section. Thus the
channel depth would be 3.016 m.

4.2.2 Flexible-lined channels

Flexible-lined channels include rock riprap and vegetative linings and are considered flexible because
they can conform to change in channel slope. Flexible linings have the following advantages for
stormwater conveyance:
a) They permit infiltration and exfiltration,
b) They filter out contaminants,
c) They provide greater energy dissipation,
d) They allow flow conditions that provide better habitat opportunities for local flora and fauna,
and
e) They are less expensive.

There are two main types of flexible-lined channels: vegetative and non-vegetative. Non-vegetative
flexible channels are considered in this section.

The main design procedure for non-vegetative flexible lining is as follows:

CVS 547E: LECTURE NOTES Page 67
 1. Choose a flexible lining from Table 4.2.2.1 and note its permissible shear stress,  p .
2. Assume an appropriate range of flow depth.
3. Use Table 4.2.1.1 to find the Manning, n.
4. Compute the flow depth, ycomp, from the Manning equation using the value of n from step 3.
5. Compute shear stress for the design condition as:

 comp = RSo 4.2.2.1

where,
 comp - is the computed shear stress
 - is unit weight of water, 9810 N/m3
R - is the hydraulic radius, m
So - is the average bottom slope
If  comp   p the lining is acceptable, otherwise go to step 1 and choose a different lining.

Table 4.2.2.1: Permissible shear stresses for lining materials

Example 4.2.2.1
Design a flexible-lined trapezoidal channel for a discharge of 11.33m3/s and a slope of 0.0016. use
non-vegetative lining. Consider the best hydraulic section.

Solution
1. A gravel riprap, 1 in (2.5cm) D50, is chosen. From Table 4.2.2.1 its permissible shear stress is
 p = 1.61 kg / m 2 = 1.61 9.81 = 15.79 N / m 2 .

CVS 547E: LECTURE NOTES Page 68

 2. A flow depth of greater than 60 cm is assumed..
3. From Table 4.2.1.1, the Manning roughness factor is n = 0.03
4. The flow depth, ycomp, is computed from the Manning equation as follows:

2 Qn 11.33  0.03
AR 3
= = = 8.498
0.0016

For the best hydraulic trapezoidal section:

1
 y8  3 2
3  = AR 3 = 8.498
 4 
 y = 2.159 m

1 1
Hydraulic radius, R = y=  2.159 = 1.08 m
2 2
5. Compute shear stress for the design condition as:

 comp = 9810 1.08  0.0016

= 16.952 N / m 2

Because  comp   p , try a 5-cm gravel riprap and repeat the procedure.

1. A gravel riprap, 2 in (5cm) D50, is chosen. From Table 4.2.2.1 its permissible shear stress is
 p = 3.22 kg / m 2 = 3.22  9.81 = 31.59 N / m 2 .
2. A flow depth of greater than 60 cm is assumed..
3. From Table 4.2.1.1, the Manning roughness factor is n = 0.034
4. The flow depth, ycomp, is computed from the Manning equation as follows:

2 Qn 11.33  0.034
AR 3
= = = 9.631
0.0016

For the best hydraulic trapezoidal section:

1
 y8  3 2
3  = AR 3 = 9.631
 4 
 y = 2.263 m
1 1
Hydraulic radius, R = y =  2.263 = 1.132 m
2 2
5. Compute shear stress for the design condition as:

 comp = 9810 1.132  0.0016

= 17.769 N / m 2

CVS 547E: LECTURE NOTES Page 69

 Because  comp   p , accept a 5-cm gravel riprap .

The best hydraulic trapezoidal section is half a hexagon, thus:

z = tan 30o = 0.577 .
A = y2 = 2.2632  = 8.87 m 2
Therefore:
(Bw + zy )y = 8.87
(Bw + 0.577  2.263)2.263 = 8.87
 Bw = 2.614 m

The velocity is given as:

Q 11.33
v= = = 1.277 m / s
A 8.87
This is greater than the minimum permissible velocity for silt deposition.

6. An appropriate freeboard can be added.

CVS 547E: LECTURE NOTES Page 70