Design Flood

NATIONAL INSTITUTE OF TECHNOLOGY

ROURKELA

K C PAT R A
Professor in Civil
Engineering
National Institute of Technology, Rourkela, Orissa, India-769008,
[email protected]
Former Head and Dean
(AR)
1
 INTRODUCTION
• A flood is an unusual high stage of a river that overflows the
natural or man made banks spreading water to its flood plains.
• The first information required to predict a flood at a particular
place and time is the measurement of all the floods for
maintaining a good record. Analysis of flood records gives an in-
depth knowledge based on which flood prediction and protection
measures can be carried out.
• A design flood to be considered for a structure is dependent on a
large number of factors, but the importance of the structure vis-a-
vis its desired objective has to be kept in mind.
• A flood accepted for the design of a structure is based on
(i) importance of the structure,
(ii) economy,
(iii) probable effect at its downstream due to its sudden damage,
(iv) life expectancy of the structure,
(v) inconvenience it can cause to traffic,
(vi) population density of the down stream area,
(vii)submergence of mineral, industrial and other strategic areas
and
(viii)economic condition of the people of the affected area.
• Damages due to small structures like minor irrigation projects,
small causeways or bridges create temporary disruptions of the
area. Loss to life and property from such damages are small.
DESIGN FLOOD
• A flood used for the design of a structure on considerations of its
safety, economy, life expectancy and probable damage
considerations is called the design flood.
• The flood selected for design of such structures should probably
be the highest one. For other structures, some probability of
failure can be allowed.
• Depending on the size of a water resources project, any of the
following types of flood can be estimated.
Frequency Based Flood (FBF)
A design flood estimated using flood-frequency analysis for an
accepted return period is called FBF. Sometimes frequency analysis
of rainfall data is carried out and a suitable rainfall-runoff model
generates the required FBF.
Probable Maximum Flood (PMF)
US Army Corps of Engineers (1979) defined it as extreme flood,
which is physically possible in a region due to the most severe
combination of critical meteorological and hydrological factors that
are reasonably possible over the region. PMF is used for design of
important structures with virtually no-risk criterion. Failure of dams
can create immense loss to a nation. Virtually 100 % safety against
such failures should be ensured. PMF is associated with probable
maximum precipitation (PMP).
Standard Project Flood (SPF)
Guidelines by CWC, India (1972) for selection of design flood of a
project depending on its size and the importance are
Investigators have proposed regression type equations to suit the
specific area under consideration. Computation of design flood for a
structure is carried out under the following methods.
(1) For un-gauged catchment (2) For gauged catchment
(a) Rational approach (a) Unit hydrograph approach
(b) Empirical equations (b) Flood frequency analysis.
(c) Envelope curves
Rational Method
The original formula in FPS unit in which one acre-inch of rainfall for one hour
produces one cusec of flow from the area.
Runoff gradually increases from zero to peak when rainfall duration reaches
the time of concentration tc and thereafter it becomes constant for the period
of rainfall excess ( t–tc ), i.e., from time tc onwards. After the cessation of rain,
the runoff recedes gradually to become zero at time tc from the end of the
peak.
In rational formula a certain percentage of rainfall is considered as runoff. The
peak discharge in the original FPS unit is given by
Qp= C I A or in SI units Qp= 0.278 C I A.
where Qp is the peak discharge in m 3/sec , C the runoff coefficient
representing a ratio of runoff to rainfall, A the catchment area in km² and I
the rainfall intensity in mm/h whose duration should at least be equal to the
time of concentration of the basin. The value of coefficient C varies from 0.05
to 0.95. It represents the total cumulative effect of the watershed losses.
Factors affecting the value of C are (i) initial losses, (ii) depression storage, (iii)
nature of the soil, (iv) surface slope, (v) degree of saturation, (vi) rainfall
intensity, (vii) geology of the catchment and (viii) geo-hydrological
When a watershed of area A has distinct zones of C then divide the
area into sub zones of A1, A2 ,..., An, with their runoff coefficients
C1, C2, ..., Cn and a weighted C is computed as CW = S(CiAi )/A. The
To find intensity of rainfall I at a place any of the following formulae
can be used depending on their suitability. The general form
correlating intensity -duration-return
K T a period is given as
I
(t c  b ) n
where I is the intensity of rainfall (cm/h), T the return period (years),
tc the time of concentration of the watershed in (h), K, a, b and n are
constants. For different places of India these constants can be
evaluated and used (Table 8.3).
Limitations of Rational Formula
1. The formula is applicable to small catchments up to 50 km².
2. Duration of rainfall intensity should be more than the time of
concentration.
3. It gives the peak of hydrograph but does not provide the
complete hydrograph.
4. It plots a straight line relation between Qp and I with intercept
zero whereas nature does not follow such a simple equation.
5. Rainfall intensity must be constant over the entire watershed
during the time of concentration.
6. Coefficient C is assumed to be same for all storms which means
the losses are constant for all storms.
The first step in using the rational formula is to compute tc for the
watershed. Once tc is known all previous rainfall records can be
checked and the rainfall durations exceeding tc are analyzed to get
rainfall intensities I for which return period T is computed.
Regression Formulae
A regression relation of the form Q = CAn can be developed for the
region under consideration, where Q is the discharge rate in m3/sec,
A the drainage area in sq.km, C and n are constants of regression
which are to be evaluated for the region.
Flood Peak Formulae
(1) Dicken‘s Formula (1865): Dicken proposed the catchment
flood peak relation for the regions of Central and North India
catchments Qp = C A3/4
where A is the catchment area in km², Qp the maximum flood
discharge (m3/sec) and C a coefficient varying from 2.8 to 28 or
even up to 40 for the Western Ghat hills. The range of C for various
regions of India are given in Table 8.4.
Table 8.4 Dicken's Coefficient for Various Regions of India
Sl. No. Area Description Value of C
(1) (2) (3)
1. For flat, sandy or cultivated plains of north India 2.8 to 6.0
2. Undulating area with hard soil for north Indian hills 11.0 to 14.0
3. Undulating country for central India region 14.0 to 28.0
4. Catchment covered with precipitous hills as of Orissa, Andhra 22.0
to 28.0
5. Western ghat regions of India. 20.0 to 40.0
For the estimation of flood peaks for small structures, the coefficient
C varying between 25 and 28 is widely used for Orissa catchments.
(2)Ryve's Formula (1884) : Ryve modified Dicken's equation to
suite the catchments of Tamil Nadu and proposed relation as
Qp = CA2/3
Ryve suggested the following three values of C
(i) C = 6.8 for areas within 75 km from the coast
(ii) C = 8.5 from 75 to 175 km from the coast
(iii) C =10.2 for a limited areas near hills.
(3)
The Inglis Foormula
value of (1940)
coefficient : Inglis
exceeding 40studied a good
have been number
reported for of
catchments
certain river in the earlier Bombay state, parts of west Deccan
projects.
124 A
plateau and suggested the following relation Q pbetween
 catchment
( A  10 . 4 )
area and peak discharge

(4) Ali Nawaz Jung Bahadur Formula: Qp= C (0.3906A)0.925–(1/14) log

0.3906A

where C varies from 49 to 60.

(5) Coutagne Formula : The formula is used in France Qp =
150(A)0.5
and is applicable for area between 400 - 3000 0.08
km².
Q p  130(0.386 A) 0.894( 0.386A )
(6) Meyer Formula (1926) : The formula is extensively used in USA
Qp =175 (A)1/2
For use of this equation the area should be higher than 10 km².
(7) Fanning (USA) Formula : Qp = 2.64 (A)0.8
(8) Creager (USA) Formula :
(9) Boston Society of Civil Engineers Formula : Qp = C (A)1/2
Empirical Relations with Return Periods
(1)Horton's Equation : qp = 71.2 ( T)1/4 (A)–1/2
qp is the flood peak in m3/s/km2, A the drainage area in km², T the
return period in years
(2) US Geological Survey (1955) : The mean annual average
flood (Q2.33 m3/s) with a return period of 2.33 years can be computed
as Q2.33 = 0.0147 C A0.7
where C is a constant whose value varies from 1 to 100. The
equation follows Gumbel's extreme value distribution which has the
property of T = 2.33 years for average value of the series when data
length is large, A is the catchment area in sq. km.
(3) Fuller's Equation (1914) :
QPT = Qav (1 + 0.8 log T)
or QPT = C A0.8 (1 + 0.3474 ln T)
and Qmax=QPT (1+2.66A-0.3)
where QpT is the maximum 24 h flood in m3/sec with return period T
years, A the catchment area in km2, C a constant varying between
0.03 to 2.8, QaV the yearly average flood, Qmax the maximum
instantaneous discharge.
(4) Pettis Formula : Qp=C(P×B)5/4  1185 
 
 ( a  6)   p year return period in
where Qp is the peak flood
p   discharge   100 of 100
 ( A  a ) 
m3/sec, P the one day rainfall of 100 year return period in cm, B the
SCS Curve Number Method
• Also known as hydrologic soil cover complex number method,
developed by Ogrosky and Mockus (1957) for determining peak
rate of runoff from small watersheds.
• A runoff curve number (CN) is developed through field studies by
measuring runoff from different soils at various locations.
• The antecedent moisture condition and the physical characteristics
of the watershed are correlated to give hydrologic soil groups.
• Soil of any watershed can be classified into the following four
hydrologic groups.
Group A: A low runoff potential group with very high infiltration
rate. From such soils, even under wet condition, the runoff
expectations are low. Infiltration rate is 8-12 mm/h.
Transmission rate is very high for such soils.
Group B: Moderately low runoff potential soil groups with
moderate rate of water transmission. Soil textures vary
from fine to moderately course. Final infiltration rate is 4 -
8 mm/h.
Group C: Moderately high runoff potential with low infiltration
rates with moderately good to well drained soils. Texture is
moderately fine to moderately coarse with slow rate of
water transmission. Final rate of infiltration is 1 - 4 mm/h.
Group D: Very slow infiltration rate when thoroughly wet. Clay
soils form such groups. The final infiltration rate for such
soils vary from 0 to 1 mm/h. Such soils have very low rate
AMC - I= Lowest runoff potential. Soils are dry enough for
satisfactory cultivation to take place.
AMC - II= Average condition.
AMC - III= Highest runoff potential. Watershed is practically
saturated from antecedent rain. The AMC group is
determined using five day antecedent rainfall.

• A value of CN for AMC -II condition is selected from Table 8.5

depending on the land use pattern.
• If the area consists of patches of land uses then a composite curve
number (CN) for the watershed can be obtained by weighing
them in proportion of the area.
• For example for a watershed of 100 km², if soil of 70 sq. km (70%
of area) has CN of 60 and 30 sq. km (30%) has CN 80, the
weighted CN for the catchment is 0.7 × 60 + 0.3 × 80 = 66.
• The value of CN is lower for soils with high infiltration, than for
soils with low infiltration.
• A poor condition refers to heavily grazed pastures with almost
no vegetation.
• A good condition refers to pastures, lightly grazed with 75% or
more land covered with plants.
where Qd is the runoff depth in cm over the basin, P is the mean
rainfall in cm over the basin, S the potential maximum retention in
cm, Ia the initial losses consisting of interception, depression
storage and infiltration.
For black soil Ia is taken as 0.1 S for AMC II and III type, Ia is taken ( P  0 .3 S ) 2
Qd 
as 0.3 S for AMC - I type. For all other types, Ia is 0.2 S. In general
( P  0 .7 S )
( P  0 .1 S ) 2
the value of Ia is take as 0.2
Qd S.
( P  0 .9 S )
For all soil regions except black soil of AMC II and III, Qd is
calculated as
For black soil of AMC II and III
Example: A cross drainage structure is to be constructed at a place
in Orissa 65 km away from the Bay of Bengal. Calculate peak
discharge for the catchment area of 15 km2. Use any four empirical
relations and tabulate their results. Calculate the coefficients for
Ryve's and Ali Nawaz Jung Bahadur formula if they are to be used
to give flood peak of same magnitude as Dicken's formula. Also
calculate the flood peaks for return period of 50 years if the flood is
considered as the result of 24 h rainfall.
Solution
(i) Dicken's formula QP = C A3/4
For Orissa, C = 28 (maximum for Orissa region)
QP = 28 × (15)0.75 = 213.4 m3/sec.
(ii) Ryve's formula = C A2/3
For use of Ryve's formula in Orissa, the value of C is selected as 35
to match with flood peak of Dicken's formula.
QP = 35 × (15)2/3 = 213.4 m3/sec.
FLOOD ESTIMATION FOR GAUGED CATCHMENT
Computing the peak rate of runoff from a watershed introduces an
element of probability. Basing on the availability of data the two
approaches for predicting the peak flow from a given gauged
watershed are :
(i) Approach - I : When the peak river discharges are available.
(ii) Approach - II : When complete runoff hydrograph along with the
corresponding rainfall data and the PMP over the watershed
are available.
Approach I is solved by the use of flood-frequency analysis and II by
unit hydrograph
8.4.1 Flood Frequency Analysis
The method of analysis and predicting flood from the data of runoff
peaks is called flood frequency analysis.
It gives only the magnitude of flood peak of desired recurrence
Example: Annual flood peaks (m /s) at a gauging site from 970-90
3
interval or return period, but doesn't provide information on
are given below.
complete hydrograph or the flood volume.
(i) Determine flood magnitudes for 100 and 1000 years by various
distributions.
(ii) What is the recurrence interval of flood magnitude of 1000
m3/s?
Year 1970 71 72 73 74 75 76 77 78 79 80 81 82 83
84 85 86 87 88 89 90
Flood 1065 645 1005 1350 860 150 2260 650 2840 990
870 910 750 930 750 1070 830 1095 384 2230 3210
Solution
Method of frequency factors are used to determine flood peaks of
return periods of 100 and 1000 year. The statistical parameters are
calculated in Table 8.9 that gives

Standard deviation = 786.2

Coefficient of variation = 0.665
Coefficient of skew = 1.478
Chow's general equation for flood frequency analysis is given as
XT = Xav + K s
where K is the frequency factor, T the recurrence interval in
years, XT the flood magnitude of return period T years, Xav the
mean of the maximum series, s is the standard deviation. Using
the statistical parameters, the values of flood peaks for 100 and
1000 year return periods for various methods are discussed
Gumbel's
below. Method
(a) From the frequency factor table for Gumbel distribution the
frequency factors for return periods of 100 and 1000 years for the
sample length of 21 are read as
K100 = 3.815 and K1000 = 5.96
X100 = 1183 + 3.815 × 786.2 = 4182 m3/sec.
X1000 = 1183+ 5.960 × 786.2 = 5869 m3/sec.
(b) Using the relation of Gumbel modified by R.V. Powel (1943)
K100 = -Ö6/p [0.5772 + ln.ln(T/T-1)] = -ÖÖ6/p [0.5772+ln.ln
(100/99)]
= -Ö6/p (-)4.023 = 3.137
K1000 = -Ö6/p [0.5772+ ln.ln(1000/999)] = -Ö6/p (-6.33) = 4.935
The values of frequency factors using Gumbel - Powel equation is
less than the frequency factors given by Gumbel's table because
of the data length. Powel method is used when data length is large
(more than 100 years). Using Gumbel-Powel relation, 100 and 1000
year return period floods are computed below.
X100 = 1183 + 3.137 × 786.2 = 3649 m3/sec
Pearson
X1000= Type
1163 -+III4.935 × 786.2 = 5063 m3/sec
The sample has the coefficient of skewness Cs as 1.478. The
frequency factors for 100 and 1000 year return periods with Cs as
1.478 is obtained from Table 2.8A.
T = 100 , K100 = 3.32 ; X100 = 1183 + 3.32 × 786.2 = 3793
m3/sec
T = 1000, K1000 = 5.26; X1000 = 1183 + 5.26 × 786.2 = 5318
m3/sec.
As per US practice Cs may or may not be multiplied by the factor
(1 + 8.5/N). In the present problem the factor is not multiplied to
the values of Cs.
Log-Pearson Type III
For log-Pearson type -III distribution, the series are first transferred to
logarithmic scale. The mean, standard deviation and coefficient of
skewness are computed from the log transferred series given in
Table 8.9.
Mean (yi = log xi) = 2.987626
Standard deviation = 0.2913
coefficient of variation Cv = 0.2913/2.9876 = 0.0975
Coefficient of skewness Cs =–0.613
Corresponding to Cs =–0.613, the frequency factors are read from
the Table 2.8B as
K100 = 1.8704 and K1000 = 2.2587
Flood peaks are calculated as follows
K100 =1.8704 : Y100= 2.987626 + 1.8704 × 0.2913 = 3.5324735
K1000=2.2587 : Y1000= 2.987626 + 2.2587 × 0.2913 = 3.6455853
taking antilogarithms X100 =3408 m3/sec and X1000 = 4422
m3/sec.