Flood Risk Assessment for the Design of Hydraulic Structures –

A Case Study of Greater Zaab River, Kurdistan Region, Iraq

Sarfraz Munir(1), Rubar Dizayee(2), Ghamgin K. Rashid(3)

(1) University of Kurdistan- Hewler, Erbil. Iraq.

00 964 771 916 5526, s.munir@ukh.edu.krd

1. Introduction – Hydraulic structures are installed in streams and rivers in order to harness water for

beneficial uses, economic development as well as for flood protection. Occurrence of floods is common

in rivers, almost in every part of the globe, which results in losses to human life, deterioration of

environment, damage to economy as well as failure of hydraulic structures. Recent flood in Kurdistan,

Iraq has caused losses of millions of dollars. Heavy rain fall in the region has destroyed the infrastructure,

many villages have been damaged, and most of the countryside roads and bridges have been blocked and

damaged due to landslides and rock fall. On an average, floods cause around 140 deaths and cost

approximately $6 billion annually in United States of America [1]. Flooding in Mississippi and Missouri

Rivers in 1993 caused around $20 billion damages. Generally, it is impractical to stop floods from

occurring, but structural and nonstructural strategies can be devised to reduce the flood risks and

vulnerabilities [2]. [3] suggested that the development of economically efficient and rational plans require

good estimates of the risk of flooding.

Structures in rivers are generally designed on the basis of peak discharge, in order to ensure safety of the

structure, during its service (design) life. Peak discharges for the design of structures are generally

decided on the basis of Extreme Value Distributions Type-I and Type-III like Gumble Extreme Value

Type-I (GEV-I) distribution, log-normal distribution and Log-Pearson Type-III (LP-III) distribution.

These distributions are well explained in many engineering hydrology text books like [4-8]. [9] found that

Gumbel Distribution is the most suitable for the Tigris River, also for other rivers such as Greater Zaab

and Lesser Zaab. [10] provides the standard procedures for flood estimation in United States, which

recommends the usage of Log-Pearson type 3 (LP3) for flood estimation. [11] compared several extreme

value distributions for flood analysis in Greater Zaab River and recommended the GEV-I and LP-III

Distribution for Greater Zaab River.

Though the methods are available to estimate the peak flood for various return periods, however, it is

often difficult to decide the peak discharge, which guarantee the safety of the structure during its service

life. The overestimated peak discharges make structures very costly, whereas the underestimated

discharges cause catastrophic failure of the structures. A risk based approach may help in this regards.

Researchers like [12-16] suggested to adopt a risk based approach in the design of hydraulic structures

and to consider as many as uncertainties involved in the process. [17] found that the simultaneous

consideration of construction cost and failure risk in design of hydraulic structures is a long-standing

problem in water management. They developed an innovative evolutionary-computation-based multi-

objective model as a remedy to shortcomings of existing common methods and tested on Bakhtiari Dam

in Iran. According to [18], flood management should be based on risk analysis, including the estimation

of probability of exceedance and then the consequences of flood occurrence.

The current study is focused on understanding the occurrence of floods with various return periods and

risks of exceedance associated with them for Greater Zaab River at Eske Kelek. Analysis is performed by

using 50 years annual peak flow data at this gauging station. Based on probability distribution of

historical flood data, the flood magnitude for various return periods ranging from 5 years to 10000 years

are estimated. Furthermore, under risk analysis, the flood magnitudes of various return periods are

suggested in order to ensure the safety of a structure under a certain degree of risk. Moreover, considering

a service life of 100 years of a structure, simulation are performed for the exceedance of floods ranging

from 5 years to 10000 years. The analysis helps in decision making about safe return period and

magnitude of design flood in order to ensure safety of the hydraulic structure during its service life.

Water-19, Paris, 22-24 July 2019 Pag. 51

2. Experimental -

The study was conducted at Greater Zaab River, Eske Kelek gauging station. Greater Zaab River is a

major tributary of Tigris River with a

mean annual flow of 420 m3/s and

catchment area of 20,500 km2 [19].

Eske Kelek is located at 36o 16’ 00” N

and 43o 39’ 00” E. Image 1 presents the

river network of Iraq, showing the

location of Eske Kelek at Greater Zaab

River. The gauging station T9,

encircled, represents the Eske Kelek.

Annual peak flow data were collected

from Hydrologic Survey of Iraq,

Ministry of Development, Iraq, for 50

years of the Greater Zaab River at Eske

Kelek Station from 1933 to 1982.

Future Flood Estimation and Risk

Analysis

Before determining the probability

distribution of flood data, the outliers

test was performed using Chauvnet’s

Method [6]. Then the probability

distribution of extreme flood data was

determined by plotting flow data versus

reduced variate. It was found that the

data follows Extreme Value Type-I

distribution as it follows a straight line, Image 2. Then the future floods for various return period were

estimated using Gumbel Extreme Value Type-I Distribution.

The probability distribution function for EV-I is given by [7]:

Where is the probability of exceedance, parameters and are scale and location parameters,

which are given by: and . A reduced variate can be defined as . Then

becomes: . Solving this equation for y, it becomes:

Where . Then for a given return period, , yT is given

as: , where . For EV-I distribution,

is related to as:

Where is the flood magnitude for a given return period, .

If, a certain risk factor is adopted during the planning and design phase of the structure, the design flood

can then be estimated based on the adopted risk factor. The exceedance or non-exceedance (success or

failure) of a flood in any year follows Bernoulli process so that the probability of occurrence of an

event in independent trials and if is the probability of an occurrence in a single trial, the probability of

Image 1. River network in Iraq (Source: [19])

Water-19, Paris, 22-24 July 2019 Pag. 52

exceedance is given by [8]:

;

Design return period for a defined risk factor can

be estimated by using the above expression,

where , gives the probability of

exceedances in a number of years (n trials) under

consideration. From this equation, the return

period can be found for adopted risk factor as

given by:

where probability, = Return period and is

the risk factor.

Simulation of Flood Exceedances

The Binomial process can be simulated using uniform variates [6]. For N experiments, with each having n

trials and an outcome A with a probability of exceedance, p, and an outcome B with a probability of non-

exceedance, 1− p. For each trial in the experiment, a uniform variate, ui is generated. If ui < p, then

outcome A has occurred; otherwise, outcome B has occurred. The number of outcomes A are determined,

x, that occurred in the n trials. The uniform variates are generated by RAN() function in the Microsoft

Excel ® Software.

3. Results and Discussion -

Peak Flood Data and probability distribution

The maximum observed peak flow from 1933 to 1982 was 12130 m3/s, which was found to be an outlier.

So this number was ignored in further analysis. The next maximum discharge was 7930 m3/s in year 1942

and the minimum was 1110 m3/s at Eske

Kelek in year 1951. Most of the time the flood

remained between 1000 to 3000 m3/s. Since,

annual peak flood data follow Extreme Value

Type-I (EV-I) distribution, so the Gumbel

Extreme Value Type-I (GEV-I) Distribution

was used for future flood estimation.

Peak Flood Estimation for Different Return

Periods

The peak flood flows were estimated for

various return periods ranging from 5 years to

10000 years, using frequency factor method.

The flood flow estimated for 10000, 1000, 500

and 100 years return periods under GEV-I

become 13200 m3/s, 10400 m3/s, 9500 m3/s

and 7500 m3/s respectively. The GEV-I distribution can be used quite confidently for future flood

estimation for Greater Zaab River at Eske Kelek.

Risk based Design Return Period

Table I presents the return periods for various risk factors, ranging from 0.7 to 0.05 for a project life of 5

years to 100 years. For a structure, having service life of 100 years, in order to be 90% or 95% sure that

the design flood will not be exceeded, the return period of 950 years or 1950 should be used, respectively.

The magnitudes of floods for these return periods at Eske Kelek can be found from Image 3.

Image 2. Extreme value type-I distribution of flood

data of Greater Zaab River at Eske Kelek

Image 3. Flood estimation for various return periods at

Eske Kelek (Greater Zaab River)

Water-19, Paris, 22-24 July 2019 Pag. 53

Table I. Return periods for various risk factors and different project service durations.

Risk of flood exceedance during service life of the structure

Design Service life 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.05

5 5 6 8 10 15 23 48 98

10 9 11 15 20 29 45 95 195

25 21 28 37 49 71 113 238 488

50 42 55 73 98 141 225 475 975

100 84 110 145 196 281 449 950 1950

Probabilities of more than one exceedances during design life of the project

Image 4 presents the probabilities of more than

one exceedance during the design life of 100 years

of a project. The probabilities are estimated for

various return periods ranging from 100 to 10000

years. It can be seen that probabilities of more

exceedance for higher return periods are extremely

low. The probabilities become less and less with

the increase of return period and number of

occurrrences.

Simulation of flood exceedance

In order to see the behavior of flood exceedances

during a project service life, the flood exceedance

simulations are performed. Ten experiments are

performed to simulate the flood exceedance for

project design life of 100 years for return periods

ranging from 5 years to 10000 years. The

simulation results are presented in Table II. The number of maximum exceedances of 100 year return

period flood were 2, whereas for 500, 1000 years return periods the maximum number of exceedances

was 1. There is zero exceedance of 10000 year flood.

Table II. Flood exceedance simulation results

Experiment Number 1 2 3 4 5 6 7 8 9 10

Return Period

(Years) Number of occurrences in 100 years

5 6 15 13 16 13 7 6 11 9 9

10 3 10 8 6 5 4 3 7 3 7

20 2 4 4 2 3 1 2 5 0 1

50 1 3 4 5 1 2 0 3 0 1

100 1 2 1 1 0 2 0 1 0 0

500 0 1 0 0 0 1 0 0 0 0

1000 0 1 0 0 0 0 0 0 0 0

10000 0 0 0 0 0 0 0 0 0 0

4. Conclusions - The current study was aimed to understand the occurrences of floods during the design

life of hydraulic structures and deciding the risk based return periods in order to ensure the safety of the

structure during its design life. The study was conducted at Greater Zaab River, Eske Kelek station. The

annual peak flood data were used in the analysis. It was found that the annual flood data follow Extreme

Value Type-I (EV-I) distribution. Using GEV-I distribution the floods for return periods ranging from 5

Image 4. Probability of more than one exceedance during

design life of a project

Water-19, Paris, 22-24 July 2019 Pag. 54

years to 10000 years were estimated as 3800 m3/s to 13200 m3/s, respectively. In order to be 90%, 95%

and 99% sure that the design flood will not exceeded during the design life (100 years) of the structure,

the return periods of 950 years, 1950 years and 9950 years were found. The probabilities of 2, 3 and 4

exceedances for 1000 year 0.0045, 0.00014 and 0.0000036 for a design life of 100 years, whereas for

10000 years flood, these are almost zero. The simulation for the maximum exceedances of different return

period floods during the serve life of 100 years of a hydraulic structure showed that the 5, 50, 100, 500,

1000 and 10000 years are 16, 5, 2, 1, 1 and 0. Based on this analysis if return period ranging from 1000 to

10000 years is decided for a structure having service life of 100 years, the structure has much less risk of

failure.

5. References

[1] U.S. Geological Survey (USGS). (2006). Flood hazards—A national threat, fact sheet 2006–3026,

U.S. Department of the Interior, Washington, D.C.

[2] Interagency Floodplain Management Review Committee (IFMRC). (1994). Sharing the Challenge:

Floodplain Management into the 21st Century (Galloway Report), U.S. Government Printing Office,

Washington, D.C.

[3] Stedinger, J. R., Griffis, W. J. (2008). Flood Frequency Analysis in the United States: Time to Update.

J. Hydrol. Eng., 2008, 13(4): 199-204

[4] Naghettini, M. (Ed.). 2017. Fundamentals of Statistical Hydrology. Springer. ISBN 978-3-319-43560-

2 ISBN 978-3-319-43561-9 (eBook). DOI 10.1007/978-3-319-43561-9

[5] Deodhar, M. J. 2009. Elementary Engineering Hydrology. Dorling Kindersley Pvt. Ltd. India. Pearson

Education. ISBN 978 81 317 0805 7.

[6] McCuen, R. H. 2003. Modelling Hydrologic Change - Statistical Methods. Lewis Publishers. CRC

Company. ISBN 1-56670-600-9.

[7] Chow, V. T., Maidment, D. R., Mays, L. W. 1988. Applied Hydrology. McGraw-Hill Inc.

[8] Haan, C. T. 1977. Statistical Methods in Hydrology. The Iowa State University Press.

[9] Mujda, M.F. (1978) Flood Frequency Analysis of Tigris River at Mosul. M.Sc. thesis, Department of

Civil Engineering, Mosul University, Iraq.

[10] Interagency Committee on Water Data (IACWD) (1982). Guidelines fordetermining flood flow

frequency: Bulletin 17B (revised and corrected), Hydrol. Subcomm., Washington, D.C., 28.

[11] Abidalla (2013). Flood Frequency Analysis for Greater – ZAB. River. Journal of Babylon

University/Engineering Sciences/ No. (4) / Vol. (21): 2013

[12] Tung, Y-K. (1987) Effects of Uncertainities on optimal risk-based design of hydraulic structures.

Journal of Water Resources Planning and Management. Vol. 113, Issue 5.

https://doi.org/10.1061/(ASCE)0733-9496(1987)113:5(709)

[13] Yen, B. C. and Tung Y-K (Editors) (1993). Reliability and Uncertainty in Hydraulic Design.

American Society of Civil Engineers. ISBN 0-87262-993-7.

[14] Mays L.W. (1987) Review of Advances in Risk and Reliability Analysis for Hydraulic Structures.

In: Singh V.P. (eds) Application of Frequency and Risk in Water Resources. Springer, Dordrecht. DOI

https://doi.org/10.1007/978-94-009-3955-4_19. Springer, Dordrecht. ISBN 978-94-010-8254-9.

Water-19, Paris, 22-24 July 2019 Pag. 55

[15] Schoustra, F., Mocket, I., Simm, J. (2004). A new risk-based design approach for hydraulic

Engineering. Journal of Risk Research 7 (6), 581–597 (September 2004). Journal of Risk Research

ISSN 1366-9877 print/ISSN 1466-4461 online © 2004 Taylor & Francis Ltd

http://www.tandf.co.uk/journals. DOI: 10.1080/1366987042000192183.

[16] Winter, B., Schneeberger, K., Huttenlau, M., Stötter, J. (2018). Sources of uncertainty in a

probabilistic flood risk model. Nat Hazards (2018) 91:431–446. https://doi.org/10.1007/s11069-017-

3135-5

[17] Rasekh, A., Afshar, A. & Afshar, M.H. (2010). Risk-Cost Optimization of Hydraulic Structures:

Methodology and Case Study. Water Resour Manage (2010) 24: 2833. https://doi.org/10.1007/s11269-

010-9582-3. DOI https://doi.org/10.1007/s11269-010-9582-3. Springer Netherlands. Print ISSN 0920-

4741, Online ISSN 1573-1650

[18] Beven KJ, Hall J (eds) (2014) Applied uncertainty analysis for flood risk management. Imperial

College Press, London.

[19] Saleh, D.K., 2010, Stream gage descriptions and streamflow statistics for sites in the Tigris River and

Euphrates River Basins, Iraq: U.S. Geological Survey Data Series 540, 146 p.

Water-19, Paris, 22-24 July 2019 Pag. 56