T14 02 Jeftenic - Kolakovic - Stipic - Kolakovic - Budinski - Seslija - GNP2020

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

THE 7th INTERNATIONAL CONFERENCE

"CIVIL ENGINEERING - SCIENCE AND PRACTICE"


GNP 2020 – Kolašin, Montenegro, 10-14 March 2020

Goran Jeftenić1, Slobodan Kolaković2, Danilo Stipić3, Srđan Kolaković4, Ljubomir Budinski5,
Miloš Šešlija6

USING HYDRAULIC MODELING IN AN OPEN CHANNEL FLOW AND


VARIOUS RESTRICATIONS BASE ON IRON GATE CANYON, DANUBE
SECTION KM 1116.2 – 943.0

Abstract
In this paper hydraulic conditions of the Danube River on the section from Smederevo (km
1116 + 000) to the Iron Gate 1 Dam (km 943 + 000) were defined. Dependence between the
levels of the flow at the estuary of the Nera River into the Danube established in 1974, when
the dam was built, has been proven valid. In order to put this analysis into practice, 1D model
of unsteady flow at the aforementioned section has been made using the software package
HEC-RAS. For the upstream boundary condition it is necessary to adopt time period during
which there were no abrupt changes in the flow in order to achieve as precise hydraulic
conditions as possible. Instead of Q(h)-curve which is defined by the opening of spillways, dam
is physically removed from the model and hydrograph h(t) is inserted as a boundary condition
achieved at the dam during observed period. At the end, comparison of that computational
water level with estimations is shown that it was defined during design and erection of Iron
Gate 1 Dam from 1974.
Key words
Hydraulic analysis, 1D unsteady flow, Iron Gate, hydraulic modeling of dam

1
Teaching Assistant MSc, University of Novi Sad, Faculty of Technical Sciences, goran.jeftenic@uns.ac.rs
2
PhD, Assistant Professor, University of Novi Sad, Faculty of Technical Sciences, kolakovic.s@uns.ac.rs
3
Teaching Assistant MSc, University of Novi Sad, Faculty of Technical Sciences,danilostipic@uns.ac.rs
4
PhD, Professor, University of Novi Sad, Faculty of Technical Sciences, kolak@uns.ac.rs
5
PhD, Professor, University of Novi Sad, Faculty of Civil Engineering, ljubab@uns.ac.rs
6
PhD, Assistant Professor, University of Novi Sad, Faculty of Technical Sciences, sele@uns.ac.rs

893
Civil En g in eerin g – S cien ce a n d Pr a ctice

1. INTRODUCTION

Study on the flow in an open channel and through the hydraulic structures usually is
conducted using physical modeling. Physical modeling is based on the basic fluid mechanic
equations. Physical modeling of hydraulic structures means that a scaled laboratory model from the
prototype is constructed. This approach is a safe way to analyze the flow through or over the
hydraulic structures. Due to high cost of laboratory experiments, researchers have attempted to use
numerical simulation along the physical modeling [1, 2]. Numerical modeling of the river basin
and hydraulic structure has become unavoidable works in the hydraulic engineering projects [3].
Although very complex flow regime can be approximated for the purpose of engineering tasks by
introducing certain assumptions and simplifications. In order to make hydraulic analysis of the
river flow successful, it is necessary to be familiar with processes that take place in the river
system. Knowledge of the flow characteristics and the river bed morphology are basic. Globally,
rivers and their basins have been progressively impacted by human activity, as part of the wider
modification of the global hydrological cycle through the Anthropocene [4]. In the Danube River
Basin multiple pressures affect the river system as a consequence of river engineering works,
altering both the river hydrodynamics and morphodinamics [5].
The Danube River is the 21st largest river globally, and the second largest river in Europe
[6]. The Danube is the most international river basin globally, spanning 19 countries, and arguably
it is one of the most complex basins in which the European Water Framework Directive
2000/60/EC and other EU directives have to be implemented. Along its course, the Danube flows
through a series of alternating basins and gorges: below the confluence of the Danube and the
Morava, the river enters the Devin Gate, below which the Danube forms an internal delta as it
starts to traverse the Pannonian Plain. Here the flow of the ‘Middle’ Danube is augmented by the
Drava, the Tisza, and the Sava rivers: tributaries that rise in the Southern Alps, the Western
Carpathians, and the Julian Alps respectively, highlighting the degree to which the Danube is
dependent upon flow generated in alpine areas. The Middle Danube (mean slope of about 0.1‰) is
~929 km long and surrounded by the Hungarian plain and ends in the incised Iron Gate Gorge
which has a higher slope [5].
The construction of the Iron Gate 1 dam was completed in 1972, and it created one of the
largest reservoirs in Europe, with the backwater effect extending 310 km further upstream, and
with the total drainage area of 577,000 km2. In 1985, a second dam was built 80 km downstream
from the first one, for additional power production and more flexibility of the joint operation of the
two power plants. The Iron Gate System induced considerable modifications of natural river
regime, and raised a number of questions concerning water management decisions, such as:
reduced sediment transport capacity, followed by sediment deposition; raise of ground water table,
endangering many settlements, industrial, municipal and transportation facilities, as well as the
agricultural production in the riparian belt; inadequacy of the existing flood protection structures;
decrease of ice transport capacity at the end of backwater zone [7]. This study targets new flow
regime because of the presence of a dam in the system. It means changes in water level line in the
longitudinal profile, various morphological changes and decreasing flow velocity which implies
large amounts of sediment settlement.

894
GNP 2 0 2 0

2. STUDY AREA

The section of the Danube River, upstream from the estuary of the Nera River, going
through our country is distinguished by its lowland character. Downstream from the estuary of the
Nera River, the Danube penetrates through the branches of the Carpathian Mountains, making the
Iron Gate, the largest and the longest canyon in Europe. The canyon stretches out from the town of
Golubac (at 1048 km of length of the Danube) to the town of Kladovo (941 km length). This part is
characterized by large and concentrated downs with very steep canyon-like slopes, with plenty of
rocky sills, rapids, meandering riverbeds and narrow passages. The route of the mentioned section
begins at the gauge station near Smederevo (1116.2 km) and passes through Banatska Palanka,
Veliko Gradiste and Donji Milanovac, and ends with the Hydro Power Plant (HPP) Iron Gate 1
Dam (943 km). The route is mostly located in the Iron Gate canyon, and as a result of dam
erection, backwater section has occurred upstream.
Since the sections are located in the canyon, and in the rock mass, water spillage into
floodplain is not expected, and plane flow model could not have been taken into consideration. At
the 1091 kilometer from the estuary, the Morava flows into the Danube on the observed section.
Apart from the Morava, on this section of the Danube, there are also the Mlava, the Nera and
Danube-Tisa-Danube channel (section Banatska Palanka-Novi Becej) that flow into the river,
whose discharges are negligible compared to the Danube flow, therefore not considered in
hydraulic analysis.

3. DEFINING BACKWATER REGIME AT THE HPP IRON GATE 1


DAM

Because of the presence of a dam in a system, water surface does not behave naturally (with
increase of the discharge, water level increases), in the part of the section affected by the dam it
reacts dictated (level decreases with the increase of discharge) as a result of the opening of
floodgates (Figure 1).

Figure 1. Water surface levels for different discharges at water gauges

Before erection of the dam, the Danube had a natural flow regime along its entire river flow
through Serbia (water levels at cross-sections increase with the increase of discharge), this function
is defined by hydraulic-morphological conditions, for the point of the Nera flowing into the

895
Civil En g in eerin g – S cien ce a n d Pr a ctice

Danube, curve 1 (Figure 2.). The function is based on perennial measurements of dependence
between water level (h) and discharge (Q) in this section (km 1077). After the erection of the dam,
flow regime of the river is completely changed, with interstate agreement with Romania which
stipulates that depending on the discharge, river level at the dam should be regulated (by
selectively opening the spillways), and in case of a flood wave with discharge over 12 000 m3, all
spillways should be opened thus achieving a natural regime at the estuary of the Nera.

Figure 2. Computational curves of discharge for backwater regime 69.50/63

As aforementioned, the result of such prescribed regime is that water level of the profile at
estuary of the Nera slowly increases (compared to natural regime) with the increase of discharge
because of the water level decrease at the Iron Gate 1 Dam. This is the first part of function 2, the
second part of the function 2 is parallel with natural regime (function 1) with a slight increase due
to formed dam backwater while the spillways are completely opened. Function 3 is a result of these
water level conditions prescribed at the estuary of the river (interstate agreement), and it represents
a dictated level that is try to carry out at the Iron Gate 1 Dam. According to the dam backwater
regime "69.50/63" water level at the dam is held in line with the regression curve from point 69.50
(for the low Danube discharges) up to point 63.00 (for discharges greater than 11,000 m3/s), where
the water level at the estuary of the Nera is higher than 69.50 and it is defined by the curve 2. Since
Figure 2 is practically borderline for the upstream part of observed section through the Iron Gate
canyon, and it is also downstream borderline for the upstream part of the Danube, curve 3 will be
considered with higher precision. Curve 3 represents the dependence of prescribed water level on
discharge at the dam. With discharges up to 11 000 m³/s, curve 3 is descending (with discharge
increase water level decreases, due to successively opening of the spillways at the dam in case of
discharge increase), after this flow level at the dam remains the same. This function directly affects
curve 2 which compared to the natural regime (curve 1) has significantly higher water levels for the
same discharge as a result of the formed backflow, used for achieving level separation between
"upper" and "lower" water necessary for production of electricity. For discharges greater than 11
000 m³/s, Q (h) curve at the estuary of the Nera after the dam erection (curve 2) becomes parallel
to the curve at the estuary of the Nera in natural conditions (curve 1), and the difference between
curves 1 and 2 is a result of formed backwater in the flow of the Danube due to dam erection when
all spillways are opened.

896
GNP 2 0 2 0

4. HYDRAULIC MODEL DESIGN

In practice and according to recommendations of Kovacs [8], for hydraulic calculations of


uniform and non-uniform flows in a network of open flows, with an irregular geometry, the tool
which is often used is HEC-RAS, the same tool was used for the purposes of this study
accordingly. The advantage of this software tool is primarily in the fact that it practically became
globally standardized, thus its use achieves compatibility with other users in the Danube basin.
Another important advantage lies in compatibility with the other, now standardized software tools
which provide us with input data for the hydraulic model. It should be noted that the authors of this
paper have not specifically explained the equations that are used in calculations of uniform and
non-uniform flow in the software package HEC-RAS as they are generally known and verified
throughout large number of examples in the world. The general principles of modeling unsteady
flow in systems of open channels in the HEC-RAS environment are given in [9-12].
In terms of hydraulics, model is limited at one side with the hydrological station of
Smederevo, as the upstream borderline, taken from the Republic Hydrometeorological Service
bulletin, (Q(t) hydrograph near Smederevo). At the other side, model is confined with downstream
borderline at the dam. There is the third borderline, inner condition, which is not critical for the
model, but provides a more realistic picture of flow in this section, and that is the inflow at the
confluence of the Morava and the Danube downstream from Smederevo. Discharge of the Morava
is not negligible and neither are discharges of the Nera, the Mlava and the DTD channel, hence
hydrograph of the Morava, Q (t), is taken into account as a lateral point flow at the location of the
Morava actually flowing into the Danube.

4.1. DEFINING THE UPSTREAM BOUNDARY CONDITIONS


For the upstream boundary condition it is necessarily to adopt time period during which
there were no abrupt changes in the flow in order to achieve as precise hydraulic conditions as
possible. For that reason, hydrograph near Smederevo was adopted for a period of 6 days. Given
the fact that the Danube has a large range of discharge values, for the purpose of calibration of the
model, six time series with different discharge values were adopted (in ascending order). For this
purpose, the year 2006 served as a basis since it had the required range.

4.2. DEFINING THE MORAVAS INFLOW


This borderline describes inflow of the Morava into observed section. Although, in reality
twenty kilometers away from the confluence with the Danube, Ljubicevski Most gauging station is
the closest station to the confluence of the Morava and the Danube rivers which can provide data
about discharge of the Morava. Adhering to the presumption of continuity, it is logical to conclude
that the flow through the gauging station at Ljubicevski Most must arrive to the confluence, hence
the input data for establishment of the borderline, the curve Q(t) at mentioned gauging station, is
for the same dates as data for upstream borderline.

897
Civil En g in eerin g – S cien ce a n d Pr a ctice

4.3. DEFINING DOWNSTREAM BOUNDARY CONDITIONS


The downstream boundary condition (condition at the dam) is very complex. Experience
and personal judgment are also important for operation of the spillway gates during an incoming
flood. For example [13], state: “The discharge from a storage reservoir is regulated by gates and
valves operated on the basis of the judgment of the project engineer”. Similar comment: “... for the
extremely big flood, a reservoir operator has to control the gates to protect the reservoir and the
downstream reference point by relying on his judgment.” [14]. Because there are many options for
partial gate openings and their timings during a flood wave, identification of the most suitable
operation of the spillway gates is a complicated problem [15].
Given data include computational Q(h)-curve (Figure 2.-Function 3.) which is defined by
the opening of spillways. This is the reverse Q(h)-curve (descending with increase of discharge),
gives us lower water levels, and as such "reverse" curve it cannot be recognized by hydraulic
software tool HEC-RAS. Once this curve is set as a boundary condition, the software tool identifies
it as an error. The first idea for to overcome this problem was to model the dam with all its
elements and to simulate it. This was done. HEC-RAS offers two options for setting the boundary
conditions, when it comes to dams:
- ‘Elevation controlled gates’
In case of this boundary condition, depending on water level achieved on the upstream
contour of the dam, the program opens or closes the spillways and as a consequence, certain level
of water surface along the entire route is established. The problem arises due to the fact that
spillways at Iron Gate 1 Dam are very complex, they have more positions than just fully
opened/closed as it is offered by the program, but can take any position, even the position where
water simultaneously runs below the outlet and beyond the flood level, and their manipulation is
too complex for software tool HEC-RAS to simulate. In accordance with aforementioned, the idea
was to simplify dam action. Simplification was made by indirectly transferring Function 3 from
Figure 2 to the downstream boundary condition through openings at the dam. Since there is a
dependency which behaves like an exponential function up to 11,000 m³/s discharges (water level
changes depending on discharge), for discharges greater than 11000 m³/s, the water level remains
the same for each discharge value, 11 spillways have been modeled, with dimensions and spillover
coefficients adopted so that for discharges of 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000,
9000, 10000 and 11000 m³/s (one fully opened spillway for each discharge, Figure 3.), provide
exactly the same water levels defined by Function 3 - Figure 2. The idea is to simulate complex
geometry of the spillways and their complex action with simpler model which will behave in the
same way (in a hydraulic sense) as the actual spillways. The results of this simulation are presented
in Figure 4.

898
GNP 2 0 2 0

Figure 3. 11 simplified spillways

Figure 4. Discharge above spillways

- green - the number of fields openned is equal to the discharge;


- red - the number of fields openned is higher by one field relative to the discharge;
- purple - the number of fields openned is smaller by one field relative to the discharge.
Water levels marked in green were registered directly upstream of the dam when the
number of fields opened is equal to the discharge (for example, 5 fields opened and discharge of
5000 m³/s). Water levels in red are obtained when the number of fields opened is higher by one
field relative to discharge (for example, 5 fields opened and discharge of 4000 m³/s). While levels

899
Civil En g in eerin g – S cien ce a n d Pr a ctice

marked in purple color are obtained when the number of fields opened is smaller by one (for
example, 5 open fields and discharge of 6000 m³/s). With the increase of discharge of 1000 m³/s
another spillway successively opens, so, for example, discharge of 5,000 m³/s correspond to 5 open
spillways; discharge of 9,000 m³/s corresponds to 9 open spillways; etc. The results of this analysis
were satisfactory. However, it gave valid results only for cases with discharges 1000, 2000, 3000,
..., 11000 m³/s). In case of a number between these discharges (for example, 2500, 3300, 7857...
m³/s), it could not be stated with certainty that the results of the model reflect actual state of the
water surface level. The reason is that water level could take any position between the red (for
example, 5 spillways opened-discharge of 4000 m³/s) and purple (for example, 5 spillways opened-
discharge of 6000 m³/s) peak value in Figure 4. This will be explained through an example: at one
point discharge of 7000 m³/s was reported, and 7 spillways were opened (which correspond to
green peak value). It should be noted that this is not the real situation when it comes to opening of
spillways at the dam, but they are modeled this way in order to simulate the authentic state of the
level for particular discharge. Going forward, the discharge increases and it reaches a value of
7500 m³/s, when a problem arises. It is unknown whether a new spillway will be opened so the
level would decrease (level would take position between green and red peak value), or the number
of spillways will remain 7 and the level would increase (level would take position between green
and purple peak value). What is certain, when discharge reaches 8000 m³/s, new spillway will be
opened and the level from curve 3 - Figure 2. will be achieved (level is at green peak value), but
everything that happens between 7000m³/s and 8,000 m³/s stays uncertain. In the end, the idea is
good, but not useful because it works correctly only for some isolated cases and does not reflect the
actual situation for non-uniform flows and we passed to the idea with hydrograph h(t) as a
boundary condition.
- ‘Time series gate openings’
Finally, dam has been physically removed from the model and hydrograph h(t) has been
inserted as a boundary condition achieved at the dam during observed period. These boundary
conditions replace physical existence of the dam by describing hydraulic characteristics that the
dam generates above the level at the dam during observed period. In order to set up this kind of
downstream boundary condition, data about the levels of the dam in the year 2005, 2006, 2008,
and 2014 were required.
For this kind of downstream boundary, model is working for steady, as well as unsteady
flow. This model was calibrated for unsteady flow, where any inaccuracies resulting from geometry
approximation, geodetic surveys, neglecting the discharges of the Mlava, the Nera and the DTD
channel, etc. were compensated in the process of calibration.

5. CALIBRATION AND VERIFICATION

The model is calibrated for non-uniform flow, for six different discharges occurred during
2006. Periods of 6 days in which discharges did not differ significantly were requested in
hydrological bulletin. For comparison of the obtained results and real situation, water levels during
these periods on each of available gauging stations on the given section were used. Gauging
stations located along the route are: Smederevo, Banatska Palanka, Veliko Gradiste and Donji
Milanovac. The first iteration was assigning Manning’s coefficient of 0.025 m1/3/s to entire
section [16], but results indicated that the model would not work in this way (deviation of model
results from recorded values at the gauging stations were up to one meter at some points, even for

900
GNP 2 0 2 0

two adjacent stations levels in the model were higher on one and lower on another station). The
higher the discharge the more detectable deviation from control values is. Because of this
particular problem, the section was divided into 4 stages. It is important to be noted that the exact
geometry of the riverbed is unknown (profiles are quite remote and the riverbed configuration
changes drastically between them). Consequently, the fact that Manning’s coefficient is the only
variable and that it includes all mistakes that were made, due to geometry estimations, geodetic
observations, had to be taken into account.
Verification of the particular problem was done as a simulation of non-uniform flow in the
year 2005, 2008 and 2014, from geometry and Manning’s coefficients which were adopted for
2006 in the process of calibration. Considering that in 2008 there had not been floods, nor the
discharge exceeded 8000 m³/s, the verification confirmed Manning’s coefficients which can be
applied only to discharges up to 8000m³/s. Discharge rates greater than 8000 m³/s were verified for
2005. The interval of time between verification and calibration should not be longer than a couple
of years. It ‘moves’ and eventually it leads to various morphological changes in the riverbed.
Verification is carried out also for the year 2014, but the purpose of this operation was not
confirmation of calibration results (as was the case with data from 2005 and 2008), but an attempt
to show that the river over time (in this case period of 8 years) changes geometrical characteristics
of the riverbed which is reflected in the increasing deviation of the calculated levels from the ones
recorded at gauging points. In this case it is a period of 8 years, where verification of the results for
2014 points out aforementioned assumption of morphological changes. These results show great
deviations and the fact that Manning’s coefficients determined in 2006 cannot be applied to
hydrological data from 2014 so that deviation remains acceptable.

6. RESULTS OF MODELING

Results of the model set this way show water level at the estuary of the Nera into the
Danube River. Their comparison with estimations that had been defined during design and erection
of the Iron Gate 1 Dam (1974), which are taken from the HPP Iron Gate, is shown in Figure 5.

901
Civil En g in eerin g – S cien ce a n d Pr a ctice

Figure 5. Level comparison at estuary of the Nera

Figure 5 clearly shows that the level that occurs at the mouth of Nera, due to existence of
the dam, is higher in 2006 than it was in 1974. The accuracy confirmation of the ‘red’ curve is
provided by calibration and verification of the model. In addition, this graph shows that the two
curves have the same tendencies which further confirms accuracy of the ‘red curve’. The difference
in water levels between ‘red’ and ‘blue’ curve is approximately ten centimeters for smaller
discharges, whereas this difference is rapidly increasing for discharges greater than 10,000 m³/s,
and for discharges of 12000 m³/s, the difference is about seventy centimeters. This difference is
explained by the time passed between the two measurements. Namely, in those thirty years the
river has rather changed in terms of morphology, the accumulation volume is reduced due to
sediment subsidence made over the years due to backwater. On the other hand, at a higher flow
rate, the model is much more sensitive (small irregularities for higher discharges lead to major
errors), therefore it is possible that deviation of seventy centimeters at a higher discharge rate does
not reflect the real state.

7. CONCLUSION

First a foremost, the whole model depends on the availability of geodetic database since it
represents geometry of the whole model. It is recommended that the whole section should be
recorded using some ‘more precise’ geodetic method throughout the whole year, because the river
constantly changes shape of the riverbed in some sections. In addition, cross-sections should be
recorded in every 150-200 m, otherwise hydraulic model will incorporate numerous imprecise
geometric approximations, which mostly lead to deviation of the hydraulic model from its natural

902
GNP 2 0 2 0

form. Manning’s coefficients were ultimately adopted by dividing section into four stages, and
iteratively assigning particular a value of this coefficient to each stage, for each discharge, until
achieving reasonable deviation of recorded and calculated values. It is assumed that further
fragmentation of the problem (for example, dividing sections into multiple sub-sections) would
result in more accurate results. This would lead to the extensive and complicated process and, on
the other hand, obtained results are sufficiently useful from a practical point of view.
The analysis performed in this study using software tool HEC-RAS show that it is designed
in a way that any change in the simulation occurs slowly enough so that scheme on which this
software tool is based on remains stable. This implies that the geometric changes along the flow are
mild, and that changes in non-uniform flows take place slowly enough. Otherwise, generating of
errors in the flow may happen, which is reflected in the abrupt changes in water level line in the
longitudinal profile. In cases where the geometry of the riverbed is such that there are large
alterations, recordings of the riverbed should be more detailed and computational time step should
be reduced, i.e. dense the computational mesh.
Verification of the model is completed based on data for the years 2005 and 2008. These
include all information about discharges for which the calibration was carried out. Even in the
verification conducted for 2014, applied Manning’s coefficients were defined on the basis of
discharge and water level from 2006, differences in the level were increasing. This has occurred as
a result of various morphological changes constantly occuring on the river. This primarily refers to
the fact that large amounts of sediment settlement due to created backflow, respectively decreased
flow velocity, which implies reduced hydraulic profile.

ACKNOWLEDGEMENTS

This work is supported by Ministry of Education and Science, Republic of Serbia (Grant
No. TR 37003).

LITERATURE

[1] Ettema, R., Arndt, R., Roberts, P., Wahl, T.: Hydraulic Modeling Concepts and Practice, ASCE
manuals and reports on engineering practice, New York, 2000.
[2] Hager, W.,H., Pfister, M.: Hydraulic modeling – an introduction: principles, methods and
applications, JOURNAL OF HYDRAULIC RESEARCH 48 (2010) 4, pp 557–558.
[3] Dehdar-Behbahani, S., Parsaie, A.: Numerical modeling of flow pattern in dam spillway’s guide wall.
Case study: Balaroud dam, Iran, ALEXANDRIA ENGINEERING JOURNAL 55 (2016) 1, pp 467-
473.
[4] Vörösmarty, CJ., Pahl-Wostl, C., Bunn, S.,E., Lawford, R.: Global water, the anthropocene and the
transformation of science, CURRENT OPINION IN ENVIRONMENTAL SUSTAINABILITY 5 (2013)
6, pp 539–550.
[5] Habersack, H., Heinb, T., Stanica, A., Liska, I., Mair, R., Jäger, E., Hauer, C., Bradley, C.:
Challenges of river basin management: Current status of, and prospects for, the River Danube from a
river engineering perspective, SCIENCE OF THE TOTAL ENVIRONMENT 543 (2016), pp 828-
845.
[6] The Danube River Basin District Management Plan, International Commission for the Protection of
the Danube River, 2015.

903
Civil En g in eerin g – S cien ce a n d Pr a ctice

[7] Babic, Mladenovic, M., Kolarov, V., Damjanovic, V.: Sediment regime of the Danube River in
Serbia, INTERNATIONAL JOURNAL OF SEDIMENT RESEARCH 28 (2013), pp 470-485
[8] Kovacs, S.: Modeling the River Tisza by using one-dimension hydro-dynamic model, 3rd
International Symposium on Flood Defense, Nijmengen, The Netherlands, 2015.
[9] Barkau R. L.: UNET One - Dimensional Unsteady Flow through a Full Network of Open Channels,
User’s Manual, US Army Corps of Engineers, Hydrologic Engineering Center, Davis, California,
USA, 1997.
[10] Knight, D., W., Shiono, K., Pirt, J.: Prediction of depth mean velocity and discharge in natural rivers
with overbank flow, International Conference on Hydraulic and Environmental Modeling of Coastal,
Estuarine and River Waters, Gower Technical, University of Bradford, pp. 419-428, 1989.
[11] Fletcher, C., A., J.: Computational Techniques for Fluid Dynamics, vol. I, Springer, Berlin, Germany,
1991.
[12] Chadwick, A., Morfett, J.: Hydraulics in Civil and Environmental Engineering, Spon, London, UK,
1999.
[13] Linsley, R.,K. et al.: Water resources engineering, New York, Mc-Graw-Hill, 1992.
[14] Sakakima, S., Kojiri, T., Itoh, K.: Real-time reservoir operation with neural nets concept,
International conference on applications of artificial intelligence in engineering—AIENG/92.
Southampton, UK: Computational Mechanics Publications, pp 501–514, 1992.
[15] Kolaković, Sl., Fabian, Đ., Kovacs, S., Budinski, LJ., Stipić, M.: Exploitation of documented
historical floods for achieving better flood defense, Hindawi Publishing Corporation, ADVANCES
IN METEOROLOGY (2016), Article ID 2317252, 9 pages.
[16] Chow, V.,T.: Open Channel Hydraulics, Mc Graw-Hill, New York, USA, 1959.

904

You might also like