Scale Modeling of Riga HPP Stilling Basin

 Danica Starinac1, Predrag Vojt1, Dimitrije Mladenović1, Dragiša Žugić1, Radomir Kapor2, Ljubodrag Savić2

 

1 Jaroslav Černi Institute for the Development of Water Resources, Jaroslava Černog 80, 11226 Pinosava-Belgrade, Serbia; E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

2 University of Belgrade - Faculty of Civil Engineering, Bulevar Kralja Aleksandra 73, 11000 Belgrade, Serbia

 

Abstract

Riga HPP was built in 1974, as the third and most downstream hydropower plant in a system of cascade dams on the Daugava River (Latvia). According to the design criteria at that time, all structures were designed to safely pass a discharge of Q0.1%=7800 m³/s. Considering the fact that the capital city is located some 15 km downstream from the dam, it is considered a high-hazard dam. Current engineering trends recommend the application of a more stringent design criterion – all structures should be able to safely pass a discharge of QPMF=12800 m³/s. Scale model tests were aimed at assessing the possibility of passing the higher flood discharge through existing structures, by determining all the important hydraulic parameters in the study area. The project included a combination of numerical modeling of the downstream rating curve and physical modeling, and was carried out using a model on a scale of 1:60. The project provided not only important results, which should be very helpful in future maintenance, but also recommendations for improving the existing situation.

Keywords: scale model, spillway, Riga.

Introduction

In 2016, the Hydraulics Laboratory of the Jaroslav Černi Institute successfully completed an important project – Scale Modeling of the Riga HPP Stilling Basin (Starinac, 2016). The opportunity to lead the project came from winning an international tender of Latvenergo AS, the biggest electricity trader in the Baltics. The project was carried out with the assistance of EC Profecto (Latvia) and consulting services were provided by the University of Belgrade – Faculty of Civil Engineering.

 

Riga Dam Overview

Riga HPP is one of three hydropower plants on the Daugava River, Latvia, and its dam is the most downstream in a system of cascade dams. Riga Dam is located near Salaspils; the distance downstream to Old Riga is approximately 15 km and to the Baltic Sea some 30 km (Figure 1).

 

fig01
Figure 1: Location of Riga HPP and reservoir.

 

The dam comprises a powerhouse, spillway and long embankment dam. The main road, A5, runs along the crest of the left embankment dam.

The spillway of the Riga Dam consists of six openings with radial gates. The width of each opening is 20 m and the spillway crest elevation is 7.5 m BAS (Baltic 1977 Normal Height System BAS-77; hereinafter BAS). The dam is 160.15 m long and is split by an expansion and contraction joint into two blocks. The maximum height of the dam is 30 m.

The installed capacity of the HPP is 402 MW, generated by six turbine units.

The HPP layout is shown in Figure 2 and typical cross-sections of the spillway and HPP in Figures 3 and 4.

The Riga HPP energy dissipation structures include a horizontal stilling basin with two rows of buffer blocks, arranged in a chessboard pattern, and a solid wall (Figure 5).

 

fig02
Figure 2: Riga HPP layout.

 

fig03
Figure 3: Cross-section of the spillway.

 

fig04
Figure 4: Cross-section of the HPP.

 

fig05
Figure 5: Stilling basin layout.

 

Purpose of the Model Study

According to the design criteria applicable at the time the dam was built, all the structures have been designed to safely pass a discharge of Q0.1%=7800 m³/s. The stilling basin design provides for drowning of the hydraulic jump, verified by testing on a 1:80 scale model before the dam was constructed. Based on the tests, the flow conditions in the stilling basin were acceptable for Q0.1%=7800 m³/s and a reservoir water level at 18.00 m BAS (FRL), as well as for Q0.01%=9450 m³/s and a reservoir water level at 18.80 m BAS (MRL).

Current engineering trends call for a more stringent design criterion for high-hazard dams: all structures should be able to safely pass a discharge of QPMF=12800 m³/s. Therefore, the purpose of the model was to study the possibility of passing the Probable Maximum Flood (PMF) of QPMF=12800 m3/s over the existing spillway, by elevating the reservoir water level to 20.9 m BAS.

 

Methodology

Scale Model

The physical model of Riga HPP (length scale 1:60, Figure 6), was built in February 2016 at the Hydraulics Laboratory of the Jaroslav Černi Institute for the Development of Water Resources, Belgrade, Serbia.

The model reproduced a 380 m long section of the reservoir, the spillway spans with the stilling basin and transition area, the powerhouse inlet and outlet structures with the stilling area, retaining walls, and a 610 m long downstream river channel (Figure 7).

The spillway and powerhouse were made from fine concrete, whereas the walls and stilling basin were fabricated using transparent Plexiglas plates mounted on a special metal structure (Figures 8-9). The gates were made from metal. The turbines were represented on the model by special metal valves, which guaranteed that the discharge/upstream water level/downstream water level were similar to those of the prototype.

The reservoir section upstream from the dam and the river section downstream from the dam were made of sand, formed according to the topographic maps provided by the Employer. The surface of the sand was covered with a cement layer to improve the stability of the modeled terrain.

The present state of deposition and scouring in the transition area was represented on the model with a fixed riverbed. Upon completion of the first testing phase, a part of the fixed riverbed (250 m downstream from the stilling basin) was replaced by a movable riverbed, to analyze deposition and scouring.

All the components of the model reflected the design dimensions and the roughness of the materials met all similarity requirements to prototype conditions.

 

fig06
Figure 6: Scale model of the Riga HPP at Jaroslav Černi Institute for the Development of Water Resources, Hydraulics Laboratory.

 

fig07
Figure 7: Area represented on the model (red dashed line).

 

fig08
Figure 8: Spillway and stilling basin, view from the downstream end.

 

fig09
Figure 9: Spillway and powerhouse - view from the downstream end.

 

Measurements on the Model

Different types of measurements were undertaken as part of the model testing. The methods applied for the various parameters are described below.

 

Water Level and Depth

Staff-level gauges (Kern, with nonius scale) were used to measure water levels in the reservoir, in the stilling basin and in the downstream section. The measurement accuracy was ±0.1 mm.

The water levels in the reservoir and at the downstream boundary cross-section were also measured continuously, by a magnetostrictive water level sensor (Nivelco Nivotrack and Balluff Micropulse), with an accuracy of ±0.25 mm. These measurements were performed in a stilling well, to reduce water level fluctuations. Continuous measurement was supported by a suitable data acquisition unit (HBM QuantumX).

 

Velocity

Water velocities along the section downstream from the stilling basin were measured by means of a laboratory-type current meter (OTT C2 and Seba M1, accuracy ±1%).


Flow rate

Flow rates were measured by means of weir boxes of the Bazin and Thomson type, where the head was measured by staff gauges (Kern, accuracy ±0.1 mm), resulting in a flow measurement accuracy of ±0.5%. Steady flow was provided by means of the Laboratory's water supply system. The flow rate was also measured continuously, by an ultrasonic flow meter (Krohne UFM), with an accuracy of ±1.5%. Continuous recording was supported by a suitable data acquisition unit (HBM QuantumX).

Discharge through each turbine was defined as the product of the cross-sectional area and the velocity measured by a current meter (Jaroslav Černi Institute, accuracy ±1%). Such measurement was made possible by the pressure flow conditions achieved at the turbines.

 

Pressure and Hydrodynamic Load

Hydrodynamic pressures were recorded at several points on the bottom of the stilling basin. Pressure sensors (GE Druck Unik 5000) with an accuracy of ±0.1% were used, along with suitable synchronous data acquisition units (HBM QuantumX, up to 19200 Hz).

A standard analysis, previously validated in other scale-model tests (Hajdin, 1982; Muškatirović and Jović, 1982; Starinac, 2013 and 2014), was applied to assess hydrodynamic loads in the stilling basin. The main points of the mathematical analysis are described below.

The instantaneous pressure pi (t) at point i in the stilling basin can be expressed as the sum of the average pressurei and the fluctuating portion of the pressure pi' (t):

for01          (1)

For the time-averaging period Т, standard deviation is an indicator of pressure fluctuation, which is defined as the square root of variance:

for02          (2)

Past measurements on scale models and structures have shown that the probability density functions for pressures can be approximated rather well by normal (Gaussian) distribution. In such cases, extreme pressure fluctuations are determined by the expression:

for03          (3)

The value of factor K depends on the probability of occurrence. For a 99.9% probability it amounts to 3.09, meaning that extreme pressure will be exceeded only 0.1% of the time.

 

Model Calibration

The objective of calibrating the model is to ensure similarity between prototype flow and the flow on the scale model. Hence, appropriate boundary conditions need to be established: the flow rate at the upstream boundary of the model and the water level at the downstream boundary.

The flow rate boundary condition was achieved by the model's water supply system, via weir boxes from which a pre-defined amount of water was supplied to the model.

The water level boundary condition is generally achieved by adjusting the water level at the most downstream point, based on rating curves (obtained from a numerical model or field measurements). In this case, the rating curve obtained from the numerical model was used.

 

Downstream Rating Curve Calculation

In order to provide the downstream boundary condition on the model, it was necessary to calculate the rating curve in a cross-section between D8 and D9 (Figure 10).

 

fig10
Figure 10: Downstream rating curve cross-section of the model.

 

The following input data were used to calculate the rating curve in the respective cross-section:

  • Baltic Sea water levels (provided by the Latvian Environment, Geology and Meteorology Center; Figure 11 and Table 1), and
  • Cross-sections of the Daugava River downstream from the Riga HPP (Starinac, 2016; Figure 12).

fig11
Figure 11: Maximum and minimum water levels of the Baltic Sea (gauge readings).

 

Table 1: Maximum and minimum water levels of the Baltic Sea, Daugavgriva Station.
tab01

 

fig12
Figure 12: Cross-sections of the Daugava River downstream from the Riga HPP.

 

HEC-RAS software developed by the US Army Corps of Engineers was used for rating curve calculations. HEC-RAS is capable of performing 1D water surface profile calculations for steady gradually-varied flow in natural or constructed channels (Brunner, 2010).

Water surface profiles were computed from one cross-section to the next by solving the energy equation, with an iterative procedure called the standard step method (Starinac, 2016).

Two boundary cases were analyzed to address the influence of the Baltic Sea water level on the downstream rating curve:

  1. MIN: Baltic Sea water level = -1.01 m BAS (min. water level from Table 1), nmch=0.020 m-1/3s, nob=0.100 m-1/3s, and
  2. MAX: Baltic Sea water level = +2.11 m BAS (max. water level from Table 1), nmch=0.030 m-1/3s, nob=0.100 m-1/3s,

where

nmch = Manning's roughness coefficient of the main river channel, and

nob = Manning's roughness coefficient of the overbanks.

The values of Manning's roughness coefficients were chosen according to recommendations (Brunner, 2010). The value of nob=0.100 m-1/3s is generally used for overbanks covered by vegetation, which is the case in the Daugava River floodplain. Selected nmch values are applied for relatively clean and straight channels, such as that of the Daugava. In the first boundary case (MIN), the smallest value of nmch was chosen to obtain minimum energy head losses for the analyzed river section, and therefore the lowest water levels at the downstream boundary cross-section on the model. Analogously, in the second boundary case (MAX), the highest value of nmch was chosen to obtain maximum energy head losses for the analyzed river section, and thus the highest water levels at the downstream boundary cross-section on the model. This way, a wide range of possible water levels (and flow conditions) was covered. Both curves are shown in Figure 13.

 

fig13
Figure 13: Min. and max. rating curves at the cross-section where water level was to be measured on the model.

 

The rating curve was calculated for the cross-section where the water level was to be measured on the model (Figure 10). The downstream boundary condition was thus achieved on the model, by adjusting the water level on the model to be similar to the calculated value.

 

Scale Model Results

Scale Model Tests

Scale model tests included testing of different combinations of spillway and turbine operating conditions (scenarios), at the original design flow rate of Q0.1%=7800 m³/s and the recommended flow rate of QPMF=12800 m³/s (Table 2). All the scenarios were tested at maximum and minimum water levels in the downstream measurement cross-section.

 

tab02

 

During the scale model tests, water levels were measured on the following locations:

  • In the reservoir upstream from the spillway and the HPP – Zres;
  • In the stilling basin downstream from the HPP (Figure 14) – Zds,hpp;
  • In the downstream cross-section for which the rating curve was calculated – Zds.

 

Design Flood Criteria

The use of the probable maximum flood (PMF) as the design flood represents the most stringent design criterion that can be applied in the case of high-hazard dams. According to ICOLD recommendations and general international practice, when sizing a spillway the turbine discharge should not be deducted from the spillway design flood, for the following reasons:

  • Intake trash racks can be blocked by wood, grass and other debris.
  • High tailwater levels may flood the powerhouse downstream.
  • High winds can damage transmission lines, requiring the HPP to be shut down.
  • Heavy rainfall can damage access roads or other important infrastructure, making access to the power station impossible.

 

fig14
Figure 14: Location where Zds,hpp was measured on the prototype.

 

The hydraulic capacity of the spillway can be reduced because of:

  • Gate operating difficulties (due to power failure, equipment breakdown or jamming caused by uneven structural settlement);
  • Blockage of spillway openings by ice or floating trash;
  • Threat of uprooted trees and other debris blocking the spillway gates in extreme flood situations;
  • Some gates out of operation for maintenance.
  • Permissible water level in the reservoir.

In this regard, the so-called (n–1) rule is common in international guidelines for defining spillway capacity. Namely, one gate is assumed out of operation when the design flood is passed.

The existing spillway structures have been sized using the design flood of Q0.1%=7800 m³/s. The previously-mentioned criterion is fulfilled in that case (Scenario 1, Figures 15-16) – the total discharge can be safely passed without turbine discharge. The (n-1) rule is also followed, but in such a way that the reservoir water level is 19.25 m BAS, which is above MRL=18.8 m BAS but there is no threat of overtopping.

The probable maximum flood, QPMF, is 5000 m³/s (64%) larger than the applied design flood, so the existing structures would not be expected to fulfill the above mentioned criteria to the extent that the design hydraulic conditions are achieved.

 

fig15
Figure 15: Scenario 1, Q0.1%=7800 m³/s, MAX Zds.

 

fig16
Figure 16: Scenario 1, Q0.1%=7800 m³/s, MIN Zds.

 

As anticipated, hydraulic tests of Scenario 1 at the new flow rate (QPMF=12800 m³/s) indicated that the HPP would be overtopped. Figures 17-18 show the observed flow conditions. The hydraulic jump is located in the stilling basin, but there are also additional hydraulic jumps forming in downstream river sections, which result in rather unfavorable hydraulic conditions.

 

fig17
Figure 17: Scenario 1, QPMF=12800 m³/s, MAX Zds.

 

While passing QPMF=12800 m³/s over the spillway, without turbine discharge, the reservoir water level was 22.3 m BAS, which exceeded the dam crest elevation of 22 m BAS. This situation will undoubtedly result in overtopping of the HPP, pose a high hazard for dam security and is therefore not permissible. This means that the criteria were not fulfilled.

 

fig18
Figure 18: Scenario 1, QPMF=12800 m³/s, MIN Zds.

 

Characteristic Rating Curves when the HPP is Not Operating

For the case where the HPP is not operating, scale model measurements were carried out over the entire range of discharge, in order to determine rating curves at characteristic cross-sections.

Figure 19 shows the spillway rating curve when all 6 gates are fully open. It is apparent that the capacity of the spillway is approximately 7900 m³/s at FRL=18.0 m BAS, and 8800 m³/s at MRL=18.8 m BAS. For a discharge of 12800 m³/s, the water level in the reservoir would be 22.3 m, which exceeds the dam crest elevation of 22 m BAS. The graph shows the values of both minimum and maximum downstream water levels; as expected, they are the same because the downstream water level does not affect the function Q (Zres).

On the same graph, the green line is the existing rating curve, provided by the Employer. This rating curve shows higher-than-measured discharge, at the same water level. Considering the fact that this rating curve was a result of scale model tests carried out for the purposes of the main design for Riga HPP, the difference was not expected. On the other hand, it should be kept in mind that previous model tests were conducted more than 40 years ago on a 1:80 scale model, when measurement equipment was not of such a high-accuracy as it is today. For example, if the same flow were to be represented on different scale models, the scale factor for the flow rate on the larger model (1:60) would be 602.5=27885 (Table 1), and on the smaller model (1:80) it would be 802.5=57243. This means that the same prototype flow rate would be lower on the smaller model by a factor of approximately 2 (57243/27885= 2.06), which makes it more difficult to ensure precise measurement. Again, considering the substantial increase in measurement accuracy over the past 40 years, the results of the present scale model tests should be considered more reliable.

 

fig19
Figure 19: Spillway rating curve.

 

The main conclusion of the tests was that the current spillway capacity is not sufficient to pass QPMF=12800 m³/s at permissible reservoir water levels.

Figure 20 shows the rating curve for a single spillway gate. Single-gate rating curve measurements were conducted for spillway gates 1 and 3, in order to check the difference in capacity. The difference was so small (up to approx. 50 m³/s at the same reservoir water level) that it could be disregarded. The conclusion was that all gates have approximately the same capacity. The capacity of a single spillway gate is about 1450 m³/s at FRL=18.0 m BAS and 1650 m³/s at MRL=18.8 m BAS (Figure 20).

 

fig20
Figure 20: Rating curve of a single spillway gate.

 

Figure 21 shows the rating curve in the cross-section downstream from the HPP, based on scale model test results when the HPP is not operating, for the entire range of discharge. It is apparent that the water level does not exceed 8.1 m BAS, which was observed at the maximum downstream water level and QPMF. The model indicated that the downstream water level of 9.1 m BAS proposed in the project's terms of reference was unrealistic and impossible to achieve. The graph shows certain differences in the downstream minimum and maximum water levels, within limits of 1-2 m. The same graph also shows the existing rating curve (result of previous numerical simulations conducted by Enbiko, 2007), provided by the Employer.

The existing downstream rating curves (for minimum and maximum levels of the Baltic Sea), provided by the Employer, are also presented in Figure 21 (orange lines, minimum (-1 m) – continuous line, maximum (1.5 m) – dashed line). These lines agree rather well with the rating curves based on the scale model test results – the lines are mostly within the bounds of MAX Zds,hpp and MIN Zds,hpp. The rating curve range on the scale model was wider, which guaranteed that the scale model test results covered a broader range of extreme values.

 

fig21
Figure 21: Rating curve in the cross-section downstream from the HPP (Figure 14).

 

The differences between the calculated and measured values were obvious and attributable to different methods used, as well as to the influence of the Baltic Sea level fluctuations on the water levels downstream from the dam (which had not been taken into account in previous calculations).

 

Rating Curves when the HPP is Operating

A summary rating curve (Figure 22) was determined for the case where the HPP is operating according to Figure 9, based on scale model tests conducted over the entire range of discharge. The measurements performed at the Riga HPP during the 2013 spring flood (when the gates were opened) are also shown on the graph (orange squares). Since the operating mode of the turbines and gates at that time was not known, it can only be stated that the reservoir water level was between minWL and FRL and that the flow rates did not exceed 5000 m³/s. Figure 23 shows the results of the same measurements in 2013, but in the cross-section downstream from the HPP. These values can be compared to those from Figure 21. The obvious conclusion is that the measured values were within the boundaries of the MAX Zds and MIN Zds rating curves, which means that the scale model was properly calibrated.

 

fig22
Figure 22: Summary rating curve, comparison with in situ measurements.

 

fig23
Figure 23: Rating curve in the cross-section downstream from the HPP, comparison with in situ measurements.

 

Possibility of Passing a Higher Discharge Through Existing Structures

None of the modifications at the downstream end of the spillway, as proposed in the ToR (additional strengthening of the transition area over a longer section, construction of additional energy dissipaters, reconstruction of some elements of the stilling basin, deepening of the riverbed, changing (rounding) the geometric shape of the spillway gates) can compensate for the insufficient capacity of the spillway to pass QPMF=12800 m³/s. Spillway discharge is determined by the reservoir water level and discharge coefficient (rating curve), which are upstream conditions, not affected by the downstream side. Therefore, the proposed measures can only be considered supplemental, aimed at protecting the downstream area.

There are two possible solutions that will ensure safe passing of QPMF=12800 m³/s: to consider a less stringent design criterion or to build an additional spillway structure.

The consideration of an additional spillway structure was not a project task, so the focus was on a less stringent design criterion. More specifically, turbine engagement was investigated.

According to the scale model test results (Figure 21, Figure 23), the maximum measured water level in the cross-section downstream from the HPP was 8.1 m, which meant that the downstream limit of 8.2 m

BAS, at which it would not be possible to operate the turbines, was not reached in any of the tested scenarios. Therefore, this requirement was not an obstacle.

Hydraulic tests of Scenario 13 (Table 3) at the new flow rate (QPMF=12800 m³/s) indicated stable flow conditions in the reservoir and relatively stable flow conditions in the stilling basin, at both downstream water levels (Figures 24 and 25). According to visual observations, the flow conditions downstream from the stilling basin were in the transient regime, so the locations of the beginning and end of the hydraulic jump were not clearly defined. From this point of view, it looked like the hydraulic jump was repelled, but without significant threat. Various detailed measurements (hydrodynamic loads in the stilling basin, downstream water levels, velocities, and scouring and deposition tendencies) were undertaken to check this scenario more closely.

 

fig24
Figure 24: Scenario 13, Qpmf=12800 m³/s, MAX Zds.

 

fig25
Figure 25: Scenario 13, Qpmf=12800 m³/s, MIN Zds.

 

Hydrodynamic Loads in the Stilling Basin

Hydrodynamic loading of the stilling basin was assessed at the design flow rate of Q0.1%=7800 m3/s (Scenario 1) and the new flow rate of QPMF=12800 m3/s (Scenario 13). The assessment focused on the points in time when the pressure fluctuations were the highest, corresponding to maximum flows down the spillway. For both flow rates, pressures were measured for MIN Zds and MAX Zds at 52 points on the bottom of the stilling basin. According to the explanation given in Methodology – Model Measurements, calculations were carried out to determine the average pressure and extreme pressure fluctuations, acting on the bottom of the stilling basin at predefined points. The results obtained for MIN Zds (less favorable) at Q0.1% are shown in Figures 26 and 27, and at QPMF in Figures 28 and 29. The maximum values are given in Table 4.

 

fig26
Figure 26: Average pressures [KPa] at Q0.1%=7800 m3/s, HPP not operating, MIN Zds.

 

fig27
Figure 27: Extreme pressure fluctuations [KPa] at Q0.1%=7800 m3/s, HPP not operating, MIN Zds.

 

fig28
Figure 28: Average pressures [KPa] at QPMF=12800 m3/s, HPP operating, MIN Zds.

 

tab03

 

In all four tests, extreme pressure fluctuations were generally proportional to the average pressures at all points. Extreme pressure fluctuations were significantly smaller than the average pressures. None of the tests indicated a threat of negative pressures in the stilling basin.

A flow rate increase from Q0.1%=7800 m3/s to QPMF=12800 m3/s (HPP operating) caused an increase in average pressures of 12-15%, but also a reduction in extreme pressure fluctuations of 23-24%.

Therefore, the conclusion was that there were no significant differences in pressure conditions in the stilling basin and, most importantly, no threat of negative pressures in any of the tested cases.

 

fig29
Figure 29: Extreme pressure fluctuations [KPa] at QPMF=12800 m3/s, HPP operating, MIN Zds.

 

Water Levels Downstream from the Stilling Basin

Water levels downstream from the stilling basin were measured in the cross-section at the stilling basin centerline, in order to clearly show the form of the downstream waves. These measurements were made at Q0.1%=7800 m3/s (Scenario 1, Figure 30) and QPMF=12800 m3/s (Scenario 13, Figure 31), for both downstream conditions MAX Zds and MIN Zds. The distances indicated in the figures are from the end of the stilling basin.

According to visual observations, it was clear that the main hydraulic jump occurred in the stilling basin and was drowned locally. Also, it was noted that smaller (secondary) hydraulic jumps appeared in the downstream river section. Figures 30 and 31 corroborate these phenomena. Due to the terrain geometry of the downstream area, the downstream water level was simply not high enough to ensure stable supercritical flow (Fr˃1). This was noted even at the lowest flow rates. At both maximum and minimum downstream boundary water levels, the flow from the spillway and the downstream flow were connecting through a few smaller hydraulic jumps and flow conditions were in the critical (transient) regime (Fr≈1). Such conditions promote erosion and are very hard to avoid.

 

fig30
Figure 30: Water level downstream from the stilling basin at Q0.1%=7800 m3/s.

 

fig31
Figure 31: Water level downstream from the stilling basin at QPMF=12800 m3/s.

 

Velocities in the Downstream River Section

The velocities in the downstream river section were measured at the design flow rate of Q0.1%=7800 m3/s (Scenario 1) and the new flow rate of QPMF=12800 m3/s (Scenario 13), for both MIN Zds and MAX Zds. The velocity trends were similar in all four tested cases. The lower downstream water level was less favorable in the sense that the velocities (and therefore the scour potential) were slightly higher at MIN Zds than MAX Zds. At Q0.1%=7800 m3/s, the maximum velocity was 6-7 m/s for MAX Zds and 8 m/s for MIN Zds (Figure 32). At QPMF=12800 m3/s, the maximum velocity was 6 m/s for MAX Zds and 7 m/s for MIN Zds (Figure 33).

 

fig32
Figure 32: Downstream velocities at Q0.1%=7800 m3/s, HPP not operating, MIN Zds.

 

fig33
Figure 33: Downstream velocities at QPMF=12800 m3/s, HPP operating, MIN Zds.

 

Scour and Deposition in the Downstream Riverbed

When discussing potential damage to dam structures due to changes caused by riverbed morphology alteration (scour and deposition), the following facts need to be kept in mind: In scour analyses, the most important areas are those close to the bottom, because that is where water comes into contact with bed material. Movement of the particles that form the reservoir bed and riverbed depends on the water velocity close to the bottom. When this velocity is low (in the reservoir), the particles will not move. The velocities are the highest in the stilling basin and the area downstream. Stilling basins are generally constructed from concrete, in such a way as to withstand these velocities and pressures and protect all structures from potential vibrations and stability issues. The downstream riverbed is made up of incoherent material (variably fractured dolomite), which cannot handle such high velocities. As a result, riverbed material moves and causes scour and deposition. The main problem of scour pits forming is not the scour itself, but the effect of scouring on the structural stability of the stilling basin, and in turn of all other dam structures. Considering this fact, it is important to prevent scouring in the area close to the stilling basin and to ensure that washouts do not affect the stilling basin structure. Velocities in the downstream river section are reduced along the flow and the scour potential is balanced by the river itself. All these facts are the reasons why scour on scale models is usually analyzed in a limited area (250 m downstream from the stilling basin in this case).

Scale modeling in scour analyses is constrained because it is not possible to achieve total similarity of the model and prototype. The use of similar incoherent riverbed material as in the prototype would require impossibly small fractions on the model (for example, if the average diameter of the riverbed material on the prototype is 1 mm, the average diameter on the model would have to be 1/60=0.017 mm). These small particles (even if such material exists) would be lighter than water and completely mixed with water (forming a suspension), so scour analyses would not be successful. Therefore, by applying fractions of material that is available and not so light-weight (sand), the model can provide results of qualitative, but not quantitative similarity to the prototype. That means that scour tendencies and threatened locations will be determined, but not the dimensions of the scour pits. Even if limited, a scale model is still the best method for studying scouring, compared to other (numerical) methods.

Scour and deposition were tested on the movable-bed model, with two different incoherent materials, at Q0.1%=7800 m3/s, HPP not operating, MIN Zds (Scenario 1), and QPMF=12800 m3/s, HPP operating, MIN Zds (Scenario 13).

Upon completion of the previous test phase, the model was modified to include a movable bed. The cement layer was removed from the river channel downstream from the stilling basin over a length of 250 m. In the first part of the subsequent tests, the screed was replaced with loose material – sand. That particular part of the scale model was filled with sand (grain size 1 mm) and shaped per original design documents.

The idea was to test the behavior of the riverbed at Q0.1% and QPMF and note any propensity for scouring, and/or detect any critical areas, for the purposes of identifying suitable protection measures. The photo taken upon completion of the first test phase at Q0.1% (Figure 34) shows that some of the material was washed-out. The photo taken after QPMF (Figure 35) shows that more material was washed-out, so the scouring intensity was higher. The isolines in the figures are given with reference to BAS.

 

fig34
Figure 34: Scouring downstream from the stilling basin at Q0.1%, sand.

 

fig35
Figure 35: Scouring downstream from the stilling basin at QPMF, sand.

 

fig36
Figure 36: Scouring downstream from the HPP at QPMF, gravel.

 

In the second step, the model was filled with larger material – gravel Dav,mod=15-20 mm. This material was selected according to velocity and depth measurements, as the grain size that would not move at the maximum measured velocity of 8 m/s. According to Shields, the critical tractive force on the bed is given by (Smith, 1995)

for04          (4)

where:

τc – bed shear stress at incipient motion of the bed particles [N/m²]

τ – critical Shields stress value (0.048, recommended by Meyer-Peter and Mueller)

γ – unit weight of water (9810 N/m2)

Sg – relative density of the stone particles (2.65, standard value for most gravels)

dm – median size of the stones [m].

Using the above values in brackets, Eq. 4 can be transformed into:

for05          (5)

For an average flow depth of 9 m, tc = pgHI = 699 N/m², so from Eq. 5, dm = 0.9 m. According to the similarity between the model and the prototype, the average diameter of the incoherent material on the scale model was 15 mm.

The photo taken after the tests at Q0.1% shows no washout of the material. The photo taken after QPMF (Figure 36) also shows that virtually no material was washed-out. The conclusion was that downstream scouring will be avoided if protective material with a diameter of 0.9-1.2 m is used.

These analyses are considered merely basic, to be used as inputs for the design of downstream protection measures. Precise technical measures need to be evaluated taking into account the technical solutions, but also economic aspects. Reducing the area where protection is necessary could significantly decrease the cost.

 

Possibility of Operation at Permissible Reservoir Water Levels

The presented case of passing QPMF=12800 m³/s through the 6 spillway gates (fully open) and all turbines would result in a reservoir water level of 19.25 m BAS. The respective maximum water level in the cross-section downstream from the HPP would be 7.9 m BAS.

For the same flow rate and turbine operating conditions, closing any of the spillway gates would also ensure permissible reservoir water levels, of 20.6-20.7 m BAS. For example, if all gates are more than 8.75 m open, the permissible reservoir water level will be up to 20.8 m BAS.

According to the spillway rating curve (Figure 19), the maximum permissible reservoir water level of 20.9 m BAS will result in a maximum spillway capacity of 11247 m³/s. Therefore, at QPMF=12800 m³/s the turbines will have to be engaged in order to cover the necessary portion of the discharge of 12800-11247=1553 m³/s. By extrapolating the turbine rating curve (Starinac, 2016), at the reservoir water level of 20.9 m BAS, the capacity of the turbines is 3669.5 m³/s and of a single turbine 611.6 m³/s. The conclusion was that it is possible to pass QPMF=12800 m³/s at a reservoir water level of 20.9 m through the spillway (6 gates) and 3 turbines. The corresponding maximum water level in the cross-section downstream from the HPP will be about 8 m BAS.

If one spillway gate is closed, the reservoir water level of 20.9 m BAS will result in a spillway capacity of 9744 m³/s. Therefore, to pass QPMF=12800 m³/s, the turbines will have to be engaged to cover the necessary portion of the discharge of 12800-9744=3056 m³/s. This means that it will be possible to pass QPMF=12800 m³/s at the reservoir water level of 20.9 m, through the spillway (5 gates) and 5 turbines. The corresponding maximum water level in the cross-section downstream from the HPP will be about 8 m BAS.

 

Conclusions

According to the scale model tests, the conclusions were as follows:

  1. The spillway capacity is not sufficient to pass a flood discharge of QPMF=12800 m³/s at permissible reservoir water levels.
  2. The spillway capacity cannot be increased by downstream modifications.
  3. There are two possible solutions to ensure safe passage of QPMF=12800 m³/s: to consider a less stringent design criterion or to build an additional spillway structure.
  4. It is possible to pass QPMF=12800 m³/s at a reservoir water level of 20.9 m BAS, through the spillway and 3 turbines.
  5. It is possible to pass QPMF=12800 m³/s at the reservoir water level of 20.9 m BAS, through 5 spillway gates and 5 turbines.
  6. Passing of QPMF=12800 m³/s through the 6 spillway gates (fully open) and all the turbines will result in a reservoir water level of 19.25 m BAS.
  7. There will be no threat of negative pressures in the stilling basin while passing QPMF=12800 m³/s through the 6 spillway gates (fully open) and all the turbines.
  8. Hydraulic conditions in the area downstream from the stilling basin are problematic.
  9. Erosion in the area downstream from the stilling basin could be significantly reduced by implementing riverbed protection measures, using D=0.9-1.2 m stones as protection material.

 

Acknowledgment

The authors express their gratitude to the Ministry of Education, Science and Technological Development of the Republic of Serbia for its financial support to the Technology Development Projects TR 37005, 37009, 37010 and 37014.

 

References

Brunner G.W. (2010). HEC-RAS, River Analysis System Hydraulic Reference Manual, US Army Corps of Engineers HEC, CPD 69, January 2010.

http://www.hec.usace.army.mil/software/hec-ras/documentation/HEC-RAS_4.1_Reference_Manual.pdf

Hajdin, G. (1982). Prilozi za procenu fluktuacionog opterećenja na granične površine fluidne struje - na osnovu izmerenih pritisaka u nekoliko tačaka površine [Annexes for calculating hydrodynamic force fluctuations on the boundary surfaces of the flow current - based on measured pressures at several points of the surface]. Proceedings from 8th Conference of Yugoslav Hydraulic Society, Portorož, December 1982.

Мuškаtirоvić, Ј. and S. Јоvić (1982). Аnаlizа hidrоdinаmičkоg оptеrеćеnjа slаpištа prеlivnе brаnе [The analysis of the hydrodinamic forces in the stilling basin of an overfall dam]. Proceedings from 8th Conference of Yugoslav Hydraulic Society, Pоrtоrоž, December 1982.

Smith C.O. (1995).Hydraulic structures. University of Saskatchewan printing services and Universal Bindery, Saskatoon, Canada. p.8-1.

Starinac, D. (2013). Hidrаuličkа mоdеlskа ispitivаnjа brаnе Buzinа - Kоnаčаn izvеštај /1203/[Scale model analyses of the Bouzina dam – Final Report]. Jaroslav Cerni Institute for the Development of Water Resources (JCI), Belgrade, June 2013.

Starinac D., Vojt P., Damnjanović M., Žugić D., Savić Lj., Kapor R., Zindović B. and R. Glišić. (2014). Scale Modeling of the Bouzina Dam Flood Mitigation Structures, Journal of Serbian Water Pollution Control Society "Water Research and Management", ISSN 2217-5237, Vol. 4, No. 1, pp. 31-42, 2014.

http://www.wrmjournal.com/images/stories/casopis/No_13/04.pdf

Starinac D., Kapor R., Savić Lj., Vojt P., Žugić D., Damnjanović M., Zindović B. and P. Đajić. (2014). Air-Water Flow on a Labyrinth Spillway, Journal of Serbian Water Pollution Control Society "Water Research and Management", ISSN 2217-5237, Vol. 4, No. 3, pp. 11-20, 2014.

http://www.wrmjournal.com/images/stories/casopis/No_15/02.pdf

Starinac D., Vojt P., Damnjanović M., Žugić D., Kapor R., Savić Lj., Zindović B. and P. Đajić. (2015). Scale Modeling of Beni Slimane Dam, Journal of Serbian Water Pollution Control Society "Water Research and Management", ISSN 2217-5237, Vol. 5, No. 1, pp. 9-21, 2015.

http://www.wrmjournal.com/images/stories/casopis/No_17/02.pdf

Starinac D. (2016). Scale modeling of HPP Riga spillway stilling basin – Final Report /1243/. Jaroslav Cerni Institute for the Development of Water Resources (JCI), Belgrade, September 2016.