A PHYSICAL MODEL RESEARCH ON THE EFFECT OF SEDIMENT SCOUR IN LOWERING WATER LEVEL OF THE LIUJIAXIA RESERVOIR

Huang Jinchi and Huang Yongjian

China Institute of Water Resources and Hydroelectric Power Research

Fuxinglu, A-1, Beijing, 100038, China 

Abstract: Physical model is an effective approach in studying sediment problems such as the degradation and aggradation in a reservoir. For reservoir sediment problems, the deposited material is usually fine and the coherence force plays an important role in the evolution process. Therefore, it is usual in physical model designing to consider some special properties of fine sediment transport and the corresponding riverbed evolutions, especially for the sediment erosion processes. A dry weight scale was introduced into present model designing for the erosion problem research of fine sediment particle, in which the coherence force is the main resistance to erosive processes. The necessary steps for such a physical model research and, some fundamental principles for the designing of a reservoir erosion model are discussed. A case of Liujiaxia Reservoir, which is located at the upper reach of the Yellow River, is employed to verify present approach of model designing. A wonderful recurrence in the verification tests and satisfactory agreement with the data site-observed illustrated that the method introduced here is a reasonable one.

1    INTRODUCTION

Reservoir sediment problems are becoming more and more severe for those projects built on the rivers with heavy sediment laden flows as the situation in the Yellow River in China. A reasonable sediment disposal measure must be found for project operation. Experience has shown that it is an effective measurement to lower the water level to such an appropriate value that large amount of the pre-deposited material can be scoured away in a relatively short time, and then certain perpetual storage can be kept for utilization over a long period of time (Chian , 1989, Du & Zhang, 1989). Clearly, the effects of this methods closely depends on many factors such as the processes of water level lowering, the initial bed profile, water discharge, and bed material composition. Unfortunately, it is not so clear on the physical mechanics of sediment transport and river bed evolution during such a erosion process. Up to now, the reliability of a designed operation scheme is usually determined according to the physical model research.

In recent decades of years, large number of investigations have contributed to the physical mode designing of an erosive process but most of them concentrated their works on non-coherence sediment problems. However, for some rivers with hyperconcentrated sediment laden flows, the particle size is very fine and the coherent force controls the whole erosion processes. Theoretical and experimental researches have shown such sediment has a fully distinguished properties compared with the non-cohesion one in the erosive processes in sediment flushing operation of a reservoir. In the scale model designing of such situation, it should be carefully considered on some special aspects of the sediment transport and river bed evolutions. One of these special properties is its kinetic and dynamic parameters, such as the initial velocity, erosion rate etc., being closely related to the dry weight, rather than the diameter of the sediment particles. Wang and Zhang [1989] have used this concept to design a scouring model but no verification was made on the model and thus, the result only can be taken as qualitative one.

2    MODEL DESIGNING

(1) Basic principles

Particle size distribution in the Liujiaxia Reservoir

Many researchers have illustrated (Smerdpm & Bensley, 1959) that, for the sediment with the particle size less than 0.05 mm, the main resistance to erosion is not the diameter of the grain itself but the coherent forces among the particles. Further experiments have shown that for such kind of sediment, the functions of the dry weight of the eroded material is the most important. Recently, with a sediment sample of diameter of 0.004 mm and specific gravity 2.75, a series initial velocity tests were conducted. The relation between the initial velocity and dry weight is shown in Fig.2. The fairly little scatter shows that the conclusion is confirm that the initial velocity strongly depends on the dry weight of the deposited material in a reservoir. The figure can be numerically expressed as:

where, Vc is the initial velocity, and γ' is dry weight of sediment deposition.

For the erosion rate similarity, problem is more complexity. There is little information on the quantitatively estimation of the erosion rate of the modeling material. Based on a conceptual understanding, the erosion rate of the cohesive sediment depends on different material composition and hydraulic conditions such as coherence, water stability, density,

                                (1)

specific gravity, moisture before exposure to flow, plastic properties, chemical composition and salinity, vegetable cover and species of plants, meteorological conditions, chemical compositions of water and its sediment concentration, geometry of channel in plan and cross-section, flow depth, turbulence characteristics, roughness, etc. Clearly, all these factors can not be fully reflected in a determined formula. The similarity can not be justified unless a series of verifications are made.

From the mass conservancy of sediment transport, with the assumption that a direct exchange exists between the suspended sediment and the bed material and the movement of bed load can be neglected, a sediment continuity equation can be expressed as:

                              (2)

where, Q=discharge of water flow; s=sediment concentration; x=distance; B=river bed width: z=river bed elevation; and t=time. If the erosive rate of sediment particles is defined as E=z/t, then the similarity scale of the erosion rate can be written as:

                                 (3)

where, E=the erosion rate scale; λ_V=velocity scale; λ_H=vertical scale; λ_c = sediment concentration scale; λ_L=length scale; and, λ_γ' =dry weight scale. All these similarity scales can be determined according to some basic principles of hydraulic and sediment movement except forλ_γ' andλc, because there is no extensively-accepted quantitative expressions available for the calculation of the dry weightγ' and sediment concentration c. some investigators have argued that the scaleλ_c can be replaced byλ_c*, which is the transport capacity scale, based on the sediment movement requirement, while the scaleλ_c* can be calculated out by employing some transport formula such as Celik and Rodi's work (Celik & Rodi, 1988). Then γ' is a free variable in Eq.2 and can be artificially determined according to selected model sediment material. This idea has a remarkable neglect on the function of the boundary condition. As well known, physical modeling depends on dynamic similarity. For erosive problem, as discussed here, the crucial aspect is on the water shearing force acting on the movable bed and the resistant force of the bed. Controlling factors for the sediment concentration is not only the transport capacity of the water flow but also, even more important in present case, the erosion rate. Experience in some reservoir sediment flushing processes illustrated that most of the flushing processes are in non-sutured conditions and the sediment transport capacity is much larger than actual concentration. Thus during the flushing processes, the variation of sediment concentration mainly depends on the erosive rate. This is clearly not enough to meet the requirements of a physical model designing. For a special problem, the most reliable treatment is through trial and error to test what is a suitable dry weight of the model material which is just in the similarity with the case of prototype. Thus, a verification test is also necessary before the formal experiments are carried out.

(2) Steps of the model designing

The first problem faced by a physical model researcher is how to select material used in model. Experience has shown that such a work must be carefully taken. Usually, selection of the model material is influenced mainly by following factors:

① the availability and the cost of the material used in the model;

② the requirements of the properties for appropriately modeling the studied phenomena; and

③ the stability of the physical and chemical properties of the material.

For a reservoir scouring model, the coherence resistance force must be considered on the model material selection. Thus, for the model designing of a water level lowering caused erosion process, the steps can be summarized as:

① according to the space condition, a basic geometry scales are preliminarily selected;

② based on the requirements of water flow similarity, some basic water flow scales are determined;

③ according to the suspended sediment similarity requirement, a suitable model sediment material and diameter scale is preliminarily selected;

④ according to the results in flume tests, verifying if the similarity of the initial conditions and erosion rate can be meted.

⑤ perform a series of verification tests based on the site-observed data. Check if the natural phenomena such as the water flow properties and sediment movement can be represented in the model. If the answer is no, go back the step iii, if the answer is yes, the model designing is thought appropriate and a formal experiment can be started.

3    A CASE STUDY

Fig. 2    Plan view of the liujiaxia reservoir

series of problems for the management of the project. For safety usage of the project and full utilization over a long time period, some technical measures must be taken to keep the storage capacity of the reservoir.

It is assumed that the model is conventional. According to the task requirements and the space conditions, distortion on geometrical scale is inevitable. Based on the water flow similarity requirements, some basic similarity scales are preliminarily selected as shown in Table 1.

Table 1    Selected model scales for water flow

For scouring similarity on sediment transport, following relationship on initial velocity must be satisfied:

                                  (4)

Lending Eq.2, the dry weight scale of the present model should be 1.778. The dry weight of the deposited sediment material of Liujiaxia reservoir is about 1.10 and thus the model material in the river bed should be 0.62.

Through a series of detailed comparison and analysis on the physical and chemical properties of some usual model material and certain previous practice, a coal powder with specific gravity of 1,61 was used in present research.

The verification tests include two aspects of research work, one is to study the initial properties of light model material and the other is the erosion rate of such a material in certain water flow and boundary conditions. At present, the former was conducted in a flume and the later was performed in scale model.

        Tests on the similarity of sediment initial condition

Previous discussion shows that the most of the present formula on initial condition is on the natural sediment material with a specific gravity varying in a range of 2.5-2.8. A typical relationship can be expressed as:

                    (5)

where, H is the water depth, d is the sediment diameter,   and  are specific gravities of water and sediment respectively, , _k,Δare some special constants witch can be previously determined according to laboratory test results. For much lighter material it is necessary to test the characteristics on initial condition. Thus a experiment was conducted in a flume of the Institute of water Conservancy and Hydroelectric Power Research (IWHR) [Huang Yongjian & Huang Jinchi, 1992]. A total of 8 runs of experiments were conducted and the results are plotted in semi-log paper as shown in Fig.3 (the value has been transferred to the natural condition based on the similarity scale). A curve calculated with the formula presented by Du Guoren [Du guoren, 1983] for the sediment initial condition of fine sediment particles is also plotted in the same figure. The comparison shows that the exact similarity is just in one depth value of about 2.0 m. In other depth conditions, the initial condition in model will deviate more or less from the prototype. Actually this is a common existing problem that the initial conditions can not reach similarity in all depth variations. A common treatment is to adjust the model material composition to make the exact similarity point being in the experimental range of a special studied problem. In present model research sediment erosion usually takes place in the water depth ranging from 2-5 m. Fig.3 shows that the exact similarity depth point of the selected model material is just in this range.

Fig. 3    A relation between the initial velocity and water depth for coal powder

Tests on the erosive rate similarity

With an available observed data on erosion processes of water level lowering in 1988 and 1981, 6 runs of tests were made of which, 3 runs are for the process of 1981 and the others are for that of 1988. Here the repetition of same case is for testing the stability of the experimental results. Fig.4 is the input water discharge hydrograph in Taohe River and the water level process at the dam site for these tow typical scouring practices. A final dry weight scaleλ_γ' was selected as 2.02 with theγ' values of the model material of 0.76 and the natural material of 1.10. Table 2 is a comparison between the observed and the calculated sediment scouring data in these two cases of the verifications. The agreements illustrated that the present model designing is reasonable. Then the final selected sediment similarity scales, which will be used in predicting scheme tests, are listed in Table 3.

Fig. 4    Input conditions of the physical model

Table 2    Test results of the total deposition in the reservoir    unit:  10⁶ m³

Table 3    Finally Selected Model Scales for Sediment Transport

4    CONCLUSION

Present experimental research shows for the research on the erosion problems with fine sediment deposition in reservoir the dry weight scale in a physical model designing is a very important aspect. Sediment concentration similarity can not be reached for an erosive model unless the erosion rate similarity can also be met while this can be realized by adjusting dry weight scale of the deposited material ready to be modeled. For any physical model designing, the verification tests are necessary owing to the uncertainty on some basic principles of sediment transport and water flow.

References

Chain Min Wu, (1989) Hydraulic properties of reservoir desilting. XXIII Congress of IAHR, Ottawa, Canada, August 21-25.

Celik, I. and Rodi, W. (1988). Modeling Suspended sediment transport in Non-equilibrium Situations, Journal of Hydraulic Engineering, ASCE, 114(10).

Huang Yongjian, Huang Jinchi, (1992) A Research on Sediment Flushing by Current Flow in Liujiaxia Reservoir. National Conference on Basic Theory of sediment Transport, Beijing, China (in Chinese).

Laassen, G. J. K., (1992) Experience from Physical Model for a Bridge Across a Braided River with Fine Sand as Bed Material, 5th International Symposium on River Sedimentation, Vol.1, Karlsruhe.

Smerdpm. E.T., and Bensley, R.P. (1959) The Tractive Force Theory Applied to Stability of Open Channels in Cohesive Soil. Research Bulletin 715, University of Missouri, Oct..

Wang Zhaoyin, Zhang Xinyu (1989), A Model Study on Scouring of Cohesive Sediment from Reservoirs. Journal of Sediment Research, No.2 (in Chinese).