SUSTAINABLE RESERVOIR SEDIMENTATION CONTROL

A. M. Siyam, J. S. Yeoh and J. H. Loveless(Mice)

Department of Civil Engineering,

University of Bristol, Queen’s Building,

University Walk, Bristol BS8 1TR,

United Kingdom.

Tel: +44(0) 117 928 7707, Fax: +44(0) 117 928 7783,

E-mail:j.loveless@bristol.ac.uk 

Abstract: The useful life or remaining capacity of reservoirs can be predicted from their trap efficiencies. The most widely known and used method for predicting trap efficiency are the empirical curves of Brune (1953). Many researchers have since questioned Brune’s efficiency curves, which use the capacity to inflow alone to describe the dynamics of sedimentation in reservoirs, and makes no distinction made to the source or type of sediment. This paper presents a new simple and generalised trap efficiency relationship, developed by considering the reservoir sedimentation problem as similar to a mixer tank problem.

The paper also describes the work done on a study of a new system for the control of reservoir sedimentation. It presents some results of the physical and mathematical model used to simulate the proposed system consisting of a submerged intake and a weir connected to the main dam by closed conduits as shown in Figure 3. Results from the experimental tests conducted on the physical model were compared to SSIIM; a 3-D mathematical model, which has already shown encouraging, results in modelling reservoir sedimentation. It has been demonstrated in this research that the submerged dam and closed conduits for the mitigation of reservoir sedimentation shows great promise in many cases and merits further research. 

Keywords: trap efficiency, useful life of reservoirs, sustainable reservoir sedimentation control, mathematical model, physical model, laboratory experiments

1    INTRODUCTION

The widely used Brune’s (1953) trap efficiency curves are empirical curves based on data collected from 40 normally ponded USA reservoirs. Brune presents 2 curves, an upper and lower curve to fit his collected data and suggested a third middle curve for use in design. Brune included records of two semi-dry reservoirs and two desilting basins, all of which fall outside the envelope curves for normally ponded reservoirs as shown in Figure 1. Due to the generality of Brune’s curves and the claim that the curve gives only a rough estimate of reservoir trap efficiency (Heinemann, 1981) has inspired the authors to develop a new and simpler trap efficiency relationship.

Also in this paper the use of a system similar to that suggested by Singh & Durgunoglu (1991), which consists of pipelines to transport the incoming and newly deposited sediment in the reservoir is proposed. It features a submerged check dam to block the advancement of the delta formation and bypass pipelines to transport and flush the incoming sediment to the downstream of the reservoir. Laboratory tests were conducted to examine the performance of the proposed system and the results were compared to a mathematical model known as SSIIM. The research program has been carried out for the past 3 years and will be continued for the coming years at the University of Bristol.

2    NEW THEORIES

T has been found that the trap efficiency of a mixer tank is governed by a simple exponential law, which depends on a simple dimensionless parameter. The following describes the derivation of the new proposed efficiency relationship. The rate of change of C with time can be expressed as

                           (1)

where V = volume in the tank, C_in = inflow sediment concentration, Q₀ = inflow rate and

  C = uniform concentration in mixer tank.

By integrating Eq. 1 and evaluate the arbitrary condition for initial condition C =0 at t = 0, this reduces equation 1 to

                            (2)

By definition of trap efficiency,

                                  (3)

By substituting equation 2 into 3, yields

                                   (4)

As mentioned, the above trap efficiency of a mixer tank can be viewed as a function of a single dimensionless parameter which is the inflow/capacity ratio (the inverse of the parameter used by Brune 1953). In a mixer tank model, V and Q₀ are constant and efficiency, h changes with time, t. In reservoirs, the storage V also changes with time as a result of sediment deposition. However, since the siltation occurs over many years, the product Q₀t could be combined and introduced as average annual inflow, I. Equation 4 can then be rewritten as

                              (5)

The equation above was plotted against the Brune’s (1953) data but it does not fit. However there is a similar trend suggesting some kind of transitional effect and the mechanics of trap efficiency in both systems belongs to the same physical process. It is therefore suggested the term b, which is a sedimentation parameter that reflects the decay of the storage capacity of the reservoir due to the ongoing sedimentation process, is introduced into equation 5.

                             (6)

Hence equation 6 is the proposed generalised trap efficiency relationship which could be adopted for any water retaining body likely to experience sediment deposition. It should be understood that b is not a universal physical constant but is governed by some hydraulic parameters of the reservoir mainly the fall velocity of sediments, reservoir shape, reservoir area and type of reservoir operation. The following method is suggested for the determination of the factor b for existing reservoirs.

Consider a monthly average trap efficiency,h_m 

                              (7)

The parameters, monthly inflow I_m can be obtained from hydrological data while V_m can be obtained from the capacity curve of the reservoir and b_m can either be calibrated by measuring local inflow and outflow concentration data or by using information from another reservoir with similar inflow sediment characteristics and inflow-capacity ratio. The detailed steps for determining factor b from other reservoirs and a table of examples listing the factors b for different types of reservoirs can be found in Siyam (2000). If data on inflow and outflow sediment concentrations are available, the following dimensionless expression groups the factors that determine b_m.

                               (8)

where w_s is the sediment fall velocity, A_m = average effective area in m² where sediment is depositing, Q_m is the average inflow in m³/s during the month considered and a_m represents the average degree of turbulence in the whole reservoir. If this process is completed for the whole year, the average annual trap efficiency can be determined from

                           (9)

Therefore the annual sedimentation factor can be obtained as

               (10)

3    PHYSICAL MODEL

All the laboratory experiments were conducted in a rectangular flume in the Hydraulics Laboratory of the Civil Engineering Department of the University of Bristol. A schematic layout of the flume is shown in Figure 3. The dimensions of the flume can also be obtained from Figure 3 and a detailed description of the experimental apparatus can be found in Siyam (2000). Special lightweight and fine materials (Urea Formaldehyde plastic media) were used to simulate the suspended sediment and flow velocities.

4    MATHEMATICAL MODEL

The SSIIM program, which stands for Sedimentation Simulation in Intakes with Multiblock option is a 3-D Computational Fluid Dynamics (CFD) program, developed by Olsen (1997, 1999) in 1993. The program has been specifically designed to simulate/solve water and sediment flow in reservoirs and rivers. It presents graphical plots of the velocity and sediment concentration field with changes in channel geometry due to morphological processes. In this research, the latest two versions SSIIM 14 & 15 were used. The SSIIM model was used to solve the Navier –Stokes equations with the k-e model on a 3-D non-orthogonal structured grid. The program uses the Control Volume Method (CVM) for the discretisation of the equations with either the Second Order Upwind (SOU) scheme or a power law. For pressure coupling, SSIIM uses the SIMPLE (Semi-Implicit Method for Pressure-Linked Equations) algorithm developed by Patanaker and Spalding (1972).

5    RESULTS

When equation 6 was fitted with Brune’s (1953) data, it was found that b=0.0079 gives the excellent coefficient correlation r² =0.952. Also, interestingly the trap efficiencies of the two semi-dry reservoirs mentioned by Brune fall close to that of the mixer tank, with b = 0.75. The data points of the other two desilting basins, fitted very well by a low value of b=0.00012 (see Figure 1). This is best explained by the fact that semi-dry reservoirs have efficient mixing processes i.e. high turbulence that resembles the process in the mixer tank and also having a high capacity/inflow ratio. On the other hand, desilting basin have a very low mixing effect, thus act predominantly in deposition. The normal ponded reservoirs fall in between these two types.

The procedure for all type of physical tests were basically the same with exception of the sediment discharges, inflow flowrate, initial bed level profiles, reservoir water levels and discharges through bypass pipes. These parameters could be varied in different test runs. The computed bed profile by SSIIM fitted quite accurately with the measured values. However the mathematical model slightly under predicted the outflow sedimentation concentration of the bypass pipes, C_p while overestimating that in the flow over the main dam, C_d (see Figure 2). This maybe due to the extra turbulence created by the water jets resulting from the diffuser, exerting extra shear stress at the bed preventing sediment from settling or moving downstream as bed load. Conversely due to the fact that the actual flow depth over spillway is smaller than most of upper cells in the SSIIM model, the computed flows over the dam are expected to be higher than the actual values. Test 8 simulates the advanced stage of reservoir sedimentation where the delta formation has reached the submerged intake. It was observed that the upper region of delta had risen but the front region had not moved forward. This might be due to the to the presence of the submerged dam with its bypassing system.

6    DISCUSSION AND CONCLUSIONS

The proposed trap efficiency relationship is used to explain the scatter of Brune’s data on different type of reservoirs. Besides providing a theoretical justification for the use of the inverse parameter by Brune (1953), it has been shown that the general effect of all the other significant parameters to sedimentation in reservoirs produces a translational effect to the trap efficiency curves. This will result in the curves being similar and parallel to each other for different type of reservoirs. The new trap efficiency relationship has accommodated this through the introduction of a sedimentation factor b. A method for evaluating this factor is presented.

Table 1 summarised the performance of the submerged dam and bypass system for the sedimentation tests carried out. The table compared the computed overall trap efficiencies, which is the percentage of the total inflow sediment deposited in the flume with the measured trap efficiencies. The trap efficiency without the pipe was calculated assuming that the outflow concentration would be that measured or computed over the main dam. The difference between these efficiencies is the reduction in trap efficiency to the bypass system. It can be concluded that the computed trap efficiencies compared very well with the measured values. However, in these tests, both the computed and measured reduction in trap efficiencies were low. This is best explained by the fact, most sedimentation took place in the upstream region. Hence, the sediment concentrations that reached the submerged intakes were low and uniform in the vertical direction. This complies with the objectives of the use of bypass system that is to protect the region between the submerged and the main dam.

References

Heinemann, H.G. (1981) A new Sediment trap Efficiency Curve for Small Reservoirs Water  Resources Bulletin 17(5): PP. 825-830.

Loveless, J.H.& Siyam, A.M.(2000)  Model assessment of a novel system of reservoir sedimentation control Proceeding 8^th International Symposium on Stochastic Hydraulics, Beijing. A.A.Balkema.

Olsen, N.R.B (1997), A Three Dimensional Numerical Model for Simulation Movements inWater Intakes with Multiblock Option SSIIM User’s Manual Version 21.5a,Division  of Hydraulic Engineering, The Norwegian Institute of Technology, April 1997.

Olsen, N.R.B, (1999), Two-dimensional numerical modelling of flushing processes in water reservoirs Journal of Hydr. Res. Vol. 37 No1 pp 3-13.

Patankar, S.V. & Spalding, D.B (1972) A calculation Procedure for Heat, Mass and Momentum Transfer in three-dimensional Parabolic Flows International Journal of Heat Mass Transfer, Vol. 15, pp 1787.

Singh, K.P. & Durgunoglu A. 1991, Remedies for sediment buildup Journal of Hydro Review, December 1991, PP 90-97.

Siyam, A.M. (2000)  Reservoir Sedimentation Control PhD Thesis University of Bristol.

Fig. 1    Comparison of Brune’s data with the proposed efficiency relationship

Fig. 2    Comparison of measured and computed outflow sediment concentration for Test 2

Table 1    Summary of computed and measured trap efficiencies for sedimentation tests

Fig. 3    Plan View of Flume