Sediment storage requirements for reservoirs

Transcription

1 Challenges in African Hydrology and Water Resources (Proceedings of the Harare Symposium, July 1984). IAHS Publ. no Sediment storage requirements for reservoirs INTRODUCTION T, C, KABELL The Hydrological Branch, Ministry of Water Resources and Development, PO Box 8132, Causeway, Zimbabwe ABSTRACT Reservoir sedimentation is a serious problem in Zimbabwe, leading to diminished draft or even total loss of live storage capacity. The correct solution undoubtedly lies in improved conservation and land use practices, but unless and until such measures are adopted, adequate allowance for sedimentation must be included in water storage works. Sediment loads are known to vary widely between different rivers and from season to season, but there is a lack of sediment load data for Zimbabwe. It is proposed that until more reliable figures of sedimentation rates are available, reservoirs should be designed to include a provision for sedimentation based on the mean annual runoff from the drainage basin. The sediment storage requirements have been estimated from surveyed volumes of silt deposited in a number of reservoirs over a period of several years. Volumes reserves aux sédiments dans les retenues RESUME Le sédimentation des retenues présente en Zimbabwe un problème sérieux qui peut causer une diminution des apports utilisables ou même la perte totale de la capacité utile. Pour combattre ce phénomène nuisible il faut certainement améliorer les méthodes de conservation et les pratiques d'utilisation des sols, mais en attendant cette amélioration on doit prévoir dans tous les projets pour l'approvisionnement en eau une réserve suffisante pour le dépôt des sédiments. On sait que la concentration des sédiments dans les rivières est fort variable et, de plus, que la quantité change d'une saison à l'autre, mais on manque de données quantitatives sur ces concentrations. En attendant que l'on dispose de chiffres plus exacts sur la sédimentation, on a suggéré qu'il fallait déterminer le volume des retenues en y prévoyant une réserve pour la sédimentation calculée sur la quantité moyenne des apports en eau du bassin versant. La grandeur de cette réserve est estimée selon les volumes de sédiments déposés tels qu'ils sont déterminé par des relevés topographiques dans plusieurs retenues pour une période de quelques années. The rapid sedimentation of water storage reservoirs is a matter of increasing concern in Zimbabwe. Expanding population creates a 569

2 570 T.C.Kabell demand for additional secure water sources, and at the same time human and livestock pressure on the land lead to deforestation and degradation of the drainage basins, which cause accelerated erosion and more rapid rates of sedimentation in reservoirs. Although improved conservation is the basic solution to the problem, in terms of safeguarding both land and water resources, there is a need for an examination of measures that can be taken to minimize the effects of sedimentation in reservoirs formed by the construction of dams. Programmes are to be implemented to measure both the sediment loads of rivers and the volumes of deposited sediment in existing reservoirs. However, it will be a long time before meaningful data from such programmes are available, and there is an immediate need for an empirical method of estimating a realistic figure for the volume required for sediment storage or for assessing the useful life of a reservoir. REDUCING THE EFFECTS OF SEDIMENTATION There are several possible means of reducing the effects of sedimentation in a reservoir. Subsidiary dams or silt traps may be constructed upstream of the main storage dam to reduce the sediment load entering the reservoir. Alternatively, the main dam may be equipped with large diameter sluice gates or valves to introduce a scouring effect, or the dam might even comprise a fully gated barrage to facilitate flushing of deposited sediments during periods of flood flow. As a last resort mechanical removal of sediment can be carried out by earth-moving plant or by suction dredging. Unfortunately, all the methods mentioned above are expensive and of limited use. Scouring only occurs over a localized area, and fresh sediment rapidly redeposits in regions that have been cleared. There appear to be only two practical approaches to the reduction of the effects of sedimentation. The first is to avoid building small storage works on major drainage basins, since these will silt up in a very short period of time. In cases where a small water supply is required and the only suitable source is a large river, it is preferable to construct an off-river storage reservoir that can be filled annually by pumping or limited diversion from the main river. The second approach must be to allow sufficient extra capacity in the reservoir so that the deposition of sediment will not unduly affect the operation of the storage for a reasonable span of years. FACTORS AFFECTING RATES OF SEDIMENTATION The rate of reservoir sedimentation depends on the sediment input and the trap efficiency of the reservoir. Both these factors are variable, being themselves dependent on many other variables. It is therefore necessary to make some simplifying assumptions in order to develop a basis for estimating the rate of sedimentation. The two major factors will be discussed separately.

3 Sediment input Sediment storage requirements for reservoirs 571 The sediment load of a river or the sediment yield of a drainage basin is a measure of the total quantity of material carried by the river. The sediment load can be divided into two parts, namely, the suspended load (wash load) and the bed load (traction load). The suspended load consists of silt and clay particles that are carried within the flood waters of the river and kept in suspension by turbulence. The quantity of sediment carried is wholly dependent on the supply of material to the river system by erosion processes, since the transport capacity of the river is always greater than the amount of wash load it is required to carry. The suspended sediment load at any point in time will depend on the topography of the drainage basin, the effectiveness of any conservation measures, the time of year (whether vegetation cover is well established in the growing season), antecedent rainfall conditions and rainfall intensity. Of these factors, only the first can be considered constant. Measurements of suspended sediment load are carried out by taking samples at numerous points across a surveyed cross section of the river, and relating the observed sediment concentrations to the gauged flow. As pointed out above, however, there are many factors that will affect the sediment load and a given discharge may be associated with a wide range of suspended sediment concentrations. It is therefore necessary to take many sets of measurements and to attempt to correlate these with a variety of controlling variables, in order to estimate the sediment concentrations during periods when no samples are collected. The bed load consists of the coarser sand particles that are moved down the river bed under the action of the flowing water as a traction load. The rate of movement of the sand particles, and hence their significance in terms of sedimentation, is usually dependant not on the rate of supply of material into the river system, but on the transport capacity of the river, which reflects the volume of water flowing and the channel gradient. Sedimentation due to sand transport is therefore not significantly influenced by catchment conservation, but is controlled by channel hydraulics. The supply of bed material is not normally a limiting factor, and the annual load will depend on the flow conditions particular to that year. The measurement of bed load transport presents many difficulties and results are frequently of dubious accuracy. Information from various areas of the world, including the size analysis of sediment deposits in large reservoirs, has indicated that bed load commonly represents only a small proportion of the total load, usually about 10%. Direct measurement of bed load is therefore not usually attempted. It is a common fallacy to assume that rivers evidencing large sand deposits on their beds are those most prone to reservoir sedimentation problems. This is not necessarily the case. It must be remembered that one may be observing material that represents only 10% of the total sediment transport, and that this may only be evident in a particular reach of a river because of the flow regime at that point. A steep section of river with a rocky bed may look clean, but may frequently transport more sediment than a flat sandy river section.

4 572 T.C.Kabell Attempts are sometimes made to make direct measurements of soil loss from a given tract of land due to erosion processes. Such information may be invaluable to conservationists and land planners, but the data obtained cannot be readily transposed to provide values of sediment yield for a drainage basin. " Not all the soil removed from an upland area will actually enter the river system, and any drainage basin will be composed of many different land types with varying rates of erosion. Reservoir trap efficiency The trap efficiency of a reservoir is a measure of the proportion of the total sediment input that is retained. When flowing water enters the still water of the reservoir basin, there will be less force to transport the bed load and the finer particles of silt will settle out in the tranquil conditions. The trap efficiency will be dependent on several factors, including the length and shape of the reservoir at a particular level of storage, the magnitude of the storage capacity in relation to the water inflow, whether or not excess water is spilling from the reservoir, the particle size and grading of the transported sediments, and the magnitude of the sediment input. Rates of deposition for different sediment sizes and theoretical trap efficiencies can be calculated, but the accuracy of the estimates is entirely dependent on the sediment load and grading data employed, which may itself be unreliable. It is therefore questionable as to how much effort should be put into trap efficiency calculations,although they can be used to indicate the range of values that might be associated with various levels of sediment input. PREDICTING RATES OF RESERVOIR SEDIMENTATION There are few data on sediment yields in Zimbabwe. Some surveys of existing reservoirs that have been subject to heavy siltation have been undertaken, and it is planned to obtain further data during the current drought when water levels are exceptionally low. A programme to monitor concentrations of suspended sediment during flood flows in the major rivers is being implemented by the Hydrological Branch of the Ministry of Water Resources and Development, with technical assistance from Hydraulics Research Ltd (Wallingford, UK). Although there are many variables that affect the magnitude of the sediment load, the most important must be the total water runoff from the drainage basin. Other factors being equal, more erosion, and therefore greater sediment yields, will occur in high rainfall areas than in low rainfall areas. Annual rainfall and runoff in Zimbabwe are highly variable, and considerable variations in annual sediment loads must occur. However, in the absence of detailed data, it is proposed that estimates of reservoir sedimentation rates can be obtained from values of mean annual runoff (MAR). On the basis of the available data, it is suggested that drainage basins can be grouped into three categories:

5 Sediment storage requirements for reservoirs 573 <a) Basins with well developed conservation measures and moderate topograph y. Mean sediment concentration = 3000 mg 1~ <b) Basins prone to erosion through poor conservation and steeper slojpes. Mean sediment concentration = 5000 mg 1~ c) Basins highly susceptible to erosion. Mean sediment concentration = mg 1~ A wide range of concentration values between or even beyond these limi_ts will occur, but where precise measurements are absent refinement of these estimates is not practicable. T?o relate these concentration values to values of sediment yield, TabLe 1 provides estimates of sediment yield for various levels of MAR and mean sediment concentration. TABLEE 1 Values of sediment yield (t km~ year~ ) associated with varr ous levels of MAR and mean sediment concentration MAR(mn) Mean sediment concentration 3000 mg 2 _ mg T mg I Although the trap efficiency of a reservoir has been shown to be dependent on several variables, the most important factor must be the capacity of the reservoir, and in particular its capacity in relation to the magnitude of annual water inflow. If the capacity is very large and the retention time of water within the reservoir correspondingly long, complete deposition of suspended sediments may occur and the trap efficiency will approach 100%. Conversely, if the capacity is very small and the retention period short, the bulk of ttie suspended sediment will pass over the spillway before settling, and only the bed load will be deposited. In the extreme case of a very low weir, or if previous sedimentation has resulted in a small effective depth, even the bed load may be transported through the reservoir. Some workers have derived expressions for estimating trap effici^ency by relating the capacity of the reservoir to the area of the drainage basin. This approach takes no account of the climatic conditions in the catchment, and it is suggested that a better generalized method of estimation can be based on the storage ratio of the r-eservoir, that is the ratio of capacity to MAR. In this present study it has been assumed that trap efficiency varies linearly from 9% at zero capacity, representing the deposition of bed load only, to 100% at a storage ratio of There is little

6 574 T.C.Kahell evidence to support this relationship, but it appears to be consistent with the scant information currently available. If it is accepted that both the sediment load and the trap efficiency can be estimated from the MAR, it is possible to calculate the rate of sedimentation from the storage ratio of the reservoir. The only variable that must be estimated independently for an individual case is the sediment concentration. Calculations of the rate of sedimentation have been carried out for suspended sediment concentrations of 3000, 5000 and mg l" 1. In all cases the bed load has been estimated as representing an additional 10%. A density for the deposited sediment of 1.7 t m -3 has been assumed. The results of the calculations are presented in Fig.l, which shows the time taken for live storage to be reduced to half the initial value, and Fig.2 which shows the decline in the storage ratio over time. YEARS FIG.l The influence of the initial storage ratio and the mean sediment concentration of the reservoir inflow on the number of years required for 50% loss of storage. Figure 1 shows that there are three distinct sedimentation situations. With extremely small storage ratios (less than 0.025), sedimentation is due almost entirely to bed load. The rate of sedimentation is actually relatively slow, but since the storage capacity is so small, the effects are marked. In the second case associated with intermediate storage ratios (between and 0.100) the trap efficiency of the reservoir is considerably increased with consequent deposition of suspended load. However, the storage capacity is limited and the rate of sedimentation is therefore very high. For large storage ratios (greater than 0.100), the trap efficiency is assumed to be 100%, causing total deposition of bed load and suspended load. However, because the storage ratio in this case is quite high, the effects of sedimentation are not as severe in terms of percentage reduction in capacity and yield. Figure 2 demonstrates the decline in the storage ratio as a

7 Sediment storage requirements for reservoirs 575 _, 1 j 1 L_,_l 1 1 L_,_l 1 1 1,_l 1 1 L-r-j 1 1_ IS YEARS FIG.2 The decline in storage ratio as a function of time in relation to the mean sediment concentration of the reservoir inflow. function of time for the three different sediment concentrations considered. The initial point is arbitrarily shown as a storage ratio of 0.15, but the curves can be entered at any selected storage ratio. These two figures can be used to predict sedimentation rates in any reservoir, within the validity of the assumptions made in their derivation. It can be seen that any storage reservoir with a storage ratio of less than 0.1 could lose 50% of its storage within 10 years if it is in a drainage basin with poor conservation practices. The "half-life" of smaller reservoirs is extremely short. CORRELATION WITH OBSERVED DATA The methods for predicting rates of reservoir sedimentation outlined in this paper may be tested against those cases where field measurements of deposited sediment are available. Table 2 indicates the rates of sediment yield and the mean sediment concentrations associated with these reservoirs The accuracy of the method of predicting rates of reservoir sedimentation proposed in this paper will depend on the values assumed for trap efficiency and for the density of deposited sediments. However, the most important factor will be the value of mean suspended sediment concentration selected for the prediction. This is the most variable of all the factors, and will vary considerably both between different drainage basins and between different seasons. Even when measurements of the sediment concentration in various rivers are available, it will be many years before a comprehensive and reliable body of data is compiled. In the meantime it is neces-

8 576 T.C.Kabell TABLE 2 Sediment yields and mean sediment concentrations associated with measured rates of sediment deposition in several reservoirs in Zimbabwe Storage dam Years Drainage area (km 2 ) MAR (mm) Mean sediment concentration (mg l' 1 ) Sediment yield (t km~ 2 year~ l ) Austral Weir Glen Avilin Weir Marah Ranche Dam Rinette Weir Jotsholo Weir sary to make an approximate and generalized assessment. RECOMMENDATIONS FOR PROVISION OF SEDIMENT STORAGE For an average suspended sediment concentration of 5000 mg 1 the annual volume transported (and deposited when the trap efficiency is 100%) is m 3 per m 3 of inflow water. Over a period of 20 years, the total volume of sediment deposited will amount to 6.5% x MAR. It is therefore recommended that this volume of sediment storage should be provided in any new major reservoir. Bottom outlets must be located to enable water to be abstracted from below this level of dead storage, because most of the sediment will be deposited in the upper reaches of the reservoir. For a dam having a storage ratio of 0.65, sedimentation will account for a loss of 10% of the total capacity over a 20 year period. For smaller storage ratios, the percentage of initial capacity lost by sedimentation will be greater over the same period. By reference to Fig.2 it can be seen that for a storage ratio of 0.15, the effective storage after 20 years will be or 58% of the initial capacity, if the mean suspended sediment concentration of the inflow is 5000 mg l -1. For a storage ratio of 0.10 the effective storage after 20 years will be or 50% of initial capacity. It is suggested, purely on economic grounds, that no storage work should be constructed that will lose as much as half of its original capacity in less than 20 years. On this basis all dams should be built to provide a minimum storage ratio of 0.10, and if the dam is in an area of extreme degradation, the storage ratio should be at least If for any reason it is necessary to build dams with smaller storage ratios, full cognisance should be taken of the probable rate of sedimentation. ACKNOWLEDGEMENTS This paper is published with the consent of the Secretary for Water Resources and Development, Government of Zimbabwe.