Committee on Cost Savings in Dams
Bulletin : “Cost Savings in Dams”
3.4. Impact of sedimentation on Dams Designs
This chapter refers to high and low dams, and to large or small reservoirs.
High rates of sedimentation in many reservoirs and increased attention to long term sustainability have emphasized the importance of this problem. Some relevant data have been misused, especially by anti-dam organizations, for implying the early end of the worldwide usefulness of dams. It is thus important to evaluate the true impacts of reservoir sedimentation and the cost efficiency of various solutions for mitigating them. The main problems are:
– The loss of storage
– The damages to turbines
– The impact on the downstream river bed.
The total world reservoirs storage is about 7.000 km² of which 3.000 km3 dead storage for hydropower; from 4.000 km3 of live storage most if devoted to hydropower and 1.000 to irrigation, drinkable water or industrial water: part is in multi purpose dams. The number of irrigation dams is higher than for hydropower but the average storage is much lower.
The annual sediment load of all world rivers is evaluated at between 20 and 40 billion tonnes for a water inflow of 40 000 km3, i.e. an average sediment content of 0.5 to 1 ton per 1 000 m3 of water but it varies enormously according to the river and to the discharge throughout the year. All rivers are not dammed and all sediments are not trapped in reservoirs but several dams may be impacted along the same river. The cumulated sediment storage in the worlds reservoirs has been very roughly evaluated as typically 1 000 km3 for dams 40 years old, i.e. per year in the range of 20 to 30 billion m3, or 0.3 to 0.5% of the total storage.
However this does not mean that all reservoirs will be completely filled by sediment within 200 years because a reservoir filled in 50 years will not be filled 4 times in 200 years and many dams will not be filled in 500 years.
Most siltation is for hydropower dams, partly in dead storage. The loss of power supply is not proportional to the loss of storage. The annual loss of power supply would therefore seem to be in the range of 0.2% of a total investment of about USD 1 000 billion, i.e. a loss of USD 2 or 3 billion per year.
The annual loss of storage at irrigation reservoirs, possibly 5 to 10 billion m3, impacts directly on the irrigation capacity. For an average investment of USD 0.5 /m3, the annual loss may be in the range of USD 3 or 5 billion.
There is also the cost of downstream damage and, for possibly 10% of hydroplants, losses of power supply and the cost of maintenance and turbine wear.
The total annual loss linked with sedimentation problems thus seems to be between USD 10 and 15 billion and therefore warrants great consideration. It should however be compared with the annual overall cost and benefit of dams, i.e:
– Some USD 40 billion for construction and USD 20 billion for operation, maintenance and upgrading (assuming 1% x USD 2 000 billion), i.e. a total cost in the range of USD 60 billion.
– Some USD 150 billion of electric power supply (assuming 3 000 TWh x USD 0,05) and USD 50 to 100 billion other benefits including especially food by irrigation from dams for over 500 million people.
For the world’s dams the total annual impact of siltation of USD 10 to 15 billion should thus be compared with the overall annual costs of USD 60 billion and overall annual benefits of over USD 200 billion.
Great care with siltation is anyway justified for the following reasons:
– The costs are high and a better knowledge based upon experience of various countries favours efficient mitigation.
– The sedimentation risk is high in many areas where many future dams will be built.
– Beyond economic optimisation, the care of long term sustainability is a key element of future dam acceptability.
Spending up to USD 10 billion/a for existing or new dams for mitigating sedimentation impacts appears reasonable. The cost of some new dams may be increased by well over 20% for mitigating high siltation risks.
The risks from sedimentation vary enormously with:
– The average sediment yield of the river, which is under 0.1 ton per 1 000 m3 of water for over half of dams, and close to or over 5 tons per 1 000 m3 for some 10% of them.
– The hydrological size of the reservoir, i.e. the ratio of the storage capacity to the annual discharge.
– The size and quartz content of materials.
– The environmental conditions downstream.
The problems and solutions are not the same for:
– Irrigation dams where the key problem is the loss of storage, but the reservoir may often be emptied every year. This favours siltation mitigation by sluicing.
– Low hydropower dams which are not managed in the same way as the low irrigation dams.
– High hydropower dams where the reservoir level may not vary much during the year and one key problem may be the turbines wear. The financial income is not reduced as much as the water storage.
– Multipurpose dams for which siltation mitigation may be more difficult.
3.4.2 Sediment in-flow data
Many past damages or losses from reservoir siltation were in fact linked with incorrect evaluation of the sediment volume and size and could have been much better mitigated by tailored design and operation.
The evaluation of sediment inflow is difficult because the sediment content varies considerably for a same river according to the flow and the season. It may be much higher during the first part of a flood and the total yield may vary over the years. It is thus essential to devote enough time and cost to this problem and to choose the right method and place of measurement. However the result cannot be very precise and the design should take into account some uncertainty. Better data will be obtained during the first years of operation and will favour optimised reservoir management.
3.4.3 General lay out of structures
Where the siltation problems may be important, these problems should be taken in account for the general lay out of the various structures, such as; water intake, bottom gates, spillways and even possibly for the choice of the dam site and of the overall river utilization. Placing the main reservoir in a tributary is sometimes the best solution.
Well adapted hydraulic model tests may be very useful for optimising layouts and levels of gates and for studying reservoir management options.
The rate of reservoir siltation is more important in reservoirs which store only a reduced part of the annual discharge. For relevant irrigation reservoirs which may be emptied every year at the end of the dry season, it is often advisable to use sluicing, i.e. to keep the reservoir empty during the first part of the flood season, operating the river as close as possible to natural conditions and thus avoiding most siltation. The extra cost for relevant bottom outlets is usually a few per cent of the dam cost but much higher if constructed only after some years of operation. This solution is flexible, environmentally friendly and may apply to many dams. The optimum size of the sluicing gates deserves a specific study for each dam according to hydrology, dam purpose and sediment data. For many dams these gates may be used for sluicing and flushing.
3.4.5 Diverting floods
Sluicing cannot apply to reservoirs kept virtually full all the time. Diverting part of the flood discharge (when significantly siltated) by a tunnel by-passing all or most of the reservoir length may then be cost efficient especially if the rock quality favours unlined or partly lined tunnels and if the slope or curves of the rivers allow rather short tunnels. Another advantage of this solution is that most of the relevant investment may be made during operation when the need can be more precisely evaluated. Alternatively such tunnels may sometimes be used also for diverting floods during construction. Such solution is well developed in Japan. Canals may be used instead of tunnels for low reaches of very large rivers.
3.4.6 Desilting the discharges to power houses
The wear of turbines by sediments may be high if the water head is over 20 m and very high for high heads, especially if the quartz content of sediments is high. In some past cases turbines have been abraded too badly for further use after just a few months of operation. This has resulted in high financial losses due to reduced power revenue and repair costs.
It is thus necessary to use desilting structures, such as settling basins, for avoiding silt or sand discharges to the power house. Corresponding huge structures, often underground, have been implemented. Many have had two drawbacks; they were expensive and their efficiency was questionable. Most have been made by hopper chambers with permanent flushing of sediments and with the target of withdrawing particles greater than 0.2 mm in diameter, with reduced water velocities of about 0.2 m/s over a length of about 200 m. Actually the true efficiency of hoppers has often been much less than anticipated; but the key drawback is the fact that 0.1 mm diameter particles may also erode turbines and the advisable water velocity in a more efficient desilting structure to deal with those would be in the range of 0.05 m/s. This would only appear cost effective if the reservoir itself were to be used as a desilting structure and designing accordingly. Further savings in turbines wear may be obtained by specific mechanical design and coating.
An in depth review of the usual design of desilting structures is thus essential.
Flushing increases water velocities through the reservoir for a while in order to scour and remove sediment deposits. The worldwide efficiency of this has varied considerably and the cost efficiency seems mainly linked with the following conditions:
– A ratio of reservoir volume to the annual discharge of under 20 or 30%
– A ratio of the reservoir volume to the annual sediment discharge under 20 for early sluicing, under 50 for late sluicing
– A significant river slope
– A rather narrow valley in the reservoir area
– An acceptable downstream impact.
Successful flushing thus applies mainly to high or medium reaches of rivers. The choice, design and level of gates for flushing is a key problem. For gates operating with high head, the problem of wear or cavitation with erosive sediments may be high. For low dams the gates should be close to the natural river bed. For high hydropower dams, the optimum level of sluicing gates may be 20 to 40 m under the reservoir maximum level, but is more generally dictated by the associated minimum operating level of the reservoir.
One drawback of flushing is often the downstream impact linked with significant changes in flow and sediments rates. Such drawbacks may prevent or limit many flushing opportunities. Usually flushing cannot avoid siltation in a part of the reservoir but may, after some years, reach an equilibrium between further sediment inflows and flushed sediment outflows.
Flushing in large hydropower reservoirs may be mainly for moving sediments from live storage to dead storage and thus keeping enough storage capacity in the upper part of the reservoir for daily power peaks. The width (in metres) of channels created by flushing in silt deposits is in the order of 10 q0,5 (where q is the discharge in m3/s) and may thus be 100 m or higher. It is smaller if flushing is made only after some years of silt consolidation.
For each site, the precise efficiency and the optimum operation of flushing may be difficult to define precisely at the time of design, but the cost of flushing outlets is often a small part of the main investment. Their discharge capacity will be small compared to total spillway capacity discharge, but adding them later could be difficult and very expensive.
– The total cost for creating reservoirs has been in the range of USD 1 500 billion for 4 000 billion m3 of live storage, i.e. an average cost under USD 0.5 /m3 and the cost of dredging is usually well over USD 2 /m3 of sediment. Dredging cannot thus be the standard solution. However the option should not be overlooked, for instance the use of hydro suction in small irrigation reservoirs and large purpose-made dredging equipment for large hydroelectric schemes.
– For small irrigation dams, dredging costs are reduced by avoiding pumping equipment when the reservoir head may be used up to 8 or 9 m. In Asian countries where the labour cost is low, the cost per m3 of sediment removal may be acceptable, comprising mainly the cost of pipes and small pontoons in calm water. The efficiency may be improved by special pipe inlets. There are however some possible limitations such as the pipe lengths, the length of reservoir to be desilted, the sediment depth and the water loss. The latter may be a factor as the sediment concentration in pipes is usually under 10%. This “hydro-suction” solution applies mainly to small reservoirs storing a small part of annual discharges, say 10 000 to 100 000 m3 of sediment per year. This solution may also be used for emptying settling basins instead of flushing them.
For hydropower the cost of dredging should be compared to savings in power and operating costs and not to storage saving. For instance for an high dam supplying power under 100 m head and an annual flow of 10 billion m3, 80% of which generates power, the annual energy generation is in the range of 2 billion KWh with a value of about USD 100 million.
If the annual loss of energy and damage due to siltation is 20% of this amount, i.e. USD 20 million, and the silt content of the river is assumed to be 2 tonnes per 1 000 m3, i.e. 20 million tonnes per year of which half could be stored in the reservoir, it may be cost effective to dredge 10 million tonnes per year if the cost of dredging is under USD 2 /tonne. Such costs may be reached by specific equipment tailored to the reservoir conditions. The cost may be low for dredging fine materials in calm water, using electric power, with heads reduced by 7 m or 8 m by using part of the reservoir head. The equipment may be used for many years after partial filling of the reservoir. A justified investment of dredging equipment may be 5 or 10% of the overall investment for the dam and power house, i.e. USD 50 million for dredging. In such cases the volume of water needed for dredging will be about 5 times the volume of sediment.
This solution could be studied for many future large hydro schemes as an alternative to costly desilting basins, which also require considerable advanced investment costs. Dredging equipment may also be designed according to the siltation measured precisely during the first years of reservoir operation. The impact on the environment is much lower than from flushing. It may apply to over 10 millions m3 sediments per dam per year. Holes could be left through the dam or banks for dredging pipes in order to reduce the dredging head required.
3.4.9 Choice of solutions
There is no standard solution and various alternatives should be studied allowing for uncertainty in their evaluation. The study should not be limited to 20 or 50 years and the long term impact and decommissioning conditions should also be considered. Associating several solutions may be the best choice. Some suggestions are given below for three usual likely future schemes.
– Most existing “large dams” are low irrigation dams built in Asia with rather small reservoirs, say a few million m3 or some tens of million m3. Many more will be built worldwide. If the reservoir stores most of the annual flow, even with sediment ratios over the world average of 0.5 or 1 tonne per 1 000 m3, the useful life of the reservoir may be well over 100 years. But for reservoirs storing 10 or 20% of the annual flow, the reservoir may be half siltated in few tens of years. In many countries, especially in Asia, the floods, and most siltation, happen in just the few months of the flood season. Keeping the reservoir empty during the first part of the flood season may more than double the reservoir life. Using hydro suction may further increase its useful life. This requires, at the construction stage, the provision of bottom gates for sluicing and/or pipes through the dam for hydro-suction. This may be also justified if there is a serious concern about the importance of siltation yield. The sluicing, often done in China, could be of use for most irrigation reservoirs prone to siltation.
– In steep parts of rivers, many hydropower schemes are made using a low gated dam, a headrace tunnel and a high head power house. The small dam reservoir is flushed for evacuating the bed load but the water diverted to the power plant often includes sand particles up to 1 or 2 mm which may cause severe damage to turbines and increase by up to 20% or more the overall cost per KWh. Large underground desilting chambers with continuous flushing have been used with a water velocity of 0.20 m/s along 200 m for limiting the particles diameter about 0.2 mm. Their efficiency has been often less than anticipated and particles of 0.1 mm may anyway be harmful, especially if including a high quartz content.
It may be more cost efficient to use the main dam reservoir itself for desilting. For instance if the flow to the power house is 50 m3/s, a natural desilting basin 500 m long may be created in the river between a downstream gated dam and an upstream ungated dam. The cross section of the reservoir, for a water speed of 0.05 m/s would be at least 1 000 m² (50/0.05), i.e. a reservoir close to 1 million m3 and a dam 15 to 25 m high. A diversion tunnel 500 to 1 000 m along the river with a cross section of about 50 m² would bypass the basin for discharges above 50 m3/s and would be controlled by an upstream gate. It would divert nearly all bed load and a large part of annual silt and sand. The sediments deposited in the basin should be flushed in a few days per year, perhaps at weekends. For a few days per year when floods exceed the capacity of the diversion tunnel, the diversion tunnel and the head race to the power house could be closed and the floods sluiced through the basin.
– Many future large hydroelectric schemes will be based upon dams 50 to 200 m high on large river and a power house using heads in the range of 100 m with a large discharge. If the reservoir stores less than 10% of an annual flow of several billion m3 the reservoir may be siltated in a few dozens years or less. The scheme is likely to be designed for operating full time for a few months of the rainy season and for supplying mainly daily peak power during the dry season. It may also be necessary to avoid silt or sand through the turbines. Large artificial desilting basins are very expensive for the large flows corresponding to these schemes and they may not retain particles less than 0.2 mm. Using about 1 to 2 km of the downstream part of the reservoir as a desilting basin, with cross sections in the range of 10 000 m², say for instance 25 m deep and 400 m wide, may be much more cost effective. Such schemes could feature specifically designed dredging equipment. In addition the early investments and uncertainties associated with a large artificial chamber are considerably reduced. Flushing by gates 25 to 50 m under the maximum reservoir level may also be efficient, at least for keeping enough storage for peaking capacity but may be prevented or very limited by environmental reasons downstream. The corresponding loss of water will be usually higher than for dredging. Associating dredging and exceptional flushing may be the best solution with a low early investment and with the later dredging investment adapted to the true siltation rates.
3.4.10 Conclusion for siltation mitigation
Reservoirs sedimentation is not globally as detrimental as sometimes claimed but is a serious problem for many reservoirs, especially in Asia where most future dams will be built. Beyond economic optimisation, long term sustainability requirements favour mitigation measures. They may vary significantly with dam site, purpose and reservoir operation. The sedimentation management may impact not only the design of the structures but also the choice of dam site and even the general layout. The studies should take care of some unavoidable uncertainty in the evaluation of the siltation.
Sluicing may be efficient for many irrigation dams. Permanent diversion tunnels, flushing and/or dredging may be cost efficient according to specific local conditions. Desilting the water inflow to power houses deserves an analysis of the true efficiency of existing structures. Using the reservoir itself for desilting may be cost efficient. Combining solutions is often suitable.
A long term sustainability analysis is advisable even if the risk is rather low. It may justify early small investments, such as bottom gates increased capacities, pipes through dams for future dredging, intakes for possible tunnels, etc. All these will considerably reduce future investments by adjusting siltation treatment to the actually measured siltation.
Progress in siltation mitigation will improve by obtaining more data and information on the true cost and efficiency of various worldwide solutions that have been used successfully and even more from difficulties and relevant changes and modifications that have had to be made to permanent works.