Sedimentation Problems with Dams

Excerpt from Silenced Rivers: The Ecology and Politics of Large Dams, by Patrick McCully, Zed Books, London, 1996

All rivers contain sediments: a river, in effect, can be considered a body of flowing sediments as much as one of flowing water. When a river is stilled behind a dam, the sediments it contains sink to the bottom of the reservoir. The proportion of a river’s total sediment load captured by a dam – known as its "trap efficiency" – approaches 100 per cent for many projects, especially those with large reservoirs. As the sediments accumulate in the reservoir, so the dam gradually loses its ability to store water for the purposes for which it was built. Every reservoir loses storage to sedimentation although the rate at which this happens varies widely. Despite more than six decades of research, sedimentation is still probably the most serious technical problem faced by the dam industry.

Professor K. Mahmood of George Washington University in Washington, DC, "roughly estimated" for a 1987 World Bank study that around 50 cubic kilometres of sediment – nearly one per cent of global reservoir storage capacity – is trapped behind the world’s dams every year. In total, calculated Mahmood, by 1986 around 1,100 cubic kilometres of sediment had accumulated in the world’s reservoirs, consuming almost one–fifth of global storage capacity.

The rate of reservoir sedimentation depends mainly on the size of a reservoir relative to the amount of sediment flowing into it: a small reservoir on an extremely muddy river will rapidly lose capacity; a large reservoir on a very clear river may take centuries to lose an appreciable amount of storage. Large reservoirs in the US lose storage capacity at an average rate of around 0.2 per cent per year, with regional variations ranging from 0.5 per cent per year in the Pacific states to just 0.1 per cent in reservoirs in the northeast. Major reservoirs in China lose capacity at an annual rate of 2.3 per cent.

Apart from rapidly filling their reservoirs, sediment–filled rivers also cause headaches for dam operators due to the abrasion of turbines and other dam components. The efficiency of a turbine is largely dependent upon the hydraulic properties of its blades, just as an aeroplane depends on the aerodynamic properties of its wings. The erosion and cracking of the tips of turbine blades by water–borne sand and silt considerably reduces their generating efficiency and can require expensive repairs.

Erratic Mud

To make a meaningful economic forecast for a planned dam, it is necessary to be able to predict its sedimentation rate with reasonable accuracy. However, it is extremely difficult to estimate how much sediment will be trapped by a reservoir. Collecting data on sediment discharge is even more expensive and difficult than gathering streamflow data, and so there is little reliable information available on the sediment carried by the world’s rivers. Sediment flows vary widely both annually and seasonally over time – far more than water flows – and so calculating an annual average needs a long run of data. According to Mahmood, dam planners should ideally have sediment statistics going back over a period equal to at least half the projected life of the dam. Such records, however, are available only in exceptional cases. As with river flows, the variability of sediment yield is greatest in arid and semi–arid climates – where the data tends to be sparsest.

The amount of sediment carried into a reservoir is at its highest during floods: in the US, for example, commonly half of a river’s annual sediment load may be transported during only 5 to 10 days flow. During and after a particularly violent storm a river may carry as much sediment as it would in several "normal" years. Mudslides caused by earthquakes and volcanoes can also have a dramatic and unpredictable effect on reservoir sedimentation. Global warming, which is predicted to cause more intense storms, will likely increase both the unpredictability and rate of reservoir sedimentation.

Despite all the uncertainties over reservoir sedimentation, it is extremely rare for a planned project to be stopped because of a lack of adequate sediment data. In fact, time and again dam planners have made hugely overoptimistic predictions that reservoirs will fill much more slowly than they actually do. Chixoy is one of a number of very expensive hydrodams built in Central America during the 1970s and 1980s with loans from the World Bank and Inter–American Development Bank despite the very high and accelerating rates of erosion in their watersheds. These dams are now rapidly filling with sediment, leaving small, impoverished countries like Guatemala, Honduras and Costa Rica with huge debts and in desperate need of building new power plants to reduce their dependence on their white–elephant dams. A team from the US Army Corps of Engineers concluded in 1993 that sedimentation could reduce the life of the 135 MW Cerron Grande Dam in El Salvador to 30 years – compared to the pre–construction prediction of 350 years.

In India, government statistics on eleven of the country’s reservoirs with capacities greater than one cubic kilometre show that all are filling with sediment faster than expected, with increases over assumed rates ranging from 130 per cent (Bhakra) to 1,650 per cent (Nizamsagar in Andhra Pradesh). A 1990 World Bank paper on watershed development concluded that in India, "erosion and [reservoir] sedimentation are not only severe and costly, but accelerating. It is now obvious that the original project estimates of expected sedimentation rates were faulty, based on too few reliable data over too short a period."

Most modern dams are designed so that they can afford to lose some storage capacity without their performance being impaired – the part of a reservoir known as "dead storage" which lies beneath the elevation of the dam’s lowest outlet. However sediments do not build up evenly along a horizontal plane, so that some "live storage" is usually lost long before the dead storage is filled. At Tarbela Reservoir in Pakistan, for example, 12 per cent of the live storage had been lost by 1992 (after 18 years of operation) while 55 per cent of the dead storage was still empty of sediment.

The actual process of sediment deposition is unique to every reservoir and is impossible to predict accurately. In general, the coarser, heavier sediments, the gravel and sand, tend to settle out at the upper end of reservoir, forming a "backwater" delta which gradually advances toward the dam. The lighter sediments, the silt and clay, tend to be deposited nearer the dam. In 1983 the crest of Tarbela’s backwater delta had advanced to 19 kilometres from the dam: according to pre–construction predictions the delta should have been 48 kilometres behind the dam at this time. By 1991 the delta crest was just 14 kilometres from the dam.

Mitigating Geology

"Watershed management" – including afforestation and the promotion of farming practices which reduce soil erosion – is frequently advocated as the best way of cutting sediment deposition in reservoirs. While these schemes may be recommended in project plans, they are rarely implemented: dam–building agencies are usually more interested in putting their funds toward building dams than planting trees and digging field terraces.

Overall, building a dam in a valley is much more likely to increase erosion than reduce it: dams open up remote areas to road–builders, developers, loggers, farmers and miners, accelerating deforestation and soil loss. When insufficient resettlement land is made available, oustee farming families may have no choice but to clear land further up the valley or hillside. In any case, deforestation and soil erosion are both increasing rapidly around the world, and it should be assumed when dams are built that soil erosion in their watershed will increase over the projected economic life of the reservoir.