Sedimentation Management in Hydro Reservoirs


21 févr. 2014 (il y a 3 années et 1 mois)

64 vue(s)

By S. Alam

Sedimentation Management in Hydro Reservoirs

Keynote speech by S. Alam

Water lndia

Delhi, February 3
4, 2004 .


"Among the many sessions ofthe Third World Water Forum held in Kyoto, Japan in March
2003, there was one t
itled 'Sedimentation Management Challenges for Reservoir
Sustainability'. Two main messages emerged from that session:

Whereas the last century was concerned with reservoir development, the 21 st century will
need to focus on sediment management; the ob
jective will beto convert today's inventory
sustainable infrastructures for future generations

The scientific community at large should work to create solutions for conserving existing
water storage facilities in order to enable their functions to
be delivered as long as possible,
possibly in perpetuity."

We must say that the above messages summarizes very weil the challenge that all engineers
involved in storage reservoir sedimentation management should keep in mind and work
towards finding reas
onable solutions.

For new projects this would mean:

Having good knowledge of the watershed sediment yield and ifnecessary and/or possible
propose long
term solutions which would either maintain the present sediment yield if
considered satisfactory , if
not, take actions which will gradually reduce the sediment yield
from the watershed area.

Confirm predicted sediment yield by actual field measurement of the total sediment load
(sand, silt and clay) that is being transported by the river.

Amongst othe
r criteria used for determining the proposed dam and reservoir
characteristics also include the impact of the expected sediment load on short and long
evolutions of the project and possible remedial measures.

Analyze reservoir sedimentation manageme
nt strategies by using numerical rnodel like
RESCON proposed by the World Bank

Incorporate in the dam design possibilities of future structural modifications or retro fit
of structural arrangements to alleviate the problems that might be created by a reservoir full of

For existing projects with severe reservoir sedimentation, it will be necessary:

To develop solutions which will enable to stop its further
degradation and carry out
remedial measures which will enable restorations of ample power generation in a manner

In this respect we think that the very small scale physical modelling technology developed
for studying the diversion concepts of
the bulk of the Mississippi River sediment load into the
wetlands in its delta could eventually be used for:

* Reproducing the actual known historical sedimentation process of the reservoir

* Develop practical and viable alternative solutions ofreservoi
r rehabilitation by flushing and

hydro suction.

* Verify the potential for optimization of water consumption for the combined flushing and

suction operation

* Limitations of flushing operation during the flood flows and hydro
suction during the


or low flows.

We believe that such design procedures have great potential for reducing the initial
construction cost, increase the project life and defer costs for structures not necessary at the
project inception and eventually completely elimi
nate structures like the de
sanding structures
for projects with a large reservoir. Where it is not required when the reservoir is free of
sedimentation and with the reservoir full of sediment the concentrations during the flood
flows are so high, that the
sander is no longer capable of trapping sand particles adequately
to prevent turbine abrasion, and plant shutdown becomes often necessary .


1 Quoted from the foreword by lan Johnson, Vice President, Sustainable Development, World
Bank for
the World Bank Publication 'Reservoir Conservation by Alessandro Palmieri,
Farhad Shah, George W. Annandale and Ariel Dinar, June 2003.

1. Watershed sediment yield assessment and its management

During the project design phase thorough investigation of
the project watershed regarding the
parameters such as nature of the soil, intensive useofland for agriculture, pastures, systematic
forestation, intensity of rainfall etc., should be carried out and documented. Based on the
distribution of the above p
arameters over the entire watershed area its long
term annual
sediment yield in tons/km²/year may be established. As general information we may indicate
that the areas with highest sediment yields may produce more than 10,000t/km²/yr and the
minimum value
could be less than 50 t/km²/yr. The precise assessment of the sediment yield
of a watershed is therefore not an easy task. Expert advice and in situ measurements may help
to obtain more reliable indications.

Another way of estimating watershed sediment yi
eld from a given particular geographical
area will be to carry out regular and precise bathymetric survey by using multi
beam echo
sounding technique in existing storage reservoirs and by collecting sediment samples at
selected locations within the reservo
ir. This way it will be possible to have fairly complete
information on the average annual or periodic volume and/or weight of sediment load
transported by the river into the reservoir. At the same time it will be possible to obtain
information such as: Se
diment particle gradation, their mineralogical composition and sand
particle form coefficient.

Based on the preliminary findings during the design stage it may either be concluded that the
sediment yield from the watershed is compatible from the stand poin
t of long term reservoir
sedimentation rates and the project life or that in order to assure a reasonable project life it
might be possible to create appropriate vegetal cover which will reduce the watershed erosion
characteristics, but this may not be pos
sible for some projects.

2.Sediment sampling

Generally the gauging stations used for discharge rating and sediment sampling are located in
the same area where the river flows are fairly uniformly distributed and the reference channel
section is constan
t over time. Discharge measurement and sediment sampling are generally
carried out at regular intervals, often every 15 days for mobilization and cost reasons.

In the past we have discussed the difficulty of assessing sediment data sufficiently accurately

by using the conventional depth integrated sediment sampling method at periodic intervals.
Regular sampling intervals combined with a fairly small number of verticals (Fig. 1) may in
our opinion induce some error in the assessment of total sediment load.
Perhaps this is one of
the reasons for which the actual reservoir sedimentation rates are often much higher than that
predicted during the initial evaluation of the reservoir sedimentation rates and the ultimate
Project life.

Typical depth integra
ted sediment sampling verticals in a river cross section

Figure 2 shows a typical correlation between the river discharge variations in a fairly large
river like the Mississippi River at Old River Control and the corresponding total sediment
load and
sand load variations based on daily sediment sampling. This figure shows clearly
that if the sediment sampling is carried out every 15 days interval that is say December 1 and
15 the average sediment load estimated will be very different than if the sampli
ngs were
carried out on December 10 and 25.

To avoid such uncertainty a sediment sampling arrangement has been developed at Old River
Controlon the Lower Mississippi River where samples are taken twice a day and at three
different structures: Sidney A. Mu
rray Hydroelectric Station and two US Army Corps of
Engineers flow diversion structures. In the past a paper has been published to give full details
of this system.²

We will therefore only briefly describe the main concept ofthe system and the need for
wing the daily variations of the sediment concentrations as shown in Figure 2 and its
importance in assessing correctly the total sediment load.

Mixture of water and sediment is pumped from the highly turbulent flow areas such as: From
the top of the turb
ine runner chamber and the energy dissipation areas of the flow diversion
structures and hand samples are collected twice a day from easily accessible installations
adjacent to the structures. Experience has shown that the vertical mixing of the total sedi
load is complete and samples collected represented the total sediment load including the
coarsest sand particles found on the river bead.

Fig. 2
Correlation between the river discharge and the sediment load variations

These sampling stations hav
e now been in operation for 13 years and have proven that their
operation is simple safe and very reliable. Only one technician is required to collect and
analyze all the samples and carrying out sediment analysis using standard United States
Geological Su
rvey (USGS) laboratory procedures.

It is also very important ta know the mineralogical composition of the sand particles.
Sediments with very high content of Quartz sand (85% or more) are very detrimental to the
turbines. Himalayan rivers often transport
sand with high Quartz content, so particular
attention must be given in the design of the structural arrangements, which will reduce the
risksas much as possible entrainment of such material into the turbine flow. This aspect is of
prime importance ifthe p
roject designers intend ta use de
sanding structures for this purpose.

It would be of interest to mention that at the Jhimruk Run
of river Hydroelectric Station in
Nepal, due to the very high concentration of Quartz sand in the river flow (maximum
t concentration recorded 23,760 PPM or 23.76 kg/m
) has produced very severe
abrasion of turbine cover, guide vanes and blades short time after starting the project
operation.³ ln this case the de
sander was designed to retain sand particle sizes equal or
greater than 0.09 mm. This would tend to indicate that the use of de
sander in certain cases is
not a viable solution. The project designers should therefore weigh carefully the need and the
cost benefit ratio of such structures before retaining them as ne
cessary project component.

For information it is interesting to know the average annual sediment discharge of the 10
major rivers of the world

River and country

Average sediment discharge, 10

1.Ganges/Brahamaputra, India


2.Yellow, China


3.Amazon, Brazil


4.Yangtze, China


5.Irrawaddy, Burma


6. Magdelane, Colombia


7. Mississippi, United States


8. Orinoco, Venezuela


9. Hungo (Red), Vietnam


10. Mekong, Thailand


"Geology, slope, climate, drain
age density, and patterns of human disturbance all affect
sediment yield, and no single parameter or simple combination of parameters explain the wide
variability in global yields."

3. Sediment management practice for various types of dam and reservoirs

Depending on the project characteristics we may divide the dams and reservoirs into the
following main categories:

3.1 Run
of river dams with no or relatively small reservoirs

3.2 Run
dams in series with excavated power discharge conveyance channe

3.3 Large dams and reservoirs with minimum pool level variations

3.4 Large dams with long reservoirs and significant variation in pool levels

3.5 Reservoirs filled with sediment

3.1 Run
river dams with no or relatively small reservoirs

In the abs
ence of reservoir water is supplied directly to the turbine water intake through a de
sanding structure. The power plant capacity is based on average flow of the stream and often
combined with a relatively high head.

Efficiency of sediment management in s
uch projects depends on the performance of the de
sanding structure. In mountain streams during high discharge the sediment source is often
from land slides and bank failures, causing sudden rise in the sediment concentration. For this
reason it is found t
hat concentrations in certain Himalayan streams could be very high as
much as 40,000
80,000 PPM or40 to80kg/m
. It has also been observed that performance of
sanding structures for sediment concentrations in excess of 2,000
5,000 PPM are not
ctory and often it is necessary to shutdown the plant. However, it is often too late and
sediments have already reached the turbines long before the decision to shutdown the plant
has been taken. This is confirmed very often by reports of serious equipment
problems from such plants. In such cases a real time and continuous sediment concentration
detector could be very useful for timely warning and plant shutdown.

For run
river projects with small reservoir it is necessary to equip the diversion
dam with
adequate gate size and number which will allow establishment of pre
dam stage discharge
conditions at the dam site. This will allow efficient flushing of the sediments accurnulated at
the upstream end of the upper pool under normal operating cond
itions. Such flushing is
necessary otherwise permanent aggradation of the river bed upstream of the diversion dam
will occur.

Marsyangsi Hydropower Station (Nepal) upper pool sedimentation profile

The sediment build up on the upstream of the di
version dam of the Marsyangdi Hydropower
is a good example where a permanent loss of upper pool storage volume has
occurred. Figure 3 shows the rise in the river bed by about 14 m out of a total depth of 21 m at
the dam. This situation is responsi
ble for producing at times extremely high sediment
concentration in the flow entering the settling basin (up to 80,000 PPM or 80 kg/m
) reducing
its trap efficiency and causing severe equipment abrasion. This is often the case with
reservoirs full of sedi
ment, at the onset of flood the saturation sediment concentration is
attained for which no de
sanding structure can be designed.

3.2 Run
river dams in series with excavated water conveyance channel to the plant

On the Rhône River in France
there is
a series of low head water diversion dams where the
plant discharge is supplied through excavated channels almost parallel to the river. The
diversion dams are spaced in such a manner that the backwater from the downstream dam
reaches the tail water level
of the plant upstream. During the flood flows when the total river
discharge is much in excess of the plant discharge, sediment is flushed out of the upper pool
to the lower pool through the gated diversion dam with a combination of flushing and
However, these projects are in relatively low sediment yield area, with sediments
containing very little or no quartz sand and the total head in the range between 20 to 25 m,
hence no abrasion problem. Average discharge is around 2000 m
/s. So we may conc
lude that
where the average river discharge is high, sediment yield not excessive, low head run
projects may be a good solution for sediment management.

3.3 Large dam and reservoir with minimum pool level variation

A large number of world hydr
o dams are in this category. The total sediment inflow in this
case is stored in the reservoir and the dam height and volume of storage is designed to assure
fairly long project life, 100 to 150 years. In such cases the correct assessment of the average
nual sediment load that is going to be transported into the reservoir is very important. After
putting the project into operation regular reservoir sedimentation survey should be carried to
compare the actual sedimentation rate with the predicted volumes.
Also the nature of the
watershed soil erosion protection should be regularly evaluated and if necessary adequate
measures should be taken to prevent its de gradation and if possible work towards its funher

The reservoir sedimentation process
in this case will create a fairly fiat top
set bed slope of
coarser material close to the pool surface, with a steeper tore
set slope, with tirne the fore
slope slowly and gradually encroach upon the earlier deposits of the finer materials (silt and
ay) as shown in Figure 4.

Fig. 4
Schematic representation of the reservoir sedimentation process with minimum pool
level variation

If the bulk of the sediment load is very fine, silt and clay, then the possibility of its evacuation
with the power fio
w should be attempted. If suitable the extraction of the coarser material
from the upstream end of the reservoir for construction purposes is also a possible solution.
Diversion of sediment from the upstream end of the reservoir to the downstream of the da
will reduce reservoir sedimentation, and may be economically justified if the reservoir is not
too long. In this regard we might site the example of the 86 m high Asahi Arch Dam in Japan

where a bypass tunnel of about 2.350 m long and diameter 3.80 m ha
s been used for diverting
the sediment load from the upstream end of the reservoir to the downstream of the dam.
According to the available information the system has produced very satisfactory sediment
diversion and has reduced the reservoir sedimentation
by about 83,000 m
after the first year
of operation. The sediment material diverted contained coarse sand and gravels and this has
produced about 550 m
of tunnel invert erosion with an average depth) of 0.62m.

If it is regularly necessary to open the s
pillways during flood flows then some fme sediment
deposited earlier may also be added to the outgoing fiow with the help of panning, dredging
and/or hydro
suction. The combined effect of the aforementioned activities could help to
increase significantly
the project life.

3.4 Large dam and long reservoir with important pool level variation

A large number of multipurpose dams (hydro, irrigation, urban water supply and navigation)
are in this category. In this case also the total sediment inflow is store
d into the reservoir and
the dam height and volume of storage is designed to assure long project life. However, due to
the important pool level variations the top
set bed slope will be steeper and the fore
set bed
slope will move much faster (Fig. 5) towar
ds the dam because at the onset of the flood flows
the pool level will be low and the flow velocities over the steeper top
set slope will be high
and create massive erosion and transport of sediment towards the dam each year. Such annual
operation may crea
te sand transport into the power and irrigation intakes long before the
whole reservoir is full of sediment. Because of the high storage dam and the very long
reservoir length it is practically impossible to achieve worthwhile results in removing
by flushing. The eventual problems related to sediment removal in such projects
should be considered at the site selection stage and also during the design phase. Amongst
possible solutions alternatives such as diversion of the sediments at a convenient lo
along the upper part ofthe reservoir to an adjacent valley or tributary and/or use of multiple
smaller diversion dams that will serve as sediment excluder and supply relatively sediment
free water to the main reservoir for power generation and irrig

Fig. 5
Schematic representation of the reservoir sedimentation process with important
poollevel variation

3.5 Rehabilitation of reservoirs filled with sediment

There are many dams and reservoirs around the world where long before the pred
icted project
life the reservoir sedimentation has reached such a stage that adequate power generation, or
irrigation and urban water supply is no longer possible. Remedial measures which would
eventually restore at least partially their initial power gene
ration and irrigation and urban
water supply capacity is worth trying.

Type of solution will depend on the individual project characteristics and the potential
environmental impacts the remedial measures might cause to the river downstream. Use of
nt such as: drag line, dredging and hydro
suction on smaller scales to solve the
problem excessive sedimentation, on relatively smaller reservoirs have been often used with
success. But for large dams and reservoirs such solutions will require considerable

mobilization, time and cost. So before undertaking such an operation it will certainly be
convenient for the decision makers to have a good understanding of the feasibility of the
proposed solutions. There is nothing better than a reliable physical model
simulation of the
actual process of sediment removal and its impact on long
term behaviour of the sediment
movement and deposition over the entire reservoir length .and also the down stream
movement of the removed sediments during flood flows or during low
flows. For large
reservoirs this might mean reproduction ofcombined river and reservoir lengths of 20, 50 or
100 km.

If we use conventional movable bed physical hydraulic models for such purpose the size ofthe
model will be huge, its cost prohibitive, an
d the testing time and conditions will simply be

Recently we have developed a very small scale physical model of the Mississippi River delta
over a length of 100 km for studying large scale water and sediment diversion in to the
marshes for
recreating land. Because of the existing Mississippi River levee system annually
about 210,000,000 tons of sediment is getting lost in the Gulf of Mexico and at the same time
Louisiana is loosing about 80km²/year. The geometrical and sedimentation scales a
re such
that it can reproduce 100 years evolution in 50 minutes. This renders the models testing
conditions very easy to manage. The model was initially proposed as a highly qualitative
model, but in the light offairly accurate reproduction of the river se
diment transport and
distribution patterns in the various passes and distributaries, and also the total sediment
balance after 100 year operation we can say that it is may be considered also as almost
quantitative in certain respect.

A similar model could
be used for simulating firstly the actual known historicalreservoir
sedimentation process, i.e., satisfactory reproduction ofthe progression of the top
set bed
slope and the fore
set bed slope through the reservoirs over the actual number of years of
ation and then study sediment removal project concepts and procedures for partial
rehabilitation of the reservoir.

This would mean that it will be possible ta recover sufficient area and volume ofthe reservoir
allowing normal power generation without the
risks of abrasive sand getting into the power
intake and adequate supply of irrigation water year round.

As in most cases the maximum percentage of sediment is transported during the onset of
flood flows, detailed information regarding the major flood hyd
rographs and the estimate of
corresponding total sediment load deposited in the reservoir should be gathered for accurate
calibration of the model.

To explain the basic concepts of reservoir rehabilitation design study we have considered the
well known ca
se of Salal Dam
reservoir with a total storage volume of 96 million cubic yard.
Figure 6 shows a schematic representation of the project layout. The reservoir is full of
sediment, to the water surface level at the upstream end and to the spillway crest le
vel at the
downstream end.

Fig. 6
Schematic layout of the Salai Dam reservoir completely filled up with sediment

Average annual sediment load is approximately 30,000,000m
of which about 25% is
and during the flood flows the concentrations are very high, most probably much more than
10,000 PPM or 10 kg/m
. So the spillway and the power intake are continuously passing sand
causing severe abrasion damage to the spillway concrete structure and
the turbine equipment.

Fig. 7
Schematic longitudinal section of Salal Dam upper pool

As fist approximation a storage volume equivalent to at least 4 times the annual sand transport
volume, i.e.,
about 30 million m
may have to be created in the upper pool by using a
combination of means such as: extraction, flushing, dredging, panning (putting bed sediments
into suspension during flood), and hydro
suction ,etc. Removing sediments buried many year
within the reservoir will also need especial care in disposal method, to avoid negative
environmental impact for the water users downstream. Thus it might be necessary to carry out
the rehabilitation work over several water years.

Assuming that during t
he low flow period the sediment volume removed from the reservoir
may be stored in the river bed between immediately downstream of the spillway and the
powerhouse tailrace tunnel outlet, in this way it will then be progressively carried away by the
y discharge with reasonable concentration.

During the flood period when the flow velocities are high it might be possible to remove bulk
of the sediment volume from the reservoir by panning. Sheet
pile groins judiciously located
(to be determined by model
studies), will also create turbulence and remove and direct the
heavier sediment particles towards the spillway maintaining the power intake area reasonably
free of sand.

The small scale physical model will confinu the feasibility of the project and then
optimization of the essential features such as:

Configuration, dimensions and the volume ofthe reservoir excavation that will be able to
trap efficiently the total sand load transported by the river during the flood flows, for annual
flood, ten ye
ar flood and 20 year flood frequencies.

What are the best maintenance procedures for removing sediment during the low flow
period and

average flows and flood flows

Best use of the river flow velocities to transport and remove the bulk of the sediment
from the reservoir .

Best way to keep the flow to the power intake, free of sand under all river discharge

Optimization of cost/benefit ratio of such operation

4. General conclusion

Reservoir sediment management, especially in sed
iment rich areas of India and other
countries around the world is becoming more and more a major problem for the hydro
projects. It is therefore important that aspects related to improved reservoir sedimentation
management is better understood and practice
d. We would strongly recommend that:

Concern about reservoir sedimentation becomes an integral part of design standards, so
that hydro and storage dams in sediment rich areas become as sustainable as possible.

Project operators and designers should tr
y to use known technologies and also when
possible advance the state of the art in sediment management by using innovative ideas and
new technology.

Due attention must be given to all the parameters related to sediment source, its
transportation and depo
sition patterns in the reservoir.

Try to develop reservoir storage volume mailtenance methods adapted to local conditions


1. Alessandro Palmieri
Farhed Shah
George W. Annandale. Ariel Dinar ., Reservoir
The World Bank, Jill
le 2003

2. Sultan Alam, Cecil Soileau and Ralph L. Laukhuff ., Sediment Transport Assessment in the
Old River Control Area of the Lower Mississippi River
Waterpower '93 Proceedings of the
International Conference on Hydropower

3. Shailendra Basnyat ., P
roceedings, Optimum use of Run
river hydropower schemes,
Seminar in

Trondheim, Norway, June 1999.

4. Gregory L. Morris, Jiahua Fan ., Reservoir Sedimentation HandJook, 1997

5. Gyanendra Prasad Kayastha ., Proceedings, optimum use of Run
river hydr
schemes, Seminar in Tronfdheim, Norway, June 1999.

6. Valérie Chabrier, Alain Comtet, Jacques Lovenq ., Evaluation of 50 year development on
the Rhône valley, Rehabilitation of the old river at Pierre Bénite., ICOLD, Beijing Congress,

7. Mino
ru Harada, Hiroshi Morimoto, Tetsuya Kokubo ., Operational results and effects of
sediment bypass system. lCOLD Beijing Congres s, 2000

8. S. Alam ., Improving sedimentation management using multiple dams and reservoirs; The

International Journal on Hydr
opower and Dams, volume nine, Issue 1,2002

9. V.K. Vanna., Virendra Johri ., P roceedings, Removal of silt through hydro
suction at Salal
Dam, India.lnternational Conference HYDRO 2003