Professional Documents
Culture Documents
006 PR Montecarlo
006 PR Montecarlo
K. Broich
ABSTRACT
Dams are designed for a certain extreme flood event. Normally this design flood will not be
exceeded, but the probability for a failure will never completely vanish. Concrete dams can
withstand overtopping, but earthen dams will fail after a certain time of overtopping. Hence
earthen dams are more endangered than concrete dams. In this paper the conditions leading to
dam erosion of an earthen dam are discussed. A security factor for dam erosion is proposed
and the procedure for the calculation of the breach discharge described using a case study.
Finally a possible integration of this approach into a more general probabilistic concept is
given.
INTRODUCTION
This paper consists of three parts. At the beginning a method for a realistic calculation of the
breach discharge due to overtopping is described. The advantage of the proposed method is
highlighted by a case study. In the third chapter this methodology is conceptionaly integrated
into a probabilistic analysis using the Monte Carlo method in order to determine the failure
probability.
The conditions during a dam break are only known for the very rare case, that a monitored
dam fails. For all other cases of interest, assumptions has to be made concerning the
A dam will fail partially or completely due to overtopping, if the incoming flood is able to fill
up the reservoir storage above the initial water elevation, if the water level raises well above
the dam crest and if it is kept there for a certain time despite of the relaxation caused by the
action of spillway outlet and overtopping. Depending on the water depth at the crest and the
local resistance dam material will be washed out after a certain time. The local resistance is
controlled by the dam coverage, which normally consists of grass. It is known, that the
quality of the grass coverage has a great influence on the dams time to resist. This has been
observed at overtoppable river dikes. The CIRIA-experiments carried out in the United
Kingdom (Hewlett,1987) give the following curves for different types of coverage (Fig. 1).
Using this additional knowledge in connection with known dam erosion models (CADAM,
1998) improves the simulation significantly, because the critical initial time period when
erosion starts is resolved in a better way. The specification of the intial breach is still
necessary to point to the place where erosion shall start, but is has no effect on the time to
start the erosion. Consequentely the breach formation starts later and the water level has to
rise higher in order to intiate erosion. Therefore the dam’s time to resist is longer and the
breach peak discharge increases compared to the former approach.
As mentioned above dams are designed for a certain flood event, whose probability is very
low; e.g. in the Unted States PMF-events are used, in Germany the DIN 19700 requires a
flood event with a return period of 1000 years.
Overtopping can only destroy a dam, if the design flood is exceeded by a certain amount.
Using the described method this amount can be calculated by scaling the design flood until
critical conditions are reached. The ratio between the flood event needed to initiate dam
erosion and the design flood might be interpretated as security factor.
Fig. 1 : Maximum overtopping velocity vs Resistance time (CIRIA
experiments (Hewlett,1987 ))
CASE STUDY
The proposed method shall be described in more detail by a case study. The examinated
reservoir has the following characteristic:
Hydraulic parameters
Full supply water level : 403 m+NN
Flood water level : 403.2 m+NN
Reservoir volume : 19.4 hm³
Combined gated outlet/spillway construction centered in the middle of the dam (2 gates)
Ratio catchment area by reservoir surface : 666 km² / 3.59 km² = 189
Area-elevation function A = 134280 (H - 396)1.67526 [m²]
Furthermore the inflow hydrograph for the design flood HQ1000 and PMF was given.
Geometry
Dam crest elevation : 405 m+NN
Dam height : 9 m
Dam length : 1600 m
Upstream slope : 1V : 2.6H
Downstream slope : 1V : 4.0H
Soilmechanic parameters
Non cohesive dam material dm : 6 mm
Cohesive core
The reservoir volume is quite smal compared to the relatively great catchment area.it is
connected to. This feature is typical for reservoirs serving for flood protection. It had been
assumed, that the breach formation most probably will take place nearby the spillway
construction. The erosion is limited and one sided due to the influence of this building.
For check flood control it is not necessary to combine all loads in the most critical way,
because here a realistic response of the existing construction to a extreme flood shall be
analysed. Therefore the n-1-rule for spillways and wind action has not necessarily to be taken
into consideration.
Inflow hydrograph and initial water elevation has to be choosen carefully. If no detailed
analysis as described in the following chapter is carried out, then it is reasonable to use the
design flood water level as initial water level and the design flood HQ1000. This method had
to be used in this study, because no other data was available.
The given design flood had to be scaled by the factor 1.7 in order to cause dam erosion. The
resulting hydrograph for the breach outflow is shown in Fig. 2. The different curves show the
influence of a variation of Stricklers-Value between 22 - 27. The wiggles are not an numerical
arte fact, but a local increase of the discharge caused by the core failure mechanism..
If alternatively the initial water level is assumed to equal the dam crest elevation, the flood
and dam break induced waves do not coincident any more. The dam erosion starts almost
instantaneously. Therefore the resulting breach outflow hydrograph posseses two maxima.
The first one is caused by the dam-break, the second one by the flood wave entering the
reservoir. Note that for this second calculation PMF (Qmax = ca. 800m³/s) instead of HQ1000
(Qmax = ca. 500m³/s) is used. Nevertheless the peak discharge is lower because the two
mentioned waves do not superpose.
Fig. 2 : Breach outflow hydrograph for initial water level = design flood level, kst between 22 - 27 and HQ1000
Breschendurchfluss Vilstalsee
1600
1000
Q [m³/s]
800
600
400
200
t [h]
Fig. 3 : Breach outflow hydrograph for initial water level = dam crest = 405 m+NN
rsp. 403 m+NN and PMF
New design concepts distinguish between design and check flood. Hable (2001) desribes a
multidimensional probabilistic concept for the design of a dam. In the following chapter a
possible integration of the above mentioned method into the concept of Hable is proposed.
Dam reservoirs are used for a long time period. After a certain time the dam safety might be
endangered by changes in construction, usage or hydrology (wrong estimation of design flood
or climatic changes). The following table Tab. 1 by Laufer (1991) shows interesting results
from revised design floods for different dams in Switzerland. For most cases the dam safety
has to be reanalysed.
CONCLUSION
Dam-break due to overtopping can only take place, if the design flood is significantly
exceeded. In order to avoid unrealistical results the initial water level should be equal to the
flood water level. Furthermore the resistance of the grass coverage should be taken into
consideration. If no probabilistic calculation is used, the design flood wave should be scaled
by a factor until dam erosion can start. This scaling factor can be regarded as a safety factor
for overtopping failure.
If enough data on past flood waves and water elevations for the investigated dam are available
a probabilistic approach using the method described by Hable (2001) might be useful. The
integration of dam-break analysis into Hable’s concept is straight forward.
LITERATURE
Hewlett, H.W.M. : “Design of Reinforced Grass waterways”, CIRIA Report 116, Nov. 1987
Laufer, F. and Brenner, R.P. : “Updating of Design Flood and Dam Safety in Switzerland”,
Proc. of the 17th ICOLD 4 p.845-860, Vienna, Austria