See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236232238

Study on Piping Failure of Natural Dam

Article · January 2011

DOI: 10.2208/jscejhe.67.I_157

CITATIONS READS

4 1,536

6 authors, including:

Ripendra Awal Hajime Nakagawa

Prairie View A&M University Kyoto University
34 PUBLICATIONS   129 CITATIONS    217 PUBLICATIONS   919 CITATIONS   

SEE PROFILE SEE PROFILE

Yasuyuki Baba Hao Zhang

Kyoto University Tetra Tech
123 PUBLICATIONS   381 CITATIONS    84 PUBLICATIONS   312 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Research on Disaster Prevention/Mitigation Measures Against Floods and Storm Surges in Bangladesh View project

Study of flow characteristics in a meandering channel View project

All content following this page was uploaded by Ripendra Awal on 15 January 2014.

The user has requested enhancement of the downloaded file.

京都大学防災研究所年報 第 54 号 B 平成 23 年 6 月
Annuals of Disas. Prev. Res. Inst., Kyoto Univ., No. 54 B, 2011

Study on Piping Failure of Natural Dam

Ripendra AWAL, Hajime NAKAGAWA, Masaharu FUJITA, Kenji KAWAIKE,

Yasuyuki BABA and Hao ZHANG

Synopsis
Natural dams may fail due to seepage or piping because they have not undergone
systematic compaction and they may have high porosities. This may cause seepage
through the dam that could potentially lead to failure by internal erosion (piping).
In-depth knowledge of the mechanism of the dam failures by seepage or piping and
measured data are still lacking. Extensive laboratory experiments are carried out to
study enlargement of the pipe due to internal erosion and resulting outflow hydrographs
by varying size of initial pipe, lake water level, lake water volume, slope and location of
the pipe. This paper highlights limitations of simplified models used to predict outflow
hydrograph due to piping failure of the dam based on experimental results and provides
data set for the validation of numerical model. Experimental results show that the initial
size, slope and location of pipes have significant effects on magnitude and occurrence
time of peak discharge.

Keywords: piping, internal erosion, natural dam, outflow hydrograph, laboratory

experiment

1. Introduction (which depends on flow condition, dam size,

Natural dam such as landslide dam and moraine
dam are made up of a heterogeneous mass of
unconsolidated or poorly consolidated material. It
differs from constructed embankment dam in that it
has no water barrier, filter zones or drain zones and
it also has no channelized spillway. Nearly all
upstream and downstream faces of natural dams are
at the angle of repose of the materials. The failure
of natural dam may occur with a variety of failure
modes which includes overtopping, seepage or
piping, and sudden sliding etc. Awal et al. (2008)
attempted to derive graphical relationships to
predict failure modes of landslide dam (overtopping
failure, instantaneous slip failure and progressive
failure) by laboratory experiments in the flume.
However these relationships are derived based on
number of assumptions. The failure modes depend
on rate of water level rise in the upstream reservoir
characteristics of dam, volume and shape of the
reservoir etc.) and strength of the dam body. Thus if
inflow discharge is very low the possibility of
failure by progressive failure is high. The
experimental result shows that peak discharge
produced by progressive failure is higher than other
failure modes. Similarly the data set of
embankment dam failure shows higher peak
discharge produced by seepage or piping compared
with overtopping for higher dam factor (Awal et al.,
2008). Therefore the study on seepage or piping
failure of the dam is essential for the prediction of
potential flood for flood risk assessment and
management.
Natural dams may fail due to seepage or piping
because they have not undergone systematic
compaction and they may have high porosities
(Schuster and Costa, 1986). This may cause
seepage through the dam that could potentially lead

― 539 ―
to failure by internal erosion (piping). Internal
erosion of natural dam takes place when water that
seeps through the dam carries soil particles away
from the dam. If the seepage that discharges at the
downstream side of the dam carries particles of soil,
an elongated cavity or "pipe" may be eroded
backward toward the natural lake through the dam
or dam bed. When a backward-eroding pipe reaches
the lake, a catastrophic breaching of the dam may
occur.
Seeps occurred on the downstream face of many
landslide dams. Several seepage points developed
before overtopping of a recent landslide dam
formed on 4th January, 2010 in Hunza River at
Attabad, Pakistan. Piping hole also developed in the
downstream face of the Tangjiashan Barrier Dam,
China in 2008 (Liu et al., 2009). Piping and
undermining caused the collapse of the landslide
dam that impounded Lake Yashinkul in 1966
(Schuster and Costa, 1986). Some other
landslide-dam failures in which seepage erosion
probably was the major cause were the 1906 Cache
Creek dam in northern California, the 1945 Cerro
Condor-Sencca dam on the Mantaro River in Peru,
and the 1973 Costantino dam on the Buonamico
River of southern Italy (Meyer et al., 1994). The
failure of Rio Toro landslide dam in Costa Rica in
1992 was due to retrogressive sliding caused by
seepage and internal erosion (Mora et al., 1993).
Allpacoma Landslide Dam (see Fig. 1) in Bolivia
failed due to piping and overtopping (Hermanns,
2005). A tunnel about 1.5 m in diameter formed
1.5 m
Fig. 1 Natural tunnel through a landslide dam in
Allpacoma valley, La Paz, Bolivia, formed due to
piping failure (Hermanns, 2005).
― 540 ―
through the dam about 2 m below the dam crest due
to erosion.
In the case of moraine dam, self-destruction of a
moraine dam may cause by the spontaneous failure
of a well weathered dam slope and seepage from the
networks of drainage conduits developed in the dam
(Yamada, 1998). Size and shape of the conduits
change from season to season and year after year by
being frozen/ disappearing during the winter and
being re-opened/ newly created during the summer.
A small conduit grows to a large one due to erosion
by water weakening the dam as a whole. Outburst
of the glacial lake in Boqu River, China in 1981
was due to piping that appeared at 20-30 m below
the water level in the base of the moraine dam
which caused the sudden failure of moraine dam
(Xu, 1988).
Most of the dam break models used to predict
flood hydrographs due to piping are based on
simplified assumptions (Fread, 1988; Paquier,
1998; Loukola and Huokuna,1988; Mohamed et al.,
2002). For example, the growth of zero sloped pipe
is assumed uniformly along its length. In NWS
BREACH model, it assumes that material above the
pipe collapses when the water level in the reservoir
falls below the top level of the pipe (Fread, 1988).
Some other model assumes that the vault of the pipe
fails when the diameter of the pipe reaches 2/3 of
the dam height (Paquier, 1998). Very few studies
are focused on physical modeling of piping failure
and resulting floods of embankment dams (Morris
and Hassan, 2005; Hanson et al., 2010). Internal
erosion and piping failures in embankment dams
have been described by Fell et al. (2001) as a
four-phase process consisting of initiation,
continuation, progression, and breach which is also
applicable to natural dams. However this study
focuses on piping failure (internal erosion) of
natural dam with an assumption that a pipe leading
from the upstream to the downstream face already
exists. The physical processes involved in the
piping failure are complex. In-depth knowledge of
the mechanism of the dam failures by piping and
measured data are still lacking. Extensive
laboratory experiments are carried out to study
enlargement of the pipe due to internal erosion and
resulting outflow hydrographs by varying size of
initial pipe, lake water level, lake water volume,
slope and location of the pipe. Fig. 2. The rectangular flume of length 500cm,
width 30cm and depth 50cm is used. The shape and
2. Experimental Methods size of the dam used in experiments for the flume
slope of 5o is shown in Fig. 2. In the case of the
The schematic diagram of the flume and other flume slope of 2.5o, the upstream and downstream
accessories used in the experiments are shown in slope lengths of the dam are 27.5cm and 22.5cm

(Unit: cm)
PC
Load cell

15
9
Servo type 4
water gauge A Qi 30
45
30 10 20 Lake length (L)
Section A-A
VC-2 15
(Pipe at different location)
Flume slope = 5o
A Pump
Initial pipe
VC-1 Enlarged pipe

Container to trap
Container to collect sediment Typical longitudinal profile of pipe
water

Fig. 2 Schematic diagram of the experimental setup.

Table 1 Summary of experiments

Expt. Channel Initial size of Location of Water depth at upstream Lake length Lake water
no. slope pipe pipe toe of dam (H) (L) volume
(o) (mm x mm) (cm) (cm) (cm3)
1 5 10x10 Side-bottom 12.5 60 21426
2 5 10x10 Side-bottom 10 60 15626
3 5 10x10 Side-bottom 7.5 60 10110
4 5 7.5x7.5 Side-bottom 12.5 60 21426
5 5 7.5x7.5 Side-bottom 10 60 15626
6 5 7.5x7.5 Side-bottom 7.5 60 10110
7 5 10x10 Side-bottom 12.5 143.4 30559
8 5 10x10 Side-bottom 10 114.7 19558
9 5 10x10 Side-4cm up 12.5 143.4 30559
10 5 10x10 Side-4cm up 10 114.7 19558
11 5 10x10 Center-bottom 12.5 143.4 30559
12 5 10x10 Center-bottom 10 114.7 19558
13 5 10x10 Center-4cm up 12.5 143.4 30559
14 5 10x10 Center-4cm up 10 114.7 19558
15 2.5 10x10 Side-bottom 12.5 286.6 57552
16 2.5 10x10 Side-4cm up 12.5 286.6 57552
17 2.5 10x10 Side-9cm up 12.5 286.6 57552
18 2.5 10x10 Center-bottom 12.5 286.6 57552
19 2.5 10x10 Center-4cm up 12.5 286.6 57552
20 2.5 10x10 Center-9cm up 12.5 286.6 57552
Note: Pipe size 10 x 10 mm is Type: A and Pipe size 7.5 x 7.5 mm is Type: B.

― 541 ―
respectively. The height of all dams is 15cm.
In order to form an eroded pipe the dam
material must be strong enough to form a roof to
keep the shape of the pipe. Dam material with
sufficient fines content (% finer than 0.075mm >
15%) and partially saturated dams are favorable
conditions for progression of erosion ability to
support a roof (Fell and Foster, 2000). Silica sand
of number 8 (D50 = 0.05mm) was used to prepare
the dam with different sized pipes in the flume. The
particle size distribution curve of sand number 8 is
shown in Fig. 3. Water was mixed with silica sand
to make initial saturation of about 16% in all
prepared dam. An artificial pipe was created in the
dam body at the bottom, 4 cm up and 9 cm up from
the flume bed at the center or side of the flume by
using different square-shaped metal rods (see Fig. 2,
Section A-A). After filling the upstream lake to the
desired level, the inflow was stopped (Qin = 0).
Then the steel rod was removed from the dam to
initiate internal erosion of the pipe. The initiation
stage of piping was not studied but attention was
focused on growth of an initial pipe. In the
beginning the pipe was enlarged due to erosion of
3. Results and Discussions
Numbers of experiments were done to study
piping failure (internal erosion of the pipe) of
natural dam and outflow hydrograph by varying
water level in the upstream lake, initial size of the
pipe, lake water volume, location of the pipe and
pipe slope. In all experiments fine sediment is used
to prepare dam, so the top part of the dam above the
pipe is stable.
3.1 Effect of different lake water levels
Fig. 4 shows the outflow hydrographs due to
piping under different lake water level. Flow
through the pipe is directly proportional to
hydrostatic head on the pipe, so the resulting peak
discharge is higher for higher lake water level. The
final shape of the enlarged pipe is bigger for higher
lake water level in the lake as shown in Fig. 5 and
Fig. 6. The enlargement of the pipe is not uniform
along its length.
1500 Expt: 1 (H = 12.5)
Expt: 2 (H = 10)
Discharge (cm3/sec)

Expt: 3 (H = 7.5)
1000
the dam material in the pipe (see Fig. 2, Typical
longitudinal profile of the pipe). The top part of the
500
dam above the pipe may collapse or stable based on
size of enlarged pipe, properties of the dam material 0
and lake water volume. In the case of stable top, 0 10 20 30 40
Time (sec)
erosion of the pipe was occurred from bottom and
sides when flow was changed to free surface flow. Fig. 4 Outflow hydrographs (Different lake water
Load cell and servo type water gauge were used level).
to measure sediment and total flow in the Dam Initial pipe Expt: 7 (H = 12.5) Expt: 8 (H = 10)

downstream end of the flume. Two video cameras Initial water

level
15
Heigh t (cm)

were used to capture enlargement of the pipe from Top

10
side and front of the dam. The summary of 5
Bottom
Section of
95 105
experiments is shown in Table 1. 0
65 75 85 initial pipe
55 l (cm)
45
sta nc e fro m en d of ch an ne
Di
100
Fig. 5 Side view of final shape (top and bottom) of
80 enlarged pipe.
Percent finer

60 Dam Initial pipe Expt: 7 Expt: 8

30
40
Distance (cm)

20
20
10
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0
45 55 65 75 85 95 105
Particle diameter (mm) Distance from end of channel (cm)

Fig. 3 Grain size distribution of sand used to Fig. 6 Plan view of final shape of enlarged pipe and
prepare dam. dam.

― 542 ―
3.2 Effect of initial size of pipes Dam Initial pipe Expt: 1 (L = 60) Expt: 7 (L = 143.4)
(
Fig. 7 shows the outflow hydrographs due to Expt: 1 (L = 60
) Expt:
Ini tial pip e
Initial water level
piping under pipes with different initial sizes. The
15

Height (cm)
size of the pipe developed after flowing full water 10
Top

Bottom
of the lake is slightly smaller than that in the case of 5
0 95 105 Section of
75 85 initial pipe
initial pipe of Type B (7.5 x 7.5 mm). Thus peak 55 65 nel (cm)
45 m end of ch an
Distance fro
discharges produced by initial pipe size of Type B
(Expt: 4 and 5) are also slightly lower than that Fig. 9 Side view of final shape (top and bottom) of
produced by initial pipe size of Type A (Expt: 1 and enlarged pipe (H = 12.5cm).
2). However, it takes more time to enlarge the
pipe in the cases of initial pipe size of Type B, so Dam Initial pipe Expt: 2 (L = 60) Expt: 8 (L = 114.7)

time-to-peak is higher in these cases.

Initial water level
15

Height (cm)
Expt: 1 (P ipe : Type A, H = 12.5) Top
1500 10
Expt: 4 (P ipe : Type B , H = 12.5) Bottom
5
Expt: 2 (P ipe : Type A, H = 10)
105 Section of
Discharge (cm3/sec)

0 85 95 initial pipe
1000
Expt: 5 (P ipe : Type B , H = 10)
65 75
55 (cm)
45 end of channel
Distance from
500
Fig. 10 Side view of final shape (top and bottom) of
0
enlarged pipe (H = 10cm).
0 10 20 30 40 50 60
Time (sec)

3.4 Effect of different vertical location of pipes

Fig. 7 Outflow hydrographs (Different initial sized
pipe).
3.3 Effect of different lake water volumes
Fig. 8 shows the outflow hydrographs due to
piping under different lake water volume. For the
same initial pipe size and lake water level, the peak
discharges produced by lakes with large lake water
volume are higher than small lake water volume.
The rate of drawdown in lakes with relatively large
lake water volume for a given water level is slow
compared with small lake water volume, which
affects the enlargement of the pipe and peak
discharge. The enlargement of the pipe in the upper
part is bigger for large lake water volume as shown
in Fig. 9 and Fig. 10.
2000
Discharge (cm3/sec)
Expt: 1 (H = 12.5, L = 60)
Expt: 7 (H = 12.5, L =143.4)
Expt: 2 (H = 10, L = 60)
Expt: 15, 16 and 17 were carried out with initial
pipes at different level. The enlargement of the pipe
and erosion of the dam at different time steps in
three cases are shown in Fig. 11, Fig. 12 and Fig. 13.
The failure processes in three cases were different.
When the location of the pipe was at the bottom of
the dam the failure was due to enlargement of the
pipe. When the location of the pipe was at 4 cm up
from the flume bed the failure was partly due to
enlargement of the pipe and partly due to headcut
erosion. Headcut erosion initiates at the outlet of
the pipe and progress upstream internally during
erosion process due to development and migration
of headcut. Physical model test of large scale
earthen embankment dam showed similar process
before collapse of the top part of the dam above
pipe (Hanson et al., 2010). When the location of
initial pipe was at 9 cm up from the flume bed the

1500 Expt: 8 (H = 10, L = 114.7)

hydrostatic head on the pipe was not sufficient to
1000 enlarge the pipe. In this case the dam was failed due
to headcut erosion of the dam due to free surface
500
flow. The erosion of downstream face proceeded
0 upstream and big flood occurred when it reached
0 10 20 30 40
near the lake water. Thus the shape of hydrograph,
Time (sec)
magnitude of peak discharge and time to peak are
Fig. 8 Outflow hydrographs (Different lake water different according to location of pipes at different
volume). level as shown in Fig. 14.

― 543 ―
 Dam 10 sec 20 sec 30 sec 40 sec
3.5 Effect of different lateral location of pipes
50 sec 60 sec 70 sec 80 sec Initial pipe
For the same lake water level, volume, initial
Initial water level pipe size and slope of the pipe the outflow
Elevation (cm)

15
Section of
10
initial pipe
hydrograph depends on different lateral location of
5
Top the pipe as shown in Fig. 15. Peak discharge is
0 Bottom
100 110
40 50 60 70 80 90 higher when the pipe is located at the center of the
Distance from end of channel (cm)
dam where enlargement is possible in all sides for
Fig. 11 Side view of temporal change (top and the pipe at the “center-4 cm up” and three sides for
bottom) of the pipe at different time step (pipe at the pipe at the “center-bottom”. Location of the
bottom) – Expt: 15. pipe significantly affect rate of the pipe
enlargement and hence flood hydrograph. However,
Dam
80 sec
20 sec
100 sec
40 sec
120 sec
60 sec
140 sec
most of the existing models are based on
160 sec 180 sec 200 sec 220 sec assumption that enlargement will occur from all
240 sec 260 sec 280 sec Initial pipe
sides of the pipe.
Section of Initial water level
15
Elevation (cm)

initial pipe 2500 Expt: 15 (Location: side-bottom)

10
Top Expt: 16 (Location: side-4cm up)
5 2000

Discharge (cm3/sec)
Bottom
Expt: 18 (Location: center-bottom)
0
80 90 100 110 1500 Expt: 19 (Location: center-4cm up)
40 50 60 70
Distance from end of channel (cm) 1000

500
Fig. 12 Side view of temporal change (top and
0
bottom) of the pipe at different time step (pipe at 0 50 100 150 200 250 300 350
4cm up) – Expt: 16. Time (sec)

Fig. 15 Outflow hydrographs (Pipes at different

Dam 20 sec 40 sec 60 sec
80 sec 100 sec 120 sec 140 sec lateral location).
160 sec 180 sec 200 sec 220 sec
240 sec 260 sec 280 sec 300 sec
Initial pipe
3.6 Effect of pipe slope
Section of
15
initial pipe Initial water level In all experiments the initial slope of the pipe is
Elevation (cm)

Top
10 Bottom
parallel to the flume bed. Fig. 16 shows the
5 comparison of outflow hydrographs for the pipe
0
100 110
slope of 5o and 2.5o. Enlargement of the pipe in the
60 70 80 90
40 50
Distance from end of channel (cm) case of steeper flume/pipe slope is faster, so peak
discharge occurred earlier and magnitude of peak
Fig. 13 Side view of temporal change (top and discharge is also higher even if the volume of
bottom) of the pipe at different time step (pipe at upstream lake water is smaller (see Fig. 16 and Fig.
9cm up) – Expt: 17. 17). The existing models assume zero sloped pipe,
so these models will underestimate flood discharge
1400 Expt: 15 (Location: side-bottom)
Expt: 16 (Location: side-4cm up) 2500 Expt: 11 (Slope = 5, Location: center-bottom)
1200
Expt: 17 (Location: side-9cm up) Expt: 13 (Slope = 5, Location: center-4cm up)
Discharge (cm3/sec)

1000
2000 Expt: 18 (Slope =2. 5, Location: center-bottom)
Discharge (cm3/sec)

800 Expt: 19 (Slope = 2.5, Location: center-4cm up)

600 1500

400
1000
200
0 500
0 50 100 150 200 250 300 350
Time (sec) 0
0 50 100 150 200 250
Time (sec)
Fig. 14 Outflow hydrographs (Pipes at different
vertical location). Fig. 16 Outflow hydrographs (Different pipe slope).

― 544 ―
 Peak discharge (Center-bottom) 4. Conclusions
Peak discharge (Center-4cm up)
Lake water volume

2500 300000 Laboratory experiments were performed to

Lake water volume .

2000 250000
Peak discharge
(cm /s)
200000
study enlargement of piping and resulting outflow
1500

(cm )
3
hydrograph. The outflow hydrographs due to piping
3

150000
1000
100000
500
failure of natural dam depends on lake water level,
50000
0 0 initial size of the pipe, lake water volume, slope
0 2.5 5 7.5
and location of the pipe. For the same lake water
Pipe slope (degree)
level and volume, the size of the enlarged pipe and
Fig. 17 Comparison of peak discharge for different
peak discharge is slightly lower in the case of
pipe slope.
smaller initial pipe size. However peak discharge
if the slope of the pipe developed in the dam is occurred just after sufficient enlargement of the
steeper. pipe, so there is time lag between occurrences of
peak discharge in different sized pipes. For the
3.7 Effect in outflow hydrograph after flow same lake water level, same size and slope of the
changes from pressure flow to free surface flow pipe the failure process depends on location of the
When flow changes from pressure flow in the pipe. Dam may fail by i) enlargement of piping, ii)
pipe to free surface flow the flow discharge may enlargement of piping and headcut erosion and iii)
increase based on progress of seepage flow in the headcut erosion based on location of the pipe at
dam, headcut erosion and size of the enlarged pipe. different level. Enlargement of the pipe in the case
As already explained, the enlargement of the pipe of steeper flume/pipe slope is faster, so the peak
depends on hydrostatic head on the pipe and slope discharge occurred earlier and magnitude of the
of the pipe. In Fig. 18 red line indicates the time at peak discharge is also higher even if the volume of
which flow changes from pressure flow to free upstream lake water is smaller for steeper slope.
surface flow in the pipe. Outflow hydrograph of Experimental results show non uniform
Expt: 9 shows sudden decrease in flow discharge enlargement of the pipe along its length, different
where as Expt: 10 and Expt: 16 show increasing magnitude and occurrence time of peak discharge
trend of flow discharge for short time period. The for different initial size, location and slope of the
simplified assumption of NWS BREACH model, pipe. Headcut erosion of the pipe occurs internally
collapse of remaining material of the dam body during the erosion process. The assumption of
when flow changes from orifice control to unrealistic failure of the material above the pipe
weir-control, always produce decrease in flood may underestimate peak discharge if the remaining
discharge for short time period is not valid for the part of the dam is stable even if pressure flow in the
dam in which the remaining material of the dam pipe changed to free surface flow. For the
body is stable. improvement of model to predict outflow
hydrograph due to piping failure of the dam we
1200
have to consider these phenomenons. The
Time at which pressure flow change to free surface flow
1000
Expt: 9 (Slope = 5, H = 12.5, Location: side-4cm up)
experimental results and data set will provide an
Expt: 10 (Slope = 5, H = 10, Location: side-4cm up) opportunity to improve existing model to estimate
Discharge (cm3/sec)

800
Expt: 16 (Slope = 2.5, H = 12.5, Location: side-4cm up)
flood hydrograph due to piping failure of natural
600
dam.
400
The piping failure also depends on the strength
200 of the dam material, so further study on different
0
dam material and dam shape is essential to study
0 50 100 150 200 250 300 350 the collapse of the top part of the dam above the
Time (sec)
pipe and resulting flood hydrograph.
Fig. 18 Outflow hydrographs (Change in flow
condition).

― 545 ―
 Acknowledgements
This work was supported by the Japan Society
for the Promotion of Science Postdoctoral
Fellowship Program (grant-in-aid P 09080). The
authors are grateful to Mr. Hiroshi Teraguchi and
Mr. Hideaki Mizutani for their help to prepare
abstract in Japanese.
References
Awal, R., Nakagawa, H., Kawaike, K., Baba, Y.
and Zhang, H. (2008): Experimental study on
prediction of failure mode of landslide dams,
Proceedings of Fourth International Conference
on Scour and Erosion (ICSE-4), pp. 655-660.
Fell, R., Wan, C. F. Wan, Cyganiewicz, J. and
Foster, M. (2001): The time for development and
detectability of internal erosion and piping in
embankment dams and their foundations, UNICIV
Rep. No. R-399, University of New South Wales.
Fell, R. and Foster, M. (2000): The internal erosion
and piping process, ASDSO-FEMA Workshop.
Fread, D. L. (1988): BREACH: An erosion model
for earthen dam failures, U.S. National Weather
Service, Office of Hydrology, Silver Spring,
Maryland (revised 1991).
Hanson, G. J., Tejral, R. D., Hunt, S. L. and Temple,
D. M. (2010): Internal Erosion and Impact of
Erosion Resistance, Collaborative Management of
Integrated Watersheds, 30th Annual USSD
Conference, pp.773-784.
Hermanns, R. (2005): All Clear for Allpacoma
Landslide Dam, MAP: GAC Newsletter, Vol.5,
No.1.
Liu, N., Zhang, J. X., Lin, W., Cheng, W. Y. and
Chen Z. Y. (2009): Draining Tangjiashan barrier
lake after Wenchuan earthquake and the flood
propagation after the dam break, Science in China
Series E: Technological Sciences, Vol.52, No.4,
パイピングによる天然ダム崩壊に関する研究
Ripendra AWAL・中川一・藤田正治・川池健司・馬場康之・張浩
― 546 ―
pp.801-809.
Loukola, E. and Huokuna, M. (1998): A numerical
erosion model for embankment dams failure and
its use for risk assessment, in Morris, M. ed.
CADAM – Concerted Action on Dambreak
Modeling, Munich, Germany.
Meyer, W., Schuster, R. L. and Sabol, M. A.
(1994): Potential for Seepage Erosion of
Landslide Dam, Journal of Geotechnical
Engineering, Vol.120, No.7, pp.1211-1228.
Mohamed, M., Samuels, P. G., Morris, M. W. and
Ghataora, G. S. (2002): Improving the accuracy of
prediction of breach formation through
embankment dams and flood embankments, in
Bousmar and Zech eds. Proceedings of the
International Conference on Fluvial Hydraulics,
River Flow 2002, Louvain-la-Neuve, Belgium.
Mora, S., Madrigal, C., Estrada, J. and Schuster, R.
(1993): The 1992 Rio Toro landslide dam, Costa
Rica, Landslide News, Vol. 7, pp.19-21.
Morris, M. W. and Hassan, M.A.A.M. (2005):
IMPACT: Breach formation technical report
(WP2), IMPACT.
Paquier, A., Nogues, P. and Herledan, R. (1998):
Model of Piping in order to compute dam-break
waves, in Morris, M. ed. CADAM – Concerted
Action on Dambreak Modeling, Munich,
Germany.
Schuster, R.L., and Costa, J.E. (1986): A
perspective on landslide dams, in Schuster, R.L.,
ed., Landslide Dams: Processes, Risk, and
Mitigation: ASCE Geotechnical Special
Publication No. 3, pp.1-20.
Xu, D. (1988): Characteristics of debris flow caused
by outburst of glacial lake in Boqu river, Xizang,
China, 1981, GeoJournal, Vol.17(4), pp.569-580.
Yamada, T. (1998): Glacier lake and its outburst
flood in the Nepal Himalaya, Japanese Society of
Snow and Ice, Data Centre for Glacier Research,
Monograph No.1.
 要 旨
天然ダムは，均等に圧縮されていない場合や高い空隙を有していると浸透流やパイピングにより崩壊することがある。
浸透流は内部浸食（パイピング）の原因となり，内部浸食はダム崩壊をもたらす可能性を有している。浸透流やパイピ
ングによるダム崩壊メカニズムについては，詳細な情報や計測データはまだまだ不十分である。本研究では，多数の実
験を実施し，内部浸食よるパイプの拡大を調査した。初期パイプの大きさや，貯水位，貯水量，斜面，パイプの位置を
変更し，流出ハイドログラフの結果について考察を行った。本稿は，簡易化された数値解析モデルでは，パイピングに
よるダム崩壊の流出ハイドロにおいて予測精度に限界があることを示し，そのパイピングによるダム崩壊を予測する数
値解析モデル構築のためのデータを提供するものである。そして，実験結果より，初期サイズ, 勾配およびパイプの位
置は，流出ハイドロのピーク流量の発生時期とその大きさに重大な影響を与えることを示した。

キーワード: パイピング, 内部浸食, 天然ダム, 流出ハイドログラフ, 実験

― 547 ―

View publication stats