Table 6.

2: Case histories and hazard or potential failure mode, key words and authors

Nº Country Dam Dam type Description Hazard or PFM Key words Authors
cutoff wall, drainage,
embankment dam,
emergency plan,
erosion, foundation
Zoned earthfill
treatment, impervious
with central clay René Gómez,
Seepage, internal blanket, incident
60 Spain La Loteta Embankment core & upstream Manuel G. de
erosion in foundation detection, internal
clay blanket and Membrillera
erosion, karst,
cut-off wall
monitoring, piezometer,
pore pressure, repair,
dam safety, seepage,
upstream blanket
monitoring, safety,
piezometer network,
drains, drainage A. Vaquero and C.
61 Spain Siurana Gravity Concrete Uplift
network, uplift pressure, Barbero
rehabilitation, remedial
work, surveillance
deformation
measurement,
René Gómez,
foundation, monitoring,
62 Spain Val Gravity RCC Uplift, deformations Manuel G. de
operation, dam safety,
Membrillera
seepage, settlement,
temperature, uplift
S. Hoppe, R. Gómez
55 dams in automated monitoring,
Diligent monitoring López de Munain and
63 Spain Ebro river All dam types - safety of dams, statistical
data analysis Manuel G. de
basin method
Membrillera Ortuño
pore pressure rise,
timber sheet pile,
Zoned fill with
leakage, geotechnical Sam Johansson,
central glacial till Seepage, internal
64 Sweden Storfinnforsen Embankment investigations, Carl-Oscar Nilsson,
core and timber erosion in dam
temperature, optical Ake Nilsson
sheet pile
fiber, stabilizing berm,
drainage system

51
 Table 6.2: Case histories and hazard or potential failure mode, key words and authors

Nº Country Dam Dam type Description Hazard or PFM Key words Authors
behaviour, monitoring,
20 high risk Improvement of Sam Johansson &
65 Sweden Several dam types - training, potential failure
dams monitoring systems Pontus Sjödahl
mode analysis
drainage, embankment
dam, hydraulic gradient,
Zoned earthfill Seepage, internal
internal erosion,
66 Tunesia Ziatine Embankment with central clay erosion in foundation N.H. Dhiab
leakage, piping, pore
core and contact
pressure, remedial work,
seepage
potential failure modes,
Homogeneous Seepage along seepage along conduit, J. Stateler and B.
67 USA Dorris Embankment
earthfill conduit effective visual Iarossi
monitoring
potential failure modes,
Homogeneous Seepage through piezometers, seepage
68 USA Ochoco Embankment J. Stateler
earthfill dam through ancient
landslide
Zoned earthfill potential failure modes,
Upstream dam slope
69 USA Steinaker Embankment with central clay upstream slope failure, J. Stateler
sliding
core responding to incident
potential failure modes,
Gravity & Concrete & sliding in body of
70 USA Wanapum Sliding of dam body K. Marshall
embankment zoned earthfill concrete dam, design
calculation error
karst foundation,
Gravity & Concrete & Seepage through seepage through B. Walker, J. Bomar,
71 USA Wolf Creek
embankment earthfill foundation foundation, water and V. Bateman
pressure data

52
 Table 6.3: Hazards or potential failure modes addressed in the case histories

6.3 Matrix case histories - hazard or potential failure mode

Improvement of monitoring systems

Erosion (foundation and dam body)

Proper opertaion and maintenance

Diligent monitoring data analysis
Settlements, deformations and

Ageing of concrete & cracking

Foundation rock deformation

Sealing membranes (cracks &

Downstream river erosion

movements of dam body

Reservoir slope sliding

Pore water pressures

Sliding of dam body

Nº Country Dam

Dam slope stability

Temperature load
Hydraulic fracture

Sedimentation
Earthquake
behaviour)

Sabotage
Seepage

Uplift
Benchmark histories
1 France Malpasset 1 1
2 Italy Vajont 1
3 Switzerland Zeuzier 1
4 USA Teton 1 1
5 USSR / Germany Dnieprostri / Möhne / Eder 1
6 USA Folsom 1
7 Mozambique Cahora Bassa 1 1 1 1
8 South Africa Zoeknog 1 1
9 Spain Tous 1 1
Case histories
1 Argentina El Chocón 1 1
2 Austria Durlassboden 1 1
3 Austria Gmuend 1 1
4 Austria Zillergründl 1 1
5 Burkina Faso Comoé 1 1
6 Cameroun Song Loulou 1
7 Canada Kootenay Canal Forebay 1 1
8 Canada WAC Bennett (trends analysis) 1
9 Canada WAC Bennett (ROV) 1
10 Canada Not reported 1 1
11 Canada Not reported 1 1
12 Colombia Porce II (assessment study) 1 1
13 Colombia Porce II (thresholds) 1 1
14 Colombia Santa Rita 1 1
15 Colombia Tona 1
16 Colombia Urrá I 1
17 Czech Republic Mšeno 1
18 Egypt El-Karm 1 1
19 France Etang 1 1
20 France Grand'Maison 1
21 France La Paliere 1
22 France Mirgenbach 1
23 Germany Sylvenstein 1 1
24 Iran Alborz 1 1 1
25 Iran Gotvand 1 1 1
26 Iran Karun 4 1
27 Iran Masjed-e-Soleiman 1 1
28 Iran Seymareh 1
29 Italy Ambiesta 1
30 Italy Isola Serafini 1
31 Italy San Giacomo 1
32 Italy MISTRAL 1
33 Japan Kyogoku Upper Reservoir 1
34 Japan Okuniikappu 1
35 Japan Yashio 1
36 Morocco Oued El Makhazine (cutoff wall) 1
37 Morocco Oued El Makhazine (culvert joint) 1
38 Morocco Tuizgui Ramz 1 1
39 Norway Muravatn 1
40 Norway Storvatn 1
41 Norway Svartevann 1 1
42 Norway Trial dam 1
43 Norway Viddalsvatn 1
44 Norway Zelazny Most (Poland) 1 1
45 Portugal gestBarragens
46 Romania Gura Râului 1
47 Romania Paltinu 1 1
48 Romania Pecineagu 1 1
49 Romania Poiana Uzului 1
50 South Africa Belfort 1 1
51 South Africa Driekoppies 1 1
52 South Africa Inyaka 1
53 Lesotho Katse 1
54 South Africa Kouga 1
55 South Africa Ohrigstad 1
56 Spain La Aceña 1
57 Spain Caspe II 1 1
58 Spain Cortes 1
59 Spain La Minilla + El Gergal 1
60 Spain La Loteta 1 1
61 Spain Siurana 1
62 Spain Val 1 1
63 Spain 55 dams in Ebro basin (DAMDATA) 1
64 Sweden Storfinnforsen 1 1
65 Sweden 20 high risk dams 1
66 Tunesia Ziatine 1 1
67 USA Dorris Dam 1
68 USA Ochoco 1
69 USA Steinaker 1
70 USA Wanapum 1
71 USA Wolf Creek 1

53
 6.4 BENCHMARK CASE HISTORIES

 Malpasset Dam (importance of monitoring engineering geological aspects)

 Vajont Dam (importance of monitoring reservoir slopes)

 Zeuzier Dam (The unbelievable effect of pore pressure relief)

 Teton Dam (value of diligent visual observations)

 Dnieprostroi, Möhne and Eder Dams (explosive loads twice during World War II)

 Folsom Dam (gate failure…tested regularly but not all the way)

 Cahora Bassa Dam (the value of diligent installations on the life of instruments)

 Zoeknog Dam (failure…predicted by pore pressure gauges, but ignored)

 Tous Dam (backup systems failure)

54
 MALPASSET ARCH DAM FAILURE AT FIRST IMPOUNDMENT

B. Goguel, France

ABSTRACT Case history category: e, b, f. The failure of Malpasset Dam was a watershed case
history that benefitted dam-engineering in more ways than one. It highlighted the utmost importance
of dam safety surveillance as well as the significance of proper geological- and geotechnical
investigations.

TECHNICAL DETAILS

Malpasset Dam was built between April 1952 and October 1954. The dam is located near Fréjus, an
old roman city on the Côte d’Azur, along the Mediterranean Sea, at a place called Malpasset (a bad
pass for passing people). The 66.5 m high arch dam with a crest length of 222 m, was constructed in a
narrow rocky section on a small river. Using only 48 000 m3 of concrete a reservoir with a capacity of
50 million m3, was created for irrigation and flood damping. The owner was the French State (Var
Département, without any dam specialist at the time). For more details on the dam see Figure 1 to
Figure 3.

The owner relied on the Engineer (André Coyne) for the design and for construction supervision, with
a loose link for evaluation after the end of construction. Due to budgetary constraints no detailed
geotechnical investigations were carried out before construction. The geology can be described as
Gneiss (old metamorphic horst), very heterogeneous and crisscrossed by joints of all scales and
directions.

SURVEILLANCE DETAILS

Customary at the time, no drainage curtain was provided; no organized leakage and no piezometric
pressures were monitored. Visual inspections were only performed by the water bailiff of the dam.
Geodetic measurements (of 27 targets on the dam d/s face) were performed in few yearly (summer)
measurements (campaigns), as shown in Figure 4. After dam completion, reservoir impoundment was
delayed due to lack of expropriation upstream of the dam, and lack of the water distribution network
downstream of the dam. Flash floods in November 1959 filled the reservoir rapidly.

Figure 1. Malpasset Dam at the end of construction, mid 1954 (left) and soon after failure in
December 1959 (right)

55
 Figure 2. Plan view with contours in meters above sea level (1 - Bottom outlet; 2 -Uncontrolled
spillway; 3 - Outlet works; 4 - Stilling basin)

Crest elevation 102.55

Maximum flood level 102.00
Spillway weir level 100.40
Normal operating level 98.50

Axis of outlet pipe 46.25

Lowest bedrock level 38.00

Figure 3. Highest cross section (the support of the 1.5 m diameter bottom outlet, fitted with a
downstream hollow valve and an upstream control gate, explains the widened foundation at this cross
section only)

DESCRIPTION OF THE INCIDENT

On 2 December 1959, following intense rains and resulting floods, the reservoir level rose rapidly
approaching the uncontrolled (non-gated, free sill) spillway level. The bottom outlet was opened after a
meeting held on site that afternoon. During that meeting none of the 10 persons present noticed or
reported any signs of distress on the dam or its abutments.

Results from the last geodetic measurement taken (see point D on Figure 4, July 1959, RWL 94.10 i.e.
8 m below the Max Flood Design Level) were not yet in the engineer’s hands. (Afterwards, he could not
say what would have been his comments on excessive movements in the lower sections of
approximately 10 mm).

56
The dam failed the same evening around 21:11 after a sudden failure of the left bank. A large wedge
of rock dislodged, below the left half of the arch dam, leaving an excavation in a dihedral form as shown
in Figure 5. Only the bases of the right bank and central concrete blocks were still there, as shown in
Figure 6. All other parts of the arch washed away; some blocks of concrete were swept away for more
than 1.5 km.

Figure 4. Reservoir level during years 1954-1959 (the box magnifies the four last days). Triangles =
dates of geodetic measurements A (late reference) to D.

Figure 5. The left bank after failure. “Dihedral” excavation with half of the upper thrust block fallen
after the failure (the exploratory adit on the lower right was excavated after the failure to perform in
situ jack tests).

The displacements of the remaining parts correspond to a rotation of the whole dam as a solid body,
the axis of which is located at the very end of the arch at the right bank (and inclined perpendicularly to
the fault plane forming the d/s face of the rock wedge which supported the dam and disappeared on
the left bank). This resulted in an open crack at the upstream heel, 0.5 m open at the crown.

57
The upstream face of the wedge comprised of several shear planes parallel to the foliation of the gneiss,
dipping towards the river. During the design and construction stages the downstream fault was not
known. It was between 15 and 40 m below the lowest excavations (Figure 7), and it surfaced unnoticed
in the overburden, more than 20 m downstream of the dam.

Figure 6. Downstream developed elevation after the failure

Figure 7. Full hydrostatic water pressure in the left flank foundation of the dam

After this catastrophic failure (resulting in 423 casualties and severe losses), several conventional back-
analyses were undertaken. However, none could lead to an explanation for the failure:

 concrete stresses were moderate and well below what could fail the dam,
 buckling of the arch was not realistic,
 sliding of the abutment block was not the initial cause, rather a consequence of the overall
rotation of the arch,
 sliding of the arch at contact with foundation was also discarded,
 sliding on the downstream fault was not possible, unless high pore pressures existed at depth
in the rock (much in excess of what could be expected from a conventional flow net analyses).

Research studies therefore focussed on the seepage flows in the foundation. These studies contributed
to a better understanding of "rock hydraulics". Seepage in rock masses, and the internal forces it
generates, is governed by the main characteristics of the rock masses, which are discontinuous, non-
homogeneous and anisotropic. Several of these new findings finally contributed in explaining the failure
mechanism:

 rock permeability may be highly sensitive to stress, depending on the degree of jointing and the
nature of joints infilling; geologic and petrographic studies established that the intensity and
nature of the fissuration and jointing at Malpasset Dam occurred at all scales; the load applied

58
 by the dam to its foundation was therefore able to induce a drastic reduction of permeability
under the structure,
 stress in discontinuous media can penetrate quite long distance following discontinuities,
particularly in the direction of the rock foliation, therefore forming a watertight barrier at depth
under the dam toe,
 upstream crack developing in the arch dam foundation provided a preferential path for the
development of full hydrostatic pressure, which could penetrate deeply when combined with
the above mentioned watertight barrier,
 deformability of the foundation was relatively high, with very low modulus of elasticity in the left
bank (as demonstrated by finite element back analyses performed during the eighties).

Figure 8. Schematic representation of the “explosive” failure phase

The watertight barrier, generated by the extremely sensitive rock of Malpasset Dam in the direction of
the dam thrust, was located deep in the left bank where it coincides with the foliation planes. The
development of high hydrostatic water pressures (uplift) consequently made sliding on the downstream
fault possible. (At the right bank, the watertight barrier could not form in the same location because the
foliation was not symmetrical with respect to the dam).

A failure scenario could therefore be reconstructed based on these observations. This has also led to
the development of the Londe method for 3D stability analysis (which became a standard in arch dam
design and safety studies).

LESSONS LEARNT

The concept of appropriate dam safety-monitoring using reliable and well located instruments has
become the standard. This was critically absent at Malpasset Dam. Piezometric measurements in the
foundation together with strain meters in the arch, and monitoring of the foundation deformations, would
probably have given ample warning at least several months prior to the failure.

59
The importance of rapid transmission of appropriate monitoring information to experienced engineers
was emphasised. The geodetic measurement system at Malpasset Dam was monitoring arch
deformations rather than foundation movements (due to the reference measurements for the
computational models not been taken). The calculation delay at Malpasset Dam was also excessive;
as well as the lack of a chain of suitable personnel to evaluate the information.

Instruments for the precise and frequent foundation deformation monitoring have since been developed.
Inverted plumblines, for example, were not yet available during the construction of Malpasset Dam.
They only appeared on the market during the very late fifties - early sixties and are nowadays common
use everywhere for the monitoring of foundations.

In the absence of preconceived potential failure mechanisms, instrumentation may not always provide
the required information. It is therefore important that monitoring (instruments and their reading
frequency) be designed pro-actively to monitor potential failure mechanisms.

Deep drainage curtains became a standard feature in arch dams (especially the flanks). Malpasset
Dam opened the eyes of dam designers on uplift acting not only below and inside the structure, but also
inside the rock mass at depth and downstream of the dam.

Rock mechanics applied to dam engineering was developed intensively after the failure of Malpasset
Dam. Systematic detailed geological and geotechnical investigations protocols for new projects have
also been developed internationally.

Pro-active surveillance of dams during their life cycle and particularly during first impoundment became
the standard. More generally, the value of an Independent Panel of Experts different from the Owner
and Engineer has been emphasized.

SELECTED REFERENCES

Londe P., 1987, The Malpasset dam failure, Engineering Geology, Elsevier, Amsterdam, 24, 295-329.

Post G., Bonazzi D., 1987, Latest thinking on the Malpasset accident, Engineering Geology, 24, 339-
353.

Carrère A., 2010, Les leçons de Malpasset, leur application aux projets de barrages d’aujourd’hui,
Revue française de Géotechnique, Paris, 131-132, 37-51.

Duffaut P., The traps behind the failure of Malpasset arch dam, France, in 1959, Journal of Rock
Mechanics and Geotechnical Engineering (2013), DOI.

60
 THE VAJONT DISASTER

R. Gómez López de Munain & J. Fleitz, Spain

ABSTRACT Case history category: e, a, b. Construction of the Vajont Dam started in 1956 and was
completed in 1960, at this time it was the highest double-curvature arch dam in the world – with a height
of 261.6 m, a crest length of 190 m and a reservoir capacity of 168.7 million m 3 water.
During the third filling, on 9 October 1963, Monte Toc, located on the left side, experienced a translational
landslide which led to 270 million m3 of rock falling into the reservoir, causing a 235 m high wave which
passed over the dam by 100 m and arrived at the Piave valley, the wave sweeping away the downstream
villages of Longarone, Pirago, Villanova, Rivalta and Fae, causing the death of 2 000 people.

DESCRIPTION OF THE DAM

Between 1957 and 1960, the Adriatic Electric Company (SADE) built the Vajont Dam, located 100 km
north of Venice (Italy). At that time, it was the highest arch dam in the world and the second highest
dam at 261.6 m. The crest length is 190.5 m, and the reservoir capacity 168.7 million m3 (150 million m3
of useful volume) with a normal maximum level of 722.50 m a.s.l. Approximately 360 000 m³ of concrete
was used in its construction. The dam was 3.4 m thick at its crest widening out to 27 m at its base. The
designer and construction manager was Carlo Semenza, an experienced and internationally renowned
civil engineer.

Figure 9. View of Vajont Dam before the event

SURVEILLANCE DETAILS

The first concerns about the stability of the valley slopes were raised during construction of the dam,
with a number of reports being compiled during 1958 and 1959. These reports identified a possible
prehistoric slide on the right bank. In view of the possibility of previous landslides and the synclinal form
of the strata there was considerable discussion on the stability of the valley walls. In 1959, before the
dam was finished, some field investigations were undertaken. A number of existing and potential

61
landslides were identified, but the analyses of them suggested that the likelihood for large-scale
movements was limited based on the following results of the field investigations:

 3 test borings had failed to identify areas of weakness;

 It was assumed that any shear plane would have a ‘chairlike’ form and exert a ‘braking effect’;
 Seismic analyses had suggested that the banks consisted of very firm in-situ rock with a high
Young’s modulus.

As a result, no action was taken to stabilize the slopes, but a monitoring program was set up. In the
following three years, the downward motion of the slide was monitored by means of surface markers.
Some of the data provided by them are plotted in Figure 10. In addition, water pressures in perforated
pipes, located in four boreholes (location shown in Figure 10), were monitored, starting in July 1961.

Figure 10. Map of the Vajont sliding area. Note the position (and comparative size!) of the arch dam
on the lower right-hand corner of the figure (Alonso et al.: 2010}

ANALYSIS OF FIELD DATA DURING RESERVOIR IMPOUNDMENT

The dam was completed in 1960 and the impoundment of the reservoir started. In October 1960, with
the reservoir partially impounded, a long, continuous peripheral crack, 1 m wide and 2.5 km in length
opened and marked the contour of a huge mass, creeping towards the reservoir in the northern
direction. The landslide was moving continually, around three to four cm/day.

Data on horizontal displacements, plotted as a function of position and time in several profiles following
the south-north direction in Figure 10, suggest that the slide was essentially moving as a rigid body.
The direction of the slide is also indicated in the figure by several arrows. Some of them (small arrows
along the peripheral crack) indicate that the moving mass was actually detaching from the stable rock,
implying no friction resistance along the eastern and western boundaries of the slide.

On November 4, 1960, when the reservoir level reached 180 m, a 700,000 m 3 lump fell off the front of
this larger landslide over a period of about 10 minutes. The speed of the slope displacement at this time
was 3-4 cm/day (Oct 1960) with a maximum displacement of 100 cm.

62
 Figure 11. Summary of monitoring data recorded at Vajont Dam: precipitation, water levels in the
reservoir and groundwater levels (measured with piezometers) and rate of movements (Belloni et al:
1987)

This event caught the attention of the site managers and they decided to lower the reservoir level down
to 135 m and the movements reduced to close to 1 mm per day. They also tried to build a slope drainage
tunnel but finally abandoned because it was impossible to progress and to sustain the tunnel due to the
strongly fractured rock material.

At that time the designers realised that the large mass of the left bank was inherently unstable. However
as there was no realistic way of arresting the slide artificially and stop the slide, or safely cause the
mass to slide down all at once, it was decided to use varying levels to try and gain control of the sliding
mass. Actually, based on the limited monitoring data available (Figure 11) a correlation of the reservoir
level with the measured vertical and horizontal displacements of a number of topographic marks
distributed on the slide surface was established (Figure 12).

The dam engineers believed that by carefully elevating the level, movement of the large landslide mass
could be started and then controlled by altering the level of the reservoir. It was calculated that, should
a sudden movement occur, providing it did not lead to a filling of the reservoir in under 10 minutes, it
would not cause over-topping of the dam.

63
 Figure 12. Relationship between water level in the reservoir and sliding velocity (Alonso et al: 2010)

As a result of such a “controlled” landslide the reservoir would be partially filled with the slope material.
Therefore, a by-pass tunnel was constructed to connect the upper and the lower part of the reservoir to
guarantee the future operation of the dam.

For most of 1961, the lake level was kept deliberately low to allow construction of the tunnel. In October
of that year the site was ready once again and filling was restarted. The plan was to slowly fill the level
of the lake while the movement of the landslide was monitored. The intention was that the landslide
would slowly slip into the lake. If the movement rate became too high the reservoir level would be
dropped to slow the movement down. Thus, the movement of the landslide would be controlled by
varying the lake level.

From October 1961 to November 1962, the level of the lake was slowly increased. Late in 1962 the
movement rate became too high, so the lake was partially emptied, whereupon the landslide effectively
stopped. Filling was then restarted in April 1963. By early September, the water depth was 245 metres.
The rate of movement of the landslide slowly increased, and in late September, the water level was
slowly dropped in an attempt to reduce the rates of movement. The maximum level achieved was 710.
The slope displacement speed was 1-3.5 cm/day. Despite the lowering of the reservoir level, the slope
displacement continued to accelerate, reaching 20 cm/day the same day of the failure.

FAILURE

At 10:38 pm on October 9, 1963, the landslide collapsed. The entire mass of approximately
270 million m3 of earth and rock slid 500 m north at up to 30 m sec-1 (110 km per hour). The huge
velocity reached by the landslide magnified the tragic consequences. When the mass fell into the lake
it had a level of 235 m and contained about 115 million m 3 of water. The landslide pushed a wave up
the opposite bank 260 m above the original reservoir level and destroyed the lower part of the village
of Casso and then over-topped the dam by up to 245m. An estimated 30 million m 3 of water fell on five

64
villages decimating them. About 2 000 lives were lost. Despite this the dam was left structurally intact,
only a metre of masonry was washed away at the top.

Figure 13. Photograph of Monte Toc taken by E. Semenza on 1 September 1959 with the
construction crane for the dam to the right. The lower line corresponds to the existing paleo-landslide
and the upper line to the 1963 landslide limit

Figure 14. Identical view taken by Daniele Rossi at the end of October 1963 just after the disaster

Figure 15. View of the municipality of Longarone before (left) and after (right) the event

65
INVESTIGATIONS OF THE FAILURE

The tragedy at Vajont Dam has spurred numerous investigations into the conditions triggering slope
collapse, some of them are mentioned in the references.

It is now widely agreed that failure occurred along bands of clay within the limestone mass, at depths
between 100 and 200 m below the surface (Hendron & Patton: 1985). These clay beds, 5 to 15 cm
thick, represented planes of weakness, which, though sub horizontal at distances less than 400 m from
the gorge, were farther away inclined at about 35 into the valley. The presence of the clay layer was
not detected by the field investigations.

Interpretation of sliding risk was essentially made on the basis of reservoir elevation and surface
displacements. There was also information on rainfall and on the levels of four piezometers. The
“piezometers” were in fact open tubes which did not reach the level of the sliding surface and only
provided average water pressures prevailing along their length. In addition, no direct information of the
position of the failure surface and, in particular, on the type of material being sheared was available.

Identification of a landslide for the purposes of estimating its evolution and of defining any remedial
measure requires information of a few key variables. Ideally, these key variables should also be used
in the formulation of a mechanical model of the motion. Early knowledge of data concerning the basal
failure surface (geometry, pore water pressure, type of material and drained strength parameters) would
have been fundamental to build a conceptual and mechanical model for the slide. In the case of
Vajont Dam, the observation that the slide velocity decreased when the reservoir level was reduced,
irrespective of the absolute level of the water, provided a reservoir filling criterion which, finally led to
the failure.

In some sense, an “observational method” was applied. The observational method, described by Peck
(1969), requires the following ingredients: a) direct observation of a key variable describing the essential
nature of the problem; b) a proper conceptual, analytical or computational model able to provide an
estimation of the risk, in a general sense, for some threshold values of the key variables and c) a plan,
defined in advance, to act in a specified manner when threshold values are exceeded. The key variables
to be interpreted were the displacement rates of surface markers and the reservoir level.

The conceptual model was essentially given by the preceding observation, illustrated in Figure 4. The
action in mind, in case of excessive displacement rate, was to reduce the water level in the reservoir. It
was accepted, despite this strategy, that a full slide was a likely event and that the expected height of
the generated wave was even estimated by model studies. However, the conceptual model was not
based on any mechanical analysis of the slide. In addition, the reservoir level did not necessarily provide
the actual pore pressures on the failure surface and the remedial plans were too simple and weakly
connected with the complex mechanisms taking place within the slide.

LESSONS LEARNT

Vajont Dam marked a change in the emphasis dam designers placed on the stability of reservoir slopes.
The Vajont Dam experience illustrates the following aspects:

 Pre-existing ancient landslides can be highly sensitive and can be reactivated when a reservoir
is formed against them.
 The importance of a thorough understanding of the geology (lithology, structure, material
properties) and hydrogeology of the reservoir rim.
 The importance of searching for recognising and evaluating precedent evidence for past
instability in the reservoir basin.
 The significance of weak seams or layers in a slope, especially if inclined towards the reservoir.

66
  The need of sufficiently reliable models for geotechnical analysis with clear acknowledgement
of their uncertainties and limitations.
 The effect of changing reservoir levels on slope stability.
 The significance of the joint effects of rapid reservoir level changes and the various influences
of rainfall on slope stability.
 The value of reliable monitoring data, prompt data evaluation and appropriate responsive
actions (emergency action plans).
 The difficulty in predicting time of failure, landslide velocity and subsequent wave size and
consequences.
 Worst case scenarios must be taken into account.

Although the sliding risk of the reservoir slope had been correctly identified by the construction team,
the assumptions regarding the magnitude, the detailed failure mechanism and therefore the estimation
of the consequences were erroneous because they were made upon very limited field investigations
and monitoring data. It should be borne in mind that this comment is made more than 60 years after the
first investigations started at Vajont Dam. The purpose is to learn from the case, not to criticize the
involved individuals who had to work with the techniques and rules of practice available at that time.

Even today, managing a very large landslide is a daunting task. It is relatively easy to extract field data
(pore water pressures, absolute deformations, “insitu” tests) in the first tens of meters of soil and rock.
Going beyond 200 m requires sophisticated, not easily available, and time-consuming efforts. A very
large landslide requires a vast site investigation and is not a matter of only a few borings.

Since the 1960s, significant progress in terms of field investigations as well as modelling and computer
calculations of slope stability has been made.

Monitoring technology applied to slope control has developed significantly, including inclinometry, radar
interferometry, GPS control, automatic topography stations, etc. Real time measurements of interstitial
pressures, displacements of the failure surface, superficial displacements, recording seismic
accelerations on the slope, including distortions and stress transmitted to other structures is feasible
today.

Emergency action plans are now mandatory in many countries and combined with adequate
surveillance programs are crucial to minimize consequences in case of failure, especially human
fatalities.

However, in the words of Carlo Semenza, the dam designer, “…things are probably bigger than us and
there are no adequate practical measures… After so many fortunate works and so many structures… I
am in front of a thing which due to its dimensions seems to escape from our hands…” (from a letter
written in April 1961; two and a half years before the disaster)

SELECTED REFERENCES

Alonso, E (2005), Las Catástrofes y el Progreso de la Geotecnia, Real Academia de Ingeniería de

España.

Alonso et al. (2010), Geomechanics of Failures. Advanced Topics, DOI 10.1007/978-90-481-3538-7_2,

© Springer Science+Business Media B.V. 2010

Belloni, L.G. and Stefani, R. (1987) The Vajont slide: Instrumentation – Past experience and the modern
approach. Engineering Geology 24, 445−474.

Carloni, G.C. and Mazzati R. (1964). Rilevamento geomorfologici della frana del Vajont. Annali del
Museo Geologico di Bologna.

67
Giudici F. and Semenza E. (November 1960). Estudio geológico sul serbatoio del Vajont. Societá
Adriatica di Electtricitá.

Gupta, H.K. (1992), Reservoir-Induced Earthquakes, Elsevier.

Hendron A.J. and Patton F.D. (1985). The Vajont slide, a geotechnical analysis based on new
geological observations of the failure surface. Technical report GL-85-5, Department of the Army, US
Army Corps of Engineers, Washington D.C. 2 Voll.

ICOLD Bulletin 124 (2002). Reservoir landslides: investigation and management - Guidelines and case
histories.

Nonveiller, E. (1987), The Vajont reservoir slope failure, Engineering Geology, 24, pp.493-512.

Pasuto, A. and Soldati, M. (1990). Some cases of deep-seated gravitational deformations in the area
of Cortina d'Ampezzo (Dolomites), The Proceedings of the European Short Course on Applied
Geomorphology, 2, 91-104.

Peck, R. (1969). Advantages and Limitations of the Observational Method in Applied Soil Mechanics.
Ninth Rankine Lecture, Géotechnique, June 1969, 19, pp. 171-187.

Petley David N. (1996). On the initiation of large rockslides: perspectives from a new analysis of the
Vajont movement record.

Rossi D. and Semenza E. (1965). Carte geologiche del versante settentrionale del Monte Toc e zone
limitrofe, prima e dopo il fenomeno di scivolamento del 9 ottobre 1963, Istituto di Geologia dell
Úniversita di Ferrara.

Rossi D. and Semenza E. (1968). La bassa valle del Vajont e lo scivolamento del 9 ottobre 1963. In
Leonardi P et al. Manfrini, Rovereto, Vol II.

Semenza Edoardo (2001). La Storia del Vajont. K-flash Editore.

Sitar, Nicolas (1997). Kinematics and Discontinuous Deformation Analysis of Landslide Movement. II
Panamerican. Symposium on Landslides, Rio de Janeiro, Nov. 13-14th, 1997. Tika, Th.E. and J.N.
Hutchinson (1999), Ring shear test on soil from the Vajont slide slip surface. Géotechnique, 49, Nº. 1,
pp. 59-74.

Torán José (1963). Vajont. Notas previas a un memorandum técnico. Revista de Obras Públicas
número 2982.

Uriel Romero, S. and R. Molina (1974), Kinematic aspects of Vajont slide. Proc. 3rd Int. Conf. ISMR,
Denver, Vol. 1-B, pp. 865-870.

Valdés and Díaz-Caneja J.M. (1964), Meditaciones sobre la catástrofe de Vajont, Vol. Nº 20 del Servicio
Geológico del M.O.P., Madrid.

Vardoulakis, 1 (2002), Dynamic thermo-poro-mechanical analysis of catastrophic landslides,

Géotechnique, 52, Nº.3, pp. 157-171.

68
 ABNORMAL BEHAVIOUR OF ZEUZIER ARCH DAM

Henri Pougatsch & Laurent Mouvet, Switzerland

ABSTRACT Case history category: e, a, b, c. In December 1978, unusual deformations were detected at
the arch dam of Zeuzier. It was established that they were the result of a modification of the
hydrogeological conditions due to the construction of the Rawil road tunnel exploration adit located 1.4 km
from the dam. The latter was significantly disrupted by ground movements. After detailed investigations,
the dam has been conditioned and made operational again in summer 1987. The Zeuzier dam mostly
gave the opportunity to be cognizant of this kind of events.

TECHNICAL DETAILS

The 156 m Zeuzier arch dam, with a crest length of 256 m and elevation at 1777 m a.s.l. in Switzerland,
was commissioned in October 1957. Its reservoir has a storage capacity of 51 million m3 (see Figures
1 and 2). The seasonal storage is mainly intended for hydroelectric power, part of it being also used in
winter by a nearby village for water supply. The dam is founded on a rocky local narrowing of a valley
shaped by Jurassic limestones. The stiff limestone layers are subdivided by many cracks. The high
permeability of the dam foundation makes the groundwater under the dam almost nonexistent. Below
the limestones, a confined groundwater aquifer does exist in a Dogger-formation protected by an
aquiclude. However, the dam site was considered as suitable for a high arch dam.

The dam behaved satisfyingly until fall 1978.

Figure 1. Plan, elevation and cross section

Figure 2. Downstream view, upstream view with empty reservoir

69
SURVEILLANCE DETAILS

The dam was very well equipped with surveillance instrumentation: 3 direct pendulums, survey of
measurement targets on the downstream face, thermometers in the dam body and seepage and
drainage flow measurement weirs. All measurements were recorded on a monthly basis. The dam
behaved as predicted, and the statistical analysis carried on clearly demonstrated the reversible
movement of the body of the dam due to variation of hydrostatic and thermal loads.

DESCRIPTION OF THE INCIDENT

At the beginning of December 1978, the monthly measurement of pendulums indicated that the
deformations obviously differed from the former elastic behaviour of the structure. Easily and rapidly
detected, the dam crown started to move toward the upstream direction although the reservoir was
nearly full (Figure 3). In January 1979, movement reached more than 5 mm in both radial and tangential
directions. End of March 1979, movement was more than 20 mm toward upstream and 15 mm toward
the abutment.

Figure 3. Radial movement at the crest, central pendulum, before and during the event [mm]
Green area: envelope for the period 1958 – 1976
Blue line: Oct. 77 to Sep. 78
Red line: since October 1978

Later on, the analysis permitted to identify the beginning of the abnormal behaviour in the last days of
September 1978 (Figure 4). The reservoir was rapidly drowned down to its minimum water level and
the frequency of monitoring measurements was increased.

70
 Figure 4. Radial movement at dam crest central section: variation between measurements and
deterministic model (in red, considered acceptable variation: +/- 1 mm

FOUNDATION MOVEMENT AND DAM DEFORMATIONS

Due to unfavourable weather conditions (heavy snow), the geodesic survey could only be carried out
in spring 1979. The results indicated a settlement (almost 10 cm) and a movement towards the
upstream (almost 9 cm) at the level of the crest as well as a narrowing of the abutment. Actually, the
valley itself was narrowing (almost 7 cm). Further to these movements, opening of vertical joints in the
top of the upstream facing and the development of cracks in the downstream facing has been noticed
(Figure 5).

Figure 5. Maximum dam deformations and crack pattern observed on both faces.
Width of main cracks (December 1980): 1 to 15 mm

71
INVESTIGATION OF CAUSES - ROCK DEFORMATIONS

To determine the reasons behind the extraordinary dam deformations, the installed measuring system
was significantly improved. Numerous instruments were installed (i.e., inclinometers, piezometers,
extensometers) particularly in the foundations. The geodetic network was considerably extended,
reinforced and adapted especially to make the measurements possible during the winter season.

The analysis of diverse hypotheses (mainly tectonic and hydrogeologic) pointed out that these unusual
deformations followed a general bowl-shaped settlement of the ground extending on 2 to 3 km. The
hypothesis of the influence of the drilling of a tunnel exploration adit for the construction of the Rawil
road tunnel was immediately considered as the tunnel was drilled only 1.4 km away from the Zeuzier
dam. This hypothesis was supported by a striking connection between the curve of the exceeding
deformations of the dam and that of the cumulative water flows drained by the above-mentioned gallery
(Figure 6). The establishment of a mathematical model representing the behaviour of a cracked elastic
saturated rock massif scientifically confirmed that the drilling of the adit was at the origin of the
settlement of the ground, (Lombardi. 1992, 2004). Consequently, the tunnelling works were interrupted
in March 1979.

Figure 6. Flow in the Rawil tunnel exploration adit (full line) and deformations shown by the pendulum
(dotted line)

REFURBISHMENT WORKS

The rehabilitation of the dam could begin only when some major conditions were fulfilled: (1) the
movement of the ground and of the dam have practically stopped. (2) the capacity of the dam to be
refurbished is ensured (i.e., the state of the foundations is checked, the ability to restore monolithism is
verified and the stress-strain conditions are controlled) and (3) a decision is made on the future works
of the tunnel exploration adit.

Once these conditions were met, the refurbishment works were applied in three different zones:

 The grout curtain: the investigations showed that the rock was of good quality and the grout
curtain was still efficient.
 The rock-concrete contact zones: based on drill holes, it was possible to detect the zones where
the rock-concrete contact had to be recovered.
 The dam body: works in this area concerned the treatment of the cracking and the vertical
joints. Specific tests were undertaken in the lower part of the dam and allowed to choose
adequate epoxy resin and injection procedure for the crack and joint grouting.
The works were successfully executed between 1982 and 1984.

72
REFILLING OPERATION

Due to the damages of the dam and its foundations, a schedule for the progressive and controlled
refilling was prepared. This schedule consisted on 6 annual steps. It included stops to perform
intermediate measurements allowing to compare with previous steps. For each step, directives such as
the kind of measurements to perform, visual inspections, etc. were previously defined in a document.

The filling operation was successfully achieved, and the normal water level was reached again by the
summer of 1987.

LESSONS LEARNT

The organization of dam surveillance

The events of the Zeuzier dam highlighted the importance and appropriateness of the multi-level dam
surveillance prescribed in Switzerland. It includes the owner’s operation staff (i.e. measurement, visual
control and operational tests), a civil engineer (analysis of the measured data, control of the structure’s
state), experts in civil engineering, geology and geodesy, as well as the high surveillance Authority.

It should be noted that the visual inspections were considered as crucial parts of the surveillance (SCD,
1997). It is also of great importance to create an archive file for the whole dam and to regularly update
it (SCD, 2005). A monograph can also be used to summarize the essential events in the lifetime of the
dam (SCD, 2002).

Monitoring system and analysis of the results

The case of Zeuzier clearly underlined the importance of the continuous supervision, the
implementation of a large monitoring system as well as the fast analysis of the results of the
measurements. It also highlighted the required means to be implemented (particularly the geodetic
tools) in order to detect, in time, any abnormal behaviour and explain the causes. The monitoring system
has to be designed to take into account measurements of loads and its consequences on the dam,
foundations and surrounding area.

In the concrete body of the dam, the pendulum is a simple and effective means. However, several other
instruments can be placed at the foundations to detect structural deformation (e.g., sliding micrometers),
seepage flows as well as drainage and infiltration. Geodesy was also very useful to explain the
deformations observed at Zeuzier. Nevertheless, this kind of specialized measurements requires
certain precautions particularly when performing the measurements during winter. Within the
development of the geodetic measurement, the GPS system offers an elegant method to integrate
stable geological points into the control network outside of the zone of deformation.

Analysis of the results

While the analysis of raw measurements of deformations allows to have good insight on the nature of
the behaviour of the dam, the comparison between foreseen deformation and observed deformation
offers the possibility to refine this analysis. The difference has to be limited within a given margin of
tolerance if the behaviour is regular (SCD, 2003).

Incidence of underground works

Underground works such as galleries or tunnels can influence the hydrogeological conditions due to
drainage in an important zone. Consequently, if such works are planned at a maximal distance of 3 to
4 km from a dam, precautionary measures must be taken (e.g., extension of the monitoring system,
increase of the frequency of the measurements). For example, such measures were taken during the
drilling of the Alpine transversals of Lötschberg and Gotthard. Deformations were noticed at the Nalps
dam (close to the Gotthard Massif), which confirms the rightness of these measures.

73
SELECTED REFERENCES

Biedermann, R. et al. 1982. Comportement anormal du barrage-voûte de Zeuzier. Abnormal behaviour

of Zeuzier arch dam. wasser energie luft – eau énergie air, Special issue to the 14th ICOLD Congress,
Rio de Janeiro, Brazil, 1974(3): 65-112. Baden, Switzerland.

Lombardi G. 1992, "The F.E.S. rock mass model - Part 1", Dam Engineering, Vol. III, Issue 1, February
1992, pp. 49-72, "The F.E.S. rock mass model - Part 2", Dam Engineering, Vol. III, Issue 3, August
1992, pp. 201-221.

Lombardi, G. 1996. Tassement de massif rocheux au-dessus de tunnels. Symposium Géologie

AlpTransit, Zurich février 1996.

Lombardi, G. 2004. Ground Water induced settlements in rock masses and conséquence for dams.
IALAD Integrity Assessment of Large Concrete Dams. Conférence in Zurich

Pougatsch, H. 1990. Le barrage de Zeuzier. Rétrospective d'un événement particulier. wasser energie
luft – eau énergie air, (82)9: 195-208. Baden. Switzerland (in French).

Pougatsch, H. & Müller, R.W. 2002. Alp Transit und die Talsperren. Sicherheit ist oberstes Gebot.
wasser energie luft – eau énergie air, (94)9/10: 273-276. Baden, Switzerland (in German).

SCD. 1997. Surveillance de l'état des barrages et check lists pour les contrôles visuels. Comité suisse
des barrages – Swiss Committee on Dams, Switzerland (in French).

SCD. 2002. Comité suisse des barrages – Swiss Committee on Dams : Monographie de barrage.
Recommandation pour la rédaction (in German, French, Italian).

SCD. 2003. Comité suisse des barrages – Swiss Committee on Dams : Methods of analysis for the
prediction and the verification of dam behaviour. wasser energie luft – eau énergie air, Special issue to
the 21th ICOLD Congress, Montréal, Canada (95)3/4: 73-110. Baden, Switzerland.

SCD. 2005. Comité suisse des barrages – Swiss Committee on Dams : Dossier de l’ouvrage
d’accumulation. Recommandations (in German, French, Italian).

SCD. 2006. Comité suisse des barrages – Swiss Committee on Dams : Dam Monitoring Instrumentation
/ Dispositif d'auscultation des barrages (Part 1 and 2). wasser energie luft – eau énergie air, (98)2: 143-
180. Baden, Switzerland.

SCD. 2013. Comité suisse des barrages – Swiss Committee on Dams : Geodäsie für die
Ueberwachung Von Stauanlagen – Empfehlungen für der Einsatz der geodätischen
Deformationsmessung bei Stauanalgen.

SCD. 2015. Comité suisse des barrages – Swiss Committee on Dams : Role and Duty of the Dam
Wardens. Level 1 surveillance of water retaining facilities. Special issue to the 25th ICOLD Congress,
Stavanger, Norway.

Pougatsch H. et al. 2011. Improvement of safety of Swiss dams on the basis of experience. Proceedings
International Symposium on Dams and Reservoirs under changing Challenges, Symposium at ICOLD
Annual Meeting Lucerne, pp. 145-152

74
 TETON DAM FAILURE

J. Fleitz, Spain

ABSTRACT Case history category: e. This article is mainly an adaptation of a paper published at
the annual ASDSO conference in 2009 on the experience made with Fontenelle Dam, Ririe Dam and
Teton Dam in the USA and its influence on technical, social and organizational innovation. The Teton
Dam failed in June 1976 and originated subsequent changes of the dam engineering community as
well as the public’s perception of dam safety. Many countries adopted or reviewed guidelines and
regulations in the field of design, construction control, surveillance and safety due to the experience
of the Teton dam breach.
KEY WORDS: Dam failure, seep, soil arching, hydraulic gradient, emergency action plan

TECHNICAL DETAILS

The Teton Dam was designed to provide irrigation, flood protection, and power generation in the lower
Teton region of southern Idaho in the USA. The dam was designed as a zoned earthfill dam with a
maximum height of 123.4 m above the lowest point of excavation. The crest length was 945 m and
there were 5 embankment zones:

 Zone 1 - Central Core

 Zone 2 - Upstream and Downstream material adjacent to Zone 1 and in a blanket under zone
3 in the river valley and abutments
 Zone 3 - Random fill downstream of zone 2
 Zone 4 - Upstream cofferdam, later incorporated into upstream toe of dam
 Zone 5 - Protective exterior upstream and downstream rockfill

Figure 16 shows a plan view of the embankment as it was constructed.

Figure 16. Plan view of Teton Dam

75
During the feasibility phase of the project, a pilot-grouting program was completed in the area of the
key trench on the upper portion of the left abutment. The results showed that above El. 1554 m, the
upper 20 m of rock was so permeable that blanket grouting was not practical from a cost standpoint.
To compensate for the high grout losses in the pilot grouting program, a key trench was designed above
El. 1554 m to connect the embankment core to the rock foundation.

To minimize costs, the trench was narrow - only 0.8 m wide at its base - and had sidewalls or side
slopes of 0.5H:1V for most of its length. On the left and right abutments, the sidewalls are near vertical
in some locations. The deep key trench was a first for Reclamation and won a design award for its cost-
effective approach, which was later rescinded.

Construction of the dam began in February 1972 and was substantially completed in November 1975.

ENVIRONMENT AND FUNDING

Before construction of the dam could begin, a group of environmental organizations filed a complaint in
Idaho District Court in 1971 to prevent construction of the dam. The lawsuit was dismissed from Federal
District Court. The legal actions continued through 1974, with the plaintiffs alleging violations of
numerous laws, including the National Environmental Policy Act of 1969 (NEPA). On December 23,
1974, the Ninth Circuit Court filed an opinion affirming the District Court's dismissal of the case,
effectively ending the lawsuit. One of the main claims made by the plaintiffs was that the costs and
benefits were misrepresented by the Government. The cost-benefit ratio for the Teton Dam project was
1.0:1.75, which is not a high ratio. Design engineers, construction forces, and other Reclamation
employees who were associated with the Teton Dam project recall being constantly reminded about
the extremely tight budget for the project. The likely source of the cost pressure was the cost-benefit
ratio and the somewhat difficult process obtaining authorization in the face of litigation.

GEOLOGY

The Teton Dam site is adjacent to the eastern Snake River Plain, a volcanic filled depression that was
formed by downwarping and downfaulting in late Cenozoic time. Older volcanic rocks are not exposed
along the edges of the plain. The Teton River incised into a portion of the volcanic upland near the
eastern end of the plain creating a steep-walled canyon at the dam site. The site is in an area of
generally low seismicity. The foundation cut-off trench was excavated into bedrock along the entire
length of the dam. The regional groundwater table is far below the river, though perched groundwater
can be found above the channel.

The canyon walls are composed of a rhyolite welded ash flow tuff. The tuff was exposed in some areas,
but talus slopewash and alluvial deposits predominantly cover it. The welded tuff is between 15 m to
180 m thick near the dam site and has prominent and abundant jointing intersecting and high and low
angles. Most of the joints are near-vertical. The major joint set, strikes N25W to N30W, and is well
developed on both abutments and in both outlet tunnels. A second joint set, striking N60W to N70W, is
well developed in the lower upstream part of the right abutment, the river outlet works tunnel, and the
downstream portion of the auxiliary outlet works tunnel. A minor set of northeast-trending, high-angle
joints is also present in the welded tuff.

Continuous high-angle joints in the right abutment have been traced for lengths of as much as 60 m,
but most are between 6 m and 30 m long. The aperture of most high-angle joints is less than 4 cm, but
many joints are as much as several decimetres or even bigger. Examples of the jointing can be seen in
Figure 17.

76
 Figure 17. Right abutment key trench showing the jointed nature of the foundation

Many joints are open; others are partially or wholly filled with clay, silt, silty ash, soil, or rubble, especially
near the natural ground surface. The permeability of the welded tuff is due entirely to the presence of
open joints. The joints are most abundant and open, and rock-mass permeability is much higher above
El. 1554 m. Many of the joints were infilled with erodible material that would soften on contact with
reservoir water.

Underlying the welded tuff are materials of lacustrine, alluvial, and pyroclastic origin. Information about
these materials has come mainly from drill holes, commonly with poor or no core recovery, and to some
extent from deep grout holes, and thus is rather fragmentary. Although there is little information about
the various units underlying the tuff, sand and gravel and variably cemented sandstone and
conglomerate are commonly present. Thick claystone and siltstone are present under at least part of
the left abutment and channel section. Thin ash-fall tuff and other pyroclastic materials were found
below the welded tuff in some core holes. The contact between the sedimentary materials and the
welded tuff is an irregular erosion surface with a local relief of at least 135 m and some slopes steeper
than 30 degrees. The permeability of most of the sedimentary materials is less than that of the intensely
ractured welded tuff, but is highly variable. The sedimentary materials are at least 120 m thick; the depth
to the materials underlying these sediments is unknown.

Basalt is present in the bottom of the Teton River Canyon and is a remnant of a lava flow that filled the
canyon to about El. 1526 m (see geologic section, Figure 18). In the dam foundation, the basalt is
restricted to the left side of the river channel section, where it has a maximum thickness of about 38 m.
It is separated from the underlying welded tuff by a deposit of alluvial material consisting of silt, sand,
and gravel from 1.2 m to 6.7 m thick. The basalt is dense to moderately vesicular and contains closely
spaced, randomly oriented joints and other fractures. In spite of its fractured nature, it is an adequate
foundation rock for the dam. Water pressure tests showed the basalt to be tight and the thin alluvial fill
between the basalt and the welded tuff to be permeable.

77
 Figure 18. Profile of the Teton embankment and foundation looking downstream

FOUNDATION PREPARATION

The cut-off trench was excavated to competent rock. Foundation grouting was completed through a
0,9 m wide concrete grout cap that ran the length of the centreline of the cut-off trench. Overhangs were
removed where possible, although excavations in the left abutment remnant found areas where
overhangs were not removed and abutment rock was not shaped. On the right abutment, shaping was
limited because the location of the spillway prevented dramatic changes to the profile. No ‘dental
concrete’ was used and no slush grouting was used on the foundation surface above El. 1585 m. Below
El. 1585 m, ‘dental concrete’ was limited to using structural concrete leftover from placements
elsewhere on the site, as there was no bid item for either dental concrete or foundation treatment. The
rock surface treatment may have hastened the failure, because the criterion for treatment was not based
on condition of rock, but on when excess concrete was available. ‘Slush grout’ was similarly taken from
leftover foundation grouting materials.

Special compaction was done at the contact between the embankment and the foundation rock. During
excavation of the left abutment remnant, it was discovered that special compaction was impossible to
perform next to some of the open jointing. Foundation grouting was difficult given the extraordinary
permeability of the Welded Tuff. Occasionally, structural concrete was poured in the large fissures and
voids in the foundation where grout was considered inadequate. There was no consistent method to
treat the rock foundations within the key trenches and there was no strong direction from designers as
to how to treat the foundation. It is apparent that none of the designers knew how to adequately treat
the foundation and that the organization did not understand the importance of doing so. Treating the
foundation would also have meant ignoring the cost and schedule pressure from the construction office.

DESIGN CONSIDERATIONS

The design for Teton Dam followed standard practices with an impervious core and progressively
coarser materials used as the zoning progressed downstream. In the maximum section, a relatively
shallow-sloped cut-off leads to a narrow 9 m wide contact with a competent foundation. The intent of
the design was to rely on the grout curtain and the cut-off to create an impermeable barrier to protect
the main core of the dam. As shown in Figure 19, there are no additional defensive measures provided
in either the main embankment or the abutment sections.

78
 Figure 19. Teton Dam maximum section

On each abutment, the side wall of the cut-off trench steepens to 0.5H:1V. However, owing to the
steepness of the abutment looking parallel to the axis, the actual sidewalls approach vertical on both
abutments, which can easily be seen in Figure 20.

Figure 20. A view looking towards the left abutment from the right abutment

79
TETON DAM PERFORMANCE

On October 3, 1975, reservoir filing began; the reservoir was then at elevation 1542 m. From January 1,
1976 to April 5, 1976 the reservoir rose from elevation 1569 m to elevation 1577 m, or about 9 cm per
day. From April 5 to June 1 the reservoir rose 36 m or about 58 cm per day. During the first 5 days of
June, the reservoir level was increasing at an average rate of 70 cm per day.

The dam behaviour seemed to be normal, but when the reservoir level approached the spillway, on
June 3, an inspection team downstream of the dam found clear water seeping from the ground at two
locations: one at 400 m and another at 450 m downstream. By next day another seep had formed 40
to 60 m downstream. All three seeps formed downstream of the north side of the dam, but none were
considered serious hazards.

Beginning at 7:00 a.m. on June 5, witnesses discovered the first seep on the dam itself and, shortly
after, found a second seep. Turbid water flowed from both leaks on the north side of the dam. At 10:15
a.m., a wet spot formed and water began to leak out and erode the dam. Engineers recalled hearing a
loud noise and rushing water at 10:30 a.m. Project Construction Engineer Robert Robison stated, “This
leak developed almost instantaneous at about 10:30 a.m. and let loose with a loud roar”. Within 10 to
15 minutes, two dozers began attempting to plug the hole and the County Sheriff was alerted to start
downstream evacuations. Behind the dam, a whirlpool formed at 11:00 a.m. and grew rapidly despite
efforts to dump rock in and seal the hole. At 11:30 a.m., drivers abandoned the two dozers on the front
of the dam as the hole expanded and swallowed both dozers. One driver recalled running to get in
another dozer to help with whirlpool efforts atop the dam. Fifteen minutes later these drivers were pulled
back. At 11:57 a.m., the entire north embankment breached and the nearly full reservoir came crashing
through the earthen wall. The final report to the DOI presents a detailed hourly account of the dam’s
failure.

Although the time when the dam failed avoided the problem that lack of attendance would have caused,
there was no one at the dam site from after midnight (12:30 a.m.) to 7:00 a.m. the day of the failure.

The dam failed with the reservoir at elevation 1616 m, 9 m below the dam crest. At the time of failure,
the reservoir contained about 310 million m 3 of water. A total of 296 million m 3 of water drained from
the reservoir within about 6 hours.

At the time the failure occurred, the power generating station was not yet complete and the auxiliary
outlet works was under construction, limiting the release capacity of the facility during first filling. As the
reservoir began to fill for the first time, a large amount of water pressure built up quickly in the lower
portion of the cut-off trench on the right abutment probably between El. 1539 m and El. 1585 m.
Untreated joints in the welded tuff in the upstream wall of the cut-off trench allowed full reservoir head
to build up on the upstream face of the cut-off and an extraordinarily high gradient to be induced across
the trench. Because of the narrowness of the cut-off trench, there were potentially low-stress areas due
to soil arching. Soil arching in the right abutment key trench was demonstrated using finite element
analyses. Following the investigations of the Independent Panel and the Interior Review Group,
Jaworski, Seed and Duncan demonstrated by laboratory testing that soil on the upstream face of the
key trenches that was adjacent to open joints would soften and collapse after wetting, allowing reservoir
water closer to the grout cap and further increased the already enormous hydraulic gradient across the
floor of the key trench.

Either in the lower portion of the cut-off or across the top of the grout cap, seepage from the high
gradients began to erode material from the downstream face of the cut-off trench and carry it through
the open jointed welded tuff to unprotected exits in the valley wall and through the pervious downstream
embankment zones. Eventually, enough material washed out allowing erosion to progress upstream
and creating an open pipe that connected to the upstream open joints and eventually connected to the
reservoir. The pipe enlarged, progressed vertically and laterally, and breached the crest of the

80
embankment causing a catastrophic failure and release of the entire pool. The uniqueness of the failure
was the very rapid progression from discovery of a rather minor seep at the abutment to the total
collapse of the dam. The speed of the progression to failure highlights the gross inadequacy of design
and treatment to control seepage through the foundation

10:45 am 11:20 am 11:30 am

11:50 am 11:55 am Shortly after noon

Figure 21. Photo sequence showing Teton Dam breach in roughly 90 minutes on June 5th, 1976

This embankment failure is the largest in the USA, based on the dam’s structural height. 800 km 2
extending 130 km downstream were fully or partially inundated. 25 000 people were displaced, 11
people were killed, and the flood caused approximately $400 million in direct and indirect damage.

LESSONS LEARNT

Many interviewees from both the Corps and Reclamation expressed surprise that a dam could exhibit
signs of distress at 7:00am and fully breach by 11:30am. Before Teton Dam, this seemed implausible
to even experienced dam designers. The rapid nature of the failure and the vast devastation
downstream changed both the dam engineering community and the public’s perception of their own
safety.

For these reasons the Teton Dam failure has been considered as a benchmark case history, although
even with the most appropriate surveillance and monitoring activities the dam break could not have
been prevented due to its fast evolution during the first filling. The experience of Teton Dam lead to a
complete review of the state of art and originated guidelines and regulations that among other aspects
of design and construction control developed comprehensive safety and surveillance procedures. Many
countries adopted or reviewed guidelines and regulations in the field of design, construction control,
surveillance and safety due to the experience of the Teton Dam breach.

The key overarching conclusion from the failure is that the design of Teton Dam was not uniquely
tailored to the site. The intense fracturing and open joints were not compatible with low-to-no plasticity
silt. The left abutment grouting program should have been an indication of how susceptible the
embankment might be to erosive forces. Instead of incorporating this into the design philosophy, the
designers attempted to change the parameters of the problem by using the key trench in place of blanket
grout.

Engineers that began their careers following the failure of Teton Dam started with an entirely new
attitude that embraced state-of-the-art ideas, external review of decisions, and significantly more

81
interaction between design and construction forces. Following the failure, Reclamation focused
significant energy on education and it thrived on in-house and external educational opportunities

The final lessons to be learned from the failure of Teton Dam are:

 Use more than one line of defence against seepage.

 Flaws can occur in man-made structures, and defence measures should be designed assuming
that flaws do occur.
 External review of designs and decisions is a key step to evaluate the safety of a structure
 Critiquing problems and discussing controversial conditions is an important step to understand
problems and planning the resolution to them.
 A central presence to facilitate communication between geologists, designers and construction
forces is important.
 Foundation approval documented by designers and geologists for each square meter of
material placed is an important consideration. Digital records of the foundation inspections
should be required in the specifications.
 Incidents and failures should be openly discussed and presented as learning tools for all dam
engineers.
 The decision structure of the organization must be continually observed and evaluated to see
that effective decisions are being made.
 Communication with the downstream population is an important step to mitigate potential
disasters.

Specific aspects regarding dam safety and surveillance that have been developed due to the experience
of the Teton Dam failure are:

 The need for failure mode oriented and dam specific design of monitoring systems, especially
the importance of an adequate seepage control in embankment dams
 Adequate and gradual first filling programs including specific inspection and monitoring
activities to control and follow-up
 Fully operative discharge facilities to reduce the reservoir level if necessary
 Implementation of emergency action plans

SELECTED REFERENCES

Snorteland, NJ, Shaffner, P & Dave Paul, D (2009): Fontenelle Dam, Ririe Dam, and Teton Dam: An
Examination of the Influence of Organizational Culture. Presented at Dam Safety 2009, held Sept. 27-
Oct.1, 2009 in Hollywood, Florida.

IDAHO Bureau of Homeland Security. 1976 Teton Dam Collapse

http://bhs.idaho.gov/Pages/History/DamCollapse.aspx.

Graham, WJ (2015). The Teton Dam Failure - An Effective Warning and Evacuation. Presentation at
the Association of State Dam Safety Officials 25th Anniversary Conference, Indian Wells, California,
September 2015. http://damfailures.org/wp-ontent/uploads/2015/07/075_The-Teton-Dam-Failure.pdf.

Rogers, JD & Hasselmann, KF. Retrospective on the Failure of Teton Dam.

http://web.mst.edu/~rogersda/teton_dam/Retrospective%20on%20Teton%20Dam%20Failure.pdf

82
 INTENTIONAL DAM BOMBING ACTIONS DURING WORLD WAR II: MÖHNE, EDER
AND DNIEPROSTROI DAMS

Pierre Choquet, Canada

ABSTRACT Case history category: e. This article collects publicly available information about
intentional dam demolition actions on dams by air bombing raids or bombing actions during World
War II.

MÖHNE AND EDER DAMS

The Royal Air Force of Great Britain conducted air bombing raids on a number of dams in the Ruhr
region of Germany in the night of 16-17 May 1943.

Three dams were mainly targeted:

 Möhne Dam (masonry gravity completed in 1913, 40.3 m high, 777 m crest length, 7.6 m crest
width and 30.5 m width at the base).
 Eder Dam (masonry gravity completed in 1914, 48 m high, 393 m crest length, 6 m crest width
and 36 m width at the base).
 Sorpe Dam (earth embankment completed in 1935, 61 m high, 640 m crest length).

Möhne and Eder Dams were hit by 2 bombs each and were breached (77 m wide breach for the Möhne
Dam and 50 m wide breach for the Eder Dam) in addition to their powerhouse sustaining heavy damage,
leading to wide-scale floods and casualties of approximately 1 650 people, in addition to hampering
Nazi Germany’s industrial capacity for the war effort. Although hit by 2 bombs, Sorpe Dam sustained
only minor damages but no breach. Figure 1 and 2 show photos of the breached dams.

The operation was codenamed “Operation Chastise” and its success was a big boost to the morale of
the Allies. A specially developed bomb called “bouncing bomb” was used to attack the dams. It relied
on a 500 rpm back spin applied prior to launching to a barrel-shaped bomb, about 1.5 m (5 feet) long
and 1.2 m (4 feet) diameter, so that it would bounce on the surface of water when launched from a low
altitude of about 18 m (60 feet) and thereby avoid protective torpedo nets that had been deployed
upstream of the dams. Upon reaching the dam upstream face, the spin would help the bomb to go down
in water while remaining near the dam face and explode at depth by means of a hydrostatic fuse.

More details on these operations, methods used, aircraft, and subsequent damages can be found easily
online, including in the list of references below. The operation was also the subject of a 1955 movie
called “Dam Busters” which was very popular at that time.

83
 Figure 22. Breached Möhne Dam (Source: Bundesarchiv Bild 101I-637-4192-20, Zerstörte
Möhnetalsperre)

Figure 23. Breached Eder Dam (Source: Bundesarchiv Bild 183-C0212-0043-012, Edertalsperre,
Zerstörung)

84
DNIEPROSTROI DAM

Dnieprostroi (Dnjeprostroj) Dam is a 60 m high concrete gravity hydropower dam with a 800 m crest
length, completed in 1932 by the Soviet Union on the Dnieper River in Ukraine.

In September 1941, retreating Soviet troops reportedly detonated 90 tons of dynamite in a gallery of
the dam. The upper part of the dam, about 200 m long was breached (Figure 3) and the resultant outflow
of water caused casualties of between 20 000 and 100 000 people. The exact number is still discussed
by historians, as little information is publicly available on the true sequence of events and its
consequences. Some of the available online references are listed below. More references in Russian
language can be found online. The pressure of the blast caused also heavy damages in the
powerhouse. Additionally, retreating German troops dynamited the dam in 1943.

Figure 24. Breached Dnieprostroi Dam (Moroz & Bigg: 2013)

LESSONS LEARNT

To those who wonder if dams are invincible, the answer is unfortunately no. Dams can be damaged or
destroyed by natural forces (internal erosion, uplift forces, overtopping, slope failure, etc...), but can
also be damaged by man created forces. In both cases, consequences can be extremely damaging.

While the dam demolition actions reported in this case history refer to actions that have taken place
during World War II, and have caused very high casualties, it should be noted that hope exists that such
dramatic events will never occur again. Namely, the “Protocol Additional to the Geneva Conventions
of 12 August 1949, and relating to the Protection of Victims of International Armed Conflicts” dated
8 June 1977 states the following in Article 56 entitled “Protection of works and installations containing
dangerous forces”:

“Works or installations containing dangerous forces, namely dams, dykes and nuclear electrical
generating stations, shall not be made the object of attack, even where these objects are military

85
objectives, if such attack may cause the release of dangerous forces and consequent severe losses
among the civilian population.”

SELECTED REFERENCES

1- Möhne, Eder and Sorpe Dams:

https://en.wikipedia.org/wiki/Operation_Chastise

http://www.raf.mod.uk/history/bombercommanddambusters21march1943.cfm

http://www.thedambusters.org.uk/index.html

2- Dnieprostoi Dam:

https://en.wikipedia.org/wiki/Dnieper_Hydroelectric_Station#WWII_and_Post-war_reconstruction

Moroz, D. and Bigg, C., August 23, 2013. Ukrainian Activists Draw Attention To Little-Known WWII
Tragedy,. http://www.rferl.org/content/european-remembrance-day-ukraine-little-known-ww2-
tragedy/25083847.html

3- Other:

Jansen, R. B., 1980 (reprinted 1983). Dams and Public Safety, U.S. Department of the Interior, Bureau
of Reclamation, 1980 332 p., Part IV, Significant Accidents and Failures, Dnjeprojstoj dam, p. 135; Eder
dam, p.137; Möhne dam, p. 164.

Protocol Additional to the Geneva Conventions of 12 August 1949, and relating to the Protection of
Victims of International Armed Conflicts (Protocol I), 8 June 1977, https://www.icrc.org/ihl/INTRO/470

86
 FOLSOM DAM SPILLWAY GATE FAILURE

Pierre Choquet, Canada

ABSTRACT Case history category: c, d, e. This article collects publicly available information about
the Folsom Dam spillway gate failure that occurred in 1995 and the subsequent re-evaluation
programs on spillway gates that were conducted all across the United States in the subsequent years
and led to improved design guidelines.

FOLSOM DAM

Folsom Dam is a concrete gravity dam located approximately 20 km northeast of the city of Sacramento,
California on the American River. It was completed in 1956 by the U.S. Corps of engineers and is now
operated by the U.S. Bureau of Reclamation and as part of the Central Valley Project . The dam consists
in a central concrete gravity dam 104 m height and crest length of 427 m and two embankment wing
dams of 44 m height and 640 m crest length for the left wing dam and 44 m height and 2040 m length
for the right wing dam (Figure 1).

The spillway of the dam is located at its center and is of the gated overflow type. It is divided in eight
equal sections separated by piers. Flow through the spillway is controlled by five service spillway Tainter
gates (also called radial gates) and three emergency Tainter gates. The maximum discharge capacity
is 14,940 m3/sec. Outlet works sluiceways consisting of eight 1.5 m x 2.75 m rectangular conduits are
located in two tiers through the bottom part of the service spillway.

Figure 25. Folsom Dam and the Folsom Lake (Source: United States Army Corps of Engineers,
Michael Nevins - U.S. Army Corps of Engineers photo ID 040316-A-3200N-112)

The Bureau of Reclamation later designed and constructed a power plant with three Francis units for a
total capacity of 198 MW. The three power penstocks, visible on Figure 25, pass through monoliths on
the left-hand side of the service spillway. Their intakes are one the upstream dam face.

87
SPILLWAY GATE FAILURE

On the morning of July 17, 1995, the Folsom Dam power plant was shut down and Spillway Gate 3 was
opened to maintain flows in the American River. The failure occurred while the gate was operated
(Figure 26). There was no warning of structural distress prior to the failure. No one was injured, even
though there was a sustained release of 1,132 m 3/s into the Lower American River.

Nearly 40 percent of Folsom Lake drained out past the broken gate before it could be repaired. Luckily,
no major flooding occurred as a result of the failure, and the Folsom Dam was fully repaired.

Figure 26. Failed Tainter gate No. 3 at Folsom Dam (Source:

http://www.pbs.org/wgbh/buildingbig/wonder/structure/folsom.html)

OTHER EVENTS SUBSEQUENT TO GATE FAILURE

A large flood occurred while the damaged gate was being replaced. This required one of the emergency
gates to be operated to provide the discharge capacity lost by the damaged gate. Operation of the
emergency gate resulted in damage to the impact area downstream of the flip bucket.

Additionally, as a result of the large flood and the ongoing repairs to the spillway gates, the outlet works
sluiceways were used more extensively to pass flows. This required all outlets to be used concurrently
for an extended period of time, which was not a normal operation. At the conclusion of the extended
operation, severe cavitation damage was found in two of the outlet works conduits, and minor cavitation
damage was found in other conduits. All these damages were subsequently investigated and repaired
(Boyer et al. 2013).

In early 2000s, a comprehensive dam safety review of the Folsom Project was initiated. This review
resulted in the replacement of the spillway gate and reinforcement of the spillway piers (Boyer et al.
2013).

88
INVESTIGATION OF THE SPILLWAY GATE FAILURE

A multi-disciplinary, multi-agency forensic team was immediately formed in 1995 to investigate the
spillway gate failure. All of the remaining gates were thoroughly inspected for signs of structural
degradation, apart from some corrosion, nothing detrimental was found. The failed gate was removed
and then thoroughly examined to determine the mode of failure (Todd, 1999). This examination
determined that a diagonal brace joint, adjacent to the trunnion was the initial point of failure. Theoretical
finite element models substantiated this study when a trunnion friction coefficient of 0.25 was
considered when simulating the forces due to gate lifting. This value had been confirmed also from
friction tests performed on the actual trunnion from the failed gate. It was therefore concluded that
trunnion friction moment was the key factor, and it had been omitted in the original design calculations.

As a general working principle, the hydraulic load on Tainter gates is transmitted from the cylindrical
skin plate, which is in contact with the reservoir, through a number of struts to a convergence at the
trunnion hub. The hub collects the load from the struts and transfers it across an interface to the trunnion
pin, which is stationary and is connected to the dam. When the gate is operated, the hub rotates around
the pin. The struts are primarily compression members, but friction at the pin-hub interface induces a
bending stress during gate operation. Typically, and in the case of Folsom Dam as well, the struts are
oriented such that the trunnion friction stress is applied to the weak axis of the struts. In order to better
handle these loads, the struts are connected with diagonal braces that take the stress as axial loads.

At Folsom Dam, the failure initiated at a diagonal brace between the lowest and second lowest struts.
Increasing corrosion at the pin-hub interface had raised the coefficient of friction and, therefore, the
bending stress in the strut and the axial force in the brace. The capacity of the brace connection was
exceeded and it failed. This caused the load to redistribute and the failure progressed, eventually
buckling the struts.

A comprehensive re-evaluation program of spillway Tainter gates was initiated in August 1995 by the
California Division of Safety of Dams and was conducted on 239 gates on 57 different California dams
and involved physical inspection of each gate, finite element modeling of each design, as well as
determination of proper loading and acceptance criteria (Schultz et al., 2007). The Federal Energy
Regulatory Agency (FERC), in coordination with state dam safety official, launched a similar re-
evaluation effort on a nation-wide scale. The U.S. Army Corps of Engineers (USACE) and the U.S.
Bureau of reclamation (USBR) also embarked on similar programs to evaluate Tainter gates within their
dam inventories.

LESSONS LEARNT

The main lesson learnt is that a coefficient of friction must be considered at the interface of the trunnion
hub and trunnion pin in the structural analysis of the lifting of a spillway gate. The generally accepted
value, after considerable debate, examination of several trunnion pins and based on a testing program
involving strain gauges and lasers is 0.3 and has been adopted by most U.S. agencies involved in
Tainter gate analysis (Schultz et al., 2007)

The various re-evaluation programs that were initiated after the Folsom Dam failure lead to a much
better understanding of their structural behavior and was at the source of a number of technical articles
and design guidelines (Todd, 1999; Schultz et al., 2007; U.S. Army Corps of Engineers, 2000; U.S.
Bureau of Reclamation, 1996).

The various studies conducted as part of the re-evaluation program helped also to emphasize the
importance of taking into account hydrodynamic loads in the design process, both flow-induced and
seismic, and led to technical articles such as U.S. Bureau of Reclamation (2011)

89
 SELECTED REFERENCES

https://en.wikipedia.org/wiki/Folsom_Dam

Modification of Existing Dams-Concrete-Hydrologic, Folsom Dam. In Achievements and Advancements

in U.S. Dam Engineering, Edited by Douglas D. Boyer, Richard L. Wiltshire, Glenn S. Tarbox, U.S.
Society on Dams, 2013, 792 p.

Todd, R., 1999. Spillway Tainter Gate Failure at Folsom Dam, California. In Waterpower '99: Hydro's
Future: Technology, Markets, and Policy. Edited by Peggy A. Brookshier

Todd, R., 2002. Determining Earthquake Loading on Spillway gates, Hydro Review,

Schultz, M., Jones, S. and Huynh, P., 2007. Summary of Results from the California Tainter Gate
Reevaluation Program. In Proceedings 27th Annual USSD Conference, Philadelphia, PA, pp. 459-468

U.S. Army Corps of Engineers, 2000. EM 1110-2-2702, Design of Spillway Tainter gates.

U.S. Bureau of Reclamation, 1996. Forensic Report on Spillway Gate 3 Failure, Folsom Dam.

U.S. Bureau of Reclamation, 2011. Seismic Induced Loads on Spillway Gates , Phase I - Litterature
review. Report DSO-11-06

U.S. Bureau of Reclamation, 2014. Design Standards No. 14: Appurtenant Structures for Dams
(Spillways and Outlet Works), Chapter 3: General Spillway Design Considerations.

90
 CAHORA BASSA DAM MONITORING

Ilidio Tembe and Chris Oosthuizen, Mozambique

ABSTRACT Case history category: a & e. The chain of monitoring aspects from conception
onwards as it has unfolded at Cahora Bassa Dam during the past 40 years is briefly described. Their
attention to these aspects, namely, design, installation, observation, maintenance, upgrading,
observations, data management and evaluation constitute best monitoring practice for large arch
dams.

TECHNICAL DETAILS

The Cahora Bassa Dam, located downstream of Kariba Dam in the Zambezi River, is the largest
hydroelectric scheme in Southern Africa. The basic statistics of the double curvature concrete arch dam
are as follows: 171 m high, 303 m crest length, 23 m thickness at the base, 4 m thickness at the top, 8
mid-level radial spillway gates and 1 surface gate. The reservoir created by Cahora Bassa dam has a
maximum volume of 65x109 m3 (the fourth largest in Africa with a net capacity of 52x109 m3). This
massive reservoir, approximately 270 km long and 30 km wide covers an area of 2 900 km2. The present
South bank underground power house has an installed capacity of 5 x 415 MW and the addition of the
planned North bank power house is presently under investigation.

DOWNSTREA M VIEW
RB LB
(m) 22 20 18 16 14 12 10 8 6 4 2 0 1 3 5 7 9 11 13 15 17 19
326.00
330

310
296.00

290
271.00
270

2
248.50
250

224.75
230 D8 D6 D4 D2 D1 D3 D5 D7

210 203.50

190 183.50
PERIP HERAL GALLERY 171 .00 FOUN
FOUNDATION GALLERY
170

145.00
150

VERTICAL CROSS SECTION VERTICAL CROSS SECTION

BY THE CENTRAL CANTILEVER BY THE ORIFICE SPILLWAYS
(m)
RWL = 326.00 RWL = 326.00
330
CREST DETAIL

310

290
331.00

270

244.30
250
231.00
326.00
230

210

2 .0 0 2 .0 0
190

170

150

Fig. 3 – Cahora Bassa dam. Downstream view, vertical cross-sections and crest detail.

Figure 27. Downstream view, vertical cross sections and layout of Cahora Bassa Dam
T he dam was built between September 1972 and March 1975 (Fig. 4). T he first filling
began on 7 December 1974, before the completion of the dam’s body concreting sequence

Construction commenced in 1969 and impoundment of the dam started in December 1974. The
(Fig. 5 ). T he first filling took place between December 1974 and June 197 6 and the
commercial operation of the scheme started in March 1977.
Laboratorio Nacional de Engenharia (LNEC) was directly involved with the design and installation of
the monitoring system and with the surveillance of the dam and Power station until 1978. Thereafter
the dam was basically operated by HCB staff and dam safety surveillance was mainly carried out by
the local HCB staff. LNEC was involved approximately every 5 years with visual dam safety inspections.
The original design addressed design assumptions and the ability to monitor the structure’s behaviour
LNEC - Proc. 0402/001/17055; 0403/541/317 5

in the long term. During construction special care has been taken by the instrumentation crew to install
the instruments by an experienced team from Portugal assisted by dedicated local personnel.

91
SURVEILLANCE DETAILS

A concise overview of the instrumentation is given in Table 5. The upgrading from time to time
addressed issues that were not foreseen during the original design, such as swelling of the concrete.
Only a few instruments stopped functioning after 40 years.

Table 5. Details of the monitoring system (parameters, measuring interval & recoding method)

Position Installa- Number & Parameter How

tion interval
Dam & found

Environment
Powerhouse

Automated
Frequency
Measuring

Response
Upgrades

Integrity
Number
Original

Manual
points

Loads
Monitoring equipment
Water Level
Gauge Plates X X 2 2 d X X
Recorders X X 2 3 2w X X
Limnometers X X 1 1 2w X X
Climate
Air temperature X 2 4 d X X
Relative humidity X 2 2 d X X
Rain gauge X 1 1 d X X
Temperature
Electrical thermometers X X X X 174 174 2m X X
Electrical strain gauges X X 6 12 2m X X X
Electrical stress gauges X X 11 22 2m X X X
Electrical creep gauges X X 4 8 2m X X X
Electrical joint meters X X 5 10 2m X X
Porewater pressures
Piezometers X X X 62 62 2m X X
Mechanical pore pressure X X 6 12 2w X X
Seepage X X X 664 664 m X X
Geodetic surveys
Triangulation X X X 49 147 6m X X
Levelling X X X X 33 33 6m X X
Traversing X X X 2 40 - X X
Vibrations
Accelerometers X X 13 19 c X X X
Seismometers X X X 5 15 c X X X
Displacements
Pendulums X X 7 28 2w X X
Manual extensometers X X X 64 64 2w X X
Electrical extensometers X X X 314 314 2m X X
Electrical joint meters X X 5 10 2m X X
Mechanical deformeters X X X X 184 552 3m X X
Mechanical convergence X X X 100 100 3m X X
meters
Deformation
Electrical strain gauges X X 6 12 2m X X X
Electrical stress gauges X X 11 22 2m X X X
Electrical creep gauges X X 4 8 2m X X X
Other
Anchorage X X 12 20 3m X X
d – daily, w – weekly, m – monthly, c - continuous

92
DESCRIPTION OF THE INCIDENT

During the mid-1990’s the swelling effect of AAR became noticeable and was studied at length for
several years by LNEC. It must be noted that the dam was designed and constructed prior to the
awareness in Southern Africa of the effect of Alkali Aggregate Reaction (AAR) on dams.

The original monitoring system has been meticulously designed and installed by the designers (LNEC).
It was well maintained by trained local personnel but had to be upgraded to address the swelling issue
that was not foreseen during the original design. Instruments were therefore required to monitor the
rate of swelling. Rod extensometers were installed along the crown cantilever during the late 1990’s.

The observed strains and strain rates in the dam wall were carefully assessed and came to the following
conclusions as far as the strain rates are concerned (Manitoba Hydro International, 2013):

 The rod extensometer strain rates correlate extremely well with the vertical strain rate results
of the originally installed Carlson strain gauges;
 The annual strain rates observed by the convergence meters (between 35-40 and 60 micro
strain) are similar to the estimated strain rates of the gate trunnions (around 40 micro strain);
and
 The precision levelling results recorded annual vertical swelling strain rates at the crest of
between 35 and 40 micro strain at the left flank that decreases to around 22 micro strain on the
right flank of the dam wall.
 From the results of the Carlson strain gauges the following is evident:
o When analysing the average “no stress” strain rates it is evident that the average
“stress free” swelling strain rates across the dam is approximately 25 micro strain (see
Table 2);
o The effect of confinement/stress is also evident through the dam section with the
average radial strain rates of around 25 micro strain close to both and downstream and
upstream faces decreasing to a round only 9 micro strain at the centre as well as
decrease of the “no stress” strain rates from 25 to 16 micro strain in the centre of the
section
o It is important to note that in general there is no significant change in the vertical strain
rates through the section of the dam;
o The apparent “stress free” strain rates as expected decreases with depth to around 20
micro strain at level RL 203 m.

Table 6. Carlson strain gauges: Average strain rates

Average strain rate (micro strain/year)

Strain direction (instrument
number) 1 m from 1 m from
Centre of block
downstream face upstream face

Vertical 16.7 17.0 15.6

Upstream/downstream (Radial) 25.8 9.33 26.5

“No stress” 24.0 16.5 25.7

93
 Figure 28. Carlson strain gauges: Strain: RL 254.75 m: Block 18-20: Centre: SACODA results

Figure 29. Carlson strain gauges: Strain: RL 254.75 m: Block 18-20: Centre: Recalculation of results
using original calibration constants since 2002

The static monitoring system was enhanced with a dynamic monitoring system in 2010 to measure the
modal parameters. Ambient excitation forces such as the natural vibrations of the hydropower plant,
outlets, wind, and waves. These systems are usually collectively referred to as Ambient Vibration
Monitoring (AVM). The relative positions of the AVM sensors as well as the seismographs are shown
in Figure 30.

94
 Figure 30. Downstream view and elevation showing the relative positions of the continuously
monitored AVM system installed at Cahora Bassa Dam

The AVM (Ambient Vibration Monitoring) system monitors the vibrations near the crest of the wall
continuously. The system was upgraded when the internal batteries of the existing seismometers had
to be replaced in 2013. The 3 additional tri-axial accelerometers installed in the dam foundation (shown
as larger red dots in Figure 30). These accelerometers give the system the capability not only to monitor
the dam’s response but would also provide the ‘reference’ seismic loads at the existing dam prior to the
envisaged future construction of the left bank power station.

The data collected from the accelerometers is recorded by means of a local data logger at the dam.
The data is saved in a file and at hourly intervals sent the central computer located in the dam safety
office using the so-called File Transfer Protocol (ftp) via a fibre optic cable. These files are generated
in ASCII format (*.txt) in the acquisition system and are processed, managed and stored in binary format
(*.bin) in the central computer using the SACODA data management software developed for arch dams
(Tecnobar, 2010). The hourly readings are converted to adjusted physical units of acceleration (g). The
Frequency Domain Decomposition (FDD) algorithm is used to identify the natural frequencies and
related modals parameters, i.e. modal shape and damping. Finally, the software stores these hourly
results in binary (*.bin) files. Typical results are presented in Figure 31. Mode shapes during relatively
low water levels on 2016-12-26 at 04:00 using FDD method. The dynamic results provide useful
information not only to detect minute changes in the behavior of the dam but also for the dynamic
calibration of the static Finite Element Models.

Mode 1 (f = 1.953
Hz) Mode 2 (f = 2.380
Hz)

Mode 3 (f = 2.942 Mode 4 (f = 3.516

Hz) Hz)

Mode 5 (f = 3.894 Mode 6 (f = 4.614

Hz) Hz)

Mode 7 (f = 5.139 Mode 8 (f = 5.664

Hz) Hz)

Mode 9 (f = 6.189 Mode 10 (f = ----- Hz) Not identified!

Hz)
Figure 31. Mode shapes during relatively low water levels on 2016-12-26 at 04:00 using FDD method

95
LESSONS LEARNT

Dynamic monitoring is a valued technique to monitor changes in the behaviour of arch dams subjected
to swelling and Continuous dynamic monitoring (AVM) is considered the ultimate real-time monitoring
of the response of arch dams. The state-of-the-art system operational at Cahora Bassa Dam is proof
of the high premium placed on dam safety monitoring.

Value of a well –designed, -installed, -maintained, -observed, -analysed and evaluated monitoring
system has been underlined. Monitoring is an interdependent chain of activities. If any link of the dam
monitoring chain is missing or weak, the whole chain is affected. The phenomenal quality of the results
of the old Carlston equipment in the studying the behaviour of the dam has been clearly demonstrates
the statement.

Value of diligent dam safety monitoring/surveillance personnel (especially instrumentation technicians).

The present dam safety surveillance staff are making a career of dam monitoring and with the necessary
management support their diligence is clearly reflected at Cahora Bassa Dam since construction. Skills
cannot be developed by training, only through practical experience. The dam safety surveillance team
is directly involved with all aspects at Cahora Bassa Dam, from the design, manufacture, and installation
of improvements up to the evaluation of the results.

Value of institutional Memory. Institutional memory is one of those less valued aspects of dam safety
surveillance. Dedicated personnel making a career of dam safety surveillance are precious sources of
institutional memory. A member of the dam safety surveillance team that was part of the original dam
instrumentation team at Cahora Bassa Dam, retired in 2015.

SELECTED REFERENCES

Carvalho, E.F. and Tembe, I. (2012): On-Line Dynamic Monitoring of Cahora Bassa Dam, published as
part of the Proceedings of the Symposium titled “Dams for a Changing Worlds” held during the ICOLD
2012 Congress in Kyoto, Japan

Carvalho EF, Masinge BT & Oosthuizen C (2016). Monitoring system of Cahora Bassa Dam….the past,
present and way forward. Proceedings of the ICOLD 2016 Symposium titled “Appropriate Technology
to Ensure Proper Development, Operation and Maintenance of Dams in Developing Countries”,
Sandton, South Africa.

Laboratorio Nacional de Engenharia (1978) Definicao dos sistemas de observacao e constants

characteristicas da aparelhagem. Proc. 43/2/3637. Lisbon, Portugal

Laboratório Nacional de Engenharia Civil (2009). Cahora Bassa Hydroelectric Scheme: Dam behaviour
analysis, interpretation and prevision. Relatorio 341/2009. Lisboa, Portugal.

Manitoba Hydro International (2013) Cahora Bassa Dam: Evaluation of Monitoring System and
Assessment of Dam Performance, Report authored by LC Hattingh & C Oosthuizen. May 2013. Songo.
Mozambique.

TECNOBAR (2010). Guide to SACODA package – dynamic analysis module. Lisbon, August 2010.
(Portuguese).

96
 ZOEKNOG DAM FAILURE

Chris Oosthuizen, South Africa

ABSTRACT Case history category: e. The installation of the instrumentation system of Zoeknog
Dam was done meticulously by an experienced instrumentation expert. He monitored and evaluated
the results during first filling (although it was not required from him). His alarm of a potential dam
failure was ignored and the dam failed within 2 weeks of his first alarm.

TECHNICAL DETAILS

Zoeknog Dam is located in the North Western corner of the Mpumalanga Province in South Africa. The
dam comprises an embankment with morning glory spillway and associated outlet works. It was
constructed primarily for irrigation of coffee plantations (for the Lebowa Homeland at the time). The
concrete works were constructed by a private contractor and the embankment by the Lebowa
government (a construction team with mainly road construction experience). The design and
supervision were done by a private consultant. As it was at the time a former “independent homeland”
dam it strictly speaking did not fall under the jurisdiction of the South African Dam Safety Regulations.
However, the owner undertook to comply with the regulations but did not submit all the necessary
documentation required timeously to the Dam Safety Office.

Construction started during 1990 on the 40 m high dam (above lowest foundation level). It was designed
as a zoned earthfill section with a central clay core and outer fill zones. The upstream slope varies from
1:4 at the bottom to 1:2,5 at the top. The downstream slope varies from 1:2,5 at the bottom to 1:2 at the
top. The central clay core had a slope of 0,8:1 and the 600 mm wide geotextile and sand chimney drain
was built in steps. The blanket drain was specified as a “Bidim U44” layer at the bottom with a with a
layer of sand and 300mm of gravel and a “Bidim U34” layer at the top. However, the blanket drain as
constructed comprised of coarse aggregates up 38mm surround by geotextile.

The embankment material comprised of leached granite from the basin area. Dispersive tests done
prior to and during construction tested non-dispersive. Apparently tests were always done on the upper
leached layers and not on the deeper lying material. The deeper layers of the borrowed material was,
however, highly dispersive. ASHTO specifications for optimum moisture content (OMC) were used and
not Proctor values. The dam was therefore constructed with highly dispersive material compacted at
dry of optimum moisture content (a fatal combination).

Figure 32. Zoeknog Dam started impounding December 1992 and failed in January 1993

SURVEILLANCE DETAILS

The layout of the monitoring system (Figure 2) was designed to study the behaviour of the right flank of
the embankment comprised the following instruments:

 Piezometers installed at two chainages (460 and 540) with three levels of instruments in each
line to measure the phreatic line from upstream of the clay core to downstream of the chimney

97
 drain. Both lines were on the right bank, the first 30 metres from the outlet conduit and the
second line of instruments 80 metres further
 Three settlement cells were added to the second line to measure the settlement at foundation
level.

The installation was done by an experienced private contractor (ex-Department of Water and Sanitation
(DWS), Fil Filmalter (Kop-Kop). With the installation of the first row of instruments, he discovered that
the blanket drain was not on actual founding level. The contractor apparently planned to install it on the
excavation level drawn on the drawings and not on the actual excavated level. Based on his experience
at earthfill dams Filmalter apparently complained regularly about the OMC of the clay material as he
had to use makeshift “dust masks” during installations when they were placing earthfill material.

Figure 33. Zoeknog Dam instrumentation drawing

DESCRIPTION OF THE INCIDENT

The events before and after the incident can be summarised as follows:

 Impoundment started just before the December 1992 construction break;

 When the works resumed early January, the instrumentation technician realised on 11 January
1993 that the pressure recorded by the piezometer upstream of the core at chainage 540 m
has shot up to the water level in the reservoir;
 The site staff treated it as a false alarm as the earthfill according to them could not be that
permeable;
 On 12 January Fil Filmalter was in the author’s office. The author told him that the Dam Safety
Surveillance teams of DWS (headed by the author) is unfamiliar with the dam as it does not fall
under his jurisdiction and referred him to the Dam Safety Office (the dam safety regulating
authority) that was indirectly involved. Fil stepped into the passage and shouted in Afrikaans

98
 the equivalent of “ I am telling you Zoeknog Dam is going to fail and you don’t want to do
anything” and left;
 On Sunday evening 24 January 1993 around midnight the security guard heard water flowing
and cautiously investigated. He found that the flow was not through the outlet pipe but on the
left-hand side of the outlets. Six hours later the dam was breached.
 Early morning of 25 January the dam was breached with no lives lost and fortunately no
significant damage to other infrastructure.
 Dam Safety Surveillance teams of DWS visited the dam on 2 and 4 February to investigate the
failure. It was suspected that it is a piping failure along the conduit and/or conduit failure leading
to the piping failure;
 A week later a taxi driver gave a photo (that he had taken when he passed by early on the
morning of the failure) to the site supervising staff. A photo that narrowed the forensic engineers’
task;
 The DWS team returned for another full-scale investigation (Barker et al 1993).
 The investigation team did their investigations and finally lodged their report to the Office of the
State Attorney on 25 May 1995. As far as this case history is concerned the following remark
is of importance “No blame for the dam failure can be directed at Kop-Kop Instrumentation” and
´The failure….to carry out regular dam inspections during initial filling of the dam may also have
contributed to the dam failure, as signs of distress or seepage could possibly have been
detected at an early stage during such inspections, prior to the failure.” (Keller & van Schalkwyk,
1995);
 The dam was abandoned after the slopes of the breach have been stabilised.

Figure 34. Zoeknog Dam on 2 February 1993 on the left (photo courtesy of Willie Croucamp) and the
photo that narrowed the investigation on the right (photo courtesy of an unknown taxi driver)

LESSONS LEARNT

The main lesson learned is to act on warnings of diligent instrumentation personnel. In the case of
Zoeknog Dam it would not have mattered. The obvious solution would have been to drain and breach
the dam, remove all the earthfill and start again from foundation level.

Value of appropriately- and well experienced instrumentation staff. That is basically a prerequisite for
the installation of a well-installed monitoring system.

99
 Figure 35. Abandoned Zoeknog Dam -note the clear signs of dispersive soils

……..In hindsight several design lessons could be learned…:

Dispersiveness of in situ weathered Granitic soils increase with depth (from non-dispersive to highly
dispersive). Only surface materials have been tested prior to construction and not thereafter.

Use Procter and not AASHTO compaction specifications for earth-fill embankment dams. The designers
used the latter specifications resulting in a less pliable clay core.

Backfill trench excavations with concrete instead of hand tamped clay. (excavation next to the outlet
conduit at Zoeknog Dam).

Ensure that the blanket drain is constructed on founding level and not on “assumed foundation levels”
indicated on design drawings (the actual levels were up to 5 metres lower in this case).

Use natural sand filters for chimney and blanket drains (not concrete aggregate wrapped in geotextiles
as was the case at Zoeknog).

SELECTED REFERENCES

Keller, H and van Schalkwyk, A. 1995. Failure of Zoeknog Dam: Investigation concerning the possible
liability of parties involved in the design, construction and supervision of the project. Report No
1561/K0001.Office of the State Attorney. Pretoria, South Africa.

Barker, MB. Oosthuizen,C and Elges, HFWK 1993. Zoeknog dam: Inspection following breaching of
the dam which occurred on 25th January 1993. Department of Water Affairs and Forestry (presently
DWS). Pretoria. South Africa.

100