Document Actions

ijege-13_bs-petaccia-natale.pdf

background image
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
437
DOI: 10.4408/IJEGE.2013-06.B-42
DESIGN FLOOD ESTIMATION:
LESSONS LEARNT FROM SELLA ZERBINO DAM-BREAK
G
abriella
PETACCIA & l
uiGi
NATALE
University of Pavia - Department of Civil Engineering and Architecture - Via Ferrata, 3 - 27100 Pavia, Italy
five victims per 100 million bodies (K
alustian
,
1995):
this statistics is clearly dominated by major catastro-
phes, such as South Fork and St. Francis in the USA,
Bouzi and Malpasset in France, Gleno, Zerbino and
Vajont in Italy, Banqiao in China, Tigra and Machhu
II dam collapse in India.
Many cases of dam failure were analyzed and sim-
ulated (b
osa
& P
etti
, 2011; n
atale
et alii, 2008; P
i
-
lotti
et alii, 2011; V
aliani
et alii, 2002; r
oGers
, 2006;
b
eGnudelli
& s
anders
, 2007). Research projects, like
CADAM (M
orris
, 2000), have been promoted and in-
ternational conferences focused on dam safety (n
atale
et alii, 1998; d
e
a
lMeida
& V
iseu
, 1997; d
ouGlas
et
alii, 1998; D.S.C., 1991) have been held.
The main causes of failure in concrete gravity dams
are listed in table 1 (i
cold
, 1987; i
cold
, 2005).
Analysing a sample of 100 failures of concrete
dam, J
ohnson
& i
lles
(1976) pointed out that the main
causes of collapse are: overtopping (35%), problems
concerning foundations (25%) and various causes due
to design and/or construction errors, poor materials,
earthquake and war actions (40%).
The number of masonry dam failures is high, even
though the worst disaster in the history happened when
a clay-core earth fill dam, the Banqiao Reservoir Dam
ABSTRACT
In 1925 two dams were constructed across the
Orba River to store water in Ortiglieto reservoir. The
13
th
of August 1935 the flood spillways of the main
dam - Bric Zerbino - were unable to discharge the flood
and both dams were overtopped. The main dam was not
damaged while secondary dam - Sella Zerbino - col-
lapsed. This paper describes in detail the rainfall event
that caused the dam break and critically compares the
historical peak discharge to the design discharges of the
existing dams located in Northern Apennine region.
K
ey
words
: dam safety, design flood, spillways design, Sella
Zerbino dam break
INTRODUCTION
Dams can be considered among the most reliable
structures, as indicated by the centuries-old experi-
ence in their construction and operation. Nevertheless
dam failures, including those with many casualties,
occur all the same. Failures and accidents of many
large dams in Europe and in USA have been recorded
(I
cold
, 1974; u
s
cold
, 1988): 500 accidents of various
severity occurred (200 of which were cases of failure)
out of a population of about 9000 dams constructed up
to 1965 (I
COLD
, 1991).
The failure probability is in the order of 10
-3
per
dam year (c
henG
, 1989; t
hanG
& Y
en
, 1991) and the
risk of victims due to the failure of dams of any type is
estimated to be 5.1 10
-8
persons/year, corresponding to
Tab. 1 - Main Causes of Failure in Concrete Gravity Dams
background image
G. PETACCIA & L. NATALE
438
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
At 7:00 a.m. the rain intensity increased and kept on
without interruptions until 3:00 p.m. The most intense
rain fell between 7:00 and 8:00 a.m. and between 2:00
p.m. and 3:00 p.m.
The synoptic representation of that very day can be
summarised as follows: the warm and humid wind com-
ing from south east over the Tyrrhenian Sea came across
the cold front descending from north, as confirmed by
the wind direction recorded during the event (a
lfieri
,
1936; c
oYne
,
1937; V
icentini
,
1936). The humid front
streamed towards the three gaps in the water divide of
the Apennine indicated by dashed lines in Figure 2 and
very intense showers hit Orba and Stura watersheds.
The records of daily rainfall at the gauging sta-
tions shown in Figure 2, indicate that the rainfall of
August 13
th
was: (1) the maximum ever recorded by
the rain gauges denoted by the symbol (○); the maxi-
mum in year 1935 for that indicated by the symbol
(■). The rainfall recorded elsewhere was not remark-
able (+). The area hit by the thunderstorm is limited to
an extension of 350 km
2
as indicated by the ellipse in
Figure 2 which is 39 km long and 14 km wide.
Due to the exceptional intensity of the rainfall, the
rain-gauge pans in the area were not emptied at 9:00
a.m. as prescribed by the Italian Hydrographic Service
operational protocol; as a consequence, the measure
of the cumulative rainfall height is assigned to very
day of the storm. Instead, the records of 6 rain-gauges
located along the shoreline, southward of the critical
area, assign the rainfall of the first part of the storm to
the previous day, the 12
th
of August.
in China, failed in August 1975 due to overtopping and
caused more than 171000 casualties. The design flood
discharge of this dam was 1-in-1,000-year while the
estimated return period of the 1975 flood was 2000
years (s
i
, 1998; G
rahaM
,
1999, X
u
et alii, 2008).
Also the masonry dam of Sella Zerbino broke due
to an extreme flood that the spillways were not able
to evacuate. The dam was overtopped and the falling
jet eroded the foundations of the dam. The dam break
drowned 111 folks (n
atale
et alii, 2008). After the event
the designers of the dam were prosecuted. The criminal
trial against the owner and the designers ended on the 4
th
July 1938 and the defendants were sentenced not guilty.
DESCRIPTION OF ORTIGLIETO RESER-
VOIR
The first plan to store water from Orba River,
see Fig. 1, date back to years 1895-1899. The origi-
nal project was soon modified to bring the reservoir
capacity from 12 to 18 hm
3
by increasing the height
of the main dam and obstructing a secondary saddle,
named Sella Zerbino, with a smaller dam. Accordingly,
two masonry gravity dams were built. The main dam,
called Bric Zerbino, is 44.0 m high and 145.5 m long.
The secondary dam, that was 14.5 m high and
120.0 m long (a
nonYMous
, 1925), was inappropriately
founded on a poor bedrock (n
oVarese
, 1938; P
eretti
,
1936; a
ccusani
, 1936; d
e
M
archi
,
1940). The flood
outlets of the main dam consisted in: 12 Heyn siphon
spillways, a gated side spillway and a bell-mouth
pressure outlet. The design discharge of these outlets
amounted to 860 m
3
/s in total. The secondary dam had
no spillways. The reservoir went in operation in 1925.
DESCRIPTION OF THE CRITICAL
STORM
After a long dry period, at 6:15 a.m. of 13
th
August
1935 an exceptionally heavy rain hit the Orba basin.
Fig. 1 - Orba and Stura Basins
Fig. 2 -
Orba and Stura watersheds, position of the gaug-
ing stations and area hit by the storm
background image
DESIGN FLOOD ESTIMATION: LESSONS LEARNT FROM SELLA ZERBINO DAM-BREAK
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
439
based on statistical distribution of standard deviation
and skewness coefficient.
To satisfy the requirement that only independent
data should be included in the regional sample, only
one among multiple data recorded in different rain
gauges in the same day was retained. The final sample
includes 572 values.
The parameters of the parent distribution are
estimated by the Probability Weighted Moment
method to obtain:
P
(h)
=
exp
{-
[1-k
(h-u)
/
α]
1/k
}
being α = 0.286058; u = 0.792868; k =-0.130374.
The return period of the 13/8/1935 rainfall event
was then estimated for the gauging station in the area
In Tab. 2 the return periods for the gauging sta-
tions having the 13/8/1935 rainfall depth as the
maximum ever registered, are highlighted in yellow.
The isohyetal map of Figure 3 shows that the total
rainfall, averaged on the Orba basin area from 9:00 of
August 13
th
to 9:00 o’clock of the day after, was 366
mm during the 8 hours event; d
e
M
archi
(1937) and
M
anGiaGalli
(1937) estimated 389 mm.
The storm hyetograph used in this study and given
in Figure 4 was reconstructed from the one measured
at the Lavagnina Centrale raingage (a
lfieri
, 1935).
ESTIMATION OF THE EXCEPTIONALI-
TY OF THE EVENT
The Authors analysed the measures of annual
maximum daily rainfall in 39 gauging stations in the
area, recorded in the period 1930-2011. The position of
the rain gauges is shown in Figure 2. The number of el-
ements in each one of the samples ranges from 9 to 75
First of all, Studentized Statistics non-parametric
test (K
otteGoda
& r
osso
, 2008, page 307) was applied
to detect outliers in the samples: only for Lavagnina
Centrale and Lavagnina Lago stations, the 1935 rainfall
could be considered an outlier at 5% significance level.
The Authors used the regional quantile estimator
to evaluate the return period of the rainfall event under
consideration (c
unnane
, 1989) assuming that all the
samples included in the same statistically homogene-
ous region, have the same Generalised Extreme Value
parent distribution, when data are normalised by di-
viding the measures by the average of their sample.
Rossiglione, whose rain-gauge was seriously dam-
aged during the event, and Lerca samples were re-
jected according to the chosen homogeneity criterion
Fig. 3 - Isohyetal map of the 13/8/1935 event
Fig. 4 -
Reconstruction of the 1935 hyetograph on Orba basin
Tab. 2 - Return periods of the 13/8/1935 events estimated
from the GEV distribution
background image
G. PETACCIA & L. NATALE
440
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
the main characteristics of the basin, taken from a dig-
ital elevation model (DEM) of 20m square grid.
The effective rain was evaluated using SCS meth-
od (S.C.S., 1985). The basin Runoff Curve Number
was determined from the land use and geologic maps
(www.clc2000.sinanet.apat.it) edited by Piedmont
and Liguria Regions, and considering the soil initially
dry (CN I) since the rainstorm occurred after a long
droughty period. An initial abstraction loss of 80%
was considered in a time period of 1.5 hours.
The rainfall runoff process was simulated using the
Instantaneous Unit Hydrograph (IUH) method. A semi
distributed model based on the Geomorphological Unit
Hydrograph (r
odriGuez
-i
tube
& V
aldez
,
1979; r
osso
,
1984; J
ain
et alii, 2000) was determined. The basin was
divided into 4 sub basins closed at the reservoir. To
evaluate the flood discharge also the rainfall fallen di-
rectly over the reservoir water surface was considered,
whose extension was in 1935 of 14 km
2
. All the contri-
butions were then added. The parameters of the GIUH
were estimated from the available DEM. The estimated
base time of the IUH, shown in figure 6, is t
c
=
2 hours.
The IUH’s shape reflects the contribution of the
four different sub-basins.
The rain falling on the water surface and on the
shores of the reservoir is directly added to the inflow
excluding losses and surface routing.
Figure 7 shows inflow and outflow hydrograph
of 1935 event. Reservoir outflow was evaluated tak-
ing into account all the outlets that worked during the
event (the siphon battery and the side spillway) and
the flow running over both the dams but the dam break
wave (n
atale
& P
etaccia
, 2012).
The reservoir level hydrograph is reproduced in fig-
ure 8 that shows the good agreement between the calcu-
lated values and the data collected by the dam keeper dur-
ing the event, before the collapse of the secondary dam.
Among these 8 stations only for Lavagnina Centrale
and Rossiglione the return period exceeds 1000 years.
We can conclude that the storm that caused the failure
of Sella Zerbino dam can be considered as exceptional
only in a very confined area.
The VAPI project (d
e
M
ichele
& r
osso
,
2001) had
the target to determine a uniform procedure to evaluate
flood discharges on the Italian territory. Some gauging
stations that recorded the 13/8/1935 rainfall are in the da-
tabase of this project as well. For some of these stations
the return period of the 13/8/1935 event was calculated,
and the estimates are higher than our ones. The estimat-
ed return period of Piampaludo rainfall exceeds 10,000
years, 588 years for Masone, 343 years for Piancastagna.
This can be explained since the region that was consid-
ered by VAPI to estimate the parameters of the probabil-
ity distribution is very wide and has characteristics that
are almost different from the limited zone interested by
the thunderstorm. As a matter of fact the VAPI procedure
overestimates the exceptionality of the rainfall.
FLOOD WAVE RECONSTRUCTION
The Orba river basin, shown in Figure 1, has an
extension of 141 km
2
(V
isentini
, 1936). The basin is
shaped like a fan, as shown in Figure 5, since its four
sub-basins join in Ortiglieto reservoir. Table 3 shows
Fig. 5 - Orba Basin, closure sections of the 4 sub basins
and the reservoir
Tab. 3 - Orba Watershed characteristics
Fig. 6 - Orba basin IUH
background image
DESIGN FLOOD ESTIMATION: LESSONS LEARNT FROM SELLA ZERBINO DAM-BREAK
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
441
dams, listed in table 4, located in the region between
the Ligurian Apennine and the sea (a
nidel
, 1952).
Looking at the envelope curve shown in Figure 9,
the unit discharge of 21.46 m
3
/s /km
2
corresponding
to the event that brought to Sella Zerbino failure (see
the symbol ▲ in Fig. 9), is almost three times higher
than the value given by envelope curve. Referring to
an event of these characteristics all the dams built be-
fore (see the symbol ○ in Fig. 10) as well as after (see
the symbol ■ in Fig. 9) the 1935 event would have
suffered since the spillways are designed to release
discharges significantly lower.
More updated techniques, like the one proposed in
VAPI project, would have determined for the 1935 event
a discharge lower than the one occurred. For the station
Erro at Sassello the index discharge method gives a unit
discharge, for a return period of 1000 years, of 9.59 m
3
/s /
km
2
, which is under the envelope curve.
Concluding, the event of 1935 would still be
critic for the any existing dam in the region: the hy-
draulic design on the dam spillways should be re-
considered in order to avoid future risk.
At 13:20 when the secondary dam collapsed, as in-
dicated by the dotted line in Figs. 7 and 8, the flood dis-
charge into the reservoir was about 1900 m
3
/s - between
2200 and 2500 m
3
/s for some others authors - and was
still increasing. Some reports discussed in the trial (l
el
-
li
,
1937; d
e
M
archi
, 1937; M
anGiaGalli
,
1937) state
that after the dam break the incoming discharge kept on
increasing to the value of 2500 m
3
/s and De Marchi cal-
culated the peak discharge to be around 3000 m
3
/s. From
our calculation the peak discharge is about 3300 m
3
/s.
CONCLUSIONS
From the statistical analysis carried out on the max-
imum daily rainfall depths we can assume that the 1935
event was really severe only for some stations close to
the Ortiglieto reservoir. The peak discharge of the flood
wave reaching the lake exceeded 3000 m
3
/s; for such a
discharge the spillways were under designed.
Now consider the flood discharge capacity of 20
Fig. 7 - Discharge entering and leaving the reservoir dur-
ing the 13/8/1935 event
Fig. 8 - Ortiglieto reservoir levels: comparison between
historical data and simulation
Tab. 4 - Characteristics of the 20 dams analysed in this study
Fig. 9 - Design discharge of the dams in the region of the
study and envelope curve
REFERENCES
a
ccusani
G. (1936, e
ds
.) - Relazione di perizia sulle cause che il 13 agosto 1935 determinarono il crollo della diga secondaria
di Sella Zerbino del serbatoio di Ortiglieto (Molare). Technical Report, Turin, Italy.
background image
G. PETACCIA & L. NATALE
442
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
a
lfieri
S. (1936) - Indagini idrogeologiche sul nubifragio dell’agosto 1935 nel bacino del torrente Orba. Annali dei lavori
pubblici, 9: 675 683, Technical Report, Turin, Italy.
a
nidel
(1952, Eds.) - Le dighe di ritenuta degli impianti idroelettrici italiani. Milan, Italy.
a
nonYMous
(1925) - L’utilizzazione del torrente Orba in comune di Molare da parte della Società Officine Elettriche Genovesi.
L’Energia Elettrica, XII (1): 1178-1219, Milano, Italy.
b
eGnudelli
L. & s
anders
B.F. (2007) - Simulation of the St. Francis dam-break flood. Journal of Engineering Mechanics, 133
(11): 1200-1212.
b
osa
s. & P
etti
M. (2011) - Shallow water numerical model of the wave generated by the Vajont landslide. Environmental
Modelling & Software, 26: 406-418.
c
henG
S.T. (1989) - Statistics on dam failure. Seminar 3 on assessment of risks and uncertainties in the design of hydraulic
structures. Proc. of the 24
th
congress of the Int. Association for Hydraulic Res., Delft, the Netherlands.
c
oYne
M. (1935) - La catastrophe de Molare, Italie, 13 agosto 1935. Memoires et Documents, 5: 133-144.
c
unnane
C. (1989) - Statistical distributions for flood frequency analysis. WMO, 718, Geneva, Switzerland.
d
arlYMPle
T. (1960) - Flood frequency analysis. US Geological Survey Water Supply Paper, 1543 A.
d
e
a
lMeida
a.b & V
iseu
t. (1997, e
ds
.) - Dams and safety management at downstream valleys. Balkema, Rotterdam, The
Netherlands.
d
e
M
archi
G. (1937, Eds.) - Sulla rottura della diga di Sella Zerbino (Molare) 13 agosto 1935. Relazione tecnica nel processo
penale, Milan, Italy.
d
e
M
archi
G. (1940) - Sulla rottura della diga di Sella Zerbino (Molare) 13 agosto 1935. L’Acqua, XVIII (5): 109-111.
d
e
M
ichele
c. & r
osso
r. (2001, e
ds
.) - Rapporto sulla valutazione delle piene nell’Italia Nord Occidentale. Estratto del
rapporto nazionale VAPI con aggiornamenti. Milan, Italy.
d
ouGlas
K.J., s
PannaGle
M. & f
ell
r. (1998, e
ds
.) - Analysis of concrete and masonry dam incidents. Report No. R-373 August
1998. The University Of New South Wales Sydney 2052 Australia.
D.S.C. (1991, Eds.) - Consequence categories for dams. DSC3A, Australia.
G
rahaM
W.J. (1999, Eds.) - A procedure for estimating loss of life caused by dam aailure. U.S. Department of Interior Bureau of
Reclamation Dam Safety Office Denver, DSO-99-06, Colorado, USA.
i
cold
(1974, e
ds
.) - Lessons from dam incidents (1973). Paris, France.
i
cold
(1987, e
ds
.) - Dam safety guidelines. ICOLD Bulletin, 59.
i
cold
(1991) - The identification of the possible causes and modalities of the failure of masonry dams. Idrotecnica, 2: 69-75.
i
cold
(2005, Eds.) - Risk assessment in dam safety management A reconnaissance of benefits, methods and current applications.
ICOLD Bulletin, 130.
J
ain
s.K., s
inGh
r.d. & s
eth
s.M. (2000) - Design flood estimation using GIS supported GIUH approach. Water Resources
Management, 14: 369-376.
J
ohnson
F. A. & i
lles
P. (1976, e
ds
.) - A classification of dam failures. Water Power and and Dam Construction.
K
alustan
E.S. (1995) - Lessons from failures of concrete dams on rock foundation. Hydrotecnical Construction, 29 (2): 77 81.
K
otteGodan
.T. & r
osso
R. (2008, e
ds
.) - Applied statistics for civil and environmental engineers. Blackwell Publishing, UK.
l
elli
M. (1937, Eds.) - Relazione tecnica nel processo penale per la rotta della diga di Sella Zerbino in comune di Molare, Italy.
M
anGiaGalli
L. (1937, Eds.) - Sulle cause del crollo della diga di Molare, Milano, Italy.
M
orris
M.W. (2000, Eds.) - CADAM concerted action on dambreak modelling, Final Report. Report SR 571, Wallingford,
England.
n
atale
l., P
etaccia
G. &
s
aVi
f. (2008) - Simulation of the Sella Zerbino catastrophic dam break. Proc. of the International
conference on Fluvial Hydraulics Riverflow 2008, 1: 601 607, Ankara, Turkey.
N
ATALE L. & PETACCIA G.
(2012) - Verifica sperimentale dello scarico a sifone della diga di Bric Zerbino. Proc. XXXII Convegno
Nazionale di idraulica e Costruzioni Idrauliche, Brescia, Italy.
n
atale
l., r
ossi
f. & s
iranGelo
b. (1998) - A review of dam break studies in Italy. Proc. of Dam Safety. 1: 267-274. Berga ed.
Balkema, Rotterdam, The Netherlands.
n
oVarese
V. (1938, e
ds
.) - Relazione geologica del crollo della diga di Molare (13/8/35). Roma, Italy.
P
eretti
L. (1936, e
ds
.) - Osservazioni geologiche relative al crollo della diga di Sella Zerbino presso Molare. Allegato alla
background image
DESIGN FLOOD ESTIMATION: LESSONS LEARNT FROM SELLA ZERBINO DAM-BREAK
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
443
relazione di perizia presentata dal Dr. Ing. Guseppe Accusani di Retorto al R. Tribunale di Alessandria, Arduini, Turin, Italy.
P
ilotti
M., M
aranzoni
a., t
oMirotti
M. & V
alerio
G. (2011) - 1923 Gleno dam break: case study and numerical modelling.
Journal of Hydraulic Engineering, 137 (4): 480-492.
r
odriGuez
-i
turbe
i. & V
aldez
J.b. (1979) - The geomorphic structure of hydrology response. Water Resources Research, 15
(6): 1409-1420.
r
oGers
J. D. (2006) - Lessons learned from S. Francis dam failure. Geo-Strata, 6 (2): 14-17.
Rosso R. (1984) - Nash model relation to Horton order ratios. Water Resources Research, 20 (7): 914-920. AGU, USA.
S.C.S. (1985, e
ds
.) - National engineering handbook, Section 4 - Hydrology. Water Resources Publication, Washington DC,
USA.
s
i
Y. (1998, e
d
.) - The world's most catastrophic dam failures: The August 1975 collapse of the Banqiao and Shimantan dams.
In: d
ai
Q
inG
. The River Dragon Has Come! M.E. Sharpe, New York, 1998.
t
anG
W. h. & Y
en
B.C. (1991) - Dam safety inspection scheduling. ASCE Journal of Hydraulic Engineering, 117 (2): 214 229.
usc
old
(1988, Eds.) - Lessons from dam incidents, US-II. 2
nd
Ed., Committee on Dam Safety of the U.S. Committee on Large
Dams, ASCE, New York, USA.
V
isentini
M. (1936) - Ecroulement d’un barrage sur l’Orba, en Italie, le 13 Aout 1935. Revue de geographie alpine, XXIV (II):
381-393, Grenoble, France.
X
u
Y., z
hanG
l. & J
ia
J. (2008) - Lessons from catastrophic dam failures in August 1975 in Zhumadian, China. Proc. ASCE
GeoCongress 2008- Geosustainability and Geohazard Mitigation, 162-169.
background image
Statistics