Documente Academic
Documente Profesional
Documente Cultură
DSO-99-06
September 1999
DSO-99-06
by
Wayne J. Graham, P.E.
September 1999
ABSTRACT
1
Hydraulic Engineer, Sedimentation and River Hydraulics
Group, Bureau of Reclamation, D-8540, Denver Federal Center,
Denver CO 80225-0007. E-mail: wgraham@do.usbr.gov
INTRODUCTION
1
substantial property damage.
The United States has also had major dam failures. Data for
failures occurring in the United States are provided in more
detail to provide the reader with an enhanced understanding
of the relationships between dam failure, flooding,
population at risk, warning and loss of life. The dam
failure data are then analyzed to show trends and patterns.
History shows that the loss of life from dam failure in the
United States has diminished with the passage of time. In
the late 1800’s and early 1900’s, there were several dam
failures with considerable loss of life. The loss of life
resulting from dam failure during the 1980’s and 1990’s has
been very low. The following is a summary of every dam
failure in the United States that caused more than 50
fatalities:
2
portions of the floodplain were densely populated, the
flooding destroyed the majority of buildings in downtown
Johnstown, and flooding in Johnstown preceding the arrival
of dam failure flooding made it difficult for people to
respond to the limited dam failure warnings that were
issued. The dam tender traveled by horseback to a nearby
community about 3 hours before dam failure and a message was
then telegraphed to Johnstown describing the danger, but the
warning was largely ignored.
3
difficulties in warning during the early morning hours. The
highest fatality rates, however, were in areas that were
close to the dam. For example, at Powerhouse No. 2, located
about 1.6 mi (2.6 km) downstream from the dam, the dam
failure claimed all of its occupants. In this same area
lived the dam tender who also perished in the flood. At the
California Edison Construction Camp, located about 17 mi (27
km) downstream from the dam, 89 of the 150 who had been
there perished. This is a fatality rate of about 60%.
Efforts to warn and evacuate people did not begin until a
few hours after the dam failed.
The Buffalo Coal Waste Structure, West Virginia, failed at
about 0800 hours on February 26, 1972. The structure did
not receive the engineering, design, construction and care
of a typical dam and is therefore called a structure and not
a dam. The structure, begun in 1970, was continually being
modified and enlarged as it was a waste pile used to dispose
of material extracted during coal mining. The structure was
about 46 ft (14.0 m) high and the failure released about 404
acre-ft (498,000 cubic meters) of water. This coal waste
pile structure failed as a result of slumping of the
structure face during a 2-year frequency rainfall event.
There were 125 fatalities, all occurring in the first 15 mi
(24 km) downstream. Warning of people exposed to the
flooding began after the structure failed; reaction to the
warnings was meager because there had been at least 4
previous false alarms.
4
Lake Dam, i.e, the incremental loss of life, will never be
known. It is estimated that the failure of Canyon Lake Dam
resulted in 33 fatalities. This estimate is based on the
assumption that the incremental loss of life downstream from
Canyon Lake Dam caused by dam failure was 20% of the total
loss of life downstream from the dam caused by the flood.
5
Table 1
Dam Failures in the United States Resulting in Fatalities
1960-1998
Dam Location Date of Age Cause of Failure Dam Volume Warning People Loss
Failure of Height Released Time at Risk of
Dam (m) (106m3) (Hours) Life
Note:
"Warning Time" is defined as the amount of time between the initiation of the dissemination of dam failure
warnings and the initiation of dam failure. Many of the entries in this column are zero, indicating that dam
failure warnings were not issued prior to dam failure.
"People at Risk" is defined as the number of people in the dam failure floodplain prior to the issuance of
any flood or dam failure warnings.
In the mid 1980’s there were about 5,459 dams in the United
States higher than 49 feet (15 meters) and more than 10
times as many, 71,000, that were more than 25 ft (7.6
meters) high. During the period 1960 through 1998, there
were more than 300 fatalities resulting from dam failures in
the United States. Failure of dams less than 15 meters high
(dams too small to be included in the International
Commission on Large Dams (ICOLD) Register of Dams) caused
88% of the total number of deaths occurring during this time
period. There are certain types of dam failures that have
occurred infrequently and thus information on these types of
failures and the consequences that would result from these
failures is deficient. These failures would include
concrete dams, high embankment dams or any type of dam
failing as a result of an earthquake.
9
Based on knowledge of the location of victims in 16 of the
23 dam failures (representing 87% of the fatalities) that
occurred during the 39-year period from 1960-1998:
10
failure caused by a major flood, the incremental consequence
would be the additional loss caused by the dam failure over
and above the loss that would have occurred if the dam and
reservoir had passed the reservoir inflow without failing.
11
have been accompanied by no dam failure warnings and would
have resulted in the loss of hundreds of lives.
Sources of Uncertainty
12
highlight weaknesses in the dam failure warning process and
provide some guidance on how improvements in the process
would reduce the loss of life.
13
Step 3: Determine When Dam Failure Warnings Would be
Initiated
14
Table 2
Guidance for Estimating When Dam Failure Warnings Would be Initiated (Earthfill Dam)
Dam Type Cause of Failure Special Time of When Would Dam Failure Warning be Initiated?
Considerations Failure
Many Observers at Dam No Observers at Dam
Earthfill Overtopping Drainage area at Day 0.25 hrs. before dam 0.25 hrs. after fw
dam less than failure reaches populated area
100 mi2 (260 km2)
Drainage area at Night 0.25 hrs. after dam 1.0 hrs. after fw
dam less than failure reaches populated area
100 mi2 (260 km2)
Drainage area at Day 2 hrs. before dam 1 hr. before dam
dam more than failure failure
100 mi2 (260 km2)
Drainage area at Night 1 to 2 hr. before dam 0 to 1 hr. before dam
dam more than failure failure
100 mi2 (260 km2)
15
Notes: "Many Observers at Dam" means that a dam tender lives on high ground and within site of the dam or the
dam is visible from the homes of many people or the dam crest serves as a heavily used roadway. These dams
are typically in urban areas. "No Observers at Dam" means that there is no dam tender at the dam, the dam is
out of site of nearly all homes and there is no roadway on the dam crest. These dams are usually in remote
areas. The abbreviation "fw" stands for floodwater.
Step 4: Determine Area Flooded for Each Dam Failure Scenario
16
take place. Centers of employment fill as housing units
empty and vice versa. It is important to understand the
methods used by the GIS Group to mesh flood boundaries with
census block data. There is uncertainty in the methods and
hence in the population at risk estimates.
17
When warning time is less than 15 minutes:
Loss of Life = .5(PAR)
18
Table 3
Dam failures and Floods used by Brown and Graham
19
DeKay and McClelland, supported by funding received from the
Bureau of Reclamation, expanded on the work begun by Brown
and Graham. They submitted the report entitled "Setting
Decision Thresholds for Dam Failure Warnings: A Practical
Theory-Based Approach" to Reclamation on December 31, 1991.
In 1993 they published "Predicting Loss of Life in Cases of
Dam Failure and Flash Flood" in the publication Risk
Analysis. The events used by DeKay and McClelland, shown in
Table 4, are the same as those used by Brown and Graham.
DeKay and McClelland also included a few events that were
not used by Brown and Graham. The DeKay and McClelland
procedure demonstrated that loss of life is related to the
number of people at risk in a nonlinear fashion. They also
found that loss of life is greater in situations where the
flood waters are deep and swift. DeKay and McClelland have
a separate equation for high and low force conditions.
Their equation, as it appears in Risk Analysis, for high
force conditions, i.e., where 20% or more of flooded
residences are either destroyed or heavily damaged is:
PAR
Deaths = ----------------------------------------------
1 + 13.277(PAR 0.440)e[2.982(WT) - 3.790]
Their equation for low lethality conditions, i.e., where
less than 20% of flooded residences are either destroyed or
heavily damaged is:
PAR
Deaths = -----------------------------------------------
1 + 13.277(PAR 0.440)e[0.759(WT)]
20
equations should not be applied to cases like Vajont, in
which a massive landslide displaced nearly the entire
contents of a reservoir. The Brown and Graham procedure as
well as the DeKay and McClelland procedure both conclude
that loss of life is much greater in those areas that
receive little warning compared to those areas that receive
more than an hour or so of warning. The value of adequate
dam failure warning in reducing loss of life from dam
failure can not be overemphasized.
21
Table 4
Dam Failures and Floods used by DeKay and McClelland
22
Limitations of Loss of Life Estimating Equations
The empirical data set used by Brown and Graham, and DeKay
and McClelland, in developing the loss of life estimating
equations did not include some types of events and warning
scenarios. Most of the dams in the data set were smaller
structures. Only 7 of the dams used in developing the
equations were more than 49 ft (15 m) high. The data set
included many more earthfill dams than concrete dams. The
data set included no dams that failed due to an earthquake.
The equations may not be applicable for use with dam sizes,
dam types, failure causes, flood severity and warning
scenarios not reflected in the data set.
23
realistic with very catastrophic flooding. The loss of life
is going to be directly related to the number of people who
do not receive the warning or ignore the warning and remain
in the risk area. The same percentage (80 to 100%) of the
people remaining would likely become fatalities if exposed
to the type and severity of flooding that occurred
immediately downstream from St. Francis Dam or Vajont Dam.
It may be very difficult to determine how many people will
not evacuate. If a warning does not reach people or if
people do not believe the warning or if they do not believe
that their life is at risk, then these people are more
likely to remain in the danger area.
When Brown and Graham originally developed their life loss
estimating equations, they thought that it was logical for
the fatality rate (the number of fatalities as a fraction of
the population at risk) to decrease as the population at
risk increased. The assumption was that as the population
increased, or became more dense, warning and communication
facilities would be more advanced. Probably what was
observed, unknowingly, is that as the population at risk
increased, the area under consideration was increasing in
size and was therefore including areas where the flooding
was less lethal. The Brown and Graham as well as the DeKay
and McClelland data bases probably contain many cases
demonstrating that there is an inverse relation between
population at risk and flood lethality. This means that as
the population at risk increased, the flood lethality (or
flood severity) decreased. Large populations do not fit
into narrow canyons - hence larger populations are situated
in the flatter areas where the lethality is usually reduced.
Some questions regarding the validity of the equations
developed by Brown and Graham as well as Dekay and
McClelland remain. Do the equations give accurate results
when large numbers of people are exposed to truly
catastrophic flooding? Should the fatality rate vary so
much for different population sizes? Does adequate warning
time result in low fatality rates? - Or is adequate warning
most likely to occur for benign floods and these floods are
not very lethal, regardless of the warning?
24
400 dams that failed in the United States from 1985 to 1994
would also fit this category.
25
Flood Severity along with warning time determines, to a
large extent, the fatality rate that would likely occur.
The flood severity categories are as follows:
26
flooding about to occur and may not evacuate at all or not
as quickly as they should. This factor will come into
consideration only when there is some or adequate warning.
27
an example, if there were 3 cases for one particular
category, and the fatality rates were 0.01, 0.09 and 0.11,
the average was shown as .07 and the range was shown as 0.01
to 0.11.
28
(
7
$
5
<
7
,
/
$
7
$
)
6
+
7 D
$
Q
(
'
5
2
) .
6
0 ,
$ 5
' 7
$
0
(
2 /
5 3
)
2
6 (
( 3
/
,
0
)
7
2 $
5 ( .
( D
/ 6
% 3 ,
5 Q
0 2
8 (
1 3
+
7
'
,
: V
D D D D D D
4
W
' I Q Q Q Q Q Q
4
2
2
/
)
G \ H
W U
O Q L H "
O D Q V K H
L
O H W
K L W X O Y
V K L K L
J I R K X P W P V W Y
6 R V W E G P H U
7 Q P W S D L S
L S R H H W H X
V G R X O V
1 G
O P L H H F R G U G
( L R U G W O R W D X
W ) E G V X G
X E F H H R
0 W H G D Y D R Q V R
E R Q D \
0 G R W H W R Z R
E R L N O R F O G
2 G O Q O V D I U R I O
V W \ I R H L
& H E U W D G Z W X
Q D F E G J
L \ W H K R U H R R
D I R G R U W W O S R Q H I R F
H · D I I
R U V H R R H R K H
P W L N I \ Q Y P W
V X U V /
U Z
H R F D G D D U H R
< U H X R L U W D
7 G + E U 0 G Z W $ I Z +
, J
5 Q
L
7DEOH
( K
9 W
( R
6 Q Q R
J G
" R H W
' Q ( H H L Q P V Q H H
W W W L D L W W
2 D 5 H H S G X U H H
2 H U U Z G D H L U U
O 8 X R R R
/ F / F F U R O H F G F F
) , Q V Q H O I L Y Q Q
W H I Q G O U
$ R R L V R R
+ S ) F < F F G R H F F
H L G X P G V
Q H R Q
* O H
, Z 0 V V D V P I D U V V
V H H V O H H
+ $ F X H
< < O D R < <
V ' R < \
L F
9 1 E
D
H
U
$
(
7
$
'
/ 1 D
$ 2 F
L
5 , U
( 7 H
$ D
1 & L P
( E $
2 H
* / P K F
Q
L X W \ \ Q
D O X O O
& D D D
S $ R R W W U $
6 1 & & 6 , , ) &
Q
R
V
L
G
7 (
1
( O
9 D
& S
( P P
D P
D D
' ' P & P P
D U D
D D D
U Z ' Q '
R R K '
H O V L P V
7 O L W D P D W L
D F F / D H F
H ' V
Q X R ' V Q
' : D U U W D
U W D Q D U
U V H S
D ) Y R O )
J D Q P M
H D D
H W R U W D W
9 % 6 & $ 6 9 0 6
< *
7 1
, ,
5 '
( H H
9 1 H V H V
$ X L X L
( F F
6 7 J J
6 D H D H
U U
' 5 9 3 9 3
2 (
2 '
/ 1
) 8
(
0
, H
7 Z W
R H
/ D
* X
1 P T
, H R
Q 6 H
1 R G
5 1 $
$
:
29
(
7
$
5
<
7
,
/
$
7
$
)
6
+
7
$
(
'
5
2
) .
6
0 ,
$ 5
' 7
$
0 "
(
2
/
5 3
)
2
6 (
( 3
/
,
0
)
7
2 $
5
( .
( / 6
,
% 3
5
0 2
8 (
1 3
+
7
'
,
: V
D D D D
4
W
I Q Q Q Q
4 '
2
2
/
)
J J V
Q Q X
H L L M
V Q H \
H F Q H W
Q U U U L Q
J W J D ) U P
H D U Q R Q U H
X 6 G V U H Z J M X H H
I H H Z O D Z R
H 7 L J I Q D U G W E
U V D K R L P X H J H
1 G Q H H F U P O I G
L U W Q L Q V H
H ( U Q G O H P L H I D
X H H G W R D G Q
G 0 H U Q I H H G G K
L D Q G Q H Q F U O
Y S L L H Z Y D X H V
0 Q Y L D O \ U
R P Z Y L R Q P H F O
2 P D H L Z Y X P
7DEOH&RQWLQXHG
U D O R K W D F [ Q M L
S & U K F J V J G H V G H W
F D F H L R R Q
S Q Q U H L F G
V J P U U K K L H W R L H
H H U H W W K Y
V Q U , R V R W W R G X K K W Q
R W L M V I - R R Q : S H U
< X X I 5 D R I P L J J
7 R R W R H R W L L H K D
, R Q O
K < , : 1 0 0 $ I E 6 P $ 1 1 ' 7 Z
5
( G
9 H
( K
6 V
X
' U H
F "
2 ( W H H
V V W W
2 5 D H H
/ H U U V
H 8 Z
) U / F F H
, V O R Q V R Q V V V V V R < V
7 H D H H H H H H H
0 $ 1 R 1 R 1
) < R F < F < < < < < R <
8
, G F 1
V V
' H 0
\ $ V H H
( H < <
R '
0 U <
W
V
H
G
H
U
D
V
J
Q
L
G
O (
L 7
X
$
%
'
$
0
/ 1
2 2 H $ J $ $
$ 2 + F '
, & & 3 U 3 3 6
5 2 Q X 5
7 D
( 7 H N N Q E Q Q $ 2 \
$ $ U U U V W
1 & 8 G D D ) Z Z Z * U L
( 9 L 3 R R R H &
V 3 3 P D
* 2 D Q V W W W Q
/ : \ L V D V V R G R
Q G W V V X L S L
Q H H M Q O Q Q F K
Q D V W W O F S S
D D H K L K K H D
K X V V U R R R R D G
+ 0 6 $ ( ( ) - : - - 7 + 5 ,
U
P H
D Y
L
' 5
H J
W
7 V Q U
L H
D U
1 W
( D G V
: R G
9 R D
O 5 V U
( D R L R
O P G I
R G ) ' O
& P P P P D R L
P R P G
D D Q D P D D ' R
N R D R O :
' O ' R ' D ' ' V )
H ) V ' R
H ' H O V
U U P H S W N N Q Q ) O
H H D U U U O P
& N H X U L D
H G D P V R H R 5 U
L '
R ) Y ) D H
' R V / V L O + '
O Q K % Q
\ L D K 5 K H N Q
H D G W Q 7 S W W U \ S
O I O O O F R
W I D V Z J X O X X O S D W
W X X L D R L R D H H O H
L K D
/ % 6 $ / % 0 6 0 6 / . + % 7
< *
7 1
, ,
5 '
( H
9 1 H V
$ X L
( F
6 7 J
6 D H
U
' 5 9 3
2 (
2 '
/ 1
) 8
(
0
,
7 Z
R H
* /
P
1 H
, R
Q 6
1 R
5 1
$
:
30
(
7
$
5
<
7
,
/
$
7
$
)
6
+
7
$
(
'
5
2
) .
6
0 ,
$ 5
' 7
$
0
(
2 /
5
) 3
2
6 (
( 3
/
,
0
)
7
2 $
5 ( .
(
/ 6
% 3 ,
5
0 2
8 (
1 3
+
7
'
,
: V
4
W
' I
4
2
2
/
)
R
W \
H D
J S G
X K
X 6
I V V J
H 7 J
U K X
1 W Q D
H ( S L
G H Q N P
L 0 U U H
G
Y 0 D R Q
R Q ) O
2 L Z R D
7DEOH&RQWLQXHG
U & U
S & D H K
O O W W H
V S W S X V Q
H G H L D L
H W R
V R I O 6 ( 0
< X R X
7 R O W W W
, K ) 0 $ $ $
5
( G
9 H
( K
6 V
X
' U "
F
2 (
2 V 5
/ H
H 8
) U /
, R V V V V
7 H H H H
0 $ 1
) < < < <
8
, G
' H 0
( \ $
R '
0 U
W
V
H
G
H
U
D
V
J
Q
L
G
O (
L 7
X $
% '
/ 1 $
$ & $ $ $
2 3 3 3
5 , V
( 7 H Q Q Q
$ 2 O
1 & & H Z Z Z
( J R R R
* 2 R Q W W W
/ O V V V
E $
Q Q Q
H V K K K
X R R R R
3 / - - -
G
Q
D
P
D
G
Q
H
H
Z
7 W
H
1 G E
(
R
9 R
( O P
) D
U ' P P P
H H D D D
Y V Y
L O L ' ' '
O
5 L U
N N N
V + ' U U U
R R R
D Q ] ) ) )
V L H
Q Z K K K K
D G F W W W
N O Q X X X
U D D R R R
$ % 6 6 6 6
< *
7 1
, ,
5 '
( H
9 1 H V
$ X L
( F
6 7 J
6 D H
U
' 5 9 3
2 (
2 '
/ 1
) 8
(
0
, H
7 W
D
* X
1 T
, H
1 G
5 $
$
:
31
(
7
$
5
<
7
,
/
$
7
$
)
6
+
7
$
(
'
5
2
) .
6
0 ,
$ 5
' 7
$
0
(
2
/
5 3
)
2
6 (
( 3
/
,
0
)
7
2 $
5 ( .
(
/ 6
% 3 ,
5
0 2
8 (
1 3
+
7
'
,
: V
D D D D D D
4
W
I Q Q Q Q Q Q
'
4
2
2
/
)
W P "
R X G
H Q L H
V G \
S W H
G X G G R
6 G D P U
7 H H O H E H H W
O \ W W V
1 \ J R Q
R D R G D D L H
( F U H O H O
U H H U H U H G
W P O W U G
0 O V Q H U H U H V
V D L G U X
V 0 J J H H H O Z H Z H F J
Q Q G P G D L O O D Q
Q 2 L L G J V L L O L
7DEOH&RQWLQXHG
U Z D V
R & Q Q V Q V D I K E K E S G
L R L H V W R R \ O
W U U H M V P W W L
D D D D G P D D \ L
P X D P P P O U X
G H L D H H U
Z Z R P L R G W G R R H E
Q G W G W D
R R K G K F H Y R
X L I X X H
< R 1 1 9 R D D 1 V 1
7 I
,
5 I
I
( R
9
( G
6 H
K "
' V (
D 5
2
Z 8
2
/ W / V V V V V V
R , R R R R
) $ H H H H H H
Q < < 1 < < < 1 1 1 <
)
: H
U 0
2 D
/ $
V '
J
Q
L
G
O
L
X
%
(
7
$
'
/ 1 6
$ $ 2 . 7
2 7 &
5 , $ 8
7 8 : 7 \
( & 0 N W H
$ O U L ; J
1 X D &
& I L K H D 7 U
( L O F H V R
2 W D L 3
* / / D Q H
Q U V L
W Z W H
V W *
X Q U V W Q V
R R
H $ D V D X ; W
% & 3 1 ( ( . $ 7 6
W
V
L
'
S
P
,
U
H
W G
D U G G
7 L R
R R Q R P
1 : X O D
( Y R
O )
\ U R '
9 W H ) P U
V G \
( Q \ D J U
H W R W H
X ' S R N
5 Q P O Q L
R N P
& O X D ) X '
O U P D R
L R D ' G
V & D & N & N
L + 3 H R
Y ' H Q O H
\ \ N H R
O O H
D U Q Q H H U L U
P D D O & )
' D D H N / J + &
J D Q
K ' Q K Q K L V O
W L J H / Q H V W L
K D D
X P H H Z S X V [
O R U X
R H O H D V X H
6 6 $ 0 / / $ % $ 7 4
< *
7 1
, ,
5 '
( H
9 1 H V
$ X L
( F
6 7 J
6 D H
U
' 5 9 3
2 (
2 '
/ 1
) 8
(
0
,
7 Z
R H
* /
P
1 H
, R
Q 6
1 R
5 1
$
:
32
(
7
$
5
<
7
,
/
$
7
$
)
6
+
7
$
(
'
5
2
) .
6
0 ,
$ 5
' 7
$
0 "
(
2
/
5 3
)
2
6 (
( 3
/
,
0
)
7
2 $
5 (
.
( / 6
,
% 3
5
0 2
8 (
1 3
+
7
'
,
: V
D D D D D D D D
4
W
I Q Q Q Q Q Q Q Q
'
4
2
2
/
)
G W
H Q
G U N S H
\ H H X L
R J D G \ V
U \ Z H H D
W Q U R R U Q
L R O G \ G D
V Q U E R R
6 H W V U
H U V V R U V W
7 G J D J K P O W J
H D I V
1 V D Z Q G F H Q J
( H L X G U S L
J P V G O V R G G Q
J O \ M H Q L
0 Q D U R H H P H V R U G
G L X Q D D
V 0 L G R Z P H G H R R
U Q R O V
D O
Q 2 U K I G R P W P I Z R
7DEOH&RQWLQXHG
E K R P H O
R & D F K K K R K H I
L U H O
W Z X V H Q V I K V Q
D L O U L V L L S
W P R J L H J H J L J
G I K G W W G W L
R R W J E Q J J O G
Q J D D R P Q H
L R R S X R R X X X H L
X O K O O R R O
< R 1 * 1 8 6 ! P : W I 6 W 6 6 0 % G
7 I
,
5 I
I
( R
9
( G
6 H
K "
' V (
D 5
2
Z 8
2
/ W / V V V V V V
R , R R R R R R
) $ H H H H H H
Q 1 < < < 1 1 1 1 < 1 < <
)
: H
U 0
2 D
/ $
V '
J
Q
L
G
O
L
X
%
(
7
$
'
V
/ 1 D $
$ 2 V &
2 & -
5 , Q 6 V
( 7 = N 1 D H
$ 2 8 O
$ U Q . H
1 & 7 D & W 7
( [ U \ V J
L & 3 U H O Q 8
* 2 H U H Q
/ Q [ K Q H
H V Y W Z $ D
H H R W W
R V W Q U V G V O
K V V 2 H R $ D L R H
'
3 ( ( & ' 1 3 ( , 0 / '
]
H
P V K
O F
R O
U D Q
I V ) D
H
6
P Q
Q Q
D J
H $ P H
U G $ H
W G R H Z
7 V Q R W
Q R R
O D W H
1 R
Z O ) F J E
( P ) L
D R Q U U
9 G L U X P
( R ' G U P
H V X E X
R D D
O G Y + [ H
) P L % H ' V
Q 5 I L
R D N U 5 V O
D U P P U R O R
H 3 ' D H H R O
D W H G L &
U O H W Y U Y P
O 3 ' L I L P R +
$ L N D 5 G D
O W O J 5 D R
[ + D D F O Q Q '
L / 3 F Q V ' I L D
Q H L L
Q \ R S K D D Q W Z H '
H K Q L W G V D
V V R R G Y $
R V Z W R X V Q W H O L
K X D D U R D R D H U D U 0
O
3 % / 1 3 6 3 ) . 7 * % ' '
< *
7 1
, ,
5 '
( H
9 1 H V
$ X L
( F
6 7 J
6 D H
U
' 5 9 3
2 (
2 '
/ 1
) 8
(
0
, H
7 W
D
* X
1 T
, H
1 G
5 $
$
:
33
Table 6
Fatality Rates Derived from Case Studies
(Use Table 7 for selecting fatality rates)
35
Another method that can be used to separate low severity
flooding from medium severity flooding is to use the
parameter DV where:
Qdf -Q2.33
DV = ------------
Wdf
And:
36
downstream from a dam for two main reasons. First, these
downstream areas receive warning that usually is much better
than the warning, if any, issued in areas nearer the dam;
and second, the energy exhibited by the flood is lessened,
the flood rises at a slower rate and the leading edge of the
flooding usually moves at a slower rate in these downstream
areas. Based on these empirical data and recognizing that
the failure of some large dams could result in loss of life
patterns or characteristics that are not observable with
this same data base, loss of life studies should extend
downstream from a dam for 30 mi (50 km). There may be some
very high dams, or those storing very large quantities of
water, where severe flooding could extend for 100 miles (161
km) or more downstream from the dam. In these cases, loss
of life studies may be extended more than 30 mi (50 km)
downstream from the dam. In general, however, life loss
more than 30 mi (50 km) downstream from a dam should be very
small compared to the life loss estimated for the areas
nearer the dam. It is not anticipated that the life loss
downstream from mile 30 (50 km) would change the results of
a dam safety recommendation.
37
Table 7
Recommended Fatality Rates for Estimating Loss of Life Resulting from Dam Failure
precise floodplain.
no warning not applicable 0.15 0.03 to 0.35
vague 0.04 0.01 to 0.08
15 to 60
MEDIUM precise 0.02 0.005 to 0.04
vague 0.03 0.005 to 0.06
more than 60
precise 0.01 0.002 to 0.02
no warning not applicable 0.01 0.0 to 0.02
vague 0.007 0.0 to 0.015
15 to 60
LOW precise 0.002 0.0 to 0.004
vague 0.0003 0.0 to 0.0006
more than 60
precise 0.0002 0.0 to 0.0004
Closing Comments on the Flood Severity Based Method
39
The fatality rate in areas with medium severity flooding
should drop below that recommended in Table 7 as the warning
time increases well beyond one hour. Repeated dam failure
warnings, confirmed by visual images on television showing
massive destruction in upstream areas, should provide
convincing evidence to people that a truly dangerous
situation exists and of their need to evacuate. This should
result in higher evacuation rates in downstream areas and in
a lowering of the fatality rate.
40
fatality rates shown in Table 7, 2) when the flooding at a
particular area falls between two categories (it is unclear
if the flood severity would be medium or low, for example)
the loss of life estimates can be developed using the
fatality rate and range of rates from all categories touched
by the event and 3) the events cataloged in table 5 can be
evaluated to see if there are any that closely match the
situation at the site under study.
SUMMARY
41
REFERENCES
42
Hatem, Georges Antoine, "Development of a Data Base on Dam
Failures in the United States: Preliminary Results," A
thesis submitted to the Department of Civil Engineering of
Stanford University, December 1985.
Qing, Dai, The River Dragon Has Come! The Three Gorges Dam
and the Fate of China’s Yangtze River and its People , M.E.
Sharpe, Armonk, New York, 1998.
43