Embankment dam deformations caused by earthquakes
J. R. Swaisgood, P.E., C.P.G.
Swaisgood Consulting, Conifer, Colorado, U.S.A.
ABSTRACT: An extensive review of case histories of embankment dam behavior during
earthquake was undertaken after several major embankment dams were severely shaken by
the 1990 Philippines earthquake. The objectives of the study, which continues to date, were to
determine if there is a “normal” trend of seismic deformation that can be predicted and if
there are certain factors that consistently have an effect on the amount of damage and
deformation incurred during earthquakes. Nearly 70 case histories have been reviewed,
compared and statistically analyzed in this effort. The results of this empirical study have
shown that the most important factors that appear to affect dam crest settlement during
earthquake include the peak ground acceleration at the site and the earthquake magnitude. A
chart has been prepared to summarize the relationship between the amount of measured
settlement and the peak ground accelerations experienced in the incidents that were studied.
In addition, an empirical equation was formulated and a graph developed as an aid in
estimating the amount of deformation to be expected.
1 INTRODUCTION
An evaluation of case histories of embankment dam behaviour has been in progress since 1990 with
two objectives in mind:
°
Providing a tool for immediate assessment of a structure that has undergone seismic
loading and
°
Creating a method for estimating how much an embankment dam will deform based on
actual dam behaviour during past earthquakes.
The findings from these ongoing empirical studies were last presented four years ago. Since that
time, the research has continued, increasing the data base by nearly 30 percent. This paper presents
the results of the extended examination and analyses of the entire data base.
2 CASE HISTORY DATA BASE
2.1 Previous work
During the 1990 Philippines earthquake, a review of incidents of seismically-induced deformation
of embankment dams was initiated to aid in evaluating the damages exhibited by several major
dams during that event (Swaisgood and Au-Yeung, 1991). These studies continued on with the
results last presented in 1998 (Swaisgood, 1998). At that time , the screening efforts had produced
54 incidents that had been described with sufficient quantified data for meaningful comparative
studies and statistical analyses
2.2 Updated version
Continuing research has yielded an additional 15 case histories, making a total data base of 69
incidents. Pertinent details of all 69 of these incidents are presented in Table 1. The new additions
include nine located in California – Case Nos. 10, 25, 26, 36 to 40 (Tepel, et. al. 1996) and 19
Paper Number 014
(ICOLD 2001); four in Chile – Nos. 28, 66,67, and 68 (Pinos 2000); one in the Philippines – No. 46
(ICOLD 2001); and one in Peru (So. Peru Copper Corp. 2001). The entire data base is plotted in
Figure 1 where the crest settlement is shown in relation to the peak ground acceleration at the site.
Table 1. Earthquake induced settlement of embankment dams
GENERAL
ID
No
INFORMATION
CREST
DAM
DH
CL
AT
NAME OF DAM
UPPER MURAYAMA
LOCATION
TYPE
m
m
m
Japan
E-HF
24
320
3
ONO
CHATSWORTH
NO.2
Japan
E
41
309
3
California
HF
12
610
4
MALPASSO
Peru
ECRD
78
152
5
COGOTI
Chile
CFRD
85
159
6
SOUTH HAIWEE
California
HF
25
7
HEBGEN
Montana
E
8
MIBORO
Japan
ECRD
9
MINASE
Japan
CFRD
10
California
E
11
UVAS
U. SAN FERNANDO
California
HF
12
OROVILLE
California
ECRD
13
LA VILLITA
Mexico
ECRD
14
EL INFIERNILLO
Mexico
15
EL INFIERNILLO
Mexico
16
TSENGWEN
17
1
EARTHQUAKE
DATE
DATA
M
D,km
SETTLEMENT
PGA, g.
m
*
%
**
RELATIVE
DEGREE
OF
DAMAGE
1
Sep
23
8.2
18
0.32
e
0.20
0.74
11
1
Sep
23
8.2
98
0.30
e
0.27
0.53
Serious
?***
30
Aug
30
5.3
1
0.40
e
0.08
0.63
Moderate
30
10
Oct
38
VI+
n/a
0.10
e
0.08
0.07
Minor
0
6
Apr
43
7.9
89
0.20
e
0.38
0.44
Minor
457
38
21
Jul
52
7.7
151
0.05
e
0.02
0.04
Minor
25
213
10
17
Aug
59
7.6
0
0.71
e
1.69
4.82
Serious
130
444
0
19
Aug
61
7.0
20
0.15
e
0.03
0.02
Minor
67
210
?
16
Jun
64
7.5
145
0.08
e
0.06
0.09
Minor
32
335
?
18
Dec
67
5.3
11
0.20
e
0.02
0.06
Minor
25
390
18
9
Feb
71
6.6
2
0.55
e
0.91
2.11
Serious
235
1707
0
1
Aug
75
5.9
7
0.10
r
0.01
0.004
None
60
427
75
15
Nov
75
7.2
20
0.04
r
0.02
0.02
None
ECRD
146
340
0
15
Nov
75
7.2
23
0.09
r
0.02
0.02
None
ECRD
146
340
0
11
Oct
75
5.9
79
0.08
r
0.04
0.03
None
Taiwan
ECRD
131
n/a
?
14
Apr
76
5.3
8
0.16
e
0.04
0.03
n/a
EL INFIERNILLO
Mexico
ECRD
146
340
0
14
Mar
79
7.6
95
0.12
r
0.13
0.09
Minor
18
LA VILLITA
Mexico
ECRD
60
427
75
14
Mar
79
7.6
108
0.02
r
0.05
0.03
Minor
19
VERMILION
California
E
50
1290
50
27
May
80
6.3
22
0.24
r
0.05
0.05
None
20
LA VILLITA
Mexico
ECRD
60
427
75
25
Oct
81
7.3
31
0.09
r
0.14
0.11
None
21
EL INFIERNILLO
Mexico
ECRD
146
340
0
25
Oct
81
7.3
55
0.05
e
0.06
0.04
None
22
NAMIOKA
Japan
ECRD
52
265
0
26
May
83
7.7
145
0.08
r
0.06
0.11
None
23
COYOTE
LEROY ANDERSON
ELMER J. CHESBRO
California
E
43
299
0
24
Apr
84
6.2
0
0.63
e
0.08
0.18
Minor
California
ECRD
72
427
0
24
Apr
84
6.2
2
0.41
r
0.02
0.02
Minor
California
E
29
220
0
24
Apr
84
6.2
22
0.18
e
0.02
0.05
Minor
26
UVAS
California
E
32
335
?
24
Apr
84
6.2
29
0.14
e
0.02
0.08
Minor
27
MAKIO
Japan
ECRD
77
264
29
14
Sep
84
6.8
5
0.57
e
0.50
0.47
Minor
28
AROMOS
Chile
ECRD
43
220
9
3
Mar
85
7.8
45
0.25
e
0.09
0.177
Minor
29
EL INFIERNILLO
Mexico
ECRD
146
340
0
19
Sep
85
8.1
76
0.13
r
0.11
0.08
Minor
30
LA VILLITA
Mexico
ECRD
60
427
75
19
Sep
85
8.1
43
0.13
r
0.33
0.24
Minor
31
LA VILLITA
ECRD
60
427
75
21
Sep
85
7.5
61
0.04
r
0.12
0.09
None
32
MATAHINA
Mexico
New Zealand
ECRD
86
400
?
2
Mar
87
6.3
9
0.33
r
0.12
0.14
Moderate
33
NAGARA
Japan
ECRD
52
n/a
?
17
Dec
87
6.9
29
0.27
r
0.02
0.04
n/a
2
24
25
Moderate
34
AUSTRIAN
California
E
56
213
0
17
Oct
89
7.1
2
0.57
e
0.85
1.51
Serious
35
LEXINGTON
California
E
63
253
0
17
Oct
89
7.1
3
0.45
r
0.26
0.41
Minor
36
UVAS
California
E
32
335
?
17
Oct
89
7.1
10
0.40
e
0.02
0.06
None
37
STEVENS CREEK
California
E
37
305
?
17
Oct
89
7.1
16
0.30
e
0.02
0.04
None
38
ALMADEN
California
E
32
140
?
17
Oct
89
7.1
9
0.44
e
0.03
0.10
Minor
39
CALERO
California
E
30
256
?
17
Oct
89
7.1
13
0.38
e
0.01
0.03
None
40
RINCONDA
California
E
12
73
?
17
Oct
89
7.1
9
0.41
e
0.02
0.15
Minor
41
California
E
43
204
0
17
Oct
89
7.1
10
0.42
e
0.20
0.45
Minor
42
GUADALUPE
ELMER J. CHESBRO
California
E
29
220
0
17
Oct
89
7.1
13
0.42
e
0.11
0.39
Moderate
43
VASONA
California
E
10
149
8
17
Oct
89
7.1
9
0.37
e
0.05
0.27
Minor
2
GENERAL
ID
INFORMATION
CREST
DAM
DH
CL
AT
LOCATION
TYPE
m
m
m
44
LEROY ANDERSON
California
ECRD
72
427
0
17
Oct
45
SAN JUSTO
California
ECRD
40
340
14
17
Oct
46
AMBUKLAO
Philippines
ECRF
120
450
5
16
47
MASIWAY
Philippines
E
25
427
3
16
48
PANTABANGAN
Philippines
ECRD
114
732
0
49
AYA
Philippines
ECRD
102
427
50
DIAYO
Philippines
ECRD
60
201
51
CANILI
Philippines
ECRD
70
52
MAGAT
Philippines
ECRD
53
California
54
COGSWELL
ROBERT MATTHEWS
55
WIDE CANYON
56
57
No
NAME OF DAM
E A R THQUAKE
DATA
SETTLEMENT
m
%
**
D,km
89
7.1
21
0.26
r
0.04
0.06
Minor
89
7.1
27
0.26
r
0.04
0.07
None
Jul
90
7.7
10
0.49
e
1.10
0.880
Serious
Jul
90
7.7
3
0.68
e
1.06
3.79
Serious
16
Jul
90
7.7
6
0.58
e
0.28
0.24
Moderate
0
16
Jul
90
7.7
6
0.58
e
0.20
0.20
Minor
0
16
Jul
90
7.7
18
0.38
e
0.07
0.11
Minor
351
0
16
Jul
90
7.7
18
0.38
e
0.04
0.06
Minor
100
1296
0
16
Jul
90
7.7
81
0.05
e
0.01
0.006
None
CFRD
81
200
0
28
Jun
91
5.8
7
0.37
e
0.04
0.051
Minor
California
E
46
192
0
25
Apr
92
6.9
64
0.07
e
0.00
0.007
None
California
E
26
678
?
28
Jun
92
7.5
30
0.20
e
0.01
0.048
Minor
YUCAIPA No. 1
California
E
13
128
9
28
Jun
92
6.6
28
0.15
e
0.01
0.028
Minor
California
E
15
146
9
28
Jun
92
6.6
28
0.15
e
0.00
0.019
Minor
Arizona
E
13
247
1
29
Apr
93
5.5
77
0.02
e
0.00
0.004
None
California
HF
25
390
18
17
Jan
94
6.7
10
0.42
e
0.44
1.021
Serious
60
YUCAIPA No. 2
UPPER LAKE
MARY
U. SAN FERNANDO
L. SAN FERNANDO
California
E-HF
38
537
6
17
Jan
94
6.7
9
0.44
e
0.20
0.460
Serious
61
LOS ANGELES
California
E
47
671
0
17
Jan
94
6.7
10
0.43
r
0.09
0.188
Moderate
62
California
E
36
427
0
17
Jan
94
6.7
10
0.43
e
0.03
0.089
Moderate
63
NORTH DIKE [LA]
LOWER FRANKLIN
California
HF
31
152
?
17
Jan
94
6.7
18
0.30
e
0.05
0.146
Moderate
64
SANTA FELICIA
California
E
65
389
0
17
Jan
94
6.7
33
0.18
e
0.02
0.030
Minor
65
COGSWELL
California
CFRD
81
200
0
17
Jan
94
6.7
53
0.10
e
0.02
0.026
Minor
66
PALOMA
Chile
ECRD
82
1000
14
14
Oct
97
7.6
45
0.23
e
0.14
0.141
Minor
67
COGOTI
Chile
CFRD
83
160
0
14
Oct
97
7.6
45
0.23
e
0.25
0.302
Moderate
68
SANTA JUANA
Chile
CFRD
113
390
19
14
Oct
97
7.6
260
0.03
r
0.02
0.015
None
69
TORATA
Peru
CFRD
120
600
0
23
Jun
01
8.3
100
0.15
e
0.05
0.042
Minor
59
PGA, g.
*
M
58
DATE
RELATIVE
DEGREE
OF
L E G E N D
DH = dam height
M = earthquake magnitude, surface-wave scale: M S
D = distance from nearest ground rupture or epicenter, whichever is closest
PGA = peak horizontal ground acceleration; e = estimated, r = recorded
HF =
E=
Hydraulic Fill
Earthfill
ECRD =
Earth Core Rockfill Dam
CFRD =
Concrete Faced Rockfill Dam
NOTES:
* - Settlement shown is the single maximum reported or is an average from upstream, downs tream and centerline readings
** - Determined as a percentage of combined dam height and alluvium thickness
*** - If alluvium thickness unknown (?), it is considered to be 0 for % settlement calculations
3
DAMAGE
% STTLMT = ---------------- x 100
DH + AT
DH
AT
SERIOUS
10
0.1
0.01
NONE
CFRD
ECRD
HF
Earthfill
RELATIVE DEGREE OF DAMAGE
MINOR
MODERATE
CRESTSETTLEMENT,in%(DH+AT)
1
0.001
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
PEAKGROUND ACCELERATION,g
Figure1. Settlement of embankment dams during earthquake
3 ANALYSIS OF DATA
3.1 General
Similar to the previous studies, crest settlement was selected as the parameter to represent
earthquake related deformation because it was the most often mentioned quantified measurement of
damage presented in the case histories. It also appears to be directly related to the severity of
deformation and cracking, i.e., as the percent of crest settlement increases, the extent of deformation
and cracking that occurs also increases. The ranges of the relative levels of damage are summarized
in Figure 1.
The data base of case histories was analyzed using statistical regression techniques for the purpose
of identifying those factors that have a major influence on the deformation and damage of
embankment dams during earthquakes. These statistical studies were performed using the percent of
crest settlement as the dependent variable and the other factors to be evaluated as the independent
variables.
4
From these regression analyses, it was found that the only factors that had major, statistically
significant effects on the amount of crest settlement included peak ground acceleration and
earthquake magnitude.
3.2 Peak horizontal ground acceleration
The peak horizontal ground acceleration (PGA) experienced by an embankment dam has a major,
direct influence on the amount of crest settlement. This relationship is apparent in the plot shown in
Figure 1. In general, dams that experience greater PGAs undergo greater deformations and
damages. In this study, it was found that serious levels of damage were reported only in instances
where the PGA exceeded 0.2g. This finding supports one of the findings of an earlier investigation
in which it was concluded that “there is ample evidence that well-built dams can withstand
moderate shaking with peak accelerations up to at le ast 0.2g with no harmful effects” (Seed,
Makdisi, and DeAlba, 1978).
3.3 Magnitude
The amount of crest settlement is also directly related to the magnitude (M) of the earthquake. As
the magnitudes increase, settlements increase. This relationship held true even at sites where the
PGAs were identical because of the longer duration of strong motion shaking associated with the
greater magnitude event.
3.4 Other factors considered
Several other independent variables were analyzed statistically and were found to have only
minimal relational effects on the amount of crest settlement. These factors included dam type,
distance from seismic source to dam site, dam height, ratio of crest length to dam height,
embankment slope angles, and reservoir water level at the time of the earthquake.
4 RESULTS OF REGRESSION ANALYSES
The regression analyses also provided a mathematical relationship between the crest settlement and
the two factors, PGA and M. This relationship can be expressed as:
% Settlement = e
(6.07 PGA + 0.57 M -8.00)
(1)
where % Settlement = the amount of settlement of the crest of the dam (in meters) divided by the
height of the dam plus the thickness of the alluvium (in meters) times 100 (see. Fig 1); PGA = peak
horizontal ground acceleration of the foundation rock (in g) recorded or estimated at the dam site;
and M = earthquake magnitude (in surface-wave scale: MS).
This relationship is illustrated in Figure 2.
5
10
% STTLMT = e (6.07 PGA + 0.57 Ms + 8.0)
(6.07 PGA + 0.57 Ms + 8.0)
ESTIMATED CREST SETTLEMENT, in %(DH + AT)
% STTLMT = e
9
1
8
7
0.1
6
5
0.01
Earthquake Magnitude - M s
0.001
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
PEAK GROUND ACCELERATION (PGA), in g
Figure 2. Chart for estimating crest settlement
5 OTHER OBSERVATIONS
5.1 Calculated vs. actual crest settlements
Using the regression equation, crest settlements were calculated for each of the 69 case histories
included in the data base. Calculated settlement values are compared to the actual values in Figure
3. It is noteworthy that the statistical fit of actual to calculated values was found to be similar to that
for acceleration attenuation data from recent well-instrumented earthquakes including the Loma
Prieta earthquake (Governor’s Board of Inquiry 1990) the Landers earthquake (Boore et al. 1993),
and the Northridge earthquake (Finn et al. 1995). These statistical similarities suggest that
prediction of crest settlements cannot be improved unless the prediction of site-specific ground
accelerations can be improved. Also, this observation supports the prudent use of the mean-plusone-standard-deviation value of the PGA for estimating crest settlements of critical, high-hazard
structures.
10
Actual Settlement, % (DH + AT)
1
Actual settlement is
MORE than calculated
0.1
Actual settlement is
LESS than calculated
0.01
0.001
0.001
0.01
0.1
1
Calculated Settlement, % (DH + AT)
Figure 3. Actual vs. calculated settlements
6
10
5.2 Suitability of Newmark method for settlement calculations
Currently, it is common practice to use one of several analytical procedures based on the Newmark
method of analysis (Newmark 1965) to calculate theoretical crest settlements of embankment dams
subjected to earthquake loadings. This method is founded on the basic assumption that a rigid block
of soil slides downward along a definite shear surface whenever a critical “yield” horizontal
acceleration is exceeded.
There has been some concern expressed by others that the Newmark method may not correctly
model crest settlement caused by earthquake. Day (Day 2002) demonstrated that it is theoretically
possible for dry granular slopes to settle and spread laterally without earthquake accelerations
exceeding yield values to initiate slides. He says that the Newmark method may prove to be
unreliable in some instances. Matsumoto (Matsumoto 2002) described centrifuge shake table tests
and supporting nonlinear analyses for modelled accelerations up to 0.7g that revealed only shallow
ravelling with no deep shear surfaces in the core zones and no definite slip surfaces anywhere in
rock fill dam models. Accordingly, he says that the hypothesis of deep slide surfaces in the
Newmark approach “may be somewhat erroneous”.
Evidence from this case history study also refutes the settlement mechanism assumed in the
Newmark procedure. Personal inspection (Swaisgood & Au-Yeung 1991) and review of many
photos of earthquake damages to dams disclosed that crest settlements and deformation (for
structures not subject to liquefaction) seem to be from slumping and spreading movements that
occur within the dam body without distinct signs of shearing displacement. This appears to be true
for earth fill embankments as well as for rock fill dams. Longitudinal cracks along the crests have
the appearance of tension cracks with little or no vertical offset. An example of these crest cracks is
shown in Figure 4.
Figure 4. Tension cracks on Cogoti Dam crest after 1997 earthquake (Case No. 67)
6 CONCLUSIONS
Conclusions from this empirical study of embankment dam settlement and deformation during
earthquake include:
°
The vertical crest settlement experienced during an earthquake is an index of the amount of
deformation and damage incurred by the embankment
°
The amount of crest settlement is related primarily to two factors: peak ground
acceleration at the dam site and magnitude of the causative earthquake.
°
An approximate estimate of the amount of crest settlement that will occur due to an
assumed earthquake can be made by using mathematical formulas that relate deformation to
the peak ground acceleration and earthquake magnitude.
°
Deformation of a dam’s crest caused by earthquake is principally settlement and spreading;
apparently, there is no slide failure along a distinct shear plane.
7
REFERENCES:
Boore, D.M., Joyner, W.B., and Fumal, T.E. 1993. Estimation of response spectra and peak accelerations
from western North American earthquakes: an interim report, United States Geological Survey, Menlo
Park, California, Open File Report No. 93-509.
Day, R.W. 2002. Geotechnical earthquake engineering handbook. New York: McGraw-Hill.
Finn, L.,Ventura, C.E., & Schuster, N.D. 1995. Ground motions during the 1994 Northridge earthquake.
Canadian Journal of Civil Engineering, Vol 22, 300-315.
Governor’s Board of Enquiry on the 1989 Loma Prieta Earthquake: George W. Housner, Chairman. 1990.
Co mpeting Against Time, a report to Governor George Deukmejian.
ICOLD 2001. Design features of dams to resist seismic ground motion. Bulletin 120.
Matsumoto, N. 2002. Evaluation of permanent displacement in seismic analysis of fill dams. In Proc third
US-Japan workshop on advanced research on earthquake engineering for dams, San Diego, 22-23 June
2002.
Newmark, N. 1965. Effects of earthquakes on dams and embankments. Geotechnique, Vol 15 (2) 139-160
London.
Pinos S. F. 2000. Instrumentación de presas de tierra, aplicaciones para evaluar la respuesta sísmica de presas
chilenas. University of Chile (Universidad de Chile). Unpublished thesis presented to obtain the degree of
Civil Engineer in Construction and Structures (Ingeniero Civil en Construcción y Estructuras).
Seed, H.B., Makdisi, F.I., and DeAlba, P. 1978. The performance of earthfill dams during earthquakes.
Journal of the Geotechnical Engineering Division, ASCE, Volume 104, No. GT7, pp. 967-994.
Southern Peru Copper Corp 2001. Unpublished settlement monitoring data – Torata Dam.
Swaisgood J. R. 1998. Seismically-induced deformation of embankment dams. In proceedings of sixth
national conference on earthquake engineering. Seattle, Washington, U. S. A. May 31 – June 4 1998.
Swaisgood, J.R. and Au-Yeung, Y. 1991. Behavior of dams during the 1990 Philippines earthquake.
Presented at the ASDSO 1991 annual conference, San Diego, 29 Sep- 2 Oct 1991.
Tepel, R.E.; Nelson, J.L. & Hosokawa, A.M. 1996. Seismic response of eleven embankment dams, Santa
Clara County, California, as measured by crest monument surveys. In Seismic design and performance of
dams; Sixth annual USCOLD lecture, Los Angeles, 22 -26 July 1996.
8