You are on page 1of 11

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/253432676

Topographic site effects and slope stability


under dynamic loading

Article April 2003

CITATIONS READS

0 43

2 authors:

Cline Bourdeau Jean-Alain Fleurisson


Institut Franais des Sciences et Technologie MINES ParisTech
13 PUBLICATIONS 125 CITATIONS 32 PUBLICATIONS 52 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Jean-Alain Fleurisson on 11 September 2017.

The user has requested enhancement of the downloaded file.


ESG2006, Grenoble, 30/08-01/09/2006

Third International Symposium on the Effects of Surface Geology on Seismic Motion


Grenoble, France, 30 August - 1 September 2006
Paper Number: 151

NUMERICAL SIMULATIONS OF GROUND MOTION AMPLIFICATIONS


DUE TO ELEVATED TOPOGRAPHY AND SURFACE GEOLOGY
Jean-Alain FLEURISSON1, Cline BOURDEAU2
1 Ecole des Mines de Paris Centre de Gosciences, Paris, France
2 Itasca Consultant, Ecully, France

ABSTRACT Seismic recordings of numerous earthquakes worldwide have shown that


elevated topographic features cause highly amplified ground motions, especially at the top
of slopes. These topographic site effects give rise to many engineering problems dealing
with structure design in seismic mountainous regions. Topographic site effects are indeed
considered in most of the Seismic Design regulations but using very empirical approaches.
In order to improve the scientific knowledge in this field, 2D numerical simulations were
performed using the FLAC software with the objective at analysing the sensitivity of ground
motions along slope to geometrical, geological and seismic parameters. Modelled surface
amplifications were compared to the amplification factor used in the French seismic
design code. The results show that the calculated amplification factor is much lower than
the value of factor recommended in the regulations. In addition, the part of the slope
subjected to high amplifications is generally larger than recommended in the regulations.
Moreover, the simulations mainly highlight the effects of subsurface geology which are
added to the topographic effects and can therefore contribute to higher values of
amplification which are not considered in the regulations.

1 Introduction

It is well known from seismic recordings of numerous earthquakes worldwide, such as


Lambesc in France in 1909, Irpinia in Italy in 1980, California in 1987 and 1990, or Greece
in 1995 that topographic structures can lead to high amplifications of the ground motion
especially at the top of the slope.
More recently, such phenomena were observed in 2001 when El Salvador was struck
by a M=7.6 earthquake with an epicentre located about 100 km offshore south-southeast
of the capital city of San Salvador. This earthquake induced more than 500 landslides
which were responsible for a large part of the damages to buildings and roads and more
than 500 deaths out of a total of 700. As reported by Murphy (2002), the location of most
of these landslides appears to have been dominated by topographic and geological
conditions more than the distance between source and sites. An illustration of these site
effects is given in Figure 1. Seismic recordings show that while horizontal accelerations on
a rock site at Berlin, 80 km from the epicentre, did not exceed 0.23g, they reached 0.45g in
Santa Tecla located 90 km form the epicentre but in pyroclastic deposits named Tierra
Blanca, and 0.6g in Armenia located 110 km from the epicentre, but at the top of a small
hill underlain by pyroclastic deposits a few meters thick. Assuming that Berlin correspond
to free field conditions, it may be considered that the seismic signal has been amplified by
a factor 2 in Santa Tecla due to the specific geological conditions of the site, and by a
factor 2.6 in Armenia due to the combined effects of the topography and subsurface

1
ESG2006, Grenoble, 30/08-01/09/2006

geology. These amplified ground motions were supposed to have contributed to the
triggering of some of the major landslides.

Armenia U A R M lo n g it u d in a l

0,6g
0 .4
(g)

0,23g
Acceleration

0 .2
A c c e l

- 0 .2

- 0 .4
1 0 1 5 2 0 2 5
T im e
Time (s)
3 0 3 5 4 0 4 5
Armenia

Santa Tecla H S R F lo n g it u d in a l
Santa Tecla
Berlin
0 .4 0,45g
(g)

0,23g
Acceleration

0 .2
A c c e l

0
Epicenter
- 0 .2

- 0 .4

1 5 2 0 2 5 3 0 3 5 4 0 4 5
Time
T i m e (s)

Figure 1: El Salvador map and seismic recordings in Armenia and Santa Tecla.

As underlined by this example, the topographic and possible associated geological site
effects give rise to many engineering problems dealing with structure design in seismic
mountainous regions, slope stability assessment and landslide triggering. Most of the
seismic design codes such as the French PS92 recommendations indeed consider
topographic site effects but only through an amplification coefficient to be used in the
design of structure, along and at the top of the slope. This coefficient is however based on
empirical approaches without any assurance about its value, and the corresponding
formulations give results which sometimes seem far from the post seismic observations. A
better scientific knowledge in the field of topographical site effects is then required to
provide authorities the necessary elements which can be used in renewed and improved
regulations. For this purpose, various numerical simulations using FLAC software were
performed on standard 2D infinite uniform slopes. By reference to these simulations,
parametric analyses were carried out varying the geometrical parameters (height and
slope angle), seismic parameters (amplitude and frequencies of incident SV waves). In
addition, the effects of subsurface geology corresponding to superficial loose materials
such as decompressed and fissured rocks, loose weathered rocks or screes often present
along natural slopes were also considered in the simulations. Modelled surface
amplifications were compared to the amplification factor used in the French seismic
design code.

2 Numerical simulations of topographic site effects and ground motion


amplifications.

The results presented hereafter are based on a parametric study of ground motion
amplifications with respect to varying geometrical, geological and seismic conditions. They
are part of the PhD thesis of C. Bourdeau performed at Ecole des Mines de Paris under
the supervision of JA Fleurisson. Numerical models were performed using the 2D finite
difference code FLAC developed by Itasca Group.

2
ESG2006, Grenoble, 30/08-01/09/2006

2.1 Numerical model parameters

2.1.1 Seismic signal


The input signal is a vertically propagating SV wave applied at the bottom of the model. In
order to evaluate the effects of the input signal type and frequency content, two different
wave forms were used in the numerical simulation: a mono-frequency sinusoidal waveform
and a multiple frequency Ricker waveform (Figure 2).

Figure 2: Input signal waveforms: sinusoidal and Ricker

Both seismic signals are defined by their:


- peak ground acceleration (PGA) varying between 0.1 to 0.5g
- frequency content: it was varying between 1 to 6 Hz for the sinusoidal waveform, and
the central frequency of the Ricker signal was varying between 1.2 to 7.3 Hz;
- and the duration was 6 seconds for all the simulations and corresponding to the
repetition of the elementary signals presented in Figure 2.

2.1.2 Model and mesh size, boundary conditions


These different parameters were extensively studied through a preliminary parametric
analysis. As a consequence:
The mesh size is adjusted to the frequency content of the input signal so that its maximum
Vs
size is equal to where Vs is equal to the seismic velocity of the shear wave and
20 * Freq
Freq is the highest frequency of the input signal;
2*H
The model height is equal to 2*H and its horizontal length is equal to + 14 * H where
tan( )
H is the slope height and the slope angle;
Free fields boundary conditions are applied on the vertical lateral limits of the model and
quiet boundaries on the horizontal bottom limits. The objective of these boundary
conditions is to prevent artificial internal reflections of the input signal on the model limits.

2.1.3 Topographic and geological configurations of the models


As illustrated in the Figure 3, two types of slope configuration were simulated:
homogeneous slope model and slope model including a surface layer of constant
thickness corresponding to weathered or decompressed materials. These later models are
of course corresponding to very simple and schematic configurations, but most of them are
based on topographical and geological conditions of slopes which collapsed during the El
Salvador earthquake.

3
ESG2006, Grenoble, 30/08-01/09/2006

h1
H H

1
h

h1
H H

Seismic input signal: SV wave Seismic input signal: SV wave


Homogeneous slope Heterogeneous slope

Figure 3: Topographic and geological configurations

The corresponding models are defined by the following parameters:


- The slope height, which can be equal to 25, 50 or 75 m;
- The slope angle , which varies from 40 to 60
- The thickness of the subsurface layer, if any, which can be equal to 10, 20 or 30 m

2.1.4 Evaluation of the site effects


The site effects are characterized by the maximum amplification of the horizontal
acceleration calculated, in a given point, by the ratio between the maximum horizontal
acceleration at this point and the maximum horizontal acceleration corresponding to free
field conditions. This latter value is equal to two times the maximum horizontal acceleration
of the input signal

2.2 Effects of the topography

The maximum amplification of the horizontal acceleration was calculated in different points
along the surface for homogeneous slope models with different heights or slope angles.

Figure 4 give the amplification at the crest of moderate to very steep, nearly vertical
homogeneous slopes. The slope height is 50 m and two types of material corresponding to
standard basalt (model 1) and pyroclastic (model 2) deposits in Salvador (Mavrommati,
2000 and Faccioli, 1986) are considered. The input signal frequency is equal to 1 Hz.

The calculation results show that the ground motion is amplified at the crest of the slope
and the amplification ratio increases with the slope angle up to 1.5 for the steepest slopes.
The amplifications take higher value in model 2 because of a slight phenomenon of
resonance at this frequency.

4
ESG2006, Grenoble, 30/08-01/09/2006

Poisson Ratio Vp (m/s) Vs (m/s) Density


Model 1 0.25 780 450 2,0
Model 2 0.3 467.5 250 1,3
Figure 4: Influence of the slope angle on the acceleration amplification at the crest of
the slope

Figure 5 gives the maximum amplification of the horizontal acceleration along the
surface of a 25 m high homogeneous slope for different slope angle respectively equal to
40, 50 and 60, and Figure 6 gives the maximum amplification of the horizontal
acceleration along the surface of a 50 steep homogeneous slope for different slope height
H respectively equal to 25, 50 and 75 m. In both cases the input signal is a Ricker signal of
central frequency equal to 1.2 Hz referred as low frequency (LF) or 7.3 Hz considered as
high frequency (HF).
All these results clearly show the influence of the frequency on the amplification
coefficient. For LF signal, the toe of the slope is generally subjected to de-amplifications
while amplifications occurred at the top of the slope. The maximum amplification
coefficient is around 1.2. Moreover the zone affected by high amplifications largely extends
behind the top of the slope. For HF signal, de-amplification is observed all along the slope
from the bottom to the top, and the amplification peak is located behind the crest.
Figure 5 underlines the role of the slope angle: for LF signal, the amplification coefficient
slightly increases with the slope angle while the extension of the amplification zone at the
top of the slope slightly decreases. For HF signal, the amplification peak coefficient and
extension of the amplification zone highly decreases when the slope angle becomes
higher
As illustrated by Figure 6, for a given slope angle, the amplification coefficient as well as
the amplification zone decreases when the slope height increases, and these results are
the same for LF and HF signal.

5
ESG2006, Grenoble, 30/08-01/09/2006

Figure 5: Maximum amplification of the acceleration along the surface of a 25 m high


homogeneous slope subjected to a Ricker input signal of central frequency equal to 1.2
and 7.3 Hz for different value of the slope angle = 40, 50 and 60

Figure 6: Maximum amplification of the acceleration along the surface of a 50steep


homogeneous slope subjected to a Ricker input signal of central frequency equal to 1.2
and 7.3 Hz for different value of the slope height H= 25 m, 50 m and 75 m

6
ESG2006, Grenoble, 30/08-01/09/2006

As studied comprehensively in Bourdeau (2005), the amplifications are resulting from


the constructive interference between the input signal, the waves reflected on the
topography and the Rayleigh waves resulting from the diffraction phenomena on the
topography.

2.3 Associated effects of topography and subsurface geology

Figure 7 gives the maximum amplification of the horizontal acceleration along the surface
of a 25 m high and 50 steep heterogeneous slope including a subsurface layer with low
seismic velocity, such as Tierra Blanca, lying on a substratum such as basaltic rocks. The
thickness of this layer varies from 10 to 30 m, and the slope is subjected to a Ricker input
signal with a central frequency varying from 1.2 to 7.3 Hz.

h1 = 10 m (fres = 6.3 Hz) h1 = 20 m (fres = 3.1 Hz)

h1 = 30 m (fres = 2.1 Hz and fres (n=1) = 6.3 Hz)

Figure 7: Maximum amplification of the acceleration along the surface of a 25 m high


and 50steep heterogeneous slope subjected to a Ricker input signal of variable central
frequency for different thickness of the subsurface layer (a) h1=10 m, (b) 20 m and (c)
30 m

From a general point of view, the amplification coefficients are much higher than the
amplification coefficients calculated for homogeneous slope. They can reach the value of
2.7 to be compared to the value of 1.3 for homogeneous slopes. The values have the
same order of magnitude as the amplification coefficient of 2.6 calculated at the Armenia

7
ESG2006, Grenoble, 30/08-01/09/2006

site, for which topographic and geological conditions are close to those of Figure 7a.
These figures also show that the amplification behind the slope crest decreases more
quickly when the slope is subjected to HF input signal.
This amplification coefficient is mainly controlled by the geological structure of the slope
and particularly the thickness of the subsurface layer, and the frequency of the input signal.
As a matter of fact, all these results clearly show that for a given thickness of the
subsurface layer, the highest value of the amplification coefficient is obtained when the
central frequency of the signal is close to the resonance frequency of the subsurface layer.
In the considered cases, the seismic velocity of the shear waves in the subsurface layer is
equal to 250 ms-1 corresponding to Tierra Blanca deposits. For a subsurface layer
thickness equal to 10, 20 and 30 m, the resonance frequency is respectively equal to 6.3,
3.1 and 2.1 Hz, and the maximum amplification coefficient is obtained when the central
frequency of the input signal is respectively equal to 6.1, 2.4 or 3.1 and 2.1 Hz.

3 Comparison with the French regulations and consequence in terms of Seismic


design code

Pending the implementation of the future Eurocode 8, the present French regulation is the
PS92 seismic design code which represents the topographic site effect through an
amplification coefficient defined as in the Figure 8. The seismic signal considered in the
design of a structure located on the slope will result from the multiplication of the signal of
reference, in the considered area, by this coefficient.

I
D
H 10 m and i :
B C 3
C D On branch BC, (max) is equal to
B i
H A 1 = 1 for I-i 0.4
IA
O = 1 + 0.8 * (I-i-0.4) for 0.4 I - i 0.9
1 = 1.4 for I - i 0.9

1 1 H + 10
b = min (20 I, ) ;
4
H H
AB = ; CD = ;I= tan (slope angle);
a c 3 4
i = tan (angle between BD and horizontal)

Figure 8: Distribution and calculation of the topographic coefficient along the slope
surface according to the French PS92 design code.

According to the equation of Figure 8, the highest amplification zone is located at the
top of the slope along the BC branch, and the maximum corresponding value is equal to
1.4 for slopes steeper than 40, and ranges from 1.0 to 1.4 for more gentle slopes.

The amplification coefficient as well as the extension of the zones of amplifications


along the slope (lengths a, b and c corresponding to branches AB, BC and CD) proposed
in the PS92 design code were compared with the results of the various numerical
simulations performed on homogeneous slopes. The Figure 9 gives an illustration of
results corresponding to some particular configurations.

8
ESG2006, Grenoble, 30/08-01/09/2006

Figure 9: Comparison between calculated and PS92 topographic coefficients

For the analysed configurations, this figure show that the maximum topographic
coefficient BC=1.4 recommended in the PS92 design code is much higher that the values
obtained from calculation. Moreover the amplification peak may even occur far from the
crest of the slope. On the other hand, the extent of the zone affected by amplification
behind the slope is more important in the simulations than in the PS92 recommendations.
For instance, for a 25 m high and 60 steep slope subjected to a LF input signal, the
maximum amplification coefficient is equal to 1.2, but amplifications value higher than 1
may occur over a distance of 125 m behind the crest of the slope. This result would
suggest longer security distances than those recommended in the PS92 for settlement of
building at the top of slopes.
The amplification value calculated along the surface under the slope crest are always
lower than the coefficient which can not be lower than 1 and can not therefore account
for de-amplifications obtained in the numerical simulations and recorded in the field.
But, it must be remembered that the amplification coefficient may reach very high value
when the slope includes a subsurface layer with low seismic velocity. For instance, as
illustrated in Figure 7a for a slope composed of a bedrock overlaid by a 10 m thick surface
layer, the amplification coefficient may vary between 1.6 and 2.6 depending on the
frequency of the input signal, and largely exceeds the maximum topographic coefficient of
1.4 recommended in the PS92.

4 Conclusion

It is known for a long time that elevated topographic structures may lead to high
amplifications of ground motions, especially at the top of slopes. These site effects give

9
ESG2006, Grenoble, 30/08-01/09/2006

rise to many engineering problems especially dealing with structure design in mountainous
seismic regions, and they have been considered in most of the Seismic Design regulations
such as the French seismic design code PS92. The code particularly defines a seismic
amplification factor used in the design of surface structures along and at the top of
slopes. This coefficient is however described through empirical approaches without any
assurance about their value, and the corresponding formulations give results which
sometimes do not correspond to the post seismic observations. A better scientific
knowledge in the field of topographical site effects is then required to provide authorities
the necessary elements which can be used in renewed and improved regulations.
Because of the complexity of such physical phenomena which depend on topographic,
geological and also seismic parameters, numerical simulations could represent a powerful
tool to explore the answers to simple systems, made gradually more complex, for finally
approaching the natural configurations with highly complex geometries and geological
structures. Moreover, amplifications of topographic and lithological origin have been too
often separated, whereas geological, geomorphological and geophysical reality of natural
slope integrates these two aspects.
The results presented in this paper correspond to the first phase of this larger project
because they are based on numerical simulations of deliberately simplified slope
configurations. They however confirm that topography may induce ground motion
amplifications which remain relatively moderate and lower than 1.3. On the other hand,
they clearly underline the important role played by the subsurface geology which may lead
to much higher amplifications up to 2.5. Moreover these amplifications are greatly
controlled by the frequency of the seismic input signal. A low frequency signal generally
produces higher amplifications, but the intensity and distribution of the amplifications may
be greatly modified when the frequency of the seismic signal is close to the resonance
frequency of the subsurface layer.
The comparison between the numerical results and the amplification factor used in the
French seismic design code show that the calculated amplification factor is much lower
than the value of factor recommended in the regulation. But the part of the slope
subjected to high amplifications is generally larger than recommended in the regulations.
So, additional simulations are therefore necessary in order to investigate more complex
and mainly more realistic geomorphological configurations, but it is worth developing this
kind of approach combined of course with field measurements in order to better
characterize the seismic coefficients recommended in the future seismic design codes.

5 References

Bourdeau C. (2005). Effets de site et mouvements de versants en zones sismiques: apport de la


modlisation numrique. Thse de lEcole des Mines de Paris. 266 p.
Faccioli E., Battistella C., Alemani P. & Tibaldi A. (1988). Seismic micro-zoning investiagtaions in the
metropolitan area of San Salvador, El Salvador, following the destructive earthquake of October 10,1986.
Proceedings of International Seminar on Earthquake Engineering, Innsbrck, pp. 28-65
Mavrommati Z.C. (2000). Seismic behaviour of slopes in unsaturated volcanic soil. MSc Dissertation.
Imperial College. London.
Murphy W., Bommer J., Mankelow J.M. (2002). Mechanisms of slope failure in volcanic soils during
th
earthquakes. Proceedings of the 12 European Conference on Earthquake Engineering. Paper N 782,
London, 9-13 September 2002.

10
View publication stats

You might also like