Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Carlo
G.
Lai
Aussois,
October
3-5,
2013
1
acknowledgements
The case-studies presented herein are taken from scientific research, PhD
and MSc theses conducted by:
outline
MOTIVATION
CASE STUDIES
- Seismic analyses of a maritime blockwork wharf structure
- Seismic analyses of a maritime pile-supported wharf structure
- Dynamic soil-structure interaction direct approach
- Numerical modeling to support physical modeling: model of the model !
- (Seismic analysis of an embankment dam) if time permits !
3
Motivation
motivation
Constant progress occurred in latest decades on computational side and on
development of specialized software has spread also in geotechnical
engineering the recourse to advanced dynamic numerical analyses.
Mohr-Coulomb
failure criterion
bounding
surface
plasticity
motivation
Advanced numerical analyses allow to simulate very complex problems
which would be difficult if not impossible to study by means of simplified
modeling (e.g. Newmark method, pseudo-static analyses, etc).
Often soil-structure interaction problems
ff
(from Conte, 2007)
motivation
MOSE PROJECT
VENICE LAGOON,
Seismic-geotechnical
numerical analyses
peculiarities
Peculiarities of numerical dynamic analyses with FEM, SEM, BEM, FDM or
hybrids methods in geotechnical engineering, with respect to corresponding
analysis in structural engineering, are multi-faceted:
1. unbounded physical domain (e.g. deep tunnels) or semi-infinite (e.g.
above-ground structures) ! introduce tricks or stratagems to simulate
energy radiation through the boundaries of the model.
infinite elements
adsorbing layers
non-reflecting boundaries
local or integral boundaries
consistent boundaries, PML
paraxial approximation
9
peculiarities
10
peculiarities
2. definition of seismic input, represented by displacement, velocity or
acceleration time series which should take into account the differences
between the outcropping ground motion recorded at the free-surface and
ground motion recorded at depth (deconvolution ill-posed problem)
accelerazione alla base
ff
dec
dec
tempo
(s)
time (s)
dec
11
peculiarities
3. GRID DISPERSION: proper energy transmission within the domain for
different excitation frequencies requires definition of maximum size of mesh
discrete elements to achieve accurate modeling of the deformability of the
continuum in relation to wavelengths propagating through the model:
G = constant (510 FEM, FDM)
To be accurate up to a frequency fmax if the minimum shear wave velocity of
the model is VSmin at least 10 points per minimum wavelength are needed in
the finite difference method (FDM) to avoid grid dispersion.
SEM are very accurate: 3-4 points per lmin useful in large-size models
12
peculiarities
4. STABILITY: time step (t) and size (x) of discrete elements cannot be
two independent variables Courant-Friederichs-Levy stability condition :
C = CFL number
Meaning of CFL stability condition is that time step cannot be larger than the
time required for any perturbation to propagate over distance x.
If Courant-Friederichs-Levy stability condition is satisfied error magnitude is
bounded and algorithm is stable. Value of C depends on numerical method.
13
peculiarities
Soil constitutive modeling and hydraulic boundary conditions
5. geomaterials and thus soils and rocks are particulate multi-phase
systems constituted by solid skeleton interacting with one/more fluid phases.
Refined constitutive modeling of porous media requires to take into account
hydro-mechanical coupling. Mechanical response arising from seismic
input followed by hydraulic response (seepage and dynamic consolidation).
pore fluid
solid particle
inter-granular forces
(from Kramer, 2001)
14
peculiarities
Soil constitutive modeling and hydraulic boundary conditions
5. time evolution of two phenomena may be very different depending on soil
hydraulic conductivity and frequency content of seismic excitation.
There is a problem of proper simulating hydraulic boundary conditions at
the boundaries where the seismic input is applied.
15
peculiarities
Initialization of geo-static stresses & dynamic boundary conditions
6. Dynamic analysis of a geotechnical boundary value problem requires
initialization of the litho-static (or geo-static) stress tensor.
Application of seismic input which follows this phase should be accompanied
by modification of initial boundary conditions applied along the
boundaries of the model and of soil mechanical impedance.
litho-static stresses
seismic input
(from Lai and Paolucci, 2008)
absorbing boundary
16
peculiarities
Near-field effects (kinematic interaction)
7. Almost always in numerical modeling of geotechnical problems soil is in
contact with one or more structural elements.
In dynamic analyses mechanical impedance contrast caused by finite-size
geometrical components may generate multiple diffractions of seismic waves
having wavelengths comparable with the dimensions of component.
Modification of the incident wave field close to anomaly. This must be taken
into account in relation to the choice of geometric dimensions of numerical
model avoid to set boundaries close to near field.
kinematic interaction
17
peculiarities
Effects of soil nonlinearities and instability
8. Soils exhibit inelastic and nonlinear behavior even at low-strain levels. In
loose, saturated, coarse-grained materials such nonlinearities may lead to
instability phenomena (liquefaction or cyclic mobility).
Liquefied soil is not able to transmit energy associated with S waves.
This must be taken into account in defining the seismic input ! avoid to
introduce boundaries in soil zones susceptible of severe nonlinear response.
S wave
propagation of
shear waves
18
peculiarities
Computational resources and costs
9. Numerical (i.e. discrete) models of geotechnical systems have in general
relevant geometric dimensions associated to a dense computational grid.
Adoption of complex constitutive models to simulate soil hydro-mechanical
response and interface conditions between soil and structural elements.
Demand of computational resources required to carry out these
analyses may be very onerous also in terms of time duration
This aspect must be taken into account when selecting accelerograms
particularly in relation to duration and sampling frequency
19
Case study
seismic analysis of a maritime
blockwork wharf structure
20
21
Liquefaction assessment
23
Residual Horizontal
Displacement (RHD)
= 36 cm
25
26
9 Input variables:
backfill, foundation,
G backfill, G foundation,
block-foundation,
block-backfill
Calibration
FLAC input files used with
random properties for input
variables
Training
R=0.99
28
Closed-form expression
Prediction of wharf displacement for different values of Vs30 (200, 300, 400
m/s) at 9 intensity levels
29
W/H = 0.88
W/H = 1.12
32
Case study
seismic analysis of a maritime
pile-supported wharf structure
33
34
18 m
55 m
Loose sand
Silty clay-Clay
35 m
Silty sand
Bed Rock
110 m
18 m
55
m
35
m
110 m
0.63m
1m
1m
1m
grid size upward is 1m to comply with the highest ground motion frequency
36
Quiet boundary
Quiet boundary
Quiet boundary
PlasFc hinge
18 m
35 m
Pile
Element
with
interface
55 m
Beam
Element
with
axial
constraint
110 m
interaction of each structural node with soil element simulated by nonlinear coupling springs
39
41
43
44
45
46
typical response showing soil liquefied zone & plastic hinges in piles
47
methodology
48
Silty
Clay-Clay
Bedrock
Silty
Sand
Bedrock
50
sample
of
20
realiza/ons
is
sucient
to
capture
variability
in
stochas/c
model
of
target
wharf
52
53
Case study
dynamic soil-structure
interaction direct approach
54
55
56
57
58
59
60
61
62
63
64
65
66
67
.
240 811
.
723 588
/
(
)
Case study
numerical modeling to
support physical modeling
69
70
71
CENTRIFUGE MODELLING OF DISCRETE PILES USED FOR SLOPE STABILISATION Nottingham University
72
73
74
75
76
77
78
LIGHT CONCRETE
to simulate bedrock
79
80
81
82
83
84
Modeled in FLAC3D
Modeled in FLAC2D
elements
Dynamic input is simulated velocity
controlled
85
86
87
88
89
90
91
92
93
Close-up view
94
Istituto Universitario
di Studi Superiori di Pavia
Carlo
G.
Lai
Aussois,
October
3-5,
2013
95
Case study
seismic analysis of an
embankment dam
96
http://www.panoramio.com/photo/6405926
41 39 Latitude
14 03 Longitude
http://an.wikipedia.org/wiki/Imachen:Map_Region_of_Molise.svg
General characteristics
Height of crest
Above lowest foundation
34 m
25 m
Strength parameters
Free height
Above normal water level
2.5 m
2.5 m
Dam Zone
Upstream and
downstream
slope
f []
c [kPa]
24
30
Width of crest
6m
Length of crest
292 m
Core
30
40
400,000 m3
Foundation
35
50
98
(http://esse1-gis.mi.ingv.it/s1_en.php)
The ground
Intensity
Measure (IM) used in this study was taken as the Peak
Retunmotion
Periods
considered
Ground Acceleration (PGA). Spectral ordinates for other periods needed to generate
the Uniform Hazard Spectra (UHS) were obtained from INGV study [2004].
Probability of exceedance
PGA [g]
100
0.141
200
0.191
475
0.265
950
0.337
1950
0.434
2475
0.455
99
The plot shows the compatibility of the mean response spectra of the selected
natural records to the probabilistic UHS for the same return period Tr=475 years.
100
101
103
Newmark
(1965)
Yegian et al.
(1991)
Jibson (1994)
0.088
0.030
0.015
0.050
0.000
0.000
0.000
0.000
0.705
0.075
0.155
0.130
104
105
106
109
111
CoV=10%
CoV=20%
CoV=40%
4m
10m
20m
4m
10m
20m
4m
10m
20m
L/2
4m
10m
20m
4m
10m
20m
4m
10m
20m
L/3
4m
10m
20m
4m
10m
20m
4m
10m
20m
For each value of CoV nine realizations are carried out. Three for each value of the
assumed horizontal correlation distance (h=L, L/2 & L/3). Each horizontal correlation
distance is fixed for analyzing each of the three vertical correlation distances, v (4m,
10m, and 20m)
Assumption of no cross-correlation between cohesion and friction angle (r=0) is made
113
The variability of the response of the dam due to the uncertainty of the most influential
input variables becomes more evident if compared with the results obtained where the
analysis was carried out using best estimate values for the input parameters
(homogeneous soil assumption)
114
The fitted probability distributions of these two sets of simulated EDPs are relatively
close at low seismic hazard levels and become considerably different at high seismic
hazard levels
115
For this study the random modelling of soil properties in the probabilistic seismic
demand analysis INCREASES the seismic demand hazard.
Accounting for the uncertainty of soil parameters is therefore a significant factor in
assessing the seismic demand on the dam.
116