You are on page 1of 87

Developments in modelling techniques

of soil-water-structure interaction
History, examples and practical applications

Dirk Luger

10 december 2015
Contents

Introduction and main messages

History and lessons learned as a journey through time


Storm Surge Barrier Maeslantkering near Rotterdam (1995)
Palm Deira earthquake deformations (2006)
Incheon bridge ship collision protection (2006)
Earthquake amplification factors (2009)
Windjack spudcan impact study (2012)
Marsrover wheel-soil interaction (2013)
Burgum bridge pier protection (2015)

Closure

10 december 2015
Introduction and main messages

My background: more emphasis on predicting soil structure


behaviour as realistic as possible rather on calculations that
aim to prove that a certain design code or standard is
complied with. That comes later.
Another reason for that comes from my involvement in
forensic geotechnical engineering. Thats an area where
understanding what actually happened is crucial.
This requires selection of parameters fit for the job. Purpose
of the calculation and the mechanisms that develop can
determine to a large extent what the proper set of soil
parameters is.

You will seldom get the proper soil parameters of the


shelf. Youll have to make them consistent with your
engineering problem.

10 december 2015
Introduction and main messages

A voyage through time to put what we can do nowadays into


perspective

Quality and power of the tools at our disposal have increased


enormously

With that the risk that calculation results are taken for granted
has increased as well (they look nice and everything is
modeled, so it has to be OK..)

You have to keep thinking, train your engineering


judgment and learn to trust it. Simple checks can
reveal a lot!

10 december 2015
Introduction and main messages

When extrapolating beyond tested ranges or application areas


verification of our models by feedback from actual behaviour
(monitoring structures) and from model tests is
indispensable.

Whenever youre venturing in an area where you havent been


before, make sure youve done everything to verify that your
calculations are reliable.

Calculation Physical model Real Structure

Feedback loops

10 december 2015
Maeslantkering (1996)

10 december 2015
Maeslantkering
Storm Surge Barrier
Where and why
Delft
An alternative to dike
reinforcements in South Holland
is a storm surge barrier.
This option turns out to be the
most attractive in terms of cost,
environmental effects and
safety.

In 1987 the Dutch government decides


that the New Waterway Storm surge
barrier should be constructed. The Dutch
government asked building contractors to
produce a design for the storm surge
barrier. Six designs were submitted. The
Bouwkombinatie Maeslant Kering (BMK)
submitted the winning design.
Overview

360 m wide
18.5 m deep
Just for scale..

As high as the
Eiffel tower, but
twice as heavy!!
Main components, North side
Driving unit
North door
Main truss
Control building

Foundation block

Dry dock

Barrier sill
Primary sheet pile wall Ball joint Back-up sheet pile wall

Sea Rotterdam, river


Back-up sheetpile wall

Risk of a ship veering off course and colliding with the


foundation block.
Sand flowing out from under the foundation block would
cause a severe collapse of the structure.
Having a back-up sheetpile wall prevents this and
enables relatively quick repairs.
And actually, during a design meeting on site, in which some people
expressed their doubt regarding this risk we were suddenly warned
that a ship had collided with the main sheetpile wall, fortunately at a
moment and a place which did not lead to flooding of the building
pit..
Geotechnical design calculations

Traditional settlement calcs.


(level of terrain, settlement of foundation block)
2-D FEM calculations
(parallel to main loading direction, perpendicular to
sheetpile wall, before and after ship collision)
BEM calculations
(Stresses under foundation block)
Discrete element dynamic calculations
(Ship collision effects)
Interaction

Asymmetric loading and combined perpendicular and in-


plane loading of the sheet pile wall, both through soil and
via anchors.

Having to account for


interaction between:

- Foundation block
- Back-up sheetpile wall
- Main sheetpile wall
Parallel modeling (direction of main load)

Displacements
Perpendicular modeling

Loads from parallel


and perpendicular Displacements
calculations were
combined to determine
the final dimensions
3-D BEM calculations
(Stresses under foundation block)
Surrounding
Self-weight Combination of both
ballast

-0.1 MPa

0.3 MPa
+ = 0.2 MPa

0 MPa
0.2 MPa

-0.1 MPa

0.1 MPa

+ =
3-D BEM calculations
(Stresses under foundation block)

0.3 MPa

0.1 MPa

0.2 MPa

0.2 MPa

0.1 MPa

0.3 MPa
Construction of the dock at the South side
Construction of the door in the dock
The main truss
500 mm camber during supported construction
80 mm camber after removal of supports
and closed.
Earthquake induced displacements

A method developed in the context of the


Palm Deira development

10 december 2015
Question

How to verify that my +0.6

embankment structure +0.4

remains within acceptable +0.2

deformation limits if the 0.0

design earthquake occurs? -0.2

-0.4

-0.6

Seismic Risk 2008 26


Sliding Block Analysis

By means of a slip circle approach one determines for which


acceleration the factor of safety falls below 1.0.
This is the so-called yield-acceleration: ayield

ayield

For a ayield one finds F.S. 1.0

Seismic Risk 2008 27


Sliding Block

Use published graphs or


perform own integration of
selected time-histories to
determine earthquake-
induced displacement.

Seismic Risk 2008 28


Sliding block

Advantage:
Simple easy to evaluate for many time histories
Disadvantage:
Only one displacement value (for the sliding block)
Not accounting for water next to the slope

Ayield

PGA

Seismic Risk 2008 29


Sliding block

Advantage:
Simple easy to evaluate for many time histories
Disadvantage:
Only one displacement value (for the sliding block)
Not accounting for water next to the slope
Not accounting for failure in overlying layers

ay=0.2 g g
a =0.25 ay=0.1 g
y

PGA = 0.4 g

Seismic Risk 2008 30


Dynamic FE analysis

Actual acceleration time history as


boundary condition at the base of
the mesh.

+ Continuous deformation field


- CPU intensive
- One time-history is not sufficient
- Free water causes problems

Seismic Risk 2008 31


Deforming Continuum Method

Apply a constant horizontal acceleration at the base of the model


and observe what acceleration level can be transferred to
the different parts of the embankment

Each line represents 0.2 m/s2 = 0.02 g

Seismic Risk 2008 32


Excess pore pressures

Estimate on basis of standard procedures: Cyclic shear stress level and


relative density of the soil.
At the onset of the earthquake excess pore pressures a zero, by the end
they have reached their maximum value.
Current approach: use the average..

+0.6 Entering excess pore pressures in


+0.4
the model by reduction of the
material strength: at 50% excess
+0.2
pore pressure we introduce a
0.0
material that has 50% of its original
-0.2 strength:
-0.4

-0.6
new = atan(0.5 tan(org ))

Seismic Risk 2008 33


Sample

Mesh

Hor. acceleration

Vert. acceleration

Shear strains

Seismic Risk 2008 34


Accelerations and displacements

Ayield-hor [g] Verpl-hor [cm]


-0,035 ; -0,005

1 cm
-0,04; -0,02
10 cm

-0,11 ; -0,04 -0,07 ; -0,04

Ayield-vert [g]
-0,12 ; -0,065
Verpl-vert [cm]

Seismic Risk 2008 35


In short:
A nice method filling the gap?

Seismic Risk 2008 36


Incheon Bridge Ship collision prevention

10 december 2015
Incheon bridge overview

10 december 2015
Idealized prototype 20 m diameter

10 december 2015
The dolphin model

10 december 2015
Modelling the sheetpile

10 december 2015
Set-up of the model

moving mass actuator


dolphin

mounting plate

sand filled
container

water basin

assembly plate

10 december 2015
Set-up of the model

10 december 2015
After the test

10 december 2015
Forces derived from ship slowdown

10 december 2015
10 december 2015
10 december 2015
Earthquake amplification factors

10 december 2015
Earthquake amplification factors

Limits to PGA
and amplification

10 december 2015
Limit to acceleration - preliminary analysis
CP stand 0.5g...
Acceleration
4,0
Point A -394.4

Point B -397.1

Point C -399.7

Point D -401.5

2,0
Point E -404.7

Point F -410.6

Point G -427.0

Point H -441.7

0,0
Point I -456.7

Point J -464.0

-2,0

-4,0

Demonstrated mechanism but needed clearer presentation


-6,0
0 1,0 2,0 3,0 4,0 5,0
Dynamic time [s]

10 december 2015
Mechanism

M1
| Peak acceleration | < Su1 / M1
Su1 So in the top layer 2 values
M2
Su2
M3 | Peak acceleration | < (Su1 Su2) / M2
So for an intermediate layer 4 values

10 december 2015
Try out for simple signal

Input at base 1g at 1Hz

10 december 2015
Velocities make it clear
Time_vx
Vx [m/s]
2,0
Point A

Point J

1,5
Point I

Point H

1,0
Point G

Point F

0,5
Point E

Point D

0,0
Point C

Point B

-0,5

-1,0

-1,5

-2,0
0 1,0 2,0 3,0 4,0 5,0
Dynamic time [s]

10 december 2015
Velocities make it clear
Time_vx
Vx [m/s]
2,0
Point A

Point J

1,5
Point I

Point H

1,0
Point G

Point F

0,5
Point E

Point D

0,0
Point C

Point B

-0,5

-1,0

-1,5

-2,0
0 1,0 2,0 3,0 4,0 5,0
Dynamic time [s]

10 december 2015
Amplification at 1g base acc.

Input at base 1g at 0.4 Hz

10 december 2015
Amplification at 0.01 g base acc

Input at base 0.01g at 0.4 Hz

10 december 2015
WindJack

Spudcan-seabed impact interaction

10 december 2015
The WindJack JIP

Soil-Structure Interaction Modelling 06-Feb-14 59


The WindJack JIP

Soil-Structure Interaction Modelling 06-Feb-14 61


The WindJack JIP

Soil-Structure Interaction Modelling 06-Feb-14 62


The WindJack JIP

Soil-Structure Interaction Modelling 06-Feb-14 63


The WindJack JIP

Forces have to be
corrected for inertia
effects.

Note the force to set


the spudcan in motion
and the force to stop it
again.

Soil-Structure Interaction Modelling 06-Feb-14 64


The WindJack JIP

Initial analytical spudcan-


seabed interaction model
performance.

Still without hydro-


dynamic effects, inertia
and rate effects.

Soil-Structure Interaction Modelling 06-Feb-14 65


The WindJack JIP

Soil-Structure Interaction Modelling 06-Feb-14 66


The WindJack JIP

MPM calculation results

Soil-Structure Interaction Modelling 06-Feb-14 67


Marsrover wheel-soil interaction

10 december 2015
Previous work: Finite elements

Lagrangian (or ALE) based approach for wheel and soil.


disadvantage: limited to wheels without grousers and curved edge

69
Previous work: Discrete Element Method (DEM)

advantage: grousers possible, numerical stability


disadvantages:
often 2D, unrealistic soil transport (impossible to go sideways)
parameters for particles difficult to relate to physical quantities
less suitable for compactive geomaterials (powder like)

70
Example of coupled Eulerian-Lagrangian FEM

Eulerian soil model and rigid (Lagrangian) wheel.


Wheel/soil is half because of symmetry

71
Flexible wheel modeling
Diameter 25 cm, width 11.2 cm.

grousers
shell

Deformable
body
Only half of the wheel is
modeled (symmetry in FEM
model)
Rigid wheel modeling
Diameter 25 cm, width 11.2 cm.

Same features as flex


wheel, in rigid body
constraint

Only half of the wheel is


modeled (symmetry in FEM
model)
Rigid wheel 60% slip
Burgum bridge pier protection

10 december 2015
Analysis of bridge pier

10 december 2015
Little effect of meshing

10 december 2015
Soil parameters

Dr sat eini E50_ref Eoed_ref Eur_ref G0_ref 0.7 (*)


% [kN/m3] [-] [kPa] [kPa] [kPa] [kPa] [%] [degr] [degr]

50 17.0 0.60 35000 35000 105000 94000 0.0150 34.3 4.3


(2.15)

75 18.0 0.52 50000 50000 150000 111000 0.0125 37.4 7.4


(3.7)

65 17.6 0.55 44000 44000 132000 104200 0.0135 36.1 6.1


(3.05)

10 december 2015
Parameters for larger strains

10 december 2015
Interface strength @ sheetpiles

a part where soil-soil or concrete-


concrete friction is mobilized and
a strength reduction factor of 1.0
applies and
a part where soil-steel or
concrete-steel friction is
mobilized where typically a
strength reduction factor of 0.67
is applied.

Rinter = (422/1160)*0.67 + ((1160-422)/1160)*1.0 = 0.88

10 december 2015
Effect of lower dilatancy

10 december 2015
Still room for optimisation: from 22m to 18m

10 december 2015
Effect of the bridge

10 december 2015
10 december 2015
10 december 2015
Movement of the bridge

10 december 2015
Results

10 december 2015
10 december 2015
Closure main messages

Train your engineering judgment and learn to trust it.


Simple checks can reveal a lot!

Select proper soil parameters, consistent with your


engineering problem.

Verify models by feedback from actual behaviour


(monitoring of structures) and by performing model
tests.

10 december 2015
Closure - thanks

For further info on Deltares or this presentation feel free to contact:


In the Netherlands:
Dirk Luger dirk.luger@deltares.nl M:+31 6 2049 1414
In Dubai:
Geoff Toms geoff.toms@deltares.nl M:+971 4 337 8353

10 december 2015

You might also like