Modeling and Simulation of Vibratory Extraction of Piles

Innovation Intelligence®

Physical Component Modeling

in Altair ScicosPro

Application to Vibratory Pile Extraction

Franck Delcroix

European ATC - June 25th 2014, Munich

Copyright © 2014 Altair Engineering, Inc. Proprietary and Confidential. All rights reserved.

Presentation Outline

• Physical Component Modeling in ScicosPro

• The Coselica Library

• Application – Vibratory Pile Extraction

2


ScicosPro and Physical Component Modeling

ScicosPro is made for system level modeling approaches

(efficient handling of discrete time and events).

Modelica is adapted to component level modeling approaches.

But the basic formalisms are compatible, which allows for the use of

Modelica components in a ScicosPro model.

ScicosPro offers a Physical Component Modeling library based

on Modelica blocks (Coselica library)

3


ScicosPro – Hybrid System Modeling

Conventional ScicosPro blocks have clearly defined input and output

ports, where an input can be regarded as cause and an output as effect.

Models built solely with these blocks are causal models.

In contrast, most Modelica blocks have neither identifiable input nor

output ports. Instead, they have connector ports which propagate

physical quantities (e.g. electrical voltage and current), and it is a priori

undecided what quantity causes which effect.

4


Causal modeling (explicit blocks)

DuCxy

BuAxx

an explicit block

in ScicosPro

• Input and output ports are explicitly defined

• In the model there is an information flow

• The input/output behavior is expressed in C

• They are considered as black boxes by ScicosPro

5


To create a model with explicit blocks, you need

• to obtain all the equations

• to simplify the equations to obtain a differential equation (ODE/DAE)

C

CR

C

C

CRs

VOutput

RIVC

IV

VVV 0

RC

V

RC

VV CS

C

CV


model equations differential equation

6


Constructing the model using explicit blocks based on the ODE can be

time consuming, error prone, and show no resemblance to the original

diagram.

CV


7


• Have implicit ports (not a priory inputs or outputs)

• Each implicit block represents a physical component

dt

dVCI C

C

an implicit block

in ScicosPro

Acausal modeling (implicit blocks)

8


• Connected with special links (no flow direction)

• Modeling is just connecting the components

• The model looks like the physical system

• The behavior is expressed in the Modelica language

Acausal modeling (implicit blocks)

9


What is Modelica?

• Declarative instead of procedural

• Object oriented modeling language (inheritance, aggregation)

• Typed language

• Allows heterogeneous models (multi-domain models)

• Allows reuse of physical models

• Allows non-causal modeling (modeling using components)

• Equation based, i.e., using mathematical equations

• Hybrid modeling, i.e., event-based and continuous-time models

10


Modelica Applications

• Modelica is a non proprietary language and exists since 1996.

It is Standardized by Modelica Association.

• The Modelica models, being independent of the tool, can be

simulated on any Modelica simulator.

• Available tools include: MapleSim, Dymola, SimulationX,…

• Several free and commercial libraries are available.

11


Physical Component modeling blocks in ScicosPro

Coselica is library of ScicosPro blocks based on the Modelica Standard

Library.

It is essentially a library of physical components covering various physical

domains, such as mechanics, electrics/electronics, and thermodynamics.

These components are provided as ScicosPro blocks and they can be found in

one hierarchical palette named Coselica. All blocks have special connector ports

and connecting them is like assembling a physical network.

12


ScicosPro model, with Coselica blocks and standard blocks

How does it work in ScicosPro?

C

Block

13


Coselica Examples in ScicosPro

Hatch Mechanism – Mechanical/Electrical modeling Heat Transfer and Convection – Thermal modeling

CMOS – Electronic modeling

14

Innovation Intelligence®

Physical Component Modeling in ScicosPro

Application to Vibratory Pile Extraction

Study made by Dr. Dirk Reusch (Kybernetics)


Application Goals

A 23m long pile has been inserted 20m into the ground

and shall be extracted by means of a static force and a

dynamic (harmonic oscillating) force.

In order to achieve this, one has to overcome the shaft

resistance force of the pile (depends on the relative

movement of the shaft with respect to the soil, soil

properties, and the pile geometry ).

In comparison to purely static extraction the use of a vibrator is advantageous, because the

static force needed here is usually significantly smaller. However, this implies the choice of

an appropriate vibrator.

16


Vibratory Piles

Pile Properties

• Pile Length 23m

• Inserted Length 20m

• Circumference 3.1m

Pile Shaft Resistance (R)

• Test data coming from

literature

Soil Properties

• Literature data coming from

Cone Penetration Tests

17


Static Pulling

Motivation

• We are first investigating pure static pulling of the pile in order to test our shaft

resistance model. This will set the magnitude of the static force needed for

extraction.

• The pile inserted the ground would be extracted by means of a static force S

that can overcome the shaft resistance force R of the pile.

Hypothesis

• Gravity is not taken into account

• Target is moving a fully inserted pile (assumption: force to move a partially

inserted pile would be lower)

18


Pile Shaft Resistance

The shaft resistance R of the pile can be split into two parts: a friction force M

and a viscous damping force D.

Dmax and Mmax are the only parameters

19



Shaft Resistance superblock

Parameters

Mmax

and Dmax

z

translational

mechanical

flange

Speed sensor

to measure z’

Force

R=M+D

20



21


Static Pulling

We are applying a slowly increasing (quasi-static) pulling force on the pile and

measuring the pile movement.

We expect that the pile should start moving when this force becomes > Mmax .

Models and simulations are performed for both a fully rigid and an elastic pile.

22


Static Pulling – Rigid Pile

The pile is modeled here as rigid body with mass mp (via a Mass block).

There are two forces acting on the pile: the Pulling Force and the Shaft Resistance

• The Pulling Force signal is generated by a ramp block and prescribed as force acting

on a mechanical flange via a Force0 block.

• The maximum friction force Mmax and the maximum damping force Dmax are provided by

two constant blocks.

• We are measuring the force f acting on the Pile and its position s.

23



Dmax and Mmax

parameters

Shaft Resistance

model

Imposed

pulling force

4.2MN

pile mass

force sensor

24



Simulation results, i.e. pulling force versus pile movement (upwards direction is

negative), are analyzed.

As expected, a significant upwards

movement of the pile happens when

the Pulling Force becomes > Mmax.

25


Static Pulling – Elastic Pile

The pile is modeled here as elastic body by using 3 pile elements (3 lumped masses)

connected via springs.

Element Length Mass

Head L/4 mp/4

Center L/2 mp/2

Foot L/4 mp/4

• A different Shaft Resistance is acting on each element.

• The right flange of the foot (very lower end of the pile) is connected to a Free block,

because we are assuming there is no interaction between pile tip and soil.

• The elasticity of the pile is modeled by the two springs in series, where each of them

has a spring constant stiffness. The relative movement of each pile element (Head,

Center, and Foot) is measured.

26



27



foot center

head

28


Dynamic Pulling

We are replacing the quasi-static, really high, Pulling Force in the previous

model by a rather small Static Force and a Dynamic Force, which is physically

generated inside a Vibrator of mass md.

The Vibrator is rigidly connected (in the real world using hydraulic clamps) to the

pile head. The Dynamic Force in the models is prescribed using a Sine block

with parameters amplitude=V and freqHz=f

The modeling of the rigid and the elastic pile remains the very same as before.

Properties of 5

possible vibrators

29


Dynamic Pulling – Rigid Pile

30


Dynamic Pulling – Rigid Pile

Whatever Vibrator we use from the 5 available, with a rigid setup, there is no

significant movement of the pile.

31


Dynamic Pulling – Elastic Pile

32



Simulation results, i.e. the movements of the pile head, center, and foot

versus time, for each available vibrator, are compared.

2310 VM 24 VM 50 VM

Head Center Foot

33



The pile as a whole is not moving significantly using the vibrators 2310 VM

and 2319 VM.

With the bigger machines 24 VM, 28 VM, and 50 VM the pile as a whole is

moving significantly upwards.

Why ?

A close look reveals, that the 2310 VM and 2319 VM are not powerful enough

to make the pile foot oscillating, such that the averaged (over time) friction

force becomes small, whereas the others make the pile foot clearly

oscillating.

All in all, a good (cost-efficient) choice for the pile extraction job seems to be

the 24 VM.

34


Conclusions

On the results

Vibrator 24 VM seems to be a good choice for the described pile extraction job.

If a vibrator is used for pile extraction, then pile elasticity does matter. This result

is twofold:

• If pile elasticity is present, then it should be taken into account

• It is easier to pull an elastic pile than a rigid pile

On the approach and tools used

With the Coselica library, it was possible to develop geomechanical simulation

models easily in ScicosPro, and then test various configurations by changing

model parameters.

35

Date post:	06-May-2015
Category:	Engineering
Upload:	altair-engineering
View:	284 times
Download:	3 times