Upload
nguyenminh
View
226
Download
0
Embed Size (px)
Citation preview
1
Leandro S. CarvalhoJosé Renato M. S. Oliveira
Marcello G. TeixeiraMaria Esther S. MarquesCarlos A. B. VasconcellosJuliana A. Lukiantchuki
2
The Problem: Numerical modelling of a T-Bar Penetration Test in a very softclay with large deformations.
Approach: Use of Discrete Element Method (DEM) as a way to simulate thebehaviour of a very soft clay layer as a viscous fluid.
Introduction
Virtual T-bar penetrometer tests using Discrete Element Method
3
The T-Bar Penetration Test
Virtual T-bar penetrometer tests using Discrete Element Method
Origins: The T-Bar Penetration Test wasfirst developed by Stewart & Randolph(1991) as a mini tool for centrifuge models.A real size tool was developed for fieldtests also by Stewart & Randolph (1994).
Description: The T-Bar tool itself is a 50mm diameter and 200 mm length barwhich is attached to an adapted piezoconerod and pushed against the soil at a 2 cm/srate to measure the undrained shearstrength of clay deposits.
4
The Geotechnical Model
Virtual T-bar penetrometer tests using Discrete Element Method
Soil: Brazilian marine clay has a very highwater content, greater than its liquid limit.
Geotechnical Model: Mitchell (2005)presented a qualitative analysis of theattraction and repulsive forces betweenclay particles showing that when theparticles interact, there is an attractionforce increasing to a maximum value andthen decreasing until changing to repulsiveforce.
5
The VISED Program
Virtual T-bar penetrometer tests using Discrete Element Method
The Program: VISED was developed in C++ and uses the graphic libraryOpenGL©, capable of generating graphics in real time, and the OpenGL UtilityTools (GLUT), used to create special input and output features. The collisiondetection between particles uses the Munjiza-NBS algorithm which employsdirect mapping to detect contact between elements and thus compute thecollision forces.
The Algorithm Procedure: Combines four main processes: Collision detection between particles; Analysis of interaction forces between particles; Fixations and external forces calculation; New velocities and position calculation.
6
The VISED Program
Virtual T-bar penetrometer tests using Discrete Element Method
The interaction forces between particles can be divided in two components:contact forces and field forces.
Contact Forces: The contact forces between particles have a normal and atangential component.
The normal component (Fn) uses as parameters the normal stiffness (kn)associated with the elements interpenetration (un); and the normaldumping (cn) associated with the normal component of the relativevelocity between elements (vn);
nnnnn vcukF
7
The VISED Program
Virtual T-bar penetrometer tests using Discrete Element Method
The tangential component (Ft) uses as parameters the minimum valueof the force, calculated based on the tangential stiffness (kt), and the forcecalculated based on the dynamic friction (μ); The tangential stiffness isassociated with the tangential damping (ct) and the tangential componentof the relative velocity (vt), while the dynamic friction is associated withthe normal force (Fn).
nttttt FcvdtvkF ;min
The algorithm uses the Central Differences Method (Krysyl & Belytschko 1998)to solve the equations numerically.
8
The VISED Program
Virtual T-bar penetrometer tests using Discrete Element Method
Field Forces: The program allows the implementation of many different waysof interaction forces between the elements, with intensity, direction and signaldetermined by the programmer. The formulation of this interaction is based onthe interaction force due to Morse potential.
The Morse potential and the interaction force between the elementsare based on: d (the depth of the potential) that represents thedissociation energy (necessary energy to break the link between theelements); k that is related with the width of the potential; x that is thedistance between the centers of the elements; and dist that is the distancebetween the elements where the force is zero.
9
The VISED Program
Virtual T-bar penetrometer tests using Discrete Element Method
21)( distxk
p eddxE
distxkdistxkp
p eedkdx
xdExE
12
)()(
The force Fp acts in the direction of the vector that links the centers of theelements and might be positive, when the elements are being attracted, ornegative, when they are being repulsed.
10
The VISED Program
Virtual T-bar penetrometer tests using Discrete Element Method
The intensity of the field force is afunction of the distance between theelements. This force aims to emulatethe interaction between the clayparticles, which is close to theGeotechnical Model presented inMitchell (2005).
Field force with dist = 0.3, k = 1, d = 2 and x = 5.
11
The Soft Soil Characteristics
Virtual T-bar penetrometer tests using Discrete Element Method
Brazilian coastal soft soils are very compressible, organic and with very low Suvalues. The soil used for comparison with the simulations in this work is a verysoft clay from Guanabara Bay, Rio de Janeiro.
12
DEM Simulation of the T-Bar Test
Virtual T-bar penetrometer tests using Discrete Element Method
For a 4.4 m high by 1.0 m width soildomain, 5,920 discrete elements wereused with a 10 mm radius each. The T-Barwas modeled as a sole discrete element.
13
Results and Discussion
Virtual T-bar penetrometer tests using Discrete Element Method
Series 1 results show that the variation ofthe parameter d did not affect significantlythe inclination of the Su profile;Series 2 results show that the variation of kvalues conducted to superposition of theelements at the bottom of the domain;Series 3 show that the increase in h causedan increase of the inclination of the Suprofile.
T-bar virtual test with depth ratios H/D = 50%, 75%, 100% and 125%.
14
Results and Discussion
Virtual T-bar penetrometer tests using Discrete Element Method
Result of the virtual test 3.11 along with its tendency line and the real test tendency line.
15
The use of field forces was fundamental to give the model soil cohesioncharacteristics which allowed virtual tests closer to real results;
The parameters used in the analysis stays as the ma-jor problem oncethey are not classical geotechnical parameters and can be hard to evaluate;
The DEM method dealt with the large defor-mations problem without anydifficulties and show to be a promising tool for geotechnical simulation ofvery soft soils.
Conclusions
Virtual T-bar penetrometer tests using Discrete Element Method