Cat Ho Sull Geo Technique PDF

A Particulate Perspective on Soil Mechanics
2015 Gotechnique Lecture
Dr. Catherine OSullivan

Dept. Civil and Environmental Engineering
Imperial College London
cath.osullivan@imperial.ac.uk
A Particulate Perspective on Soil Mechanics Can
Reveal soil behaviour under stress states that (DEM)

cannot be attained in experiments
Explain perplexing soil behaviour (microCT)
Provide scientific rationale for empirical results (DEM+

used in design microCT)
Tools:
Discrete Element Method (DEM)
Micro Computed Tomography (microCT)

Most Cited Contributions Published in Gotechnique
Rankine Lectures
Cundall and Strack (1979)
A discrete numerical model for granular assemblies
Outlines the Discrete Element Method (DEM)
Models grains as geometrically ideal rigid bodies
Allows contacts to form and break
Grain deformation modelled as contact overlap

DEM Contact
Contact force calculated using orthogonal normal and shear springs
Spring deformation calculated from relative motion of contacting

particles
Sliding governed by Coulomb friction : Tmax=mFn

m = coefficient of friction
Fn = contact normal force
Tmax=maximum tangential (shear) force
Transient Calculation In Each Time Step:
Identify contacting particles
Determine contact forces

Update positions
Calculate resultant forces on

particles accelerations+
velocities
Use of DEM No of Publications per Year
All Scientific Disciplines Main geotechnical journals
DEM simulation of
1200 No. of Papers
semi-solid metal
Exponential Best Fit
No. of Papers per Year
1000 Yuan et al (2012)
800
600
400
200
0
1980 1990 2000 2010
Year DEM simulation of infant formula
Hanley et al (2012)
Web of Science (Oct 2015)
Use of DEM No of Publications per Year
All Scientific Disciplines Main geotechnical journals

45
No. of Papers per Year

Total
Tunnelling induced displacements
2D
Bym et al. (2013) Gotechnique
30 3D
15
0
Deformation in 2000 2005 2010
triaxial cell Yearformula
DEM simulation of infant
Cui et al. (2007) Hanley et al (2012)
Web of Science (Oct 2015) Gotechnique
Direct Shear Test
DEM model: 11700 Spheres

Grade 25 steel precision balls
Diameter 0.992 mm
Cui and OSullivan (2006)

Gotechnique Mean msurface Mean msurface
5.5o 5.7o
Direct Shear Test
0.6 0.6
Lab DEM
0.5 0.5
Stress ratio, / n
Stress ratio, / n
0.4 0.4
0.3 0.3
5
0.2 0.2
50 kPa 50 MPa
100 kPa 100 MPa
0.1 0.1
150 kPa 150 MPa
0 0
0 5 10 15 20 0 5 10 15 20
Global shear strain: % Global shear strain: %
2
2.5 2.5
2 2
Vertical strain: %
Vertical strain: %
1.5 1.5
5
1 1
0.5 0.5
0 0
1
-0.5 -0.5
0 5 10 15 20 0 5 10 15 20
Direct Shear Test
32
Lab
30 DEM
28
26 p= peak angle of shearing

p
resistance
'
24 = peak friction angle
22 e0= initial void ratio
20
18
0.45 0.5 0.55 0.6 0.65
e0
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Normalized Displacements
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Vectors
scaling:
X2 Normalized Displacements
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Contact Normal Forces

(Normalized )
0 1
0 1
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Contact
Contact Normal
Normal Forces
Forces
(Normalized )
0 (Normalized ) 1
0 1
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Contact
Contact Normal
Normal Forces
Forces
(Normalized )
0 (Normalized ) 1
0 1
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Contact
Contact Normal
Normal Forces
Forces
(Normalized )
0 (Normalized ) 1
0 1
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Contact
Contact Normal
Normal Forces
Forces
(Normalized )
0 (Normalized ) 1
0 1
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Contact
Contact Normal
Normal Forces
Forces
(Normalized )
0 (Normalized ) 1
0 1
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Contact
Contact Normal
Normal Forces
Forces
(Normalized )
0 (Normalized ) 1
0 1
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Contact
Contact Normal
Normal Forces
Forces
(Normalized )
0 (Normalized ) 1
0 1
0.45 1.75
1.5
Vertical strain: %
n
1.25
Stress ratio /
0.3
1
0.75
0.5
0.15
0.25
0 -0.25
0 5 10 15 20 25 0 5 10 15 20 25
Contact
Contact Normal
Normal Forces
Forces
(Normalized )
0 (Normalized ) 1
0 1
Diagram from
Craigs Soil Mechanics
Knappett and Craig(2012)
Direct Shear Test
Reasonable correlation with physical test
DEM gives insight into kinematics and failure mechanisms
Limited number of particles (11,700)

No. of Particles in Geomechanics DEM Simulations
6 2D-per paper
10
3D-per paper
Median 2D
D50=200 micron
Median 3D >150,000 particles
4
No. of Particles
10
1 cm
2 1 cm
10 1 cm
0
10
2000 2005 2010
Year
Use of High Performance Computing to Increase Sample Size
LAMMPS (http://lammps.sandia.gov/)
Classical molecular dynamics code that can be used for DEM
Uses spatial decomposition and MPI (message passage interface)
Can run large high performance computers (hpc) with distributed

memory
Researchers at IC (Marketos, Hanley, Shire, Huang, Otsubo) have

been working to modify the granular LAMMPS package
Validation of Granular LAMMPS
Analytical Solution
6 Serial (No Damping)
Serial (Local Damping)
5.5 Serial (Viscous Damping)
Principal stress ratio, 1/3

HECToR (National High
Performance Computer)
5
4.5
3.5
Compression of face- 2.5

centred-cubic array of
2
spheres. 0 0.1 0.2 0.3 0.4 0.5
Friction coefficient
Analytical solution
documented in Validations by
Thornton (1979) K. Hanley and X. Huang
Gotechnique
A State Parameter for Sands:
Observations from direct shear tests:

Critical state locus
Peak friction angle ('p) depends on e0 (CSL) for Toyoura sand
Verdugo (1992)
Dilatancy depends on stress level 0.95 Gotechnique
0.9
0.85
Critical state is a state where the
soil deforms at constant stress e 0.8
and constant void ratio
0.75
0.7
Been and Jefferies (1985) 100 1000 5000
Gotechnique log p' (kPa)
A State Parameter for Sands:
Loose Dense < 0

(+ve)
(-ve) q
CSL
e
Loose > 0
Dense
a
logp
e void ratio Loose > 0

q deviatoric stress q=1-'3
p' mean effective stress p'= /3(1+2'3)
1
e
(triaxial case) Dense < 0
Been and Jefferies (1985)

Gotechnique a
Variation in 'p with 0
50
Jefferies and Been
45
40 Experimental data
from Jefferies and
'p (Deg)
35 Been (2006)
30
25 p=peak angle of shearing

resistance
= peak friction angle
20
0= initial state parameter
15
-0.3 -0.2 -0.1 0 0.05
0
50
Jefferies and Been
45
40
'p (Deg)
35
30
25
20
15
-0.3 -0.2 -0.1 0 0.05 Triaxial cell at IC
0
In Situ Stresses Stress increments
R beneath a circular
footing on isotropic,
r/R linear elastic
1.5 1.0 0.5 0.5 1.0 1.5 foundation
Contours of
0.2 0.2
' 2 '3
0.15 0.15
0.5 '1 '3
0.1
0.1 1=major principal stress
0.05 0.05 2=intermediate principal
stress
0.05 0.05 3=minor principal stress
1.0
Z/R 1> 2 > 3

Along CL:
Hight (1983)
2=3
Triaxial conditions
36
Variation in Soil Strength

34
with Principal Stress Ratio
' (Deg)
Lade (1977,1984)
32 Matsuoka and Nakai (1978)
p
Satake (1975)
University of Bristol Ogawa et al. (1974)
30
Cubical Cell 36
0 0.5 1
1
Can control bb
34
' (Deg)
b= intermediate
stress ratio 2
32
'2 '3
p
b 3
'1 '3
30
b=0:Triaxial 0 1>0.5
2 > 3 1
compression b
b=1:Triaxial p= peak angle of shearing
extension resistance
b0.5: Plane strain = peak friction angle
Large Hollow Cylinder OSullivan et al. (2015)
Apparatus at IC Gotechnique
DEM Analysis of for a 3D Stress State
20,164 spherical particles

Periodic boundaries
Grading similar to Toyoura sand
Friction coefficient m=0.25
Controlled b
3 Deformed to large strain

Varied initial state parameter (0)
Most of simulations used LAMMPS
Huang et al. (2014)

Gotechnique
Variation in Peak and Critical State Friction Angles with b
23
21
22
peak,(Deg.)
O
S
20
cs,(Deg.)
M
21 Ogawa

L
Satake

e
19 e
20 Matsuoka-Nakai
e
21 Peak
Lade
Critical state
e18
e
19 =0.646, =500 kPa
0 3 e
Ogawa
18 17 Satake
200 0.2 0.4 0.6 0.8 1 0 0.2 Matsuoka-Nakai
0.4 0.6 0.8 1
b b
b=0:Triaxial b=1:Triaxial Lade
Compression eo=0.533, 3 =100 kPa
cs,
Extension
19Ogawa e0=0.612, 3 =5000 kPa
Satake '2 '3
b e0=0.625, 3=2000 kPa
Matsuoka-Nakai '1 '3 e0=0.625, 3=1000 kPa
18Lade
e0=0.646, 3=500 kPa Huang et al. (2014) e0=0.646, 3=500 kPa
Granular Matter
Triaxial extension
Triaxial compression
State Dependent Response
0.8 0.65
0.64
0.6 0.63
0.62
q/p
0.4 0=-0.0414
e
0.61
0=-0.0245
0=0.0171 0.6
0.2
0.59
0 0.58
0 20 40 60 0 20 40 60
a,% a, %
> 0 loose q=1-'3 '3=500kPa

< 0 dense p'=1/3(1+2'3) Huang et al. (2014)
(triaxial) Gotechnique
50
Jefferies and Been
45
40
'p (Deg)
35 Experimental data
from Jefferies and
30 Been (2006)
25
20
15
-0.3 -0.2 -0.1 0 0.05
0
50 Jefferies and Been

DEM b=0
45
DEM b=0.2
40 DEM b=0.3
DEM b=0.4
'p (Deg)
35 DEM b=0.5
DEM b=0.6
30 DEM b=0.8
25 DEM b=1.0
DEM Plane Strain
20
15
-0.3 -0.2 -0.1 0 0.05
0 Huang et al. (2014)
Gotechnique
Variation in 'p - 'cs with 0
16
Jefferies and Been
12
'p-'cs(Deg)
8
Experimental data
from Jefferies and
Been (2006)
4
-4
-0.3 -0.2 -0.1 0 0.05
0
Variation in 'p - 'cs with 0
16 Jefferies and Been

DEM b=0
12 DEM b=0.2
DEM b=0.3
'p-'cs(Deg)
DEM b=0.4
8
DEM b=0.5
DEM b=0.6
4 DEM b=0.8
DEM b=1.0
0 DEM Plane Strain
-4
-0.3 -0.2 -0.1 0 0.05
0 Huang et al. (2014)
Gotechnique
Variation in Dp with 0
1.4
Jefferies and Been
1.2
1
p
Dilatancy at Peak, D
0.8
0.6 Experimental data

from Jefferies and
0.4 Been (2006)
0.2
-0.2
-0.4
-0.3 -0.2 -0.1 0 0.05
0
Variation in Dp with 0
1.4 Jefferies and Been
DEM b=0
1.2
DEM b=0.2
1 DEM b=0.3
Dilatancy at Peak, Dp
DEM b=0.4
0.8 DEM b=0.5
0.6 DEM b=0.6
DEM b=0.8
0.4 DEM b=1.0
DEM Plane Strain
0.2
0
-0.2
-0.4
-0.3 -0.2 -0.1 0 0.05 Huang et al. (2014)
0 Gotechnique
Huang et al. (2014)
Gotechnique
Performance of Norsand Model
Dense Sample: Loose Sample:

3=300 kPa, 0 =-0.0823 3=500 kPa, 0 =0.0171
Reminder that Robust Constitutive Models are Needed
The remains of the village of Bento Rodrigues in Brazil following

tailings dam collapse, Nov. 2015
http://www.theguardian.com
DEM simulations can capture influence of b on strength and the

state-dependency of soil strength and dilation
DEM data indicate that the correlations between 'p-'cs and 0

and Dp and 0 are independent of particle morphology and b
DEM simulations can be used to supplement experimental

investigations when developing constitutive models for soil
behaviour in a general stress space
DEM Limitations
Spherical particles under-estimate soil strength ('cs ,'p)
Toyoura sand:
'cs=31o
Subangular grains
DEM model:
'cs=18.2o
Spherical particles
Influence of Morphology on p and cs
Rotation of non-spherical particles will be frustrated by finite

contact area
Influence of Morphology on p and cs
Force chains between non-spherical particles are more stable
'1 '1
'1 '1
DEM and Particle Geometry
Options for non-spherical particles include sphere clumps -

computational cost increase
389 Agglomerates
Cheng et al (2003)
Gotechnique
High resolution, three-dimensional images created

using X-rays
Non-destructive
Computed tomography developed by Cormack and

Hounsfield Awarded Nobel Prize in Physiology or
Medicine in 1979
First paper in Gotechnique Oda et al. (2004)

Central core
76 mm
extracted for
scanning
3-10 mm in
diameter
38 mm Fonseca et al. (2014)

Gotechnique
MicroCT scanning
X-ray
X-Ray source + typical source Sample Detector
sample for scanning
(H. Taylor, current PhD student) Image source: http://www.nikonmetrology.com

MicroCT Data Analysis
2D Slice from Binary image

mCT image
Raw output 3D attenuation map

Taylor et al. (2015)
Computers and Geotechnics
(H. Taylor)
MicroCT Data Analysis
Individual
particles and
contacts from
watershed
segmentation
(H. Taylor)
Reigate Sand: A Locked Sand
Concavo-
convex
contact
Straight
to long
contact
500 mm
200 mm
Transmitted light microscope image of thin

section (cross-polarized light) Scanning electron microscope (SEM) image
(contact configurations proposed by
Doctoral Research of Dr.
Dusseault and Morgensten (1979)) J. Fonseca
Reigate Sand Load:Deformation response
2.5 5
Average volumetric strain: %

2 4 Cresswell and Powrie
Gotechnique (2004)
1.5 3
Reconstituted sample
q/p'
1 2 pluviated to highest
possible density
Cell pressure 200 kPa
0.5 1
Dev. stress intact Observations from direct shear tests:
0 Dev. stress recon. 0
Vol. strain intact Peak friction angle ('p) depends
Vol. strain recon.
-0.5 on e0 -1
0 5 10 15
Dilatancy depends on stress level
Axial shortening: %
Axial shortening (%)
Reigate Sand Site Location
From Richards & Barton

(1999)
Block Sampling
Triaxial Tests Sample Preparation
Intact Sample
Reconstituted
Sample
(Prepared by
pouring / tamping/
tapping)
Doctoral Research of Dr.

J. Fonseca
Triaxial Test Data Intact
e0=0.48
Reconstituted
e0=0.49
Cell pressure
300 kPa
Fonseca et al. (2013)

Gotechnique
mCT Imaging Points
Scan 4
Scan1:2
Scan Scan 3 Approaching
Onset
Prior to of dilationStart
Isotropic of visible critical state
Compression
shear band
5 mm voxel size
D50 = 274 mm
Average no of
voxels per particle
in scans > 27000
MicroCT Images
mCT
resol: 0.8mm
Particle Size Distributions from microCT
100
Reconstituted
Intact
80 Sieve
60
% finer
40
20 Size in microCT taken

as intermediate
500 mm
principal axis length
0
0.1 0.2 0.4 0.6
Particle size (mm) Fonseca et al. (2013)
Soils and Foundations
Scalar Fabric Description Contact Density
Coordination Number (Nc):

Number of contacts per particle Representative Particle
Contact Index:
Contact area
Particle surface area
Contacts
Coordination Number Distribution
100
Int-1-a
80 Int-1-b
Int-2-a
% Smaller
60 Int-2-b
Int-3
Int-4
40 Rec-1-a
Rec-1-b
20 Fonseca et al. (2013)
Rec-3-a
Gotechnique
Rec-3-b
0
0 5 10 15 20
Coordination Number (N )
Contact index distribution
100
Int-1-a
80 Int-1-b
Int-2-a
% Smaller
Int-2-b
60
Int-3
Int-4
40 Rec-1-a
Rec-1-b
20 Fonseca et al. (2013)
Rec-3-a
Gotechnique
Rec-3-b
0
0 0.1 0.2 0.3 0.4
Contact Index (CI)
Influence of Contact Area on p
Force chains more stable when contact area is greater
'1 '1
'1 '1
Particle Scale Analyses of Locked Sand
Triaxial test data confirmed difference in response of intact and

reconstituted materials tested at same void ratio
Micro CT data revealed a difference between particle size

distributions
At the same void ratio the intact material had a higher

coordination number and contact index than reconstituted
material
Contact index better discriminates two materials

Tools of Particulate Soil Mechanics
DEM
Model with inherent idealizations - simplified geometries and

contact models
Information on contact forces and kinematics
microCT
Can observe topology of real materials
No information on contact forces or particle velocities

Internal Stability of Dam Filters WAC Bennett Dam
British Columbia, Canada
183 m high
Zoned earthfill dam
Constructed 1968
Cross-section through W.A.C. Bennett Dam
183 m high Blended till-like core

High capacity drain
Well graded transition and filter
BC Hydro as cited by
Muir Wood (2007)
Internal Instability
Internal instability occurs where

sand can migrate within the
interstices of a framework or primary
fabric formed predominantly of the
gravel particles
Sandy gravels having a bimodal

(gap-graded) grain-size distribution
with less than about 25% of sand
exhibit internal instability
Such materials can present serious

problems where piping is possible,
notably in dams and barrages.
Quotes from Skempton and Brogan

(1994) Gotechnique
G1 Middle
sand can migrate within the
interstices of a framework or primary
fabric formed predominantly of the
gravel particles
Sandy gravels having a bimodal

(gap-graded) grain-size distribution 100
G1 Bottom
80
exhibit internal instability
% Smaller by volume
60
problems where piping is possible, 40
notably in dams and barrages.

20
Quotes from Skempton and Brogan 0 2 3 4

10 10 10
(1994) Gotechnique Particle Size (mm)
Fonseca et al. (2014) Gotechnique

sand can migrate within the Original top of
interstices of a framework or primary 1996
dam Sinkhole at
fabric formed predominantly of the WAC Bennett
Dam
gravel particles (BC Hydro as
Engineer
cited by Muir
Sandy gravels having a bimodal Wood, 2007)
(gap-graded) grain-size distribution (a) (b)
Original
exhibit internal instability crest

problems where piping is possible,
notably in dams and barrages. Engineer
Quotes from Skempton and Brogan Gouhou Dam, China (1993)

(1994) Gotechnique 71m High; 300 Deaths
(Chang, 2011)
Empirical Filter Criteria: Kzdi (1979)
Relates to Terzaghi
filter rule
100
PSD
arbitrary split of PSD

fine fraction
80 dfine
% passing by mass
coarse fraction 85 Split PSD into

coarse and fine
60
PSDs
40 Stable if: d85fine >

(D15coarse/ 4)
20 i.e. if
Dcoarse D15coarse/d85fine < 4
15
0
0.1 0.5 1
particle diameter
% Smaller by volume
80 mCT top
Particle-Scale Basis for Kzdi Criterion
60 to
Samples K0 consolidated mCT mid
p'=300kPa Increasing Kzdi no. mCT base
Voxel size 10 mm 40
(0.005d50-0.01d50) Decreasing stability Sieve
100 20 100 100
% Smaller by volume
% Smaller by volume
% Smaller by volume
80 80 mCT - int. axis,top 80 mCT top

0 2 3 4
10 mCT - int.
10 axis,mid 10 mCT mid
60 60 60
mCT - int.Size
Particle axis,base
(mm)
G1 Top FLS mCT base
40
WG To xy (z=450/900) xz (y=450/900) 40 MINxy xz
(z=450/900) (y=450/900)
G2 Top
40
xy (z=450/900)
xz (y=450/900)
Sieve
20 20 FLS
MAX 20
WG Sieve G1 G2
0 2 3 4
0 2 3 4 0 2 3 4
10 10 10 10 10 10 10 10 10
Particle Size (mm) Particle Size (mm) Particle Size (mm)
G1 Middle G2 Middle
WG Middle
G1 Bottom G2 Bottom
WG Bottom
Fonseca et al. (2014) Gotechnique
Variation in Coordination Number with Kzdi Ratio
Increasing Kzdi no.
Nc = Coordination number Decreasing stability
No of contacts per particle
Fonseca et al. (2014)

Leighton Buzzard
Glass beads Gotechnique
Sand
Blue particle Shire and OSullivan (2013)
Blue particle
50 contacts Acta Geotechnica
20 contacts
Images from H. Taylor

Skempton and Brogan Permeameter Experiments
Skempton and Brogan (1994)
Gotechnique
Piezometers
Sample A
0.6 i=0.2:
strong general
piping of fines
Rigid wall throughout
transparent 0.2 permeability twice
cylinder v cm/s initial value
0.2
Inflow
0
0.1 0.2 0.3
i
Stress Partition - a
'v
Micromechanical hypothesis to explain icrit<-1:
Coarse matrix transfers overburden
Fines carry reduced effective stress:
= icrit / icrit(heave)
'fines = x '
Skempton and Brogan (1994)

Gotechnique
DEM Simulations to Investigate Instability
DEM code granular LAMMPS with periodic boundaries
Isotropic compression at to p' = 50kPa

Sample density controlled using interparticle friction ():
= 0.0 (Dense)
= 0.1 (Medium dense)
= 0.3 (Loose)
Shire et al. (2014) ASCE JGGE
a DEM Calculations
p'=overall mean effective stress
p'fine=mean effective stress in finer

fraction
p' and p'fine can be directly obtained from

a summation of contact forces in DEM
Skempton and Brogan Sample A: Comparison of a Values
100
DEM
80 Experimental Density aDEM
Loose 0.15
60 Medium 0.06
% finer
Dense 0.04
40
20 aexperiment=0.18
Experimental sample
0 placed moist with no
0.1 0.5 1 5 10
Diameter (mm) densification

Parametric Study on Effect of Gap Width and Fines Content
100 Skempton and Brogan A

FR7
Gap width varied to vary
18% Fines Kzdi ratio (D15/d85)
80 25% Fines from 3.4 to 11.5
35% Fines
45% Fines D15/d85 =3.4 stable
60
% Finer
D15/d85 =11.5 unstable

40
Density varied for all
samples
20
Up to 304,205 particles
0
0 0.2 0.4 0.6 0.8 1
Diameter (mm)

Variation in a with Fines Content (Ffine)
Skempton and
Brogan(1994):
Critical fines content

where fines just fill
voids:Ffine=24-29%
Finer fraction separates

coarse fraction particles:
Ffine=35%

Variation in a with Fines Content (Ffine)
Skempton and
Brogan(1994):
Critical fines content

where fines just fill
voids:Ffine=24-29%
Finer fraction separates

coarse fraction particles:
Ffine=35%

Visualization of Particle Stresses
30% Fines - Loose 30% Fines - Dense

Relative density increase
Particle-Scale Analysis of Internal Instability
Particle scale analyses (microCT and DEM) link between the

Kzdi criterion and fabric
DEM analyses confirm Skempton and Brogans a hypothesis
DEM analyses confirm Skempton and Brogans delineation of

underfilled and overfilled fabrics
DEM analyses highlight the sensitivity of transitional materials to

relative density
A Particulate Perspective on Soil Mechanics Can
Reveal soil behaviour under stress states that cannot be

attained in experiments
Explain perplexing soil behaviour
Provide scientific rationale for empirical results used in

design
Tools:
Discrete Element Method (DEM)
Micro Computed Tomography

Acknowledgements
Co-authors:
Sponsors:
Dr. Daniel Barreto, Edinburgh Napier
Prof. John Burland, Imperial College
EPSRC
Dr. Liang Cui, University of Surrey IRCSET (Ireland)
Dr. Ignazio Cavarretta, University of Surrey Royal Commission for the
Prof. Matthew Coop, City University Hong Kong Exhibition of 1851
Dr. Shane Donoghue, Queens, Belfast Institution of Civil Engineers
Dr. Joana Fonseca, City University R+D Fund
Dr. K. Hanley, University of Edinburgh
Dr. X. Huang, Tongji University
Dr. Erdin Ibraim, University of Bristol
Dr. Fiona Kwok, University of Hong Kong Additional microCT images
Prof. Peter Lee, University of Manchester provided by Howard Taylor,
Dr. Mike Long, University College Dublin Imperial College
Dr. George Marketos, Utrecht University
Mr. Masahide Otsubo, Imperial College Prof. J. Fannin, University of British
Dr. Tom Shire, Imperial College Columbia collaborated on internal
Dr. Way Way Sim, Imperial College erosion research
Prof. Ahmer Wadee, Imperial College
Gotechnique Contributions
Cui, L. and OSullivan, C. (2006) Exploring the macro- and micro-scale
response characteristics of an idealized granular material in the direct
shear apparatus Gotechnique Vol. 56, 455-468
Cui, L. O'Sullivan, C. and O'Neill, S. (2007) An analysis of the triaxial
apparatus using a mixed boundary three-dimensional discrete element
model Gotechnique Vol. 57, No. 10, p 831-844.
Donohue, S., OSullivan, C. and Long, M. (2009) Particle breakage during
cyclic triaxial loading of a carbonate sand Gotechnique Vol: 59,
Pages: 477 - 482,
Cavarretta, I. , Coop, M. , OSullivan, C. (2010) The Influence of Particle
Characteristics on the Behaviour of Coarse Grained Soils
Gotechnique 60, pp 413-423
Cavarretta, I. and OSullivan, C. (2012) The mechanics of rigid irregular
particles subject to uniaxial compression Gotechnique, 62, pp 681-
692
Fonseca, J., OSullivan, C., Coop, M.R., Lee, P.D. (2012) Quantifying the
Evolution of Soil Fabric during Shearing using Directional Parameters
Gotechnique, Volume 63, Issue 6, October 2012 pages 487 499
10.1680/geot.12.P.003
Shire, T., OSullivan, C., Barreto, D., Gaudray, G. (2012) Quantifying
Stress Induced Anisotropy using Inter-Void Constrictions
GEOTECHNIQUE, 2013, Vol:63, Pages:85-91, ISSN:0016-8505(doi)
Bym, T., Marketos, G., Burland, J.B. and OSullivan, C. (2013) Use of a
two-dimensional discrete element model to gain insight into tunnelling-
induced deformations Gotechnique Vol: 63, Pages: 791-795
Fonseca, J., OSullivan, C., Coop, M.R., Lee, P.D. (2013) Quantifying the
Evolution of Soil Fabric during Shearing using Scalar Parameters
Gotechnique Vol: 63, Pages: 818-829,
OSullivan, C., Wadee, M. A., Hanley, K. J. and Barreto, D. (2013) Use of
DEM and elastic stability analysis to explain the influence of the
intermediate principal stress on soil strength Gotechnique Vol: 63,
Pages: 1298-1309
Fonseca, J., Sim, W. W., Shire, T., and OSullivan, C. (2014) Micro-
structural analysis of sands with varying degrees of internal stability
Geotechnique Vol: 64, Pages: 405-411
Huang, X., OSullivan, C., Hanley, K. J., Kwok, C.Y. (2014) DEM Analysis
of the State Parameter, Gotechnique 64(12) 954-965
Otsubo, M., OSullivan, C., Sim, W. W., Ibraim, E. (2015) Quantitative
assessment of the influence of surface roughness on soil stiffness Vol:
65, Pages: 694-700
Additional References
Been, K. & Jefferies, M. G. (1985). A state parameter for sands.
Gotechnique 35, No. 2, 99112
Cheng YP,Nakata Y,Bolton MD (2003) Discrete element simulation of
crushable soil. Gotechnique, 53(7), 633 641
Cresswell, A. and W. Powrie (2004). Triaxial tests on an unbonded locked
sand. Gotechnique 54(2), 107-115.
Cundall, P.A. & Strack, O.D.L. (1979) A discrete numerical model for
granular assemblies. Gotechnique. 29 (1), 4765
Dusseault, M. B. and N. R. Morgensten (1979).Locked sands.Q. J. Engng
Geol. 12, 117-131.
Fonseca J, O'Sullivan C, Coop MR, Lee PD, 2012, Non-invasive
characterization of particle morphology of natural sands, Soils and
Foundations, Vol: 52, Pages: 712-722
Hanley KJ, O'Sullivan C, Byrne EP, Cronin K, 2012, Discrete element
modelling of the quasi-static uniaxial compression of individual infant
formula agglomerates, Particuology, Vol: 10, Pages: 523-531
Hight D.W. (1983), Laboratory investigation of sea-bed clays, PhD thesis,
Imperial College London.
Jefferies, M.G. & Been, K. (2006) Soil liquefaction: a critical state
approach. Taylor and Francis.
Muir Wood, D. (2007). The magic of sands - The 20th Bjerrum Lecture
presented in Oslo, 25 November 2005. Canadian Geotechnical
Journal, 44(11)
Shire T, O'Sullivan C, 2013, Micromechanical assessment of an internal
stability criterion, Acta Geotechnica, Vol: 8, Pages: 81-90
Shire, T.; OSullivan, C. ; Fannin, R.J.; Hanley, K. (2014) Fabric and
effective stress distribution in internally unstable soils ASCE Journal of
Geotechnical and Geoenvironmental Engineering, 140(12)
Skempton AW, Brogan J (1994) Experiments on piping in sandy
gravels. Gotechnique 44(3):449460
Taylor HF, O'Sullivan C, Sim WW, 2015, A new method to identify void
constrictions in micro-CT images of sand, Computers and
Geotechnics, Vol: 69, Pages: 279-290
Thornton, C. (1979) The conditions for failure of a face-centered cubic
array of uniform rigid spheres. Gotechnique. 29 (4), 441459.
Verdugo, R. (1992) The critical state of sands-discussion Gotechnique.
42 (4), 655663.
Yuan L, O'Sullivan C, Gourlay CM, 2012, Exploring dendrite coherency
with the discrete element method, Acta Materiala, Vol: 60, Pages:
1334-1345

Cat Ho Sull Geo Technique PDF

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Cat Ho Sull Geo Technique PDF

Hochgeladen von

Copyright:

Verfügbare Formate

A Particulate Perspective on Soil Mechanics

2015 Gotechnique Lecture

Dr. Catherine OSullivan

Reveal soil behaviour under stress states that (DEM)

Explain perplexing soil behaviour (microCT)

Provide scientific rationale for empirical results (DEM+

Discrete Element Method (DEM)

Micro Computed Tomography (microCT)

Outlines the Discrete Element Method (DEM)

Models grains as geometrically ideal rigid bodies

Allows contacts to form and break

Grain deformation modelled as contact overlap

Spring deformation calculated from relative motion of contacting

Sliding governed by Coulomb friction : Tmax=mFn

Identify contacting particles

Determine contact forces

Calculate resultant forces on

All Scientific Disciplines Main geotechnical journals

1000 Yuan et al (2012)

All Scientific Disciplines Main geotechnical journals

No. of Papers per Year

DEM model: 11700 Spheres

Cui and OSullivan (2006)

26 p= peak angle of shearing

24 = peak friction angle

22 e0= initial void ratio

Contact Normal Forces

Reasonable correlation with physical test

DEM gives insight into kinematics and failure mechanisms

Limited number of particles (11,700)

Classical molecular dynamics code that can be used for DEM

Uses spatial decomposition and MPI (message passage interface)

Can run large high performance computers (hpc) with distributed

Researchers at IC (Marketos, Hanley, Shire, Huang, Otsubo) have

Principal stress ratio, 1/3

Compression of face- 2.5

Observations from direct shear tests:

Loose Dense < 0

e void ratio Loose > 0

Been and Jefferies (1985)

25 p=peak angle of shearing

Z/R 1> 2 > 3

Variation in Soil Strength

20,164 spherical particles

3 Deformed to large strain

Huang et al. (2014)

> 0 loose q=1-'3 '3=500kPa

50 Jefferies and Been

16 Jefferies and Been

0.6 Experimental data

Dense Sample: Loose Sample:

The remains of the village of Bento Rodrigues in Brazil following

DEM simulations can capture influence of b on strength and the

DEM data indicate that the correlations between 'p-'cs and 0

DEM simulations can be used to supplement experimental

Spherical particles under-estimate soil strength ('cs ,'p)

Rotation of non-spherical particles will be frustrated by finite

Force chains between non-spherical particles are more stable

Options for non-spherical particles include sphere clumps -

High resolution, three-dimensional images created

Computed tomography developed by Cormack and

First paper in Gotechnique Oda et al. (2004)

38 mm Fonseca et al. (2014)

(H. Taylor, current PhD student) Image source: http://www.nikonmetrology.com