Post on 05-Oct-2020
transcript
!"#$%&'(&)*+#,(-."*'/&!#0.,1"*2*3*"%&'(&45-/#3-5&!"#$%&'(&)*+#,(-."*'/&!#0.,1"*2*3*"%&'(&45-/#3-5&!'*30�*/6&7*0.5,",&83,9,/"&:'$,3*/6!'*30�*/6&7*0.5,",&83,9,/"&:'$,3*/66 66 6
!"#$#%&#'(!"#$#%&#'(&)&)
;#5$#,&4,'",.</*.-3&!'.*,"%&='5>0<'1&?@A@;#5$#,&4,'",.</*.-3&!'.*,"%&='5>0<'1&?@A@
!"#$#%&#'(*+!"#$#%&#'(*+
75B&C,,/-&!#>#9-5-/D&75B&C,,/-&!#>#9-5-/D&* * 3 2* * 3 2E'F-/&G/*H,50*"%D&43-002'5'D&IJD&G!KE'F-/&G/*H,50*"%D&43-002'5'D&IJD&G!K
75B&!<5,,>-/"<&:-/$-%-9D&7-/*,3&75B&!<5,,>-/"<&:-/$-%-9D&7-/*,3&C-55'"C-55'"D&;-"5*.>&4*'5$-/'D&D&;-"5*.>&4*'5$-/'D&4,'56,&4,'56,&),.->,0),.->,0D&E'F-/&G/*H,50*"%D&43-002'5'D&IJD&G!KD&E'F-/&G/*H,50*"%D&43-002'5'D&IJD&G!K
75B&I*H,$*"-&7-075B&I*H,$*"-&7-0=,*$3*/6,5=,*$3*/6,5 K00'.*-",0D&L/.BD&M-925*$6,D&:KD&G!KK00'.*-",0D&L/.BD&M-925*$6,D&:KD&G!K
75B&K3--&NB&K0<9-F%75B&K3--&NB&K0<9-F%%%K9,5*.-/&G/*H,50*"%&*/&7#2-*D&GK8K9,5*.-/&G/*H,50*"%&*/&7#2-*D&GK8
:-%&AD&?@A@:-%&AD&?@A@
!"#$%&'
Background
Research Objectives
Shape Characterization and ReconstructionShape Characterization and Reconstruction
Discrete Element Modeling
Conclusions
()*+,-."&/,
Particle morphology governs micromechanical behavior
of granular mediaof granular media
Traditionally, 2-D DEM analyses adopted in studying
influence of particle shapes on mechanical response of
cohesionless soil and most analyses are limited to
circular or idealized shapesp
Limited knowledge available on 3-D discrete element
modeling of highly irregular particle shapesmodeling of highly irregular particle shapes
()*+,-."&/,Particle shape modeling techniques in 2-D using DEM
Circular/spherical discrete elements: Cundall and Strack (1979)
Polygonal discrete elements: Barbosa and Ghaboussi (1992)
Elli ti l di t l t Ti t l (1993)Elliptical discrete elements: Ting et al. (1993)
Overlapping discrete element cluster (ODEC): Ashmawy et al. (2003)( )
Particle shape modeling techniques in 3-D
3-D ellipsoid-based DEM - ELLIPSE3D: Lin and Ng (1997)
Polyhedron-based approach: Ghaboussi and Barbosa (1990)
3-D image-based DEM: Matsushima (2004)
!"#$%&'
Background
Research Objectives
Shape CharacterizationShape Characterization
Discrete Element Modeling
Conclusions
0'1')-*23!45'*#%6'15
To design and develop automated 3-D tomography
reconstruction algorithms applied to shape characterizationreconstruction algorithms applied to shape characterization
of sand particles
T i ll lid t th t ti th d bTo numerically validate the reconstruction method by
comparing with 3-D reconstructions obtained from multiple
projections of a single particle generated using optical and p j g p g g p
X-ray methods
2-D and 3-D discrete element modeling of particle shape2-D and 3-D discrete element modeling of particle shape
Evaluation of influence of particle shape on shear strength
f ilof soil
!"#$%&'
Background
Research Objectives
Shape Characterization and reconstructionShape Characterization and reconstruction
Discrete Element Modeling
Results
ConclusionsConclusions
7)#)38'#
7)9#.&)3(')*238)&/ :%*2%,)&37"&'38)&/
!"#$%&'
Background
Research Objectives
Shape Characterization and ReconstructionShape Characterization and Reconstruction
Discrete Element Modeling
Results
ConclusionsConclusions
0'1')-*23;-'<%1'
=>73?-.5'*#%.&13
.@3A>73?)-#%*$'13
=>73?-.5'*#%.&13.@3
B.<?.1%#'3;)-#%*$'B.<?.1%#'3A>73
C0'*.&1#-"*#%.&D
%&3<%E
!!"
##$
D
D
2
1
!!"
##$
D
D
2
1
!!"
##$
D
D
2
1 % &% & !
!"
##$
2
222
2
111
,
,
D
D
'(
'(
!!"
##$
D
D
2
1
!!"
##$
D
D
2
1
!!"
##$
D
D
2
1
!!!
)###
* ND
!
2
!!!
)###
* ND
!
2
!!!
)###
* ND
!
2 % &
% &!!!!
)####
*2
222
,
,
NNND '(
(
!
!!!
)###
* ND
!
2
!!!
)###
* ND
!
2
!!!
)###
* ND
!
2
)* )* )* N % &)* NNN ( )* N )* N )* N
A>73
7'1*-%?#.-1
=>737'1*-%?#.-13@-.<3
:%E
=>737'1*-%?#.-13@-.<3
;)-#%*$'
A7382)?'37'1*-%?#.-1? ?=73?-.5'*#%.&3@-.<3)3
?)-#%*$'(."&/)-93.@3#2'3
?)-#%*$'
F $$ / 4 / ;'-%./%* -'?-'1'&#)#%.& .@
200
250
200
250
F&>-.$$'/34."&/)-93)13)3
@"&*#%.&
;'-%./%*3-'?-'1'&#)#%.&3.@3
4."&/)-9
100
150
100
150
0 500 1000 1500 200050
0 200 400 600 80050
8'4'1#9&3)&/3('&1.&3> C"&-.$$%&,D3)3*$.1'/3."#$%&'
A7382)?'37'1*-%?#.-1? ?
The Fourier Coefficients can then be reduced using principal components analysis (PCA) to a desired set amount
200
250
100
150 GGH
1c$ "# ! D$ "
0 500 1000 1500 200050
2
3..
cc# !# !# !# !# !
1
2
DDD
$ "# !# !* )
;BI
.
nc# !# !* ) 3D# !* )
0J37"/)K3;JL)-#K37J38#.-+K3Pattern Classification
M$$%?1.%/3:./'$3@.-3N$$"1#-)#%&,3#2'3;'-@.-<)&*'3.@3#2'3I$,.-%#2<,
0 /% % O B'&#'- .@ M$$%?1.%/0)/%"13%&3O3P
Q)-%)&*'3.@3
G%-1#3
7'1*-%?#.-
B'&#'-3.@3M$$%?1.%/3
P
A>73B..-/%&)#'3.@3
7'1*-%?#.- :')&17'1*-%?#.-
0)/%"13%&3R3>
7'1*-%?#.-3:')&1
Q)-%)&*'3.@3
8'*.&/3
7'1*-%?#.-
0)/%"13%&3S3
P Q)-%)&*'3 E
T
.@3
H2%-/3
7'1*-%?#.-
9
I$,.-%#2<3)&/3;-'<%1'3Q)$%/)#%.&Q)$%/)#%.&
Q)$%/)#'1
;-'<%1'
!?#%*)$3:%*-.1*.?' !?#%*)$3H.<.,-)?293
8 #891#'<:"$#%?$'36%'U1K3:"$#%?$'3
;)-#%*$'1
:"$#%?$'36%'U1K38%&,$'3;)-#%*$'1
Q)$%/)#'1
I$,.-%#2<
O>0)93BH3891#'<
:"$#%?$'36%'U1K38%&,$'3;)-#%*$'3V:.1#3I**"-)#'W
=>73I$,'4-)%*30'*.&1#-"*#%.&3H'*2&%X"'3@.-3B.&1#-"*#%&,3A>73Q.$"<'1Q.$"<'1
Original Projections
2-D Reconstructed Slices 3-D Volume
For a set of 2-D projections, each row of pixels is treated as a set of 1-D
projections ART is performed on each set of these 1 D projections and 2projections. ART is performed on each set of these 1-D projections and 2-
D slices of the 3-D object are created. These slices are then stacked to
form the final 3-D object
I$,'4-)%*30'*.&1#-"*#%.&3H'*2&%X"', X
The following explanation of ART
is for the 2-D case, and is how one
can reconstruct a single 2-D slice
of a 3-D model, 1, 2, ,ij j iw f p i M+ +, "pn pn+1
th
jth th
ijth
i
f = constant value in the j cell
w = fractional area of the j cell intercepted by the i ray
p = the ray-sum measured with the i ray as shown in previous slide
pi
f1 f2 fn
fn+1 f2n
pi+1pi+2
i
M = the total number of rays in all projections
A B
The ith ra!"#$"%&'()"!
w f w f w f w f p- - - - +"
"
A B
C
wji $#*"()&+",-.."/"0*-0"#$"1234"2
11 1 12 2 13 3 1 1
21 1 22 2 23 3 2 2
31 1 32 2 33 3 3 3
N N
N N
N N
w f w f w f w f pw f w f w f w f p
w f w f w f w f p
- - - - +- - - - +
- - - - +
"
"
!
"
YJ3H)&)4'K3C;-.5'*#%.&3<'#2./3@.-31.$6%&,3)31%&,"$)-3
191#'<D315
89'1%130'1"$#13V7 # ( 2W
!"#$%&"!'(")*
V7)9#.&)3(')*2W
+","# -.+ ,"#!-.
/.
/0
+"1"# -. /. /0+ 1"#!-.! /. /0
16
89'1%130'1"$#1V:% 2% 7 W
02)*23"&!!4&(
V:%*2%,)&37"&'W
+","# -.
+"1"# -. /. /0
+ ,"#!-.
/.
/0
+ 1"#!-.! /. /0
Z'U3I0H3I$,.-%#2<
5!!6%748(!9,2:
,
;,%<()$2%&=;,%<()$2%&=
;"==(:!$*,%43*!"$
)%,,(=>%&:2&3!"&37(=
?&!$*(!&(@!"73%,2$*8A!$*(!(&$2,(!!>,%<()$2%&!2=!>"==(:!
$*,%43*!"!5!!",,"#A!),("$2&3!"!B"&C!%D!=488(:!E"74(=A!
@*2)*!",(!4=(:!D%,!,()%&=$,4)$2%&
02)*23"&!!4&(!F"&:!,()%&=$,4)$2%&!4=2&3!&(@!"73%,2$*83
-. /7: /. G(@ /.-. /7:!/. G(@!/.
!&$%&'3/)#)4)1'*$$>HII@@@J,%@"&J(:4I)%77(3(=I(&32&((,2&3I)72&2)=I=*,((C
/&72&(!
3(%8"$(,2"7
KLM!N!G48B(,!%D!;",$2)7(=
3(%8"$(,2"7!
:"$"B"=(
-%8>4$(: />$2)"7
+","#
-%8>4$(:!
.%8%3,">*#
/>$2)"702)*23"&!!4&(!KOP5M
/>$2)"7!
02),%=)%>#
+","#!
$%8%3,">*#
/>$2)"7!
$%8%3,">*#F$"&:",:!0(7$!K5PQM
!,#!RQ!K5SOM
02)*23"&!!4&(!KSM
: : 7 K M
02)*23"&!!4&(!KSM
F : : 0 7 KTM
!"#$%&"!'(")*!KOPPM
F$"&:",:!0(7$!KTM
1*%:(!?=7"&:!KTM
!,#!RQ!KUM
V7" W"2 '(")* KXM
F$"&:",:!0(7$!KTM
1*%:(!?=7"&:!KTM
!,#!RQ!KUM
V7" W"2 '(")* KXMV7"!W"2!'(")*!KXM
Y"*"7"!'(")*!KTM
V7"!W"2!'(")*!KXM
Y"*"7"!'(")*!KTM
!"#$%&"!'(")*!KXM!"#$%&"!'(")*!KXM
!"#$%&'
Background
Research Objectives
Data SetData Set
Shape Characterization and Reconstruction
Discrete Element Modeling
ConclusionsConclusions
0'1')-*23:'#2./.$.,9[3
;)-#%*$'382)?'3:./'$%&,3%&3HU.37%<'&1%.&1? ,
8+'$'#.&%T)#%.&
7)9#.&) (')*2 8)&/ \-)%& 8+'$'#.&7)9#.&)3(')*238)&/3\-)%& 8+'$'#.&
0'1')-*23:'#2./.$.,9[3
;)-#%*$'382)?'3:./'$%&,3%&3HU.37%<'&1%.&1? ,
!6'-$)??%&,37%1*-'#'3M$'<'B$"1#'-3I$,.-%#2<[3!7MB>=7
8+'$'#.&3]3!7MB7)9#.&)3(')*23
8)&/ \-)%&
:./'$3;)-#%*$'
Z"<4'-3.@3*%-*$'13&''/'/3#.3*)?#"-'3#2'312)?'3^3_`
I-')3"&*.6'-'/3a3_b
8)&/3\-)%&
0'1')-*23:'#2./.$.,9[3
;)-#%*$'382)?'3:./'$%&,3%&3H2-''37%<'&1%.&1? ,
8+'$'#.&%T)#%.&
7)9#.&) (')*2 8)&/ 8"-@)*' 8+'$'#.&7)9#.&)3(')*238)&/3333333
V\-)%&3_W8"-@)*'38+'$'#.&
0'1')-*23:'#2./.$.,9[3
;)-#%*$'382)?'3:./'$%&,3%&3H2-''37%<'&1%.&1? ,
!6'-$)??%&,37%1*-'#'3M$'<'B$"1#'-[3!7MB>A7
cdb33
B.6'-'/3
7)9#.&)3(')*233
8)&/3\-)%&deb3333
B.6'-'/3
c`b333
B.6'-'/3
ceb33
B.6'-'/3
0'1')-*23:'#2./.$.,9[3;)-#%*$'382)?'3:./'$%&,3
%&3HU.3)&/3H2-''37%<'&1%.&1
I-') f Q.$"<' Z"<4'- .@ 7%1*1 Z"<4'- .@ 8?2'-'1I-')3f3Q.$"<'3
B.6'-'/
Z"<4'-3.@37%1*1
=>7
Z"<4'-3.@38?2'-'1
A>7
98% 9 14 251251 29629698% 9 – 14 251 251 -- 296296
95% 4 – 8 123 123 -- 133133
90% 3 – 6 61 61 -- 6262
85% 2 - 4 35 35 -- 3939
26
0'1')-*23:'#2./.$.,9[3
7M:38%<"$)#%.&3.@37%-'*#382')-3H'1#F
Q3V<f1#'?W
8)<?$'[37)9#.&)3(')*238)&/
5 mmI&,"$)-3MX"%6)$'.@3
=>73B%-*"$)-3;)-#%*$'1
12 mm F
Q3V<f1#'?W
I&,"$)-3MX"%6)$'.@3
A 7 8 2 % $ ; #% $
2 mm
A>738?2'-%*)$3;)-#%*$'1
4 mm
7M:38%<"$)#%.&3%&3H2-''37%<'&1%.&13.@3
7)9#.&)3(')*238)&/9
=e`
_e`
=``
ess (
kP
a)
e`
_``
Sh
ea
r S
tre
Assembly 1
Assembly 2
`
` `JA `Jg `Jc _J= _Je _Jd =J_
Shear Displacement (mm)
Assembly 3
7%-'*#382')-3H'1#3.&37)9#.&)3(')*238)&/
V! ^3=e`3+;)W
p ( )
V W
7M:38%<"$)#%.&3%&3H2-''37%<'&1%.&13.@3
7)9#.&)3(')*238)&/9
`J`g)
`J`=
`J`h
cem
ent
(mm
)
` `h
>`J`=
`
` `JA `Jg `Jc _J= _Je _Jd =J_
ert
ica
l D
isp
la
Assembly 1
>`J`g
>`J`h
Shear Displacement (mm)
V Assembly 2
Assembly 3
7%-'*#382')-3H'1#3.&37)9#.&)3(')*238)&/
V! ^3=e`3+;)WV W
7M:38%<"$)#%.&3%&3H2-''37%<'&1%.&13.@3\$)113
(')/1
_=`
g`
d`
_``
ress (
kP
a)
=`
h`
g`
Sh
ea
r S
tr
Assembly 1
Assembly 2
Assembly 3
`
` `JA `Jg `Jc _J= _Je _Jd
Shear Displacement (mm)
7%-'*#382')-3H'1#3.&38?2'-%*)$3;)-#%*$'1
V! ^3=``3+;)WV W
0'1')-*23:'#2./.$.,9[3
7M:38%<"$)#%.&3%&3H2-''37%<'&1%.&1
` `J= `Jh `Jg `Jd _ _J= _Jh _Jg _Jd
Shear Displacement (mm)
>`J`h
>`J`=
`
em
en
t (m
m
>`J`d
>`J`g
`J`h
al D
isp
lace
Assembly 1
>`J_=
>`J_
Ve
rtic Assembly 2
Assembly 3
7%-'*#382')-3H'1#3.&38?2'-%*)$3;)-#%*$'1
V! ^3=``3+;)W
0'1')-*23:'#2./.$.,9[3
7M:38%<"$)#%.&3%&3H2-''37%<'&1%.&1
A``
=``
=e`kP
a)
Spherical Particles
Daytona Beach Sand
#)&35"^3`Jd=
_e`
ea
r S
tre
ss (
k
#)&35"^3`Je`
e`
_``
Sh
e
`
` e` _`` _e` =`` =e` A``
Normal Stress (kPa)
0'1')-*23:'#2./.$.,9[3
7M:38%<"$)#%.&3%&3H2-''37%<'&1%.&1
I&,$'3.@3N&#'-&)$3G-%*#%.&3@-.<3=>73)&/3A>738%<"$)#%.&
: # % $: # % $==>>77 AA>>77
M % #M % #:)#'-%)$:)#'-%)$== 7373
8%<"$)#%.&8%<"$)#%.&
AA 7373
8%<"$)#%.&8%<"$)#%.&ME?'-%<'ME?'-%<'
Daytona
B h S d27˚27˚ 39˚39˚ –– 43˚43˚ 37.4˚37.4˚
Beach Sand27 27 39 39 4343 37.437.4
Rounded
Particles17.2˚17.2˚ 25˚ 25˚ -- 26.6˚26.6˚ 24.4˚ 24.4˚ –– 27˚27˚ ii
i O’Sullivan et al. (2004); Phillips et al. (2006)
33
j%X"'@)*#%.&38#"/%'13"1%&,37M:
34
:)#'-%)$1
:)#'-%)$ 7e` B" 8G IGV W b b
Rounded Glass Beads 1.18 1.0 0 0
V<<W b b
Michigan Dune Sand 0.33 1.5 32 9
Syncrude Tailings Sand 0 18 2 5 45 27Syncrude Tailings Sand 0.18 2.5 45 27
Daytona Beach Sand 0.23 1.4 42 19
Ott A l S d 0 27 2 2 51 29Ottawa Angular Sand 0.27 2.2 51 29
Ottawa Rounded Sand 0.53 2.4 30 7
:./'$%&,3.@3k)#'-3;-'11"-',
Undrained conditions were simulated by restricting the
volume of the particle assembly (Ng and Dobry 1994)volume of the particle assembly (Ng and Dobry, 1994)
The difference between boundary stresses and initial
fi t tt ib t d t b ild f tconfinement are attributed to buildup of pore water
pressure under cyclic loading
#$ ##!
#$v
#T
#$v
#$h
Cyclic direct shear Cyclic simple shear Cyclic pure shearCyclic triaxial
8)<?$'3;-'?)-)#%.&3> 7-93;$"6%)#%.&? ? 9
Step 1
Granular assembly
Step 2
Grains are allowed
Step 3
Excess grains areGranular assembly
is generated
Grains are allowed
to settle under gravity
Excess grains are
cropped at the top
8)<?$'3;-'?)-)#%.&3> B.&1#)Q.%/30)#%.? ?
Glass beads Ottawa rounded Ottawa angular
A single assembly is generated, and particles are transformed to
equivalent angular shapes, resulting in a similar fabric/arrangement
f ll t i l
38
for all materials
8%<"$)#'/3M@@'*#%6'38#-'113;)#23> B9*$%*382')-
5
0
q67809
Ottawa 20-70 Sand
(rounded)
at CSR=5%-5
10
0
q67809
Daytona Beach Sand
(angular)
at CSR = 16%
-10
0 10 20 30 40
at CSR = 16%
39
0 10 20 30 40
p' 67809
7'?'&/'&*'3.@3*.&#)*#3@.-*'13)&/3*..-/%&)#%.&3&"<4'-13.&3?)-#%*$'312)?'? ?
\$)113(')/1 !##)U)3-."&/'/ !##)U)3)&,"$)-
B80361 Z"<4'-3.@3B9*$'13P7-9 ;$"6%)#%.&7-93;$"6%)#%.&
0 4
0.3
0.4
\$)113(')/1
:%*2%,)&37"&'
!##)U)3-."&/'/
0.2
B80
7)9#.&)3(')*2
89&*-"/' H)%$%&,1
!##)U)3)&,"$)-
0.1
B
0
0.1 1 10 100 1000
Z"<4'- .@ B9*$'1Z"<4'-3.@3B9*$'1
B80361 Z"<4'-3.@3B9*$'13> B.&1#)Q.%/30)#%.
0.4\$)113(')/1
:%*2%,)&37"&'
80
0.3
,
!##)U)3-."&/'/
7)9#.&)3(')*2
89&*-"/' H)%$%&,1
!##)U) )&,"$)-
B8
0.2!##)U)3)&,"$)-
0
0.1
Z"<4'-3.@3B9*$'1
0
0.1 1 10 100 1000
42
8"<<)-99Individual particles from six natural, processed, and
manufactured granular materials were digitized and
t d i ft lib f DEM d listored in a software library for DEM modeling
Simulations of cyclic shear tests were carried out on a
representative volumerepresentative volume
At the pluviated void ratio, the susceptibility to
liquefaction is independent of particle shapeliquefaction is independent of particle shape
At constant void ratio, the influence of particle
morphology on liquefaction susceptibility is significant
Further studies are underway to numerically evaluate
the influence of particle shape on other engineering
tiproperties
B.&*$"1%.&1
ODEC2D and ODEC3D algorithms can model irregular 2-
D and 3-D particle shape with desired accuracyp p y
Stress-strain and volumetric behavior of simulated
material followed typical soil behavior of angular andmaterial followed typical soil behavior of angular and
rounded particles
Th l f h i i t bt i d i thThe angles of shearing resistance obtained using three
different fabrics in 3-D simulation are very close to each
other for both Daytona beach sand and rounded spheres
Angularity and particles interlocking resulted in more
shearing resistance in Daytona Beach sand compared toshearing resistance in Daytona Beach sand compared to
rounded material
I*+&.U$'/,<',
H2%13-'1')-*23U.-+3%131"??.-#'/3493#2'3Z)#%.&)$3
8*%'&*' G."&/)#%.&8*%'&*'3G."&/)#%.&3
IU)-/3l3`A=hhAm3B%6%$3]3:'*2)&%*)$3891#'<1
IU)-/3l3`A=he_d3B%6%$3]3:'*2)&%*)$3891#'<1
IU)-/3l3`h=_```3M$'*#-%*)$3]3B.<<"&%*)#%.&13) / l ` ``` '*# *) ] B. " *)# . 1
891#'<1