Les Despot Pyroclastique Et Leur Dynamisme

Mécanismes d’auto-fluidisation des écoulements
pyroclastiques : approche expérimentale

Corentin Chédeville-Monzo
To cite this version:

Corentin Chédeville-Monzo. Mécanismes d’auto-fluidisation des écoulements pyroclastiques : approche
expérimentale. Sciences de la Terre. Université Blaise Pascal - Clermont-Ferrand II, 2016. Français.
<NNT : 2016CLF22684>. <tel-01401494>
HAL Id: tel-01401494

https://tel.archives-ouvertes.fr/tel-01401494
Submitted on 23 Nov 2016
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.

!$%&' %$%(&' ' $)'$' %(*'(&'

+ , - .

%$' )/')!

,.
!"!

.
*.

#%&'(')*&+,+)'&)'#%+-('
./%-&'0+'12
..%34.#(&-

=> . >9
*-
01
( 21 3 4 .,,

!
.
5.6 )+7+8 2 +.
.,,
.
..1 2
9 ':..
&*/8 2 4. +.. ':..
1 *.

&*/8 2 4. +.. ':..
%2 &*/8 2 4. +..
;
Remerciements

!!

"

#

"

$
%

2OLYLHU5RFKH

%
&

'

("

)

&

*

)
"

%

+,&

&

"!
-

.

*

-

$
%"

&.

%

&"
%
%
%

%

/

"
*

#
0
&

&

1
&

%

#
"

%
&

2-XOLHQ
%
% -XOLHQ& 3
3LHUUH
%
/\GLH6DUDK
444$QQH
5

6 7 8 %DSWLVWH + 9

0DULQD

/ROD

%
4-XOLHWWH6LPRQ
"
$OHMDQGUD

".4 $QJpOLH 0DULQH 0DWKLDV *LDFRPR 1LQD
/

44
9DOHQWLQ / 9DOHQWLQ : 9LQFHQW 0DULH$QQH

*LR
(ORGLH 6LOYLD

" 6ZHWKD ;;;; - 0pORGLH1HLJH 6HE
$PpOLH2VFDU
<

%DSWLVWH' )HGHULFD<DQQLFN0D[&
0D[ 0 &DPLOOH WLpFDU *DE\ *DUHWK 'DYLGH $OHVVDQGUR &LDR 'DYLG
3DWULFLR0DULRQ/DXUD =
&
%
>$XGH0DQRQ=ROWDQ6LHW
&

?

&RT

%
OD&XLWH @4444
%

!

*

84

&

%

/

0
A

%"

OD)UDQJH%

*
#

"

)
/ROD
0DULQD%

#

&

B

1
*URV
%DS 4#
%

%
"

%&
%

%
OD7LTXH
"

%
!
C

? "
- .

D A

-
"

)
"

-

&

% % "

)
"

-

"

% !#

"!

!

$

"

&

"

"

%

%

"

%

%

.!
E . %

&KRXFKRX
F.
3
. /pD %

&UD]\ *UDFH
/DXUHQW $XURUH *DsWDQ *DXWKLHU 3LHUUH $PDQGLQH 6DOYDWULFH 0DQRQ "
C
3DUDFXOH.pYLQ%pQp%UDKLP&DPLOOH0DULHOOH)UDQoRLV%pMRQ
7

"

%

%RUL
(UZDQ
3DXOR "

"

%
&

/

)
-
= "
>

%
&

7KRPDV3LHUUH

%

#

&

./XGRYLF©)OH[ª

&
%
%

%
$OH[G¶$OOH[

%

"
%

&DPLOOH

)
-
-

-

25pPL(UZDQ

-HUHP

%
1LFR 3DXOR

-XOLHQ %

3LHUUH
%DSWLVWH

-HDQ*XL6XDUH]+XJR$OH[LV+HUYp
0DW/DXUHQW&

1

-

-

".

/'

"2 OD &XLWH

OH &RIIUH OH
&RWRFLUDSWRU %
%
0$ -HDQ0LFKHO 7UDYHUV OD 7LTXH .

&KRXFKRX9LQFHQW .444)DVRO0DW+XJR$OH[LV-*1LFR$EHQGDQRQ%DUG\
)DUG\ODVHUYHXVHGX+SDUN

4
$

"

GRWD

G

&

%DVWLHQ
*DXWKLHU

"

% %

=&
> &
"

%

"&
?
"

"

-

"

&
&

-HDQ /RXLV
)UXTXLqUH
&\ULOOH*XLOORW
%

%

-
-
&

&

"!

)
(ULF%UXW

%

%

$

"

&ODLUH%&

%

"

!
.

?"
-

#
"

%

$XGUH\ &pFLOH 0DULH&KULVWLQH 6DQGULQH $

6\OYDLQH
%%

%

7KLHUU\+DPHO
%

,
H

)

3DEOR
% +

-

)
"

%HQ9:G9.DULP.7KLHUU\0

5pJLV '

IJ! 3LHUUH

7LP
-HDQ/XF/H3HQQHF

%)

/DXUHQW
&KXSLQ

)
-
"

/RwF%DVWLHQ-RMR0DQX$GULHQ
*UHJ
%

%

"

H
%

K
-

%
7!L

H%

"

&

G%
"
,

%
?"

$GULHQ#
%

%
)
-HDQ0LFKHO

&

1

PDIDPLOOH

"

"

&

%

)
PDPqUHPRQSqUH
PDEHOOHPqUH

#
J %

444 )
0DWWKLHX 6LOYLD
/pD

%

%

%

)
'DYLG7RVFDQH
2WLV

)
I#

-

$

-

"

&

&

#

9

.

"

"K

%

%

"

"

=>
%

!!
"

%

"K

9&

"

"

.

%
%

?

-

"

'M!NOP

"9

"

%
"
&

.

%

%

#

"
"
-
!

!

O!OQ ""

"

.

?

9

#

B

%

/O

"

!

R

S:OOP

<

"
%
"

9

"
%
%
%
<

.
TO

/UM

""

.

"

%
!

%

) 2 V "

1

.
W 1-

X"+

+. F
-
"
"
%
"

3 " Y.Z

.
K

"

" "

3

&
%

F
"

F .
F"

KG
.
-

"

!"
F

'M!NO P
%
"
X
F
F

F

"

F

["
%
.

F
F
F
"" %
"

! K

F

F

"
F3! K

%
"
O!OQ""
"
"
F

. FG
.

%

X
F
"-

. K

%
F
"
%
.F
" /O 5

" S:OOP

%

" F
"
C

"
K
3
"

%
. F

TO /UM
C
3

""
. F""
"

! K

%
F
"

\
.F2 V ". +. F WK 1-

X"
W

6RPPDLUH

Sommaire
5HPHUFLHPHQWV
5pVXPp
$EVWUDFW
Introduction gé né rale 19
Chapitre 1
Les courants de densité pyroclastiques 25
1.1. Généralités ............................................................................................................ 26
1.2. Mécanismes de formation des courants de densité pyroclastiques 27
1.3. Mécanismes de transport des courants de densité pyroclastiques 32
1.3.1. Transport des particules par un gaz turbulent .................................. 34
1.3.2. Transport dans un mélange gaz-particules dense ............................ 38
1.4. Sédimentation des courants de densité pyroclastiques ........................ 43
1.4.1. Sédimentation à partir d’un courant dilué et turbulent ................. 43
1.4.2. Sédimentation à partir d’un écoulement dense ................................. 44
1.4.2.1. Sédimentation « en masse » ................................................................................... 44
1.4.2.2. Sédimentation par aggradation progressive .................................................. 46
1.5. Architecture des dépôts .................................................................................... 48

6RPPDLUH

1.5.1. Dépôts de déferlantes ................................................................................. 48
1.5.2. Dépôts d’écoulements pyroclastiques de faible volume ................. 50
1.6. La problématique des ignimbrites ................................................................ 52
1.6.1. Caractéristiques des dépôts ..................................................................... 52
1.6.2. Rhéomorphisme ........................................................................................... 56
1.6.3. Rapport de forme des ignimbrites : HARI et LARI ............................. 56
1.6.4. Capacité à franchir des obstacles ............................................................ 59
1.7. Mobilité des courants pyroclastiques .......................................................... 60
Chapitre 2
Fluidisation d’un systè me gaz/particules 63
2.1. Le processus de fluidisation : historique et applications ...................... 64
2.2. Fluidisation d’un milieu granulaire par un gaz ........................................ 65
2.2.1. Principe ........................................................................................................... 65
2.2.2. Vitesse caractéristiques de fluidisation (U mf, Umb) ........................... 67
2.2.3. Les différents groupes de particules ..................................................... 71
2.3. Défluidisation par diffusion de pression .................................................... 76
2.4. Quelques paramètres influant sur la fluidisation .................................... 77
2.4.1. Forces interparticulaires ........................................................................... 77
2.4.2. Autres paramètres affectant la fluidisation ........................................ 79
2.5. Fluidisation des écoulements pyroclastiques ........................................... 82

6RPPDLUH

2.5.1. Indices de fluidisation des écoulements pyroclastiques ................ 82
2.5.2. Sources de fluidisation des écoulements pyroclastiques ............... 83
2.5.3. Implications pour la mise en place des écoulements

pyroclastiques ........................................................................................................... 86
Nomenclature ............................................................................................................... 88
Chapitre 3
Mé thodes expé rimentales 91
3.1. Démarche scientifique ...................................................................................... 92
3.2. Etudes d’écoulements générés par le principe de « rupture de

barrage » ........................................................................................................................ 93
3.2.1. Travaux précédents ..................................................................................... 93
3.2.1.1. Courants gravitaires fluides ................................................................................... 93
3.2.1.2. Ecoulements granulaires secs (i.e. non-fluidisés) ......................................... 94
3.2.1.3. Ecoulements granulaires fluidisés ...................................................................... 98
3.2.2. Dispositif expérimental « rupture de barrage »

(chapitres 4 et 5) ...................................................................................................... 99
3.2.3. Les particules ................................................................................................. 102
3.2.4. Rugosité du fond du chenal ....................................................................... 102
3.2.5. Protocole expérimental ............................................................................. 104

4
3.2.6. Dimensionnement des écoulements expérimentaux ...................... 106

6
3.3. Expériences de chute de lit de particules (chapitre 6) ........................... 111
3.3.1. Dispositif expérimental ............................................................................. 111

6RPPDLUH

3.4. Capteurs de pression de fluide ....................................................................... 115
3.4.1. Caractéristiques des capteurs ................................................................. 115
3.4.2. Calibration des capteurs ............................................................................ 116
3.4.3. Utilisation dans les expériences avec matériaux granulaires ...... 118
3.5. Granulométrie des particules ......................................................................... 119
3.5.1. Granulomètre Morphologi G3 .................................................................. 119
3.5.3. Résultats .......................................................................................................... 120
Chapitre 4
Auto-fluidisation des é coulements pyroclastiques
125
se propageant sur un substrat rugueux
Chédeville and Roche 2014, Journal of Geophysical Research 127

“Autofluidization of pyroclastic flows propagating on rough substrates as
shown by laboratory experiments”
1. Introduction ............................................................................................................ 127
2. Experimental methods and device .................................................................. 128
2.1. Principles of pore fluid pressure generation and diffusion …........... 128
2.2 Experimental device ........................................................................................ 129
2.3. Scaling issues ..................................................................................................... 130
3. Results ........................................................................................................................ 130
3.1. Flow runout distance ...................................................................................... 130

6RPPDLUH

3.2. Flow front kinematics .................................................................................... 131
3.3. Flow morphology ............................................................................................. 132
3.4. Flow-substrate interaction ........................................................................... 133
3.5. Basal pore pressure ........................................................................................ 134
4. Discussion ................................................................................................................. 135
4.1. Flow mechanisms in experiments ............................................................. 135
4.2. Implication for pyroclastic flows ................................................................ 137
5. Conclusions ............................................................................................................... 137
Chapitre 5
Influence de la pente sur la fluidisation et la
141
dynamique des é coulements pyroclastiques
Chédeville and Roche 2015, Bulletin of Volcanology 143

“Influence of slope angle on pore pressure generation and kinematics of
pyroclastic flows: insights from laboratory experiments”
Introduction .................................................................................................................. 143
Experimental methods .............................................................................................. 144
Pore fluid pressure generation and diffusion in gas-particle systems 144
Experimental device and procedure ................................................................. 144
Results ............................................................................................................................ 145
Pore pressure data .................................................................................................. 145
Flow runout ............................................................................................................... 148
Flow front kinematics ............................................................................................ 149

6RPPDLUH

Flow front morphology .......................................................................................... 149
Sedimentation and deposit reworking ............................................................. 150
Discussion ...................................................................................................................... 152
Autofluidization ....................................................................................................... 152
Influence of slope angle on flow dynamics ...................................................... 152
Implication for pyroclastic flows ........................................................................ 153
Conclusion ..................................................................................................................... 154
Chapitre 6
Gé né ration et diffusion de la pression interstitielle 157
induite par la chute de particules
158
6.1. Introduction ..........................................................................................................
6.2. Résultats ................................................................................................................. 160
6.2.1. Aspects généraux ......................................................................................... 160
6.2.2. Effet de la hauteur de la porte .................................................................. 165
6.2.3. Effet de l’épaisseur du lit granulaire ..................................................... 167
6.2.4. Taille des particules .................................................................................... 172
6.2.5. Effet de la température .............................................................................. 176
6.2.6. Expériences avec l’ignimbrite .................................................................. 178
6.3. Discussion .............................................................................................................. 183
6.3.1. Sous-pression ................................................................................................ 183
6.3.2. Surpression dans le réservoir (1ère surpression) ........................... 183
6.3.3. Surpression dans le dépôt (2ème surpression) ................................. 187

6RPPDLUH

6.3.3.1. Diffusion à la base du dépôt .................................................................................... 189
6.3.3.2. Diffusion dans l’épaisseur du dépôt ................................................................... 193
6.3.4. Pressurisation thermique ......................................................................... 195
6.4. Implications .......................................................................................................... 198
6.5. Conclusions et perspectives ............................................................................ 200
6.5.1. Résumé des principaux résultats ........................................................... 200
6.5.2. Perspectives ................................................................................................... 202
Nomenclature ............................................................................................................... 204

Conclusions gé né rales et perspectives 205
1. Résumé des principaux résultats ...................................................................... 205
1.1. Ecoulements se propageant sur fond rugueux ...................................... 205
1.1.1. Chenal horizontal ........................................................................................................... 206
1.1.2. Chenal incliné .................................................................................................................. 208
1.2. Auto-fluidisation par chute de lits de particules .................................. 209
2. Implications pour la dynamique des écoulements pyroclastiques ....... 211
3. Perspectives ............................................................................................................. 213
Ré fé rences bibliographiques 215

,QWURGXFWLRQJpQpUDOH

/ŶƚƌŽĚƵĐƚŝŽŶŐĠŶĠƌĂůĞ
9

.

%

-#

"

-

%

%

"

%%-
% E

/O

%

3".
:NNU E

-

%"

%%

K

"

&

.

%

9

.

"

"K

%

9

.

-
<
2

"K
=

.
>

"K

=

.
>

-

-

#

.

.

9
%-

.

!!
-

.
!

9

.
"

%

.

K

%

"
&
K

C

%
K

<
%

-

K

%
#

-

"

"

"

" %

"

"

/O

W

:N/
+

:N] ""

.

%


G
#

:N//!
:NU

%

%

.
%
2

-
%

%

"K

.

%

9

-

"

+ %

%
"KH

9

%

%
-

.
X
.:NM'

"

!

"
"

% #

%-
E%
:NN]

#

%

+

.
2R

%

8R

"

8 9

!

"

!

#

%

" ! 8

!

!

%

8
9

%

.

"K &

+

:N]

"

-

) 3"" :NÔJ 5 :NUOJ 5 :NU'2
"

_

%
"

%

"

"

-

K

%"

%

%-


9

"

.

%
#

"
- "

9&

"

%

%

!

"

%
%

!

.
+
%

-

<

9

-

%

SNO P

%
X

/OOMJ
/OO] `

"
"

-

^

^
"
2
• 9
HU FKDSLWUH

.

.

<

"

!

.

" %
a :C
<

"

• 9
VHFRQG FKDSLWUH

"

.
"KH

"

"K% <

9

"

1

"K

.

-

"K

-

.
%
"

%

• 9
FKDSLWUH

-

-
- MO
OO

"
OMM

"

"

B

G
%
.

-


+
%

-

"

%

-
"K

9

.

"

• 9
FKDSLWUH

"!

.

-
-
-! "

K

" .

"
N

"

D
%

" 9

"

-
-
"

@B
-

%

%

%

!

%

-

%

• 9 "

-

M

'
-

%

OOQ

-

"
2

"

"

%

%

"

• 9&
FKDSLWUH

!

9 "
-

'
M

%

9

/O NO

:O!Ô

"
O!OOOP

/O!/OOQ

"

<

-

"

%

<

.


+

.

-

.

&KDSLWUH/HVFRXUDQWVGHGHQVLWpS\URFODVWLTXHV

.
9
-

<

.

&

E

!

" %

"

""

-

"


ͳǤͳǤ
±±±
9

.
+

"

"%
" "B

%

%

"

&NOOQ
"K

%

%
%

%%

/O:: &
:OO H =>
)7[

:NUO\

7
%:NUUJ3:U^G5:NUMJ

[
:NN^J\
E"
:)5
:NNM9
%

<

+
%
%

OOO:C
:OOOC 7C:NN]
<

%
-

#
%

9
+
%

:OOC
%
+

7"3 V

:NUNJ\
"
5
:NNM%

%

-

2'MOOOCb\
"
5
:NNM

!

<
-

%

"

" %
!K

< ""

"

-2 "
E $
cC. :N]'J"

3 `%

d
5 :NUMJ "

+"
E
W

:NN E "

+

"

K

C

-2 E"

W

:NNJ \C E

.

:NN^

"

7C
:NUOJ7C

:NUOJ)
%

:NN'J)
%

:NN^
9
<
+

%

%

B"
5" :NU]

%%-
%

:M!/OC
%

-+
9"

-

%

.

%

+

-

K

9
+

%

/O

3".
:NNU


ͳǤʹǤ ±±

9

.

%
#
"

%%

"

%

9 &
+
"
"

.

%

%
- %
.

-

+
%

"

%

.
"

9
-

"
+
!

(IIRQGUHPHQWGHFRORQQHpUXSWLYH)LJXUH

.

-
%

9"

%

%
[ /O
L/7

-! %

%

-

-%

9

-%

"

%

7
/O:M G %

-

"

"

!!

"K
%

"

H

9 "

"

"

#
%

-

2 %

- %

"

"
%

G%
"F
:NN^


&ŝŐƵƌĞϭ͘ϭ͗;ĂͿŐĠŶĠƌĂƚŝŽŶĚ͛ƵŶĐŽƵƌĂŶƚĚĞĚĞŶƐŝƚĠƉǇƌŽĐůĂƐƚŝƋƵĞƉĂƌĞĨĨŽŶĚƌĞŵĞŶƚŐƌĂǀŝƚĂŝƌĞĚ͛ƵŶĞƉĂƌƚŝĞĚĞůĂ
ĐŽůŽŶŶĞĠƌƵƉƚŝǀĞ͘;ďͿ/ůůƵƐƚƌĂƚŝŽŶĚĞĐĞŵĠĐĂŶŝƐĞůŽƌƐĚĞů͛ĠƌƵƉƚŝŽŶĚĞϭϵϴϬĚƵDŽŶƚ^ĂŝŶƚ,ĞůĞŶƐ͕ƚĂƚĚĞ
tĂƐŚŝŶŐƚŽŶ;ĐƌĠĚŝƚ͗h͘^͘ŐŽǀĞƌŶŵĞŶƚͿ͘
9
"

&

%

%

ÔO
H 5:N]^J7C5:N]^J7C
:NN] 9
"
"K!

%
"B

%

&

"

%
7
"

%

%

%

C
%
K

C

%

%

-

%

"

"

%

%

%

%

%

"

%

H
"

K

"
"KH

%

C

!

%
5:N]^7C
5:N]U
+
%

#

%

0C 5 :NN^J 0
.
\C
/OO/

=S\URFODVWLF IRXQWDLQLQJª

.

9

%


.

%

7 [

:NUO + :NN:

%

<

+

W" . 'ÔO 5C

:N]:J 3 :U^
$!5:NUMJV%
]N$!
.7":NU]G

#

%

%
%

%.
%

W"
::

"
%

%

9
%

"

#

"

K

"%

.

-

#
%

"

"K

"

"

7C
:NN] 9

"

!#
&

"

%

%

!

.
:NUUJ7C
:NN]
'pERUGHPHQWRX©ERLOLQJRYHUª)LJXUH

"
%

%

+

GpERUGHPHQWRX=ERLOLQJRYHUª

%

&

%

%

"
"KH

"

5 :U]U

- 1

:U]]
"

3. :NMU

:NM: ) 9" +
`%

@

+

%
S:C9

"

%

"

" %

<

-

"

@ aÔC ./OU)

G"

7C
:NUMJ
/O::

= "> #

.

7C
:NN]


&ŝŐƵƌĞϭ͘Ϯ͗ŐĠŶĠƌĂƚŝŽŶĚ͛ƵŶĐŽƵƌĂŶƚĚĞĚĞŶƐŝƚĠƉǇƌŽĐůĂƐƚŝƋƵĞ;WͿƉĂƌĚĠďŽƌĚĞŵĞŶƚŽƵͨďŽŝůŝŶŐͲŽǀĞƌͩ͘
(IIRQGUHPHQWG¶XQG{PHGHODYH)LJXUH

"
"

"

%

<

.
%

H
%

1

"

:NU^
.

"K

"
"

!

%
%

%

%

<

<

#

"

%
%
)

:NUUJ
7
:NN/J :NNU "

%

% 9
<

#

"%
-

2)E
`
F
/OOOJ?K

$7
:NN/J`CW&:NNJ7
[)

:NNU

-

-

"K

%
?K
$ 7
:NN/ ?

<

"

%

+


&ŝŐƵƌĞϭ͘ϯ͗;ĂͿŐĠŶĠƌĂƚŝŽŶĚ͛ƵŶWƉĂƌĞĨĨŽŶĚƌĞŵĞŶƚĚ͛ƵŶĚƀŵĞĚĞůĂǀĞ͘;ďͿ/ůůƵƐƚƌĂƚŝŽŶĂǀĞĐů͛ĞĨĨŽŶĚƌĞŵĞŶƚ
Ě͛ƵŶĞƉĂƌƚŝĞĚƵĚƀŵĞĚƵDĠƌĂƉŝ͕/ŶĚŽŶĠƐŝĞ;ƉŚŽƚŽhůĞƚ/ĨĂŶƐĂƐƚŝϮϬϭϬͿ͘
'pFRPSUHVVLRQEUXWDOHG¶XQFRUSVPDJPDWLTXH)LJXUH
? "

%
-

<

%

C
%

.<
%

"

"K

"

.

%

<

%

-

"
!

-
%

+
"

% =>
9

-

)7[

:NUO
0
K..
:NM^
.<

)

:NN]
0
%/OO]
<

&ŝŐƵƌĞϭ͘ϰ͗;ĂͿŽƵƌĂŶƚĚĞĚĞŶƐŝƚĠƉǇƌŽĐůĂƐƚŝƋƵĞŐĠŶĠƌĠƉĂƌů͛ĞĨĨŽŶĚƌĞŵĞŶƚĚƵĨůĂŶĐĚ͛ƵŶǀŽůĐĂŶƋƵŝ
ĚĠĐŽŵƉƌŝŵĞďƌƵƚĂůĞŵĞŶƚůĞĐƌǇƉƚŽĚƀŵĞ͘;ďͿ/ůůƵƐƚƌĂƚŝŽŶĚĞĐĞŵĠĐĂŶŝƐŵĞĂǀĞĐůĞͨďůĂƐƚͩŐĠŶĠƌĠůŽƌƐĚĞ
ů͛ĞĨĨŽŶĚƌĞŵĞŶƚĚƵĨůĂŶĐEŽƌĚĚƵDŽŶƚ^ƚ,ĞůĞŶƐ͕tĂƐŚŝŶŐƚŽŶ͕ůĞϭϴDĂŝϭϵϴϬ;ƉŚŽƚŽ'ĂƌǇZŽƐĞŶƋƵŝƐƚͿ͘


7

"K

<

%

%

" = >

<

7
:NN/J7C:NN]

"

-

<

%

" !
&
7
7V
:NO/1
:NJ)/O:OE
\FC

/O:
5HPRELOLVDWLRQG¶XQGpS{WG¶pFRXOHPHQWS\URFODVWLTXH
R
"

%

H

%

%

"

.
=

>

"

%

"

% +,

"&:OC

3

:NN^
(IIRQGUHPHQWG¶XQ©VSDWWHUFRQHª
+

%

%

<

=
>

%

+

\

/OONJ0

/O:'
ͳǤ͵Ǥ ±±

9

=

.
> "

"K

%
:NNU9
+"

-"

<

<

."
=GpIHUODQWHV S\URFODVWLTXHVª

"
S :e

%

7C
5C
:N]J :NNUJ 5 [" /OOO

<
%

&

%

=pFRXOHPHQWVS\URFODVWLTXHVª


#

K

%
W

/OOO

%-
<
+

&

% #

1

+
%
" %

%

"

-

"

"

%
.

#

%

.

"

-

"

W"
:M 9

"

"

="
!"
>

-

"
%

!

&ŝŐƵƌĞϭ͘ϱ͗ǀƵĞƐŝŵƉůŝĨŝĠĞĚ͛ƵŶĐŽƵƌĂŶƚĚĞĚĞŶƐŝƚĠƉǇƌŽĐůĂƐƚŝƋƵĞĐŽŵƉŽƐĠĚ͛ƵŶĠĐŽƵůĞŵĞŶƚ
ƉǇƌŽĐůĂƐƚŝƋƵĞďĂƐĂůƐƵƌŵŽŶƚĠƉĂƌƵŶĞĚĠĨĞƌůĂŶƚĞƉǇƌŽĐůĂƐƚŝƋƵĞ͘>ĞŶƵĂŐĞĐŽͲŝŐŶŝŵďƌŝƚŝƋƵĞ
ĂƐĐĞŶĚĂŶƚŶ͛ĞƐƚƉĂƐƌĞƉƌĠƐĞŶƚĠ͘
9
"

.

"

"K

%

"K"

%"

=

>
=

.
>


%

.
9

=

>
"

-

.
!

- -

#

9
"

%

%

X
.
.

̱/OOO

2
ρ I 8/
5H = ::

%
ɏ
%

8%

/
"

"

ȝ% .

+ "

H

%

%

%

9
-

%

"

.

+

% % #

2

X
3Q
2
8 VHWW
3Q = :/
N8 ;
%
8VHWW%

N

%\ ̱O'

-

8 %

"
ĲHɏ :H/ %
Ĳ

"

ɏ
%
.

X
:NNJ V

:NU] G
#

X

%

9

#

3Q V

:NU]73Qa/M

%

#

#

"


%

JW"
:^
.

K

K

!

<

©FKDUJHGHIRQGª
EHGORDG9K

© FKDUJH HQ VXVSHQVLRQ ª VXVSHQGHGORDG

#

K
%

9

%

"

S O:!:e %

-

.

&&

/O:/

&ŝŐƵƌĞϭ͘ϲ͗>ĞƐĚŝĨĨĠƌĞŶƚƐŵŽĚĞƐĚĞƚƌĂŶƐƉŽƌƚĂƵƐĞŝŶĚ͛ƵŶĠĐŽƵůĞŵĞŶƚƚƵƌďƵůĞŶƚ͘>͛ĠƚŽŝůĞƌŽƵŐĞ
ĐŽƌƌĞƐƉŽŶĚăů͛ŝŵƉĂĐƚĚ͛ƵŶĞƉĂƌƚŝĐƵůĞĞŶƐĂůƚĂƚŝŽŶƐƵƌƵŶĞƉĂƌƚŝĐƵůĞĞŶƐƵƌĨĂĐĞĚƵĚĠƉƀƚĞƚƋƵŝĞƐƚƌĞŵŝƐĞ
ĞŶŵŽƵǀĞŵĞŶƚ;ƌĞƉƚĂƚŝŽŶͿ͘

% %

&

#

7C
0"
0
"K/OO/J
C
/O:M
WS
6W = :
WI
%
WI

.

%
WS


ρ S − ρ I G /
WS = :'
:U
%
ɏ ɏ

%

%

"K
G

R 6W
%
O

%

%

1
%
6Waa:

%

%

9

%

"

%

%

"

E
%

-
%

f"

"

" %

"

"%
.

%

" 7C
:NN 7

"

%

%

= ! >

=

> = !
"
>"

#
%

"
%

G
F )" /O:: 9
"
%

X

"

J _K
5L = :M
8/
%
K

J¶"%

J¶g J ɏF!ɏDHɏF%

ɏD
%

+ 5L

"
%

%

= ! >
9
-.

+

&

!

1XDJHGpIHUODQWRX©$VKFORXGVXUJHVª
E " "

!

.

<
%

=>

"


"

"

%

%"

"
W
[
C
:NU/J W
:NNM

"
%
)"
+

:NO/ W
[
C

:NU/

%
) ?K

:NN: `C W& :NNJ
c
:NN

)
:NNM!/OO W"
:]J

:NNUJ

:NNNJ
/OO/

&ŝŐƵƌĞϭ͘ϳ͗ŶƵĂŐĞĚĠĨĞƌůĂŶƚĂĐĐŽŵƉĂŐŶĂŶƚƵŶĠĐŽƵůĞŵĞŶƚƉǇƌŽĐůĂƐƚŝƋƵĞ;ŵĂƐƋƵĠͿ͕^ŽƵĨƌŝğƌĞ,ŝůůƐ͕
DŽŶƚƐĞƌƌĂƚ;ƉŚŽƚŽdŚŽƐƚĞŶŽĞĐŬĞůͿ
'pIHUODQWHEDVDOHRX©%DVHVXUJHª
9

=
"
>

% "
)
:N^]

3
:N^M +
9

%
-

:N'^ 0C 9"
?7G

%

.

"

="
>
"
%

-
%

=
>

%

9
-%

"

"

"K

"


%
?

%

%

-
"K

- c":N^M)
:N^]?

"

.

"

=
>
=
> 5
K 7
:N]N

"
"

C

%

©%ODVWªODWpUDO
9

=>
%

:U :NUO ) 7 [

E
%

)

/^

:NN] - G
0
K.
:NM^
\ C 0
%
/OO]

:M'O)G"
GC 7

:NNM ""
7 [

.

"

-

% %
:/JW"
:'9

-
-#

%

"

"

+
= EODVW>)7[

W"

:'
%

-

% :MO H )
X

:NU'

&/MH

\

7
%:NUU9

" =EXUVW SKDVH>
<

"

"

"

.
K

G

#

"%

"

"% <

/O

"
) [

:NUO
:O

7
[
)
:NN]J 1 L"
/OOUJ /O:/
.

&

#

%

%

:NN/
±Ǧ
R

"

K

/O:/ 9


%

&

%

"K

7
"K

""

"

#

W"
:U1
%

"K

"
%
/

!

""

!

W"
:U7C:N]^J5:NUOJ E%
:NN]JX

/OOU

"

"

"

"
9

"

= >

= >

%

. "

!

7
"

%

H

%

"K

"

%
%

"

&ŝŐƵƌĞ ϭ͘ϴ͗ dƌĂŶƐƉŽƌƚ ĚĞƐ ƉĂƌƚŝĐƵůĞƐ ĚĂŶƐ ƵŶ ŵĠůĂŶŐĞ ŐĂǌͲƉĂƌƚŝĐƵůĞ ĚĞŶƐĞ͘ ;ĂͿ ZĠŐŝŵĞ ĚĞ ƚƌĂŶƐƉŽƌƚ
ŐƌĂŶƵůĂŝƌĞĚŽŵŝŶĠƉĂƌůĂĨƌŝĐƚŝŽŶĞƚůĞƐĐŽůůŝƐŝŽŶƐĞŶƚƌĞůĞƐƉĂƌƚŝĐƵůĞƐ͘;ďͿZĠŐŝŵĞĚĞƚƌĂŶƐƉŽƌƚƉĂƌƚŝĞůůĞŵĞŶƚ
ĨůƵŝĚŝƐĠƉŽƵƌůĞƋƵĞůůĂƉƌĞƐƐŝŽŶŝŶƚĞƌƐƚŝƚŝĞůůĞĚƵĨůƵŝĚĞƌĠĚƵŝƚůĞƐĨŽƌĐĞƐĚĞĐŽŶƚĂĐƚŝŶƚĞƌƉĂƌƚŝĐƵůĂŝƌĞƐ;ĞƚĚŽŶĐůĂ
ĨƌŝĐƚŝŽŶĞƚůĞƐĐŽůůŝƐŝŽŶƐͿ͘
9

%

"

"

%

C )" /OOU 9


"

%

"

!!

"

"B

%

-"

C
/OON

"

%
"

%

5:NUOJ
C)"/OOU

LQ
VLWX

#

%
,
"

)"
/O:O 1

+

"

E

-
*

%

"

"

""

7%"

"

E

"

"
E%

"
/OO:E
2
ρ S G /γ /
6D = :^
ρ S − ρ I JK
%
ɏS
ɏI

%

%

"K

ɀ
-

J "%

K

R

7%"

!

"

9

.

%

-

-

'D = :]
N ε S ρ Sγ
%
I % .

ɂS

N

"
+ 'D


9
- .

.
%

!

(FRXOHPHQWVGHEORFVHWFHQGUHVRX©EORFNDQGDVKIORZª

%

%
.

:O!:O^

- <

%
W"
: J
-

2
7" @
X

:N]] J ) E
0
:NN J ?K

$c
:NNJ7

[)

:NNU9

.

<
"

%

&

%

"

%

%

=
> )

:NUUJ :NNU

-

"K
%

"

<
& /O )+

%

"

)

:NUUJ 7C :NN]J 0C

/OOM

"

%

-

"

.

"
!"
!

%
:N]UJ5":NU]J:NNU
(FRXOHPHQWVGHSRQFHVHWFHQGUHV

%

%

%
W"
::
W"
:/9
"

%

"

:!:OC"HJ7C
:NN]

"B

-

"K

"
7C:NU/?

"

S : C

9 :NNJ7[

A:NUO

"

"
%
9

%
S : C "

-

.
%

.

!

H


9

.

S:O!:MC

<

%
&

,

"

K

"

SM!:OQ
9

<
.
"%
:a:OOO
C #
"

= ">

E"

^CXV
KK:NNNJ
3:U:M7

:NU'J+:NN:7
:NN] +
7"3
:UU)V

:NUNJ
@/OU)7C
:NUMJ
/O::
9
<

="
>"
=

>

""
" + C)

< .

`%

!d
) :NM G&

="
>

. "

<

%

%

.

: C

&

< "

"

"
%

E
-

.

"

[
:NN^
&

%

&

%

1

""

%
%

aMO-MO C

% "

%

%
+
-

0 3 1!?

9"V
.
%

!&%

%

< ""

%
"
"

[
):NU^J5":NU]
(FRXOHPHQWVS\URFODVWLTXHVGpULYpVGHGpIHUODQWH
R

%

%

-

.
=%>
W
[
C
:NU/JW
[
C
:NUJ

:NNNJ
/OO/


"

+H/%

:NNNJ
/OO/
(FRXOHPHQWVGHVFRULHV

%
#
"

<

%

-.

/OO^
3" 1
\

/OONJ 0

/O:' 9

.

&

"

%

9
"

.
"

ͳǤͶǤ ±±
±ǯ±

-

+

%

%-

9F

:NU/J:NNUJ0
.\C
/OO/
:9WUDFWLRQVHGLPHQWDWLRQ

K

VXVSHQGHGORDG
K

"
EHGORDG

%#
%

.

<
/
,

K

K

#

G[:NUNJ\

0
.:NNMJ0
.\C
/OO/

<< 9
< #

-

%"

"

%

#

<

/O:/


±ǯ±
ͳǤͶǤʹǤͳǤ ±Ǽǽ

-

"

!

&
%

9

."
%

=

>

#

"
9

"
%

-

<

K

%

K

=">

-

W"
:N 9
<

% ]Me

"
5:NU'JW :NN

&ŝŐƵƌĞϭ͘ϵ͗>ŝŐŶĞƐĚĞĐŽƵƌĂŶƚĐĂƌĂĐƚĠƌŝƐƚŝƋƵĞƐƉŽƵƌ;ĂͿƵŶĠĐŽƵůĞŵĞŶƚĚĞŶƐĞĐŽŵƉŽƐĠĚ͛ƵŶĞƉŽƌƚŝŽŶă
ĐŝƐĂŝůůĞŵĞŶƚĨĂŝďůĞŽƵŶƵů;ƉůƵŐͿĞƚ;ďͿƵŶĠĐŽƵůĞŵĞŶƚĚĞŶƐĞĐŝƐĂŝůůĠƐƵƌƐŽŶĠƉĂŝƐƐĞƵƌ͘
9

=
>%

.

$ :N]OJ 9F
:NU/

-

.
7C :N]^J 5 [
:NU:

"

.
0"
"

R

%

"

#

9

<

&
% # W"
::O

%

%

% "

%

%

5[
:NU:

%

< %

%


%

"

%

#

5 :NUO

""

.

="
>

"

-2

%

""

"

7C:N]^J5:NUO

&ŝŐƵƌĞϭ͘ϭϬ͗DŽĚğůĞĚĞƐĠĚŝŵĞŶƚĂƚŝŽŶĞŶŵĂƐƐĞ͘>ĞĚĠƉƀƚĞƐƚĨŽƌŵĠƉƌĞƐƋƵĞŝŶƐƚĂŶƚĂŶĠŵĞŶƚĞƚĐŽŶƐĞƌǀĞ
ƵŶĞƐƚƌƵĐƚƵƌĞŝŶƚĞƌŶĞƐĞŵďůĂďůĞăĐĞůůĞĚĞů͛ĠĐŽƵůĞŵĞŶƚ

+
-

"

"

"

"

""

"

-

W

:NNJ0
.\C

/OO/

#

<

< 1
%

*

"

%

%

-
0
.
\C
/OO/1

"
<
%

#

2
:9K
%

"

K

"
%

<
-

2 G E"
)
-
5" 5C
:N]]J
5" 5C
:NU:J [ E
5
K 7
:N]NJ "


9C
0 :NU^J + "
E

0F 0
.
/OO'J@
3 "
E
5
/O:'
/9

<
"

W"
:N

%
% =">

%
%

<

% ?
0

3 )
+
:NNO
+
7"3 -1!? [
5
:NN:

"
"

ͳǤͶǤʹǤʹǤ ±
?

%

E

< "
%

W"
::: 9
<

%

%

<

9%

<

"
"

&

%

%

< %

=SURJUHVVLYH DJJUDGDWLRQ> 0
.
\C
:NN/
%

%-
W
:N^^""

"

"
%

"

&
:NÔ ) "" %
=

"

"
"
>

%

) ""

"

9

"" "
%

%
-

G
F)"
/O:/

@
/O:O


&ŝŐƵƌĞϭ͘ϭϭ͗DŽĚğůĞĚĞƐĠĚŝŵĞŶƚĂƚŝŽŶƉĂƌĂŐŐƌĂĚĂƚŝŽŶƉƌŽŐƌĞƐƐŝǀĞ͘>ĞĚĠƉƀƚƐĞĨŽƌŵĞƉƌŽŐƌĞƐƐŝǀĞŵĞŶƚ
ĚĞƉƵŝƐůĂďĂƐĞ͘>͛ĠƉĂŝƐƐĞƵƌĂŝŶƐŝƋƵĞůĂƐƚƌƵĐƚƵƌĞŝŶƚĞƌŶĞĚƵĚĠƉƀƚŶĞƐŽŶƚƉĂƐĨŽƌĐĠŵĞŶƚƌĞƉƌĠƐĞŶƚĂƚŝǀĞƐ
ĚĞĐĞůůĞƐĚĞů͛ĠĐŽƵůĞŵĞŶƚ͘
9"" "
%

+
-

<

H<

%

"

-

G

%

<

-

""

"

%

"

"

7

%

-

-

% %

"

%

"

<
7

%

"

"

%
%

9
"

"

.
"

#

&

"

<

-

<
?

< E

%

""

-
"

0
.
\C
/OO/""

"

"

%

%
"


#

G

&

"

K

=F
F . K
>

<

0
.
\C
/OO/

"
.

%

"

"

"
%

#

X
/O:/

<

"""
%

""

%

-

%

%

%

0

<

"

&
"
%
<

ͳǤͷǤ ±Ø

<

.

%
S : C

. "

"h

±Ø±
9
<

"

.

S:

%

""

%

" "

<

"

- <

"

W"
::/J 5
K 7
:N]N 9 "

<

K

)Ȱg:!/
!!

OM
O/M

"
K ɐȰ̱:MJV

W
/OOO
-

=EODVWV>


&ŝŐƵƌĞϭ͘ϭϮ͗ZĞůĂƚŝŽŶĞŶƚƌĞůĞƐ
ĚĠƉƀƚƐĚĞƐĐŽƵƌĂŶƚƐĚĞĚĞŶƐŝƚĠ
ƉǇƌŽĐůĂƐƚŝƋƵĞƐĞƚůĂ
ƚŽƉŽŐƌĂƉŚŝĞ͘;ĂͿ>ĞƐĚĠƉƀƚƐĚĞ
ĚĞĚĠĨĞƌůĂŶƚĞƐƉǇƌŽĐůĂƐƚŝƋƵĞƐ
ͨĚƌĂƉĞŶƚͩůĂƚŽƉŽŐƌĂƉŚŝĞ͕ůĞƵƌ
ĠƉĂŝƐƐĞƵƌĂƵŐŵĞŶƚĞĚĂŶƐůĞƐ
ĐƌĞƵǆĞƚĚŝŵŝŶƵĞĂƵŶŝǀĞĂƵĚĞƐ
ƌĞůŝĞĨƐ͘;ďͿ>ĞƐĚĠƉƀƚƐ
Ě͛ĠĐŽƵůĞŵĞŶƚƐƉǇƌŽĐůĂƐƚŝƋƵĞƐ
ĚĞƉĞƚŝƚƐǀŽůƵŵĞƐƐŽŶƚ
ĐŽŶĐĞŶƚƌĠƐŵĂũŽƌŝƚĂŝƌĞŵĞŶƚ
ĚĂŶƐůĞƐǀĂůůĠĞƐ͘;ĐͿĠƉƀƚƐ
Ě͛ĠĐŽƵůĞŵĞŶƚƐƉǇƌŽĐůĂƐƚŝƋƵĞƐ
ĚĞŐƌĂŶĚǀŽůƵŵĞƋƵŝƉĞƵǀĞŶƚ
ĚĠďŽƌĚĞƌĚĞƐǀĂůůĠĞƐĞƚ
ƌĞŵŽĚĞůĞƌůĂƚŽƉŽŐƌĂƉŚŝĞ͘
9

<

%
E %

.

W"
::
%
"

%

!!

5
K 7

:N]NJ W
7 C
:NU'

%

"

%

"

%

VK
K L /OO^ 9

<

"

"

.

"

%

"
<

%

7 C

:N]J
:NN:
%

-"
G
:NU'J5":NU]


&ŝŐƵƌĞϭ͘ϭϯ͗ĠƉƀƚĚĞͨďĂƐĞƐƵƌŐĞͩĂǀĞĐƐƚƌĂƚŝĨŝĐĂƚŝŽŶƐĞŶƚƌĞĐƌŽŝƐĠĞƐ͘<ŽŬŽƌĂƚĞƌ͕,ĂǁĂŢ͕
ƚĂƚƐhŶŝƐ͘;ƉŚŽƚŽDĂƚƚWĂƚƌŝĐŬͿ͘
±Øǯ±

<

-2
?K
:NN: c
:NN

-2 9 :NN 7C

:NN]

-2 3" 0

/O:'

"

"

`

<

.

• 9
<

:]OO S ɏ S /]OO C" !J 0F G
F
/O:M
% %
<

%

• 9
<

"

%

/OOSɏS:OOOC"!J0FG
F
/O:M

"
" %

.

• 9
<

.

%

%

:OOOSɏS/OOOC"
E

%

=!&%
>

-

"


0
:NN/JW

/OOO
9
<

.

%

%

- <

<

"
E%

%

#
W"
::/ 7

"

K

<

#

H

%

I < W"
::'

""

c
:NNJ 9

/OO]J0
/O:'J0FG
F/O:M9
"

.

%

"

%

%

"

- "

:NNU 7

K

K
!

<
%

"

7C
:NN]J0FG
F/O:M

&ŝŐƵƌĞϭ͘ϭϰ͗;ĂͿDŽƌƉŚŽůŽŐŝĞƐĞŶůŽďĞƐĞƚůĞǀĠĞƐƉŽƵƌĚĞƐĚĠƉƀƚƐĚ͛ĠĐŽƵůĞŵĞŶƚƐĚĞƉŽŶĐĞƐĞƚĐĞŶĚƌĞůŽƌƐ
ĚĞů͛ĠƌƵƉƚŝŽŶDŽŶƚ^ƚ,ĞůĞŶƐ;ϭϵϴϬͿ͘;ďͿůŽďĞƐĚĂŶƐůĞƐĚĠƉƀƚƐĚ͛ĠĐŽƵůĞŵĞŶƚƐĚĞĐĞŶĚƌĞĞƚƉŽŶĐĞƐĚĞ
ů͛ĠƌƵƉƚŝŽŶĚƵ>ĂƐĐĂƌ;ϭϵϵϯͿ͖;&ĠůŝǆĞƚdŚŽŵĂƐ͖ϮϬϬϰͿ͘
9
< "

%

%
&

W"
::MJc

:NNJ :NNUJ
@

/O::J 0F G
F /O:M 9

%
#

?

%
"

%

"

%

"


)"
/O:O7

J

<=
>R

&

%

<
=

>

- "

-<

"

9
<

K

- < %

=
>
-

%

3U\VPDWLFDOO\ -RLQWHG %ORFV

"

W
:N]'

%

"

<

"
K

%"
;ĂͿ ;ďͿ

&ŝŐƵƌĞϭ͘ϭϱ͗;ĂͿĠƉƀƚŵĂƐƐŝĨĚ͛ƵŶĠĐŽƵůĞŵĞŶƚ
ƉǇƌŽĐůĂƐƚŝƋƵĞ͕dƵƚƵƉĂĐĂ͕WĠƌŽƵƉŚŽƚŽ;͘ŚĠĚĞǀŝůůĞͿ͘;ďͿ
ĠƉƀƚĚ͛ƵŶĠĐŽƵůĞŵĞŶƚĚĞďůŽĐƐĞƚĐĞŶĚƌĞƐĂƵDĠƌĂƉŝ
;ƉŚŽƚŽ^ǇůǀĂŝŶŚĂƌďŽŶŶŝĞƌͿ͘
ͳǤ͸Ǥ ±
±±Ø
9
"

%

%

"

%

W"
::^J @""


:N//J X 7 :N^:J :NNUJ 0
. \C
/OO/ 9
"

%

.

<

!&%

-2
J
7C:NU/J5C
:NUM
%
#

%:M/ 7
:NÔJ X 7 :N^:J R
X
/OO'

% "

"

<
%
%

D
E

&ŝŐƵƌĞϭ͘ϭϲ͗/ŐŶŝŵďƌŝƚĞĨŽƌŵĠĞůŽƌƐĚĞů͛ĠƌƵƉƚŝŽŶĚƵEŽǀĂƌƵƉƚĂ͕ϭϵϭϮ͖sdd^͕;ƉŚŽƚŽtĞƐ,ŝůĚƌĞƚŚͿ͘;ĂͿǀƵĞ
ĂĠƌŝĞŶŶĞĚĞůĂƐƵƌĨĂĐĞďŝĞŶƉůĂŶĞĚƵĚĠƉƀƚ͖;ďͿsƵĞĞŶĐŽƵƉĞĚƵĚĠƉƀƚ͘
9

%

%"

"

7C
:N]

=


>

5
5C
:NU/
5C

:NUJ W"
::] 7

<

9
:

%

"

%

7C
:N]
#

5C

:NU:J55C
:NU/
9
/

-

9

/ %

"

"

%

"

%

9

"

%

"

"" % "
"

"

"
%

7%"
9 :NUUJ @. 3 /OOM G!

9
/

&

"

"

%

%

"
=FRDUVHWDLOJUDGLQJ>

A
""

1

-
"

"

%

-
<

"

"
!
"

E

7C
:N]
"

=

>E

""
%

<

"%

X
/O:/%
9
"

=
>
#

%

"

"
!"

"
J
-2"

@
7C:N]^J"

X

)
-
5":NU:J03 1!?5[
/OO


"

%

K

"
G i )
-
0
. \C
:NN]J "

d"K )
- !`jk
K 0
. /OOM

#

"

%

& 0
.\C
/OO/J
!`jk
K 0
. /OOMJ 5
/O:'

H

<

%
"

:NNU

"

-

"

%

="
>

"

"
#

%

%
7C:NU/J5C
:NUM

&ŝŐƵƌĞϭ͘ϭϳ͗^ĠƋƵĞŶĐĞƐƚƌĂƚŝŐƌĂƉŚŝƋƵĞƚŚĠŽƌŝƋƵĞĚ͛ƵŶĞŝŐŶŝŵďƌŝƚĞăƵŶŝƚĠƐƐƚĂŶĚĂƌĚƐ͘ĚƌŽŝƚĞ͕ĚĠƚĂŝůƐƐƵƌ
ƵŶĞƵŶŝƚĠĚ͛ĠĐŽƵůĞŵĞŶƚƐƚĂŶĚĂƌĚ͖tŝůƐŽŶ;ϭϵϴϲͿ͘


- %
- %

%
#

"

!

K

%"

""

"B
"K %

#

-
% I < %

9
#

%
<

% "
"K

! "

"

W

/OOO9

-"

.

%

-

!<
±

"
"

%
%

"

""

""

0
. \C
:NN/J W
:NNUJ W
:NNN 9
<

#

H<

"%

5 5"
:NU:J 0
. \C
:NN/J 5 /O:O
%

< %

< W
:NNUJR
X
/OO'
ǣ
9
"

%
#

"

=DVSHFW UDWLR> J 5C

:NUOJ 5C
:NU

.

<

%
#

<
W"
::U

"

"

- "
2

[GXE ["G
XE"

9GXE 9FG
XE"


&ŝŐƵƌĞϭ͘ϭϴ͗;ĂͿZĂƉƉŽƌƚĚĞĨŽƌŵĞ;ͬ,ͿĚĞƐĚĠƉƀƚƐĚ͛ĠĐŽƵůĞŵĞŶƚƐƉǇƌŽĐůĂƐƚŝƋƵĞƐĞŶĨŽŶĐƚŝŽŶĚĞůĞƵƌǀŽůƵŵĞ͘
D^,͗DŽŶƚ^ƚ,ĞůĞŶƐ͕sdd^͗sĂůůĞǇŽĨƚŚĞdĞŶdŚŽƵƐĂŶĚ^ŵŽŬĞƐ͘;ďͿĠƚĞƌŵŝŶĂƚŝŽŶĚĞĞŶĨŽŶĐƚŝŽŶĚĞů͛ĂŝƌĞ
ĚƵĚĠƉƀƚ;ǀŽŝƌƚĞǆƚĞƉŽƵƌĚĠƚĂŝůƐͿ͘ϭĐŽƌƌĞƐƉŽŶĚăůĂƐŽŵŵĞĚĞƐĂŝƌĞƐĚĞƚŽƵƚĞƐůĞƐǌŽŶĞƐĞŶŵĂƌƌŽŶ͘
9

[GXE

"

%

"

!!

%
&

%

%

%

%
9
"

03
̱:H/MO

%

-

V337GC ̱:H'OO

:NN:
+ +
:H:/OOSS:H'UOJW"
::N

"

&ŝŐƵƌĞϭ͘ϭϵ͗ĂƌƚĞĚĞƌĠƉĂƌƚŝƚŝŽŶĚĞƐĚĠƉƀƚƐĚ͛ƵŶĞŝŐŶŝŵďƌŝƚĞ;ĂͿăĨŽƌƚƌĂƉƉŽƌƚĚĞĨŽƌŵĞ;,Z/͕WŝŶĂƚƵďŽ͕
ϭϵϵϭ͕^ĐŽƚƚĞƚĂů͘ϭϵϵϳͿ͕WŚŝůŝƉƉŝŶĞƐĞƚ;ďͿăĨĂŝďůĞƌĂƉƉŽƌƚĚĞĨŽƌŵĞ;>Z/͕dĂƵƉŽ͕ϭϴϲ͕EŽƵǀĞůůĞĠůĂŶĚĞ͕
tŝůƐŽŶϭϵϴϱͿ͘


9
9GXE I

+

"

< "

%

"

"

"

H%
9
<

%
=V
.+
E"
>
5C

:NUO

.

S :
%

#
=E"
V

>
#

%

<

W"
:/O

.
"
%

"
%

a :OO H
K

C

%
J
[
:NN^ "

3 0

K

%

"

&

<

"
%

"

X 0C5:NN^

"
%

%

%

"
X
C
L
" :H/OOOO S S :H'OOO 7
C @

:NNMJ+
7"3
̱:H:OOOO0
:NN/

"

< a :M!/O %

C
% K

C

%

&

%

1

%

X
/O:M9

9GXE
&

"

3
`%

d
̱ :H:OOOOO 5C
:NUJ 5 :NUMJ W"
::NJ

%
-

.

<
9
"

[GXE
9GXE

"

%

%

H

K

%


&ŝŐƵƌĞϭ͘ϮϬ͗ĠƉƀƚƐĐĂƌĂĐƚĠƌŝƐƚŝƋƵĞƐĚĞƐŝŐŶŝŵďƌŝƚĞƐăĨĂŝďůĞƌĂƉƉŽƌƚĚĞĨŽƌŵĞ;>Z/Ϳ͗ĠƉĂŝƐĞƚ
ŵĂƐƐŝĨĚĂŶƐůĞƐǀĂůůĠĞƐ;sĂůůĞǇWŽŶĚĞĚ͖sWͿĞƚĨŝŶĞƚƉůƵƐŽƵŵŽŝŶƐƐƚƌĂƚŝĨŝĠƐƵƌůĞƐĐƌġƚĞƐ;/ŐŶŝŵďƌŝƚĞ
sĞŶĞĞƌĞƉŽƐŝƚ͖/sͿ͖Ě͛ĂƉƌğƐtĂůŬĞƌĞƚĂů͘;ϭϵϴϬͿ͘
±

<

%

"

"
2
-

:MOO

"

3 `%

d
5 :NUM :OOO "

E
W

:NNJÔO"
E$ cC.
:N]'/OO!'OO

+ 7
:NN^

%

-
-
:

"

)
7 :N]]J 7C
:N]UJ
5":NU]

%

%
W"
:/:

%

%
#

"

JK = 8 b H / %
J "%

K

8
%

7

%

"

3

:MOO'O
C
%

% :]O H 5 :NUM

.
"

%

!%

"

-

=7

>

%ÔO

OO \
:N^^J5":NU] /9

W"
:/: J W

:NN

E

-


&ŝŐƵƌĞϭ͘Ϯϭ͗,ǇƉŽƚŚğƐĞƐƉŽƵƌůĞĨƌĂŶĐŚŝƐƐĞŵĞŶƚĚ͛ƵŶŽďƐƚĂĐůĞƉĂƌƵŶĠĐŽƵůĞŵĞŶƚƉǇƌŽĐůĂƐƚŝƋƵĞ͘;ĂͿ
ů͛ĠĐŽƵůĞŵĞŶƚăƵŶĞĠŶĞƌŐŝĞĐŝŶĠƚŝƋƵĞƐƵĨĨŝƐĂŶƚĞ͕ĐĞƋƵŝŝŵƉůŝƋƵĞƵŶĞƚƌğƐĨŽƌƚĞǀŝƚĞƐƐĞ͘;ďͿů͛ĠĐŽƵůĞŵĞŶƚĞƐƚ
ƐƵĨĨŝƐĂŵŵĞŶƚĠƉĂŝƐƉŽƵƌƋƵ͛ƵŶĞƉĂƌƚŝĞƉƵŝƐƐĞĨƌĂŶĐŚŝƌů͛ŽďƐƚĂĐůĞ͘
ͳǤ͹Ǥ ±
9
+
%

-
l+H/

%
l+

K

/ +

=
>
W"
://9

"

%
%
=
> 7
%
:NÛ

.
W
:N]'

"

.

%

%

l[H9aO:
""

.

"

"

:NNU1
%

.

.

" %
.

a :C "

l[H9 SOO/
""

-

%

"


&ŝŐƵƌĞϭ͘ϮϮ͗DŽďŝůŝƚĠĚĞĚŝĨĨĠƌĞŶƚƐƚǇƉĞƐĚ͛ĠĐŽƵůĞŵĞŶƚƐĞǆƉƌŝŵĠĞƉĂƌůĞƌĂƉƉŽƌƚ,ͬ>͘;ĂͿ'ƌĂƉŚŝƋƵĞĚ͛ĂƉƌğƐ
^ƉĂƌŬƐ;ϭϵϳϲͿăƉĂƌƚŝƌĚĞƐĚŽŶŶĠĞƐĚĞ&ƌĂŶĐŝƐĞƚĂů͘;ϭϵϳϰͿ͕,Ɛƺ;ϭϵϳϱͿĞƚ^ƉĂƌŬƐ;ϭϵϳϰͿ͖WĨƐсĠĐŽƵůĞŵĞŶƚƐ
ƉǇƌŽĐůĂƐƚŝƋƵĞƐ͘;ďͿWƌŝŶĐŝƉĞĚĞůĂĚĠƚĞƌŵŝŶĂƚŝŽŶĚĞůĂŵŽďŝůŝƚĠăƉĂƌƚŝƌĚƵĚĠƉƀƚĚ͛ƵŶĠĐŽƵůĞŵĞŶƚ͘
9

- 2 :

"K

7C :N]^J 5 :NUO &

%J /

"

%

"
"

\
:N^^J 7
%
:N^^J 7
%
:NÛ

@

G%/OO]J
01%
m/OOUJ


&

)"

.
/O:OJ W
/O:'J 0

/O:'J '

"

%

-

"
)

:N]NJ )
:NN^J )
/OOJ M

"

"

%

+
/OO^J
9
!@

/OO]

E

+H/

"

! %%

C !

%
7C :N]^ 9
" "

"

%
"

%

%

""

"

+H/

%
.

"

"

[
:NNU
%

%

+H/ %

%

%

E

$H9

%

G
%

&
9
[
:NNUJ

:NNNJW

/OOO7
9&

/OON
"

%

%

A

"

<

-"

2 :
"

"

%

<
+H/
/

"

<
%
9 G
%

""

%

"%
9

%

-

.

-

_"

-

%

%

%
2[GXE
9GXE

#

&KDSLWUH)OXLGLVDWLRQG¶XQV\VWqPHJD]SDUWLFXOHV

!

" #$

"

.

%

9

"

-

"

%

"

.

9

.
""

%

ʹǤͳǤ ǣ
9

"

.

.

.

&&
"

9

:N// 5C

"K

"K
.
5

/OOO

#

:NO

:N'O %

W )OXLG
FDWDO\WLFFUDFNLQJ "
.

"

#

G
:NU

9
F
@

&
-
"K

%

"K % #

- %

9
F
:N'NJ $"
:NUOJ 5

/OOO E "

"

%

5

/OOOJ[/O:OG

:N'O

%

%

&

9

%
#

.

W

&

"K

9

"

"K

%

9

-

"K


-

"

%

"

.

H

"
@
:NU^

"K

%

"

"

"K

#

X
/OOU
ʹǤʹǤ ǯ
ʹǤʹǤͳǤ
9 "

"K

-

"

3
K" 3
K" :N'J @

/O:O 9

"

9

-
%

%

%

%

W"
/:
%

#
-

"K

%

-
%

9
%

%
-

%#

=
>R

"K

=>
"

= >
%

.

-

&

E

#

"

&

%

""


&ŝŐƵƌĞϮ͘ϭ͗&ŽƌĐĞƐĂŐŝƐƐĂŶƚƐƵƌƵŶĞƉĂƌƚŝĐƵůĞĞŶŵŽƵǀĞŵĞŶƚƌĞůĂƚŝĨĂǀĞĐƵŶŐĂǌĂŵďŝĂŶƚ͘>ĞƉŽŝĚƐ
ĂƉƉĂƌĞŶƚĐŽƌƌĞƐƉŽŶĚăůĂĚŝĨĨĠƌĞŶĐĞĞŶƚƌĞůĞƉŽŝĚƐĞƚůĂƉŽƵƐƐĠĞĚ͛ƌĐŚŝŵğĚĞ͕ĐĞƚƚĞĚĞƌŶŝğƌĞĠƚĂŶƚ
ŶĠŐůŝŐĞĂďůĞĚĂŶƐůĞĐĂƐĚ͛ƵŶŐĂǌ͘
9

%

-
%

%-
. :UM^
-

%

8%
"

"
+

"

ο

2 8 ‫ ן‬ȟ3H+ 9

"

A
"

9

%

%

"

"

-
%

%

#
%

%

-

c" /OO

%

0C
:N//\K
. :N/]0C

+
:N/U
:N]1

%-

-

1" :NM/

"

2
Δ3 8 : − ε b ρ 8 b : − ε
= :MO + :]M I /:
+ Gb ε
G ε

"
%
- "

%
8 %

8g 4H$ %
4

$

G

ɂ
% .

ȡI
%

"

GVY

.#

-

X
/OOU
9 1"

-

%

"

X
.
0C

:N//2
ρ I 8G
5H S = //
: − ε
7
"

5HSS:!:O1" /:

!\K
. @
:NU^J c" /OOJ X
/OOU 9

%

0C
!+
:N/U

"

X
a :OOO!/OOO R

"

%

%

%

%
8UHO ߝ 8L8S%
8LOD%

8L 8ߝ
8S
%

.

%

%
#

"%

%

@
:NU^
ʹǤʹǤʹǤ±ȋǡȌ
9
"

"

%
%

%

1"&
%
"

-

2
3,QWHUVWLWLHOOH = 3/LWKRVWDWLTXH = n ρ S : − ε + ρ I ε oJ+ /
%
J "%

+
"
9
"

= >

.
%
.

=EXON
GHQVLW\>

%

=%

> 8PI W"
// +

%

SMOOP

8PI

1"
"

2

Δ3G /ε
8 PI = /'
:MOμ : − ε / +

#

Δ3
8 PI = N /M
μ+
N

ε G /
N= /^
:MO : − ε /

%

"K 1

+

.

OM8PI

7%

"K
8PI

.
"
G%

-

"

ɂ "

%

/'9
-
"

X !dC
/]

! A

.

!

-

ɂǡ

%

8
%

8W

8
= ε Q /]
8W
%
Q
-

X
. 8W
%

'^M

"

/'
"

- %
%
&

%

"K

%

=PLQLPXP
EXEEOLQJIOXLGL]DWLRQª8PE

"K

"

3
.
$
:NM/

-
"K

X
/OOU


&ŝŐƵƌĞ Ϯ͘Ϯ͗ ĠǀŽůƵƚŝŽŶ ƚŚĠŽƌŝƋƵĞ ĚĞ ůĂ ƉƌĞƐƐŝŽŶ ŝŶƚĞƌƐƚŝƚŝĞůůĞ ĞŶ ĨŽŶĐƚŝŽŶ ĚĞ ůĂ ǀŝƚĞƐƐĞ ĚƵ ĨůƵŝĚĞ ƉŽƵƌ ĚĞƐ
ƉĂƌƚŝĐƵůĞƐĚĞŐƌŽƵƉĞ͕ƉŽƵƌĚĞƐůŝƚĐŽŵƉŽƐĠƐĚ͛ƵŶĞƐĞƵůĞƚĂŝůůĞĚĞƉĂƌƚŝĐƵůĞƐ;ŵŽŶŽĚŝƐƉĞƌƐĞͿŽƵĚĞƉůƵƐŝĞƵƌƐ
ƚĂŝůůĞƐĚĞƉĂƌƚŝĐƵůĞƐ;ƉŽůǇĚŝƐƉĞƌƐĞͿ͘ŶǀĞƌƚĞƐƚƌĞƉƌĠƐĞŶƚĠůĞĐŽŵƉŽƌƚĞŵĞŶƚĚ͛ƵŶůŝƚŵŽŶŽĚŝƐƉĞƌƐĞĂǀĞĐƐŽŝƚƵŶ
ůĠŐĞƌĞĨĨĞƚĚĞĐŽŚĠƐŝŽŶĚĞƐƉĂƌƚŝĐƵůĞƐƐŽŝƚƵŶĞĨĨĞƚĚĞĨƌŝĐƚŝŽŶůŝĠĂƵǆďŽƌĚƐĚƵĚŝƐƉŽƐŝƚŝĨ͘hŵĨĐŽƌƌĞƐƉŽŶĚăůĂ
ǀŝƚĞƐƐĞŵŝŶŝŵƵŵĚĞĨůƵŝĚŝƐĂƚŝŽŶĞƚhĐĨăůĂǀŝƚĞƐƐĞĐŽŵƉůğƚĞĚĞĨůƵŝĚŝƐĂƚŝŽŶ;ǀŽŝƌƚĞǆƚĞƉŽƵƌĚĠƚĂŝůƐͿ͘
E
-

-

-

- "

9

%
5

9
- "

"K

X

+

:NNOJX

:NN9

5 :N^N

.

-

2
"K !#

9 %

"

"
%

%

-

.
% #
%

&

%

3
%

"

8ɂ

8H

-
"
1

%
8ɂ a 8H

-

"

5 :N^NJ W @:NU'Jc
:NN^

"

"
0
%.
-"

%

8PIg8PE
R

"

H

-

%
%

"

%

"

%

"K
L01 W"

// E
8PI
%

"

"

LG

1 +

#

%
"

%
%

"K

LG1

W"
//1
%

.

%

%

"
%

9
-

%

"K
%

8PI

%

8FI @
:NU^
X

/O:O
%
8PI

%NOe

-
!!

8PI

8PE
7 %

"K
%

%

! % #

&

=

> =>

.

"

% "
1

#

#

%

"

%

&

G%
c
:NU/
8 aa 8PE

""

-

%
#

%
"

9

#

.

"

%

9

%
%

"K

&

%


%

#
%

&

"
!

ʹǤʹǤ͵Ǥ±
@
:N]

%
%

ɏSɏJ"

!
W"
/

&ŝŐƵƌĞϮ͘ϯ͗ůĂƐƐŝĨŝĐĂƚŝŽŶĚĞƐƉĂƌƚŝĐƵůĞƐĞŶĨŽŶĐƚŝŽŶĚƵĐŽŵƉŽƌƚĞŵĞŶƚĚĞĨůƵŝĚŝƐĂƚŝŽŶ͖Ě͛ĂƉƌğƐ'ĞůĚĂƌƚ;ϭϵϳϯͿ͘
*URXSH$$HUDWDEOH

"

H

R 8PI

"

H
"

-
9
- %

%
&%

8PE

"
W"
/'9
-
%

"K

%

8PEH8PI G
@
:NUOG

8PE
" 8PI
- -

9

"K

9

-

%

"K


%

"
X
/OOUR

%

%

%

%

&
%

W"
/MJ@

5":NUM
*URXSH&&RKHVLYH

1
"

S /OP

V
5
%

"

#

%

"

-

"

%

"

"
7

-

%
"

K

,
=
>

-

%%
"

W"

/'

"

"

E

-

"
%

%

"
)"
:NN'J0

/OOU

%

.
"
G
- 8PI

8PE
.

R

"K

%

-

W"
/MJ @

5":NUM
*URXSH%

%

"

G

"
G

""

9

%

"K
8PI 8PIg8PE

%

- "
W"
/' 9

"

%

%

-
"
G9
%

"K

"

%

R

"K
#

%

%

W"
/M
*URXSH'

+

"

H

"
0
"
%

8PI

%

9

&

-"
%

"
%

"K

"

"K

%

%

"

"

@
:NU^ G

%

"K

"

=">

#

&

"

.
)1
:N]'


&ŝŐƵƌĞϮ͘ϰ͗ĠǀŽůƵƚŝŽŶĚĞůĂƉƌĞƐƐŝŽŶĞƚĚĞů͛ĠƉĂŝƐƐĞƵƌĚƵůŝƚĞŶĨŽŶĐƚŝŽŶĚĞůĂǀŝƚĞƐƐĞĚĞŐĂǌ͕ƉŽƵƌĚĞƐƉĂƌƚŝĐƵůĞƐĚĞ
ŐƌŽƵƉĞ͕;ŽƵͿĞƚ͘hŵĨĞƚhŵďƐŽŶƚƌĞƐƉĞĐƚŝǀĞŵĞŶƚůĞƐǀŝƚĞƐƐĞƐŵŝŶŝŵƵŵĚĞĨůƵŝĚŝƐĂƚŝŽŶĞƚŵŝŶŝŵƵŵ
Ě͛ĂƉƉĂƌŝƚŝŽŶĚĞƐďƵůůĞƐ͕W>ĞƐƚůĂƉƌĞƐƐŝŽŶůŝƚŚŽƐƚĂƚŝƋƵĞĞƚ,ϬĞƐƚů͛ĠƉĂŝƐƐĞƵƌŝŶŝƚŝĂůĞĚƵůŝƚ͘


&ŝŐƵƌĞϮ͘ϱ͗ĚĠĨůƵŝĚŝƐĂƚŝŽŶĚĞůŝƚƐĚĞƉĂƌƚŝĐƵůĞƐĚĞŐƌŽƵƉĞ͕;ĞƚͿ͕Ğƚ͘dϬĐŽƌƌĞƐƉŽŶĚĂƵŵŽŵĞŶƚĂƵƋƵĞůůĂ
ƐŽƵƌĐĞĚĞĨůƵŝĚŝƐĂƚŝŽŶĞƐƚĐŽƵƉĠĞ͘DŽĚŝĨŝĠĚ͛ĂƉƌğƐ'ĞůĚĂƌƚĂŶĚtŽŶŐ;ϭϵϴϱͿĞƚƌƵŝƚƚĞƚĂů͘;ϮϬϬϳͿ͘

ʹǤ͵Ǥ ±
9

%

"K

D

%

W"
/^

&ŝŐƵƌĞϮ͘ϲ͗ĚŝĨĨƵƐŝŽŶĚĞůĂƉƌĞƐƐŝŽŶŝŶƚĞƌƐƚŝƚŝĞůůĞĞŶĨŽŶĐƚŝŽŶĚƵƚĞŵƉƐ͘ƚϬĐŽƌƌĞƐƉŽŶĚĂƵŵŽŵĞŶƚŽƶůĂƐŽƵƌĐĞ
ĚĞĨůƵŝĚŝƐĂƚŝŽŶĞƐƚƐƚŽƉƉĠĞ͘
7

%

/O

E%
:NN]

%

∂3 ∂/3
= ' / /U
∂W ∂]
%
3

W

]
%

'

.

N
'∝ /N
ε β
N
/^
Ⱦ

+ "K
Ⱦ g :H3 3

%

/U

WGLII

+H' +

N /^
R.
-
8S8PI

/U 1
%

- "

"
G

"

""

)

/O:/

%""

%

-

!

"

%
)

/O:/

"

WGLII % #
A %

=
"
>

Fߝ
: ∂ε
Fε = /:O
ε ∂3
"

9

"
%

.
'

!

"
#

%

E.

"

-G

)

/O:/
<

.

%

ʹǤͶǤ °
ʹǤͶǤͳǤ

%
%

%
%

-

9

%

%
5

/HVIRUFHVGHYDQGHU:DDOV

.

<
!
<

9

"
%

.-

"%

.-

G"

"

*

-
&

+

"
"
.

-

.

<

"

-

"

%

%

%

S'OP

"

"

#

%
" %

%
5
)"

X

+

:NNOJX

:NN

"

-

"
G

- "

8PI

8PE % ?
"
"

%
5

"

8PE

-
"
c

/OOM
)RUFHVpOHFWURVWDWLTXHV
9

%

"

"

% "

%

"%

7

"

%

"%

%

%

"

%

%
#

"

-

%

"

.

%
"

+

"

""

%

%
5 @
:N]J X

:NN:J
7
%

:NN]

)RUFHVFDSLOODLUHV
7

^M!UOe

-
%

%
9

R

-

%

-

%

"
@
c
:NN^9

"

%

%

&

!
%
"

9

%

""

-
-

-#

[.:N]U
ʹǤͶǤʹǤ°
+

%

9

-

!

-

.
"
G 8PI ് 8PE

.

/OO]
7DLOOHGHVSDUWLFXOHV

%

"

"

J
/M
/N
"

%

"
G % W"
/

%

8PI

8PE98PI

8PE
8PE8PI
"

"

--
"

G
@
:NUOJc

/OOM
0DVVHYROXPLTXHGHVSDUWLFXOHV
?
"

%

"
8PI @
:N]JG
@
:NUO1
%
8PE

%

8PI&,

-%

.

-"
7
[

:NUOJ @
:NU^J X
/OOU 9

"
G
0

%W"
/
0DVVHYROXPLTXHGXJD]
9
%
"K

%

8PI

"
G

X
.

5H SS :O
% "

:

1" /:

%

@
:NU^ 7

"

H

"

5Ha:O

%
"K

"

8PI@
G
:N]U

-

"

"K

%

%
"K%

8PE G
@
:NUOJc

/OOM
9LVFRVLWpGXJD]

%
% "K

%

?
"

%

"K

%

:

1" 9 %

=EXEEOLQJ>8PE
"

"

8PIG
8PEH8PI

"

%
% "K

- "
@
G
:N]UJ c

/OOM?

-

"

"
0

"
G
0
@
W"
/ ?

%
%"

.
' "

%
-
/N E
-

-

% "K

"
.

(IIHWVGHODWHPSpUDWXUH
?
"

"

% "K

%

%
E.


-

"

%

-"

"
8PEH8PI

"

8PI

"

9"

.
%
5 %

%

-
"
"

-

"
"
"

%

-
%

"K

-"
"

8PE
8PI

W
/OO/J@

W /OO] +
% "K

%

#

"

'
% /U
/N

&
9

/OO
(IIHWVGHOD3UHVVLRQ
?
"

"

%
"K

%

9%
"

-

/O]O
:/'/O C+
"

"

8PE

-
-

''P
::/P $ 5

:NU]

#

"

"K

- p
@
:NNM ?
"

"

"
G ! "
0 %

"

W @:NU'J+

:NNJp
@
:NNMG

.

"
0

%

"
G

%%
"

% "K
%

ʹǤͷǤ ±
ʹǤͷǤͳǤ±
9

=
>

.

%

<

/O

W

:N/ %
&

"

V

-)
W
"
V337
"
"K

+ +

:N] "

=

>

"
+

""

"K

X
. :NM'

%

= >

"

) 3"":NÔJ7C:N]^J5:NUOJ5:NU'JX

/OO/J
/OO]
9
-
.

%

.
%

%
<

-
"

.

- %

%

C
K

C

%

"

@J
/O::J C
E"
5
:NNMJ 3 5
5C
:NU/9"

K

%

%

"
@"":N//JX7:N^:""

&

<

?

%

"

=

> 7C
:N]J 5 5C
:NU/

"

%

<

"

%

%
:

""

7C:N]^J5
:NUO

ʹǤͷǤʹǤ±
+

%

.

-
!

*D]UHMHWpSDUOHVSDUWLFXOHV)LJXUHD

-

.

""

"K %

%
+

:N]

"

+

:NO/9
"K
"

H

7C :N]U

%

O/!OUe[/L%

%
%#

K

.

SMOO
P

"

<

"

"

%

%

<

%
&

"K 5":NU]

%

%

!

%
%

%
&
"

" %

"
3
:NN' 7

-
"K

%
#

W
/OO
,QJHVWLRQG¶DLUSDUFKXWHOLEUHGXPDWpULDXS\URFODVWLTXH)LJXUHF

%

"

#
"
%

K

%

7C5:N]^J5:NUOE

%

%

9"

"

%

`CW&:NN

&ŝŐƵƌĞ Ϯ͘ϳ͗ ǆĞŵƉůĞƐ ĚĞ ƐŽƵƌĐĞƐ ĚĞ

ŐĂǌĠǀŽƋƵĠĞƐƉŽƵƌůĂĨůƵŝĚŝƐĂƚŝŽŶĚĞƐ
ĠĐŽƵůĞŵĞŶƚƐ͗ ;ĂͿ ŐĂǌ ůŝďĠƌĠ ƉĂƌ
ĚŝĨĨƵƐŝŽŶ͕ ĂďƌĂƐŝŽŶ ŽƵ ĨƌĂŐŵĞŶƚĂƚŝŽŶ
ĚĞƐƉĂƌƚŝĐƵůĞƐĚĂŶƐƵŶĠĐŽƵůĞŵĞŶƚĚĞ
ƚǇƉĞƉŽŶĐĞƐĞƚĐĞŶĚƌĞƐ͖;ďͿĂŝƌŝŶŐĠƌĠ
ĂƵ ĨƌŽŶƚ Ě͛ĠĐŽƵůĞŵĞŶƚ Ğƚ ŐĂǌ ůŝďĠƌĠ
ƉĂƌ ů͛ĠǀĂƉŽƌĂƚŝŽŶ ĚƵ ƐƵďƐƚƌĂƚ;ŝĐŝ ĚĞ
ů͛ĞĂƵͿ͖ ;ĐͿ Ăŝƌ ƉŝĠŐĠ ůŽƌƐ ĚĞ ůĂ ĐŚƵƚĞ
ĞƚͬŽƵĚĞůĂĐŽŵƉĂĐƚŝŽŶĚƵŵĠůĂŶŐĞ͘
(FKDSSHPHQWGHJD]SDUFRPSDFWLRQGHO¶pFRXOHPHQW)LJXUHF
9% "K

"
%

"

"

KLQGHUHG
VHWWOLQJ:NNM

#

%

E%

/OOO

"

H


E .

%

.
"KH

,QJHVWLRQG¶DLUDXIURQWG¶pFRXOHPHQW)LJXUHE
9 "

#

"

) 3"":NÔJ5C

:NUOJ55C
:NU/

-

:

5 5C
:NU/ %
: 9 "

#

-

.

-
=
>
"

:[J5C

:NU:E

%"

-

"

-
&

"

!
%

:++%

%
"

%

9
"

!
.

+
.

%

%

.
\
%![
K

"

-

#

%

%

"

=!>

"

!
G

:N]:J7:NU^

_

_ ̱:OOOC"!" %

%

"

G
/OON

%

%
"

(YDSRUDWLRQGXVXEVWUDW)LJXUHE
R

"

%

"

K

%"

%

9%

%

5

:NUO

-

-

<.

%

-

2%

-

0

-

"K

[/L

"

%

-%

ʹǤͷǤ͵Ǥ±
9

%-

%

.

S:OOP

%

8PI

"

L

.

"

"
%

%
&

%

%

H

CH +

.
+

%

"K

K

-

#
%

"K

"

%

.
"

"

K
a O!'O e
%

X

/OOM
%

"
E

.
"

#

"
"
5:NUOJ

/OO'

/OO]

"

-

.

%

.
"
G
- "
%

%

SM H "

%

"
O!'OOO P G

%

.

9

.
"
%

"

!!
%
-
"K

L
/OO]X

/O:O
X
/O:/

%

"K

.

% D

"

9

&
<

% %

/U
/N

<

/^ G

"

/OO]

.

#

S :O!:/ /J 5 :NUOJ
/OO]J
0"
/O:/

%

W"
/UJ
X
/O:/
9

.

-

"
"

%

%

*

"1

%

"

K

@:NNU

C )" /OOU 1
%

1

"K

#
%

"

%

.
%

"
"

?

.

%

&

"


&ŝŐƵƌĞ Ϯ͘ϴ͗ ƚĞŵƉƐ ĚĞ ĚŝĨĨƵƐŝŽŶ ƚŚĠŽƌŝƋƵĞ͕ ƚĚŝĨĨ͕ ĞŶ ĨŽŶĐƚŝŽŶ ĚĞ ůĂ ƉĞƌŵĠĂďŝůŝƚĠ͕ Ŭ͕ Ğƚ ĚĞ ů͛ĠƉĂŝƐƐĞƵƌ ĚĞ
ů͛ĠĐŽƵůĞŵĞŶƚ͕Ś͕ƉŽƵƌƵŶŵĠůĂŶŐĞĚĞƉŽƌŽƐŝƚĠߝсϬ͘ϱĞƚĚĞŵĂƐƐĞǀŽůƵŵŝƋƵĞߩсϭϬϬϬŬŐŵͲϯ;ZŽĐŚĞϮϬϭϮͿ͘

.

7E
W

WGLII

G

GVY

H%

GPpGLDQ

+

ο "

+
3

+
3/

+
7

\Q
İ

% .

+
ȡP
%

C"!
ȡS
%

C"!
ȡI
%

C"!
ȡD
%

C"!
J "%

!/
N

/
'

.
/!:
9 %

4 %
!:
$

/
8 %

!:
8L %

!:
8S %

.

!:
8UHO %

%

!:
8W %

!:
8PI %

!:
8PE %

" !:
8FI %

!:
8ߝ %

"

!:
8H %

"

!:
5HS

X
.

ߚ

+!:
Fߝ
"

+!:

&KDSLWUH0pWKRGHVH[SpULPHQWDOHV

%
& '
9

-

G%

"

-
-

'
M
%

.

%-
#

=

"
>

9

%

^

9

9.
"

-

"

͵ǤͳǤ ±
9

.

.

"%

-

%

"

"

#

"

%
#

K

.

<

%

&

"

<

%

9
<

%

.

E

%

%

-

9
-

="
>

"

.

-

%

9

%

%

-

-

"%

-

"
=
C>

B

W"

:JX
:UN/

"

-
-

.

-

"
9

-

'
M

"


"

"

"

1

%

!
%

+

!

"

-

+

&ŝŐƵƌĞϯ͘ϭ͗ǆƉĠƌŝĞŶĐĞĚĞƚǇƉĞͨĚĂŵďƌĞĂŬͩ͘;ĂͿĞƵǆŵŝůŝĞƵǆĚĞĚĞŶƐŝƚĠƐĚŝĨĨĠƌĞŶƚĞƐƐŽŶƚƐĠƉĂƌĠƐƉĂƌƵŶĞ
ƉŽƌƚĞ͕ůĞŵŝůŝĞƵϭƉŽƵǀĂŶƚġƚƌĞƵŶĨůƵŝĚĞŽƵƵŶŵĠůĂŶŐĞƐŽůŝĚĞͬĨůƵŝĚĞĂůŽƌƐƋƵĞůĞŵŝůŝĞƵϮĞƐƚŶĠĐĞƐƐĂŝƌĞŵĞŶƚƵŶ
ĨůƵŝĚĞ͘;ďͿYƵĂŶĚůĂƉŽƌƚĞĞƐƚĞŶůĞǀĠĞ͕ůĞŵŝůŝĞƵϭ͕ƉůƵƐĚĞŶƐĞ͕ĨŽƌŵĞƵŶĠĐŽƵůĞŵĞŶƚƋƵŝƐĞƉƌŽƉĂŐĞăůĂďĂƐĞĚƵ
ŵŝůŝĞƵϮ͘
9
"

%

G%

-

%-

"

%

͵ǤʹǤ ǯ±±±±Ǽ
ǽ
͵ǤʹǤͳǤ ±±
͵ǤʹǤͳǤͳǤ
9

-

"

-

%
#

%

-

2

\

" :NM]J
7 0
:N]NJ X 7 :NU G

-

%

0
&:NÛ

%

-
-
0


"

%

8g /J¶K:H/ 9

/OO^%
K

J¶
"%

J¶gJ ȡGqȡOHȡGɏ
ɏ±

%

%

"
0
&:NÛ+

=!
> =

>

%

"
%

[
7:NUOJX
7 :NUJ

/OO] R

%

-

%

"
?

"

%

%

%

%

"

X7:NUG

.
\
%![
K
%

#

%

"

-
"
%

9
-

"

#

"

"
%

%

0
K

:NN

"

"

"

.

-/OO/
͵ǤʹǤͳǤʹǤ ȋǤǤǦ±Ȍ
9

-

"

-

"

-

"%

".

.
%

"
"

.
G

-

"

<

"

<

G%

-
%-

"

.

"

"+
%

"

-

"

%

"

-

"

"

%

R

-

"

%

"

"% G

#
+
:NNN

%

+
:NNN

#

E

%

8 g8H JK:H/KHKVWRS%
K

"

K
8; = β :
KVWRS θ
%
Ⱦ

-

KVWRSɅሻ
KVWRS
ș

"

-

9 "

#

"

"

%

+"

"

"

- !R
"

%

K

<

%

%

%

"

%

"

"9"

"

<
%

K
+

"

"

"

%
"

"

B

<

"

G .

"

<

"%

% %

%

%.

9

-

"

"

& #

<

"

9

-
-

%

" %

&#

9

/OO'9

<

+HJ:H/

9&

/OOM
&ŝŐƵƌĞ ϯ͘Ϯ͗ ĐŽŶĨŝŐƵƌĂƚŝŽŶƐ ƉĂƌƚŝĐƵůŝğƌĞƐ ĚĞ ƌƵƉƚƵƌĞ ĚĞ ďĂƌƌĂŐĞ͘ ;ĂͿ͕ ĐŽŶĨŝŐƵƌĂƚŝŽŶ ͨϮͩ ŽƵ ĠĐŽƵůĞŵĞŶƚ
ƵŶŝĚŝƌĞĐƚŝŽŶŶĞů͘;ďͿĐŽŶĨŝŐƵƌĂƚŝŽŶͨϯͩŽƵĂǆŝƐǇŵĠƚƌŝƋƵĞ͕ů͛ŝŶƐĞƌƚĐŽƌƌĞƐƉŽŶĚĂŶƚĂƵĚĠƉƀƚǀƵĚƵĚĞƐƐƵƐ͘
R

"
K

-

[ [ g[ILQDOH[W"
/

D

+H[

2
[; = λ D Q /

%

ɉ

- Q

-

9&

/OO'J
9

/OO'J9&

/OOMJ9

/OOMR

S:!:M[

D:7

a:!:M
-Q

:

"
-

"

=>

W"
/[

D:H/

#

"

""

9&

/OO'R

"=/>W"

/ 9

/OOM
9&

/OOM %
[

D

0 \
F
/OOM

-
D % %

"

Ga/
"

: [;∝ DOMMrOOM

"

/O [;‫ן‬DONrO:
""

""

"

"

%

9

/OOM

%

ɉ
/
%-

%

-

"

%

[ % "

#

+

-

-

X

/OOU

%

%

+ +HJ:H/
J+:H/

DS!'

%

.

<

%

"

"

R

"

"

Ʌ S
:^Q

<

"

"

"

Ɂ

"

ș )"

.
/O:OJW
/O:'Ǥ9

"

2
[ ILQDO /σ
=
+O δ − θ
%
ɐ

-

OM)"

.

/O:O1
%

%

"


Ʌa:^Q

"

<

%

W
/O:'
͵ǤʹǤͳǤ͵Ǥ ±

:
/

.
%
"

-

"
1

@
/OOO
%

%

"

K
%

-
"

"
=>

"

"=

"
>X

/OO/

"

%%

B

E

"

%

"

"
"
%X
/OOU9

#

=
>

"2

%

"%

X

/OO'+

-

&

"

9

%

J+ X

/OOU

"

JX

/O:O
"

"
%

"

& #
X

/O:O
%

!
#

"

%

<

"
%

%

9

.

X

/OOMJ /OO^

-

!

-"

"
G S
:OOP

"

-"
0
E
%

"

Į%

"

%

%

-

"
"

A

"

"

ǤG

s̱'Oe

"

"


Ƚ g :OOe

"

"
<
""

%
#

%

%- ""

.

%

"

"

@
/OOUJ/O:O

-
-

"

:]OQ E
%

%

%

" %

8 ̱ J+

9

-
-"
&
'e
#

%

൑]!UeX

/OO'9

%

<

""

-

%

9%

""<

#

-
R
-
-"
"

%

""
<

""

͵ǤʹǤʹǤ ±ǼǽȋͶͷȌ

-

-

'
M

9

-
-
"

"

"
.
/ W"
/
9
W"

% :
/O-:O
<

%
-

9

%

/

"
:O
"

"

?
<

-"

%

"9
<

"

/

%

%

%

- <

%

%&
K?


% %

9

.

%

̱O:

"

9

-&

%9

D

%9

D

%

%

.

/OP

"

#

"

D
9 %

%

<

rO/MH+

%

'9

"B

%

.
W 7G
%

/OOO"
H

:O/'-:O/'-
9
%

Ô
UOO"
H

.

2Ô
MOO "
H

/MO!MOO "
H

-"-

/OOOUOO"
H
"

#

/'
<

9%

"

' @

-

%
"

%

K

"

%

Ô "
H

%

/OOO "
H
"

%

"

.

%

"
+ W V
F

-

%

2
•

'

"

.

"

%

'M!NOP^

-

"

: 9

:O/OMO
UO

G/OM

-
-

•

M

%

O
OQ

"
-
9

:O MO NO
:MO

G 'OO

-
-


&ŝŐƵƌĞϯ͘ϯ͗ĚŝƐƉŽƐŝƚŝĨĞǆƉĠƌŝŵĞŶƚĂůĚĞƚǇƉĞƌƵƉƚƵƌĞĚĞďĂƌƌĂŐĞƵƚŝůŝƐĠĂƵĐŽƵƌƐĚĞĐĞƚƚĞƚŚğƐĞƉŽƵƌůĂŐĠŶĠƌĂƚŝŽŶĚĞƐĠĐŽƵůĞŵĞŶƚƐ;ǀŽŝƌĚĞƐĐƌŝƉƚŝŽŶĚĂŶƐůĞƚĞǆƚĞͿ͘

͵ǤʹǤ͵Ǥ
9

%

Gg'M!NOP

%

ɏSg/MOOC"!1

"
G

@

%
/G

'
M

UOP

.

%
<

]P9
%

%
:'MOേMOC"!

%

ɂǡgOMUrOO/

"
G

%

.

/OO] 9

%

.

@
/OOUJ /O:O

-
-

"
%

#

%

""

.

.
9

%"
#

"

%

"

%
%

-

"

%

"

/MO!'/MP "
0

%

%
? ̱/] HJX

/OOM

G%

"

-

"

!
1
%

"

'M!NO P

-

!

͵ǤʹǤͶǤ ±
+
%
"
%

%

"

OO-:O +

.

%

%

.

%

"

%

%

9

%

"

9
"

W"
'

'

%#
-

%

9

%
#

"

%
"

#

-

G
"
-

%

&"

K

%

" %

"

"

-

GN"

U"

3
:

&ŝŐƵƌĞϯ͘ϰ͗ƉŽƌƚŝŽŶĚĞďĂŶĚĞƉůĂƐƚŝƋƵĞƐƵƌůĂƋƵĞůůĞŽŶƚĠƚĠĐŽůůĠĞƐĚĞƐďŝůůĞƐĚĞǀĞƌƌĞĚĞϯŵŵĞƚƋƵŝ
ĞƐƚĞŶƐƵŝƚĞƉůĂĐĠĞăůĂďĂƐĞĚƵĐŚĞŶĂůƉŽƵƌƐĞƌǀŝƌĚĞƐƵďƐƚƌĂƚƌƵŐƵĞƵǆ͘
+

-

M

%

!

%

%

9%"

"

%

"

%

"

A

%"

"

"
#

:MO
-

dĂďůĞĂƵϯ͘ϭ͗ƌƵŐŽƐŝƚĠƐĚƵĨŽŶĚĚƵĐŚĞŶĂůĚĂŶƐůĞƐĞǆƉĠƌŝĞŶĐĞƐ;ŐĂŵŵĞĚĞƚĂŝůůĞƐĚĞƐƉĂƌƚŝĐƵůĞƐĐŽůůĠĞƐͿ͘
'ĂŵŵĞĚĞƚĂŝůůĞƐĚĞƐďŝůůĞƐĚƵƐƵďƐƚƌĂƚ;ђŵͿ
>ŝƐƐĞĂ͕ď ϰϱͲϵϬ ϭϱϬͲϮϱϬ ϮϱϬͲϰϮϱ ϰϮϱͲϲϬϬ ϲϬϬͲϴϬϬĂ ϭϱϬϬ ϯϬϬϬĂ͕ď ϲϬϬϬ
ĂZƵŐŽƐŝƚĠƉŽƵƌůĞƐƋƵĞůůĞƐĚĞƐŵĞƐƵƌĞƐĚĞƉƌĞƐƐŝŽŶŝŶƚĞƌƐƚŝƚŝĞůůĞƐŽŶƚĠƚĠĞĨĨĞĐƚƵĠĞƐ͘
ďZƵŐŽƐŝƚĠƵƚŝůŝƐĠĞĚĞŵĂŶŝğƌĞƐǇƐƚĠŵĂƚŝƋƵĞăĚŝĨĨĠƌĞŶƚĞŝŶĐůŝŶĂŝƐŽŶƐ;ĐŚĂƉŝƚƌĞϱͿ
͵ǤʹǤͷǤ ±
G%

-

.

&
-
W"
M

%

W"
M
7

-

-

:O
7

-

"
"

O
+

-

^C"

+a/O

"

-

K"

9

%

W"

-

%

%

-

%

-

-
?
g :' H a ? ̱ : H
-
%
!

%

©EXEEOLQJª
-
3H3/g:

"
%
%

%

-

" "

! -
&

% 9

%M

-

9

"
B

&ŝŐƵƌĞϯ͘ϱ͗;ĂͿĐŽůŽŶŶĞƉĞƌŵĞƚƚĂŶƚůĞƐĠĐŚĂŐĞĚĞƐƉĂƌƚŝĐƵůĞƐĂǀĂŶƚĐŚĂƋƵĞĞǆƉĠƌŝĞŶĐĞ͘;ďͿĐĂƐĞǆƚƌġŵĞ
Ě͛ĂŐƌĠŐĂƚŝŽŶƉŽƵƌĚĞƐďŝůůĞƐĚĞǀĞƌƌĞůĂŝƐƐĠĞƐĂƵƌĞƉŽƐƉĞŶĚĂŶƚƉůƵƐŝĞƵƌƐũŽƵƌƐĚĂŶƐĚĞƐĐŽŶĚŝƚŝŽŶƐŚƵŵŝĚĞƐ͘;ĐͿ
ďŝůůĞƐĚĞǀĞƌƌĞƐϰϱͲϵϬђŵĂƉƌğƐƐĠĐŚĂŐĞ͘

&"

"

%

W"
^G

-

"
-

%

"

."


&ŝŐƵƌĞϯ͘ϲ͗DĠƚŚŽĚĞĚĞŵĞƐƵƌĞĚĞƐĚŝƐƚĂŶĐĞƐĚĞƉĂƌĐŽƵƌƐĚĞƐĠĐŽƵůĞŵĞŶƚƐĚĂŶƐůĞĐŚĞŶĂů͘
͵ǤʹǤ͸Ǥ ±±
?
.

#

-
X

/OOU
X
/O:/

"

.

-
-

.
1

.

"

H
E%
:NN]
E%

"
/OO: X
/O:/

:O
S 3
/

P2

C"
C" G

0 C"

.

S!P g ]

.
9

X
/O:/

3

dĂďůĞĂƵϯ͘Ϯ͗ůŝƐƚĞĚĞƐƉĂƌĂŵğƚƌĞƐƵƚŝůŝƐĠƐƉŽƵƌů͛ĂŶĂůǇƐĞĚŝŵĞŶƐŝŽŶŶĞůůĞĞƚǀĂůĞƵƌƐƚǇƉŝƋƵĞƐĚĠƚĞƌŵŝŶĠĞƐƉŽƵƌĚĞƐ
ĠĐŽƵůĞŵĞŶƚƐŶĂƚƵƌĞůƐ;ZŽĐŚĞϮϬϭϮͿĞƚĞǆƉĠƌŝŵĞŶƚĂƵǆ;ZŽĐŚĞϮϬϭϮĞƚĐĞƚƚĞĠƚƵĚĞͿ͘
9DOHXUVW\SLTXHV
3DUDPqWUHVHWGLPHQVLRQV 6\PEROHV
1DWXUH /DERUDWRLUH

G /:O!M!M:O!' 'M!NO:O!^
)
%

C"!
ρS MOO!/MOO /MOO
+ İ O]!O' O'!O'M

)
%

C"! ρI ̱: :/
V .

+ ̱:O!M :U:O!M
G "%

!/ J NU: NU:
1

K :OO!M:O: :O!/!:O!:

/ :O!:O' OM!
V

!: 8IW M!O :!M
+
.
/ N :O!:/!:O!:O :::O!::

.
/!: ' :O!'!:O/ ̱:O!/
+

Q ș O!M O!O

1RPEUHGH0DVVH0D

ɂSɏߛG

ɂɏIߛG2
ε SρS
0D = '
ερ I
%
İS %

İSg:!İ ɀ
-

8IWHK +

.

-

%
"K

0D
"

:
"

%

1
*

%

0D

:

.

1RPEUHGH)URXGH)U
9

W

"%

2
8 IW
)U = M
JK:H /
R )U S :

)U a :

9

-
-

-

1RPEUHGH%DJQROG%D
9

0"

ɂɏSߛʹʹሻ

%

ɂρߛሻ
ε S ρ Sγ G /
%D = ^
ε
+ %D a /OO

"

"

%D S /OO

%

"
!%
-
-

%D
"

/OO

""

"
!%
-9

-

%D
"

/OO

a/OQ

!!
%

"

"
!%
-H

%
%D
1RPEUHGH'DUF\'D
9

.&

:

%

-

ρߛʹȀሻ

ɂɏSߛʹʹሻ2

'D = ]
Nε S ρ Sγ
'D

%

"G'D

1RPEUHGH)OXLGLVDWLRQ)O
9

%

NJ ɏSɏIɂɂI

J/
Nε S ρ S − ρ I
( J H /)
:H /
)O = U
ε

"

:
-

"

%

&

%

"

1RPEUHGH3UHVVLRQLQWHUVWLWLHOOH3U
9

"

/J:H/

K'

/ H J :H /
3U = N
K / H '
9

%

%

-

3U

:

:O!'
:O: 9

%

-
-

H"
<%

:

-

&

.
X

/O:/
1RPEUHGH6DYDJH6D
9

7%"
"

!

<

"
E

"

ɂɏSߛʹGʹ

"

ɂ ɏSɏIJK2
ρ S G /γ /
6D = :O
ρ S − ρ I JK

+

.

-

6D

:O!M:O!
N

"

"

"

6ḎO:
G

E%
:NN]+
-

0"

7%"
%DH6D

)F

"

%

: − ε ρ S − ρ I JK
)F = ::
ε γ
E

"

&

#

%
%

-

"

%

"

%

"

-

dĂďůĞĂƵϯ͘ϯEŽŵďƌĞƐƐĂŶƐĚŝŵĞŶƐŝŽŶĚĠĨŝŶŝƐƉĂƌZŽĐŚĞ;ϮϬϭϮ͖ǀŽŝƌƚĞǆƚĞƉŽƵƌĚĠƚĂŝůƐͿĂǀĞĐůĞƐŐĂŵŵĞƐĚĞǀĂůĞƵƌƐ
ĞƐƚŝŵĠĞƐƌĞƐƉĞĐƚŝǀĞŵĞŶƚƉŽƵƌůĞƐĠĐŽƵůĞŵĞŶƚƐƉǇƌŽĐůĂƐƚŝƋƵĞƐ;ŶĂƚƵƌĞͿĞƚƉŽƵƌůĞƐĠĐŽƵůĞŵĞŶƚƐĞǆƉĠƌŝŵĞŶƚĂƵǆ
;ůĂďŽƌĂƚŽŝƌĞŐĠŶĠƌĠƐĂǀĞĐůĞĚŝƐƉŽƐŝƚŝĨĚĞƌƵƉƚƵƌĞĚĞďĂƌƌĂŐĞƵƚŝůŝƐĠĚĂŶƐĐĞƚƚĞƚŚğƐĞ;&ŝŐƵƌĞϯ͘ϯͿ͘
1RPEUHVVDQV 9DOHXUVW\SLTXHV
V\PEROHV VLJQLILFDWLRQ
GLPHQVLRQ
1DWXUH /DERUDWRLUH
R
%

`

)
0D :O/!:O :O

`

W
)U :^!O /^!/] 1

X"
!%
-
`

0" %D :OO!:O/ :O/

E
H

`

. 'D :O:!:O' :O/

`

V

)O :O!]!:O! :O!/

V

`

!' : O
3U :O !:O :O
%

-

9

`

7%"
6D :O!N!:O!U :O!^!:O!M
""

2"

͵Ǥ͵Ǥ ±ȋ͸Ȍ
͵Ǥ͵ǤͳǤ ±

&ŝŐƵƌĞϯ͘ϳ͗^ĐŚĠŵĂĚƵĚŝƐƉŽƐŝƚŝĨƵƚŝůŝƐĠƉŽƵƌůĞƐĞǆƉĠƌŝĞŶĐĞƐĚ͛ĂƵƚŽͲĨůƵŝĚŝƐĂƚŝŽŶƉĂƌĐŚƵƚĞĚĞƉĂƌƚŝĐƵůĞƐ;ĐŚĂƉŝƚƌĞ
ϲͿ͘ >ĞƐ ĞǆƉĠƌŝĞŶĐĞƐ ƐŽŶƚ ĨŝůŵĠĞƐ ĞŶ ĨĂĐĞ ĚĞ ůĂ ƉůĂƋƵĞ ͘ >ĞƐ ĚŝĨĨĠƌĞŶƚĞƐ ŚĂƵƚĞƵƌƐ ĂƵǆƋƵĞůůĞƐ ƉĞƵǀĞŶƚ ġƚƌĞ
ƉŽƐŝƚŝŽŶŶĠƐůĞƐĐĂƉƚĞƵƌƐ;ďůĞƵͿĞƚůĂƉŽƌƚĞ;ƌŽƵŐĞͿƐŽŶƚŝŶĚŝƋƵĠĞƐ͘

9

-

-

W"

]

-"
:U
:^M

%

//-//

/O-/O

9

-

%

:/M
%

%
M &
"

9

<

W"
]

:MO^M
:OO 9
.
%

:

O-/:

/O
NO

0

#

-"
G

%

%

-

?

:N-OM-OM
%

-

#

!

-

*
%

%

+%

"#

"

'

W"
U

"

%

-

S:OOP
-"

/OOP

^P%

-


&ŝŐƵƌĞϯ͘ϴ͗ĚŝƐƉŽƐŝƚŝŽŶĚĞƐĨĞƵŝůůĞƐĚĞƉůĂƐƚŝƋƵĞĚĂŶƐůĂĐŽůŽŶŶĞĚĞŵĂŶŝğƌĞăůŝŵŝƚĞƌůĂĨƌŝĐƚŝŽŶ
ĞŶƚƌĞůĞƐƉĂƌƚŝĐƵůĞƐĞƚůĂƉŽƌƚĞĞƚĨĂĐŝůŝƚĞƌů͛ĞŶůğǀĞŵĞŶƚĚĞĐĞƚƚĞĚĞƌŶŝğƌĞ͘
9

-

%

%
tg/MOOC"!

"

O!'MPOOOP 3
'J
W"
N1
^C"
^C" േOO:C"

%

-

%:OÔ ?

%

%

"

dĂďůĞĂƵϯ͘ϰ͗EĂƚƵƌĞĞƚƚĂŝůůĞĚĞƐƉĂƌƚŝĐƵůĞƐƵƚŝůŝƐĠĞƐƉŽƵƌůĞƐĞǆƉĠƌŝĞŶĐĞƐĚĞĐŚƵƚĞĚĞůŝƚƐƉĂƌƚŝĐƵůĞƐ͘
ŝůůĞƐĚĞǀĞƌƌĞ /ŐŶŝŵďƌŝƚĞ
'ĂŵŵĞĚĞ
ϬͲϰϱ ϰϱͲϵϬ ϱϯͲϭϬϲ ϭϬϲͲϮϭϮ ϮϱϬͲϰϮϱ ϲϬϬͲϴϬϬ ϯϬϬϬ ϬͲϰϬϬϬ
ƚĂŝůůĞ;ђŵ
ĚŵĞĚŝĂŶ;ђŵͿ ϯϳ͘ϰϮ ϳϮ͘ϱϰ ϴϴ͘ϭϲ ϭϲϬ͘ϯϰ ϯϯϲ͘ϵϱ ϳϬϬ ϯϬϬϬ ̱ϲϯĂ

Ă
ǀĂůĞƵƌĞƐƚŝŵĠĞƉĂƌƌƵŝƚƚĞƚĂů͘ϮϬϬϳ

&ŝŐƵƌĞ ϯ͘ϵ͗ ƉŚŽƚŽƐ ĚĞƐ ĚŝĨĨĠƌĞŶƚƐ ƚǇƉĞƐ ĚĞ

ƉĂƌƚŝĐƵůĞƐƵƚŝůŝƐĠĞƐĚĂŶƐůĞƐĞǆƉĠƌŝĞŶĐĞƐ͗ďŝůůĞƐĚĞ
ǀĞƌƌĞĚĞ;ĂͿϬͲϰϱђŵ͕;ďͿϰϱͲϵϬђŵ͖;ĐͿϮϱϬͲϰϮϱ
ђŵĞƚ;ĚͿϯŵŵ͘;ĞͿĐŚĂŶƚŝůůŽŶĚ͛ŝŐŶŝŵďƌŝƚĞ͘
+

-

%

'M!NOP

MOQ
/OOQ%

%
!

-

%

%

"

"

`

)
W

' W"
N
J@
/OOUJ @
/OOUJ /O:O 9.
"

"

%

"

"

`


/OO]
#
3
'9"

- %#

S
^ P

.

"

-

%

"
@%9

"

"

O!'MP.

UOQ%

-

#

W"
N
͵Ǥ͵ǤʹǤ ±
9

/ONO

"
9

%

%

&

%

W"
U9

W"
M

% +

-

%

%#

%

O:Q

&
%
-

9

%%

.

r M!:OQ

-

.

9%

SO/

G

-

%

%

%

!

.

͵ǤͶǤ
͵ǤͶǤͳǤ ±
9

-.

-

K

E

=:M'`!OO:@!F

>9

.

"

9
.

-

"

-.

%

!

9

%


-

%9
%

%

%
#

%

"%
!
9

%

"
r:

̱r^NC+
͵ǤͶǤʹǤ

&ŝŐƵƌĞϯ͘ϭϬ͗ĚŝƐƉŽƐŝƚŝĨĚĞĐĂůŝďƌĂƚŝŽŶĚĞƐĐĂƉƚĞƵƌƐĚĞƉƌĞƐƐŝŽŶƉĂƌǀĂƌŝĂƚŝŽŶĚĞŚĂƵƚĞƵƌĚ͛ƵŶĞĐŽůŽŶŶĞĚ͛ĞĂƵ͘
9

]MOV rOO:V

%
"

+

< [g


%

M/O'OÔ
]O

W"

:OG

%

/M

O

%

%

%

%

,
%
"

%

-

-

9
%

"]O

"

%

^NC+

"

& :OO

%

:OC+9

"

%

%

%

3gO+

.

9

-

"B

3 ɏJ+%
ɏ
%

:OOOC"HJ "%

NU:H/
+

9
%

"

.

"
W"
::9

-

9 %

"B

%

3 = X − XEODQF &FDO :/
%
X

X
.

= >
&
%

-

&FDO

-

%9
"-

-

OM
O^V

r:MO+G

<

%


&ŝŐƵƌĞϯ͘ϭϭ͗ĞǆĞŵƉůĞĚĞĐŽƵƌďĞĚĞĐĂůŝďƌĂƚŝŽŶƉŽƵƌƵŶĐĂƉƚĞƵƌĚĞƉƌĞƐƐŝŽŶ͘>ĞĐŽĞĨĨŝĐŝĞŶƚ
ĚŝƌĞĐƚĞƵƌĚĞůĂĚƌŽŝƚĞ;ĞŶƌŽƵŐĞͿĐŽƌƌĞƐƉŽŶĚĂƵĐŽĞĨĨŝĐŝĞŶƚĚĞĐĂůŝďƌĂƚŝŽŶĚƵĐĂƉƚĞƵƌ͕ĐĂů͘
͵ǤͶǤ͵Ǥ ±±
G

-

"

%

-"

U
P %

#

%

"

"
"

ÔOP%

%

" W"
:/ +

%

W"

"

%

9

"

%

'
M

^

'OO[K:'O[K

%

:M

9

"

"

-

:'O [K M

^^[K'

+

-

^%

"

!

%
"

"
? = >

%

-

E%

%
.

"
%
"

#

-

%
SOMV

&ŝŐƵƌĞϯ͘ϭϮ͗;ĂͿĐĂƉƚĞƵƌĚĞƉƌĞƐƐŝŽŶĞƚŐĂŝŶĞƉƌŽƚĞĐƚƌŝĐĞ͘;ďͿƐĐŚĠŵĂĚƵĐĂƉƚĞƵƌĚĞƉƌĞƐƐŝŽŶŝŶƐĠƌĠĚĂŶƐƐĂ
ŐĂŝŶĞƉƌŽƚĞĐƚƌŝĐĞĞƚĚŝƐƉŽƐĠăůĂďĂƐĞĚƵĐŚĞŶĂů;&ŝŐƵƌĞϯ͘ϯͿŽƵĚĞůĂĐŽůŽŶŶĞ;&ŝŐƵƌĞϯ͘ϳͿ͘
͵ǤͷǤ
±
͵ǤͷǤͳǤ
°
͵
9

%

MOOP

"B
"
=)"@> W"
:

<

=@>

O

"

%
-"

͵ǤͷǤʹǤ ±
9

%-
9

9
+

/O:M+

"

%

9

%

%

!

:OOQ
%

:/

9

W"
:

&

!


%
#
&

.
?

%

.

"

K

G

.

.

K

"

%

%

%

G

.

""

#
"
%

&ŝŐƵƌĞϯ͘ϭϯ͗ƐĐŚĠŵĂĚĞƐĐƌŝƉƚŝĨĚĞů͛ŝŶƐƚƌƵŵĞŶƚƵƚŝůŝƐĠƉŽƵƌůĞƐŵĞƐƵƌĞƐĚĞŐƌĂŶƵůŽŵĠƚƌŝĞ;ŵŽƌƉŚŽůŽŐŝ'ϯͿ͕Ě͛ĂƉƌğƐ
>ĞŝďƌĂŶĚƚĂŶĚ>ĞWĞŶŶĞĐ;ϮϬϭϱͿ͘
͵ǤͷǤ͵Ǥ ±
0

%

K"
W"
:'
"
&

W"
:'

.

: W"
:'


&ŝŐƵƌĞϯ͘ϭϰ͗ƉŚŽƚŽƐĚĞƉĂƌƚŝĐƵůĞƐƉƌŝƐĞƐƉĂƌů͛ĂƉƉĂƌĞŝůƉŚŽƚŽĚƵ'ϯ͘;ĂͿƉůƵƐŝĞƵƌƐƉĂƌƚŝĐƵůĞƐĚĞůĂŐĂŵŵĞϰϱͲϵϬ
ђŵ͕;ďͿƉĂƌƚŝĐƵůĞĚĞůĂŐĂŵŵĞϬͲϰϱђŵ͘;ĐͿƉĂƌƚŝĐƵůĞĚĞůĂŐĂŵŵĞϮϱϬͲϰϮϱђŵŶŽŶƐƉŚĠƌŝƋƵĞ;ĚĠĨĂƵƚĚĞ
ĨĂďƌŝĐĂƚŝŽŶͿ͘;ĚͿWĂƌƚŝĐƵůĞƐŵĂůƐĠƉĂƌĠĞƐĞƚŶŽŶĚŝƐƚŝŶŐƵĠĞƐƉĂƌůĞŵŝĐƌŽƐĐŽƉĞ͕ĚĂŶƐůĂŐĂŵŵĞϮϱϬͲϰϮϱђŵ͘>ĞƐ
ĂƐƐŽĐŝĂƚŝŽŶƐĚĞĐĞĚĞƌŶŝĞƌƚǇƉĞŶĞƐŽŶƚƉĂƐƉƌŝƐĞƐĞŶĐŽŵƉƚĞƉŽƵƌů͛ĂŶĂůǇƐĞŐƌĂŶƵůŽŵĠƚƌŝƋƵĞ͘;ĞͿŝƌĐƵůĂƌŝƚĠĚĞƐ
ďŝůůĞƐĚĞϰϱͲϵϬђŵƉŽƵƌƵŶĠĐŚĂŶƚŝůůŽŶĚĞϭϬϮϯϭƉĂƌƚŝĐƵůĞƐ͘


%

' §G ·
9 = π ¨ ¸ :
©/¹
%
G

@ 9
%

%

%

%

/MOOC"!
9

GPpGLDQ

MOe

MOe

9

σ=
( GU'e − G:ê ) :'
/
%
GU'e

U'e

G:ê

:ê

E
"

%

ɐHGPpGLDQ+

9

-

"
"
G

"

%

H%
.
GVY
#

!

2
Σ[
G VY = :M
Σ [L H GL
%
GL

[L
"
%

.
#

@
:NU^

!

"

H%

"

%

%

"
G

%

GVY
GPpGLDQ

`%

%
%
"

+
"

-

O!'M P
-

%

-


.

- .
9

.

3
M

W"
:M
dĂďůĞĂƵ ϯ͘ϱ͗ ƌĠƐƵůƚĂƚƐ ĚĞ ů͛ĂŶĂůǇƐĞ ŐƌĂŶƵůŽŵĠƚƌŝƋƵĞ ĚĞƐ ďŝůůĞƐ ĚĞ ǀĞƌƌĞ ĂǀĞĐ ůĞ 'ϯ͘ >Ă ůŝŐŶĞ ͨŵŽǇĞŶŶĞ ĚƵ
ĚŝĂŵğƚƌĞŵĠĚŝĂŶͩƌĞƉƌĠƐĞŶƚĞůĞƐǀĂůĞƵƌƐƵƚŝůŝƐĠĞƐƉŽƵƌůĞƐĐĂůĐƵůƐŝŵƉůŝƋƵĂŶƚůĞĚŝĂŵğƚƌĞĚĞƐƉĂƌƚŝĐƵůĞƐĚĂŶƐůĞ
ĐŚĂƉŝƚƌĞϲĞƚƋƵŝƐŽŶƚĠŐĂůĞŵĞŶƚƉƌĠƐĞŶƚĠĞƐĚĂŶƐůĞƚĂďůĞĂƵϯ͘ϰ͘
ƚĂŝůůĞĚĞƐ
ƉĂƌƚŝĐƵůĞƐ ϬͲϰϱ ϰϱͲϵϬ ϱϯͲϭϬϲ ϭϬϲͲϮϭϮ ϮϱϬͲϰϮϱ
;ђŵͿ
ŶŽŵďƌĞĚĞ
ϭϰϵϴϬ ϴϰϴϱ ϭϬϮϯϭ ϯϴϰϰ ϮϰϲϬ ϯϳϭϯ ϱϴϴϯ ϭϱϲϴ Ϯϰϰϲ
ƉĂƌƚŝĐƵůĞƐ
ĚŵŽǇĞŶ;ђŵͿ ϯϭ͘ϯ ϲϮ͘ϲϴ ϲϰ͘ϰϲ ϴϯ͘ϲϭ ϴϯ͘ϯ ϭϰϳ͘ϴϮ ϭϯϰ Ϯϵϴ͘ϯϮ ϯϭϵ͘ϰϯ
ĚƐǀ;ђŵͿ ϯϰ͘ϴϭ ϲϵ͘ϵϵ ϲϵ͘ϱϰ ϴϲ͘ϲϮ ϴϴ͘ϴϰ ϭϱϵ͘ϳϭ ϭϱϮ͘ϰϱ ϯϯϵ͘ϳϱ ϯϯϯ͘ϭϭ
ĚƐǀŵŽǇĞŶ
ϯϰ͘ϴϭ ϲϵ͘ϳϳ ϴϳ͘ϳϯ ϭϱϲ͘Ϭϴ ϯϯϲ͘ϰϯ
;ђŵͿ
ĚŵĞĚŝĂŶ;ђŵͿ ϯϳ͘ϰϮ ϳϮ͘ϴϯ ϳϮ͘Ϯϱ ϴϳ͘ϯϱ ϴϴ͘ϵϲ ϭϲϯ͘Ϭϳ ϭϱϳ͘ϲϭ ϯϰϬ͘ϱϲ ϯϯϯ͘ϯϯ
ĚŵĞĚŝĂŶŵŽǇĞŶ
ϯϳ͘ϰϮ ϳϮ͘ϱϰ ϴϴ͘ϭϲ ϭϲϬ͘ϯϰ ϯϯϲ͘ϵϱ
;ђŵͿ
DĚʔ ϰ͘ϳϰ ϯ͘ϳϵ ϯ͘ϱ Ϯ͘ϲϰ ϭ͘ϱϳ
ʍ;ђŵͿ ϳ͘ϳϭ ϳ͘ϭϳ ϳ͘ϱϴ ϴ͘ϭϮ ϴ͘ϰϳ ϯϭ͘Ϯϲ ϯϭ͘ϴϴ ϯϵ͘ϳϴ ϯϯ͘ϱ
ʍŵŽǇĞŶ;ђŵͿ ϳ͘ϳϭ ϳ͘ϯϳ ϴ͘ϯ ϯϭ͘ϱϳ ϯϲ͘ϲϰ
ʍͬĚŵĞĚŝĂŶ Ϭ͘ϮϮ Ϭ͘ϭϭ Ϭ͘ϭϬ Ϭ͘ϮϬ Ϭ͘ϭϭ

&ŝŐƵƌĞϯ͘ϭϱ͗ƌĠƉĂƌƚŝƚŝŽŶ͕ĞŶйƉŽŝĚƐ͕ĚƵĚŝĂŵğƚƌĞĚĞƐƉĂƌƚŝĐƵůĞƐƉŽƵƌůĞƐĚŝĨĨĠƌĞŶƚĞƐŐĂŵŵĞƐĚĞďŝůůĞƐĚĞǀĞƌƌĞ
ƵƚŝůŝƐĠĞƐ͘

&KDSLWUH$XWRIOXLGLVDWLRQGHVpFRXOHPHQWVS\URFODVWLTXHVVHSURSDJHDQWVXUXQVXEVWUDWUXJXHX[

*
+,

"

"'

9

.

"

%

<

&%

%
%

%

%

%

.

"

"
-
<

%

.

%
K

%"

%

" 9

H

=
>%

0
:NN/J7C
:NN]JX

/O:J
0

/O:'J 0
/O:'

%
#

-2

%

%

/O::
9

"
-

#

"

.
+

%

-

.

&KDSLWUH$XWRIOXLGLVDWLRQGHVpFRXOHPHQWVS\URFODVWLTXHVVHSURSDJHDQWVXUXQVXEVWUDWUXJXHX[

"
%

'M!NO P

"

" .

"

&

^

"

`
% "

%

/: ) /O:'

%
$ @
. X

q71

2
³&KpGHYLOOH & 5RFKH 2 $XWRIOXLGL]DWLRQ RI S\URFODVWLF IORZV SURSDJDWLQJ RQ URXJK
VXEVWUDWHV DV VKRZQ E\ ODERUDWRU\ H[SHULPHQWV - *HRSK\V 5HV ± GRL
-%´

PUBLICATIONS
Journal of Geophysical Research: Solid Earth
RESEARCH ARTICLE Autofluidization of pyroclastic flows propagating
10.1002/2013JB010554
on rough substrates as shown
Key Points:
• Flows of fine particles on horizontal
by laboratory experiments
rough substrate can be autofluidized Corentin Chedeville1,2,3 and Olivier Roche1,2,3
• Autofluidization is due to air escape
from the substrate interstices 1
Laboratoire Magmas et Volcans, Clermont Université, Université Blaise Pascal, Clermont-Ferrand, France, 2CNRS, UMR
• Autofluidization is expected to cause
long runout of pyroclastic flows 6524, LMV, Clermont-Ferrand, France, 3IRD, R 163, LMV, Clermont-Ferrand, France
Supporting Information: Abstract This study investigates the influence of the substrate surface roughness on the emplacement
• Readme mechanisms of pyroclastic flows. We carried out laboratory experiments on gravitational flows generated
• Movie S1
• Movie S2 from the release of initially fluidized or nonfluidized columns of fine particles (diameter d = 0.08 mm) in a
• Movie S3 horizontal channel. The roughness of the channel base was uniform in each experiment, created by gluing
• Movie S4 particles of diameter d0 = 0.08 to 6 mm to the base. Other things being equal, the flow runout distance
increased with the channel base roughness (d0) to a maximum of about twice that of flows on a smooth
Correspondence to:
C. Chedeville, substrate when d0 = 1.5–3 mm, before decreasing moderately at higher roughness values of d0 = 6 mm. Long
c.chedeville@opgc.univ-bpclermont.fr runout originated mainly during the late stages of emplacement as flow deceleration was strongly reduced at
high substrate roughness. This was caused by (partial) autofluidization due to an upward air flux escaping
Citation: from the substrate interstices in which flow particles settled. Autofluidization was evidenced by high pore
Chedeville, C., and O. Roche (2014), fluid pressure measurements at the base of initially nonfluidized flows and also by reduced flow runout when
Autofluidization of pyroclastic flows the interstices were initially partially filled so that less air was available. Furthermore, the runout distance of
propagating on rough substrates as
shown by laboratory experiments, flows of large particles (d = 0.35 mm), which could not be fluidized by the ascending air flux, was independent
J. Geophys. Res. Solid Earth, 119, of the substrate roughness. This study suggests that autofluidization caused by air escape from the interstices
doi:10.1002/2013JB010554. of a rough substrate is one important mechanism to explain the common long runout distance of pyroclastic
flows even on subhorizontal topographies.
Received 24 JUL 2013
Accepted 21 FEB 2014
Accepted article online 19 MAR 2014
1. Introduction
Pyroclastic flows represent the dense gravity-driven mixture of gas and solid particles at the base of many
pyroclastic density currents [Cas and Wright, 1987; Druitt, 1998; Freundt et al., 2000]. They are commonly
generated from the gravitational collapse of an unstable eruptive column [e.g., Sparks and Wilson, 1976] or a
lava dome [e.g., Cole et al., 1998], but secondary pyroclastic flows may also result from detachment and rapid
sedimentation of the dilute surge component of a pyroclastic density current [Druitt et al., 2002]. The ability of
many pyroclastic flows to propagate on subhorizontal slopes over distances of several kilometers and the
density segregation of clasts in their deposits suggests that they are gas-fluidized mixtures [Sparks, 1976;
Wilson, 1980]. In this context, the emplacement mechanisms and the runout of pyroclastic flows are likely to
be controlled principally by the longevity of the gas sources and the process of pore fluid pressure diffusion
[Druitt et al., 2007; Roche et al., 2010; Roche, 2012]. The interstitial gas can be of primary (i.e., volcanic) or
external origin. In the latter case, Wilson [1980] proposed that air can be ingested at the flow front and that
gases can be released when a pyroclastic flow travels over water or burns vegetation.
A major issue related to the hazards posed by pyroclastic flows is that the flow runout distance is difficult to
predict, because it is controlled by a large number of flow properties (volume, velocity, gas content, and clast
size distribution) as well as by topography. The latter can be particularly critical because, in addition to the
slope variation, pyroclastic flows propagate on substrates that can have a highly variable surface roughness
(Figure 1), and this may be an important factor in controlling the emplacement mechanisms (see the recent
study of Roche et al. [2013] using an erodible granular substrate). The present study addresses this issue
through a series of laboratory experiments on granular flows propagating on a nonerodible substrate of
variable surface roughness. Previous experimental studies involved flows of particles of relatively large grain
size (d ~ 0.1–1 mm) and substrate roughness of the same order [Pouliquen, 1999; Goujon et al., 2003;
Lajeunesse et al., 2004; Lube et al., 2011]. In contrast, we consider here flows of finer particles and a large range
of substrate roughness. We investigate how the latter controls the emplacement mechanisms of the flows

Journal of Geophysical Research: Solid Earth 10.1002/2013JB010554
and their runout distance in particular. This is

ash cloud surge
achieved through dam break experiments on
(dilute and turbulent) initially fluidized or nonfluidized granular flows
propagating over a horizontal rough substrate,
thus extending earlier studies involving a
smooth substrate [e.g., Roche et al., 2006, 2008;
Girolami et al., 2008, 2010].
pyroclastic flow
(concentrated) 2. Experimental Methods
and Device
substrate 2.1. Principles of Pore Fluid Pressure
Generation and Diffusion
Figure 1. Sketch of a pyroclastic density current. A basal pyro-
High pore fluid pressure, causing fluidization of
clastic flow with high particle concentration propagates over an
pyroclastic flows through drastic reduction of
irregular rough substrate and is overridden by a dilute ash cloud
interparticle frictional forces, can be generated
surge. The dashed line represents the transition (sharp or diffuse)
between the flow and the surge components. by relative vertical motion between the
interstitial gas and the particles. This occurs
when an upward gas flux is provided and/or when the particles settle in dense mixtures. In laboratory
experiments, pore pressure is generated by injecting air at the base of a static and compacted granular
column (Figure 2). Full fluidization is achieved at the minimum fluidization velocity, Umf, when the drag force
caused by the interstitial gas flow counterbalances the apparent weight of the particles, so that the pore fluid
pressure is equal to the lithostatic pressure defined as
P ¼ ½ ρs εs þ ρf ð1 εs ÞgH; (1)
where ρs and ρf are the particle and gas density, respectively, εs is the particle concentration, g is the gravity
acceleration, and H is the bed height [Rhodes, 1998]. Note that as ρf < < ρs, then equation (1) can be simplified as
P ≈ ρs εs gH: (2)
If the gas flux is no longer supplied and there is no relative gas-particle motion (Figure 2c), the pore pressure
then decreases with time (t) according to a diffusion law described by
∂P ∂2 P
¼D 2; (3)
∂t ∂z
where z is the bed height and D is the hydraulic diffusion coefficient [Iverson, 1997]. Hence, the granular material
remains (partially) fluidized during the process of pore pressure diffusion, whose time scale (Δt) increases with
H2/D and with decreasing particle grain size (i.e., low D values). At the laboratory scale, for H of the order of
1–10 cm, Δt is typically a few seconds for beds of fine group A particles (grain size < ~0.1 mm) of Geldart’s
classification [Geldart, 1986] but is almost negligible for beds of larger particles (groups B–D). Note that initial gas
velocity greater than Umf causes homogeneous expansion of beds of group A particles, which notably increases
Δt compared to nonexpanded beds owing to particle settling [Montserrat et al., 2012; Roche, 2012].
Figure 2. (a) Generation of pore fluid pressure in a bed of particles fluidized by injecting air from below. (b) Pore pressure as
a function of the gas velocity, equal to the lithostatic pressure at Umf. (c) Diffusion of pore pressure if gas supply is no longer
provided, where Δt is the duration of pore pressure diffusion. This applies to flows in our experiments since no air flux is
provided from the channel base.

Figure 3. Experimental device consisting of a (fluidization) particle reservoir and a horizontal channel whose base is rough-
ened by glued beads (not to scale). The inset shows (a) glued beads alone and (b) with their interstices partly filled by either
PEG or fine beads (gray).
2.2. Experimental Device

We carried out dam break experiments in an apparatus consisting of a particle reservoir and a horizontal channel
whose base was covered with glued glass beads (Figure 3). Removal of the reservoir gate by means of a
counterweight permitted the rapid release of 6 kg of particles, corresponding to an initial column height of
H = 20 cm, which led to a gravitational granular flow in the channel. The flows were filmed with a high-speed
video camera at rates of 60 frames/s for investigating the general flow emplacement and up to 3800 frames/s for
detailed views.
The flow particles were nearly spherical glass beads with a density of 2500 kg m3. Most experiments were
carried out with fine group A particles of mean grain size d = 0.08 mm (range 0.06–0.09 mm) to respect
dynamic similarity with pyroclastic flows [Roche, 2012]. Before each run, the particles were dried for 10 min in
a fluidization rig in order to avoid possible minor cohesion effects due to ambient moisture. In some
experiments, particles were fluidized in the reservoir by injected air through a basal porous plate. The mean
air flow velocity was 14 mm s1, above Umf of 8.3 mm s1 [Roche et al., 2006], which led to homogeneous,
slightly (~5%) expanded beds. As no gas flux was provided from the channel base, the subsequent flow then
defluidized through pore pressure diffusion during propagation (cf. Figure 2c). Complementary experiments
were carried out with group B particles of mean grain size d = 0.35 mm (range 0.25–0.43 mm). As initial
fluidization of such large particles has a negligible influence on flow emplacement at laboratory scale
because of nearly instantaneous pore pressure diffusion [Roche et al., 2006], only initially nonfluidized flows
were considered for these particles.
Variation of the surface roughness of the channel base was obtained by changing the mean grain size of the
glued beads (d0 = 0.08, 0.2, 0.35, 0.5, 0.7, 1.5, 3.0, and 6.0 mm). These particles were randomly distributed, though
hexagonal arrangements were often observed for the largest ones. The term roughness used hereafter
designates the size of the glued beads. Additional experiments with flows propagating on a smooth substrate
(without glued beads) were also carried out for comparison. For each substrate roughness, experiments were
repeated at least 7 times for initially fluidized flows and at least 3 times for nonfluidized flows. In complementary
experiments involving a substrate roughness of 3 mm, the interstices between the glued beads were filled to
about 80% of their height (Figure 3). This was achieved by adding either (i) fine beads of 0.08 mm, leveled with a
brush before each experiment, or (ii) molten polyethylene glycol (PEG) that solidified at ambient temperature.
Sensors were used to measure pore fluid pressure at the base of the flows (see Roche et al. [2010] for technical
details). They were covered by a 36 μm grid that prevented contact with the flow particles but permitted pore
pressure transmission to the sensory membrane. In such experiments, 12 kg of particles were released from

the reservoir (H ~ 40 cm) in order to obtain thick flows and hence

large pressure signals. Pressure measurements were made for both
initially fluidized and nonfluidized flows on both smooth and rough
substrates (d0 = 3.0 mm). In the latter case, the upper part of the
sensor was positioned either at the channel base (below the glued
beads) or the top of the glued beads (Figure 4).
2.3. Scaling Issues

A detailed dimensional analysis of the experiments presented in
Roche [2012] accounts for the degree of dynamic similarity between
pyroclastic flows and their laboratory analogs. The choice of particle
type (i.e., the groups in Geldart’s classification) is crucial regarding
Figure 4. Pressure sensors positioned (a) at dynamic similarity between both types of flows. Here fine particles
the base of the channel, with beads glued of group A are representative of the ash matrix of many pyroclastic
on the grid covering the sensor (method 1), flows [Druitt et al., 2007] and are used to investigate their
and (b) at the same level as the top of the emplacement. With such fine particles, dimensionless numbers for
beads glued on the channel base (method 2).
both types of flows generally have close values, or at least indicate
the same physical regimes. Experimental flows, however, are likely
to defluidize through pore pressure diffusion at a relatively faster rate than most pyroclastic flows, because
they are thinner (see equation (3)). In this paper we add the substrate roughness (d0) as another physical
parameter to the analysis of Roche [2012], which leads us to define an additional dimensionless number. We
define the roughness number
h
Ro ¼ ; (4)
d0
where h is the flow thickness. Comparing the Ro values in both nature and experiments, our analog
flows of typical thickness h ~ 3 cm simulate propagation of 1 m or 10 m thick pyroclastic flows on a
substrate having a roughness of ~0.003–0.2 m or ~0.03–2 m, respectively, which may be representative
of most natural configurations.
3. Results
3.1. Flow Runout Distance
The runout distance of both initially fluidized or nonfluidized flows of particles of size d = 0.08 mm increased
with the substrate roughness until the latter was 1.5–3 mm, at which point the runout was about twice that
on a smooth substrate (Figure 5). The runout then decreased by ~15–25% at the highest level of roughness
Figure 5. Runout distance of granular flows as a function of the substrate surface roughness (d0). Results of complementary
experiments at a roughness of 3 mm with interstices partially filled are also shown for both initially fluidized and nonflui-
dized flows of particles of size d = 0.08 mm. (a) Raw data. (b) Runout distance (xf ) normalized to the runout of flows on a
smooth substrate (xf0). Error bars can be smaller than symbols.

Figure 6. Front kinematic data of initially fluidized (filled symbols) and nonfluidized (open symbols) flows of particles of
0.08 mm on substrates of different roughness. The front position normalized to the initial column height (H) is repre-
1/2
sented as a function of time normalized by t0 = (H/g) . The inset represents a detailed view of the data for initially fluidized
flows on smooth or 3 mm rough substrates during the intermediate constant velocity phase.
investigated (d0 = 6 mm). In experiments with a roughness of 3 mm, partial filling of the interstices led to
shorter runout distances compared to a substrate with uniform-sized glued beads alone. The runout decrease
was ~20% or ~20–35% when interstices were filled with PEG or fine glass beads, respectively. In the latter
case, results were less reproducible probably because of slight changes in amount of filling between the
experiments due to the protocol used (leveling with a brush).
In contrast to fine particle flows, flows of beads of size d = 0.35 mm had runouts almost independent of
the roughness but ~20% shorter than on a smooth substrate. However, flow runout for the highest level
of roughness investigated (d0 = 6 mm) was only slightly shorter than for smoother substrates.
3.2. Flow Front Kinematics
The kinematics of flows of fine particles (d = 0.08 mm) were investigated as a function of the substrate roughness.
In Figure 6, the normalized flow front position in the channel, x/H, is shown as a function of the normalized time,
t/(H/g)1/2, where the initial column height H is the relevant length scale parameter and g is the gravitational
acceleration [Roche et al., 2008]. Earlier works showed that dam break granular flows on smooth substrates
propagated in three distinct phases according to these dimensionless parameters [e.g., Lajeunesse et al., 2004;
Roche et al., 2008]. The flows first accelerated, then propagated at nearly constant front velocity for most of their
emplacement, and finally decelerated. For flows of fine group A particles that were initially fluidized, the constant
front velocity Uf ~ (2gH)1/2 was equal to that of inertial flows of single phase fluids, reflecting sustained high pore
fluid pressure due to slow diffusion [Roche et al., 2008]. Deceleration began when pore pressure had decreased
sufficiently for particle interactions to significantly dissipate the flow energy [Roche et al., 2010].
In our experiments, the front kinematics of both initially fluidized and nonfluidized flows of fines were
independent of the substrate roughness until transition to the decelerating phase, which occurred at
t/(H/g)1/2 ~ 3.75–4 and ~2–2.5, respectively (Figure 6). However, the constant velocity of initially fluidized flows,
during the second phase of emplacement, increased slightly with the substrate roughness. For instance, it varied
from 2.15 m s1 for a smooth substrate to 2.40 m s1 for a 3 mm rough substrate (Figure 6, inset). The main
differences in flow kinematics arose at a late stage, with lower flow deceleration as the substrate roughness
increased, hence leading to longer flow runout and duration. The latter two reached a maximum at a roughness
between 1.5 and 3 mm and were about twice that for a smooth substrate. For the highest roughness investigated
(d0 = 6 mm), the front velocity, runout, and duration of both initially fluidized and nonfluidized flows were slightly
lower than for 1.5 mm and 3 mm rough substrates but remained greater than for smoother substrates.

Figure 7. Morphology of initially (a–d) fluidized and (e–h) nonfluidized flows of particles of size d = 0.08 mm, at different times
of propagation. According to Figure 6, t/t0 is the normalized time. The flow head becomes humped at the beginning of the
deceleration phase (Figures 7b and 7f), and both its length and height increase with the substrate roughness. At later stages,
the flow head stretches horizontally until the flow comes to halt (Figures 7d and 7h).
3.3. Flow Morphology

Both initially fluidized and nonfluidized flows of fine particles (d = 0.08 mm) consisted of a sliding head (i.e.,
with nonzero basal velocity) and a body, at the base of which deposition occurred incrementally [Roche et al.,
2010; Girolami et al., 2010; Roche, 2012]. As in the case of the front kinematics, the flow morphology did not
depend on the substrate roughness at early stages of propagation (Figure 7, see Movies S1–S4 in the
supporting information). However, during the decelerating phase, both the height and length of the flow
head increased with the roughness, while the shape of the flow body was little affected. During the final
Figure 8. Deposits of initially nonfluidized flows (a) on smooth substrate and (b) on 3 mm rough substrate, and of initially
fluidized flows on (c) smooth substrate and (d) 3 mm rough substrate. The flow runout distance in each case is xf.

Figure 9. Snapshots of an initially fluidized flow of particles of size d = 0.08 mm propagating over a 3 mm rough substrate,
at 100 cm from the reservoir. (a) Front arrival at time t0, (b) t = t0+ 0.008 s, (c) t = t0 + 0.016 s, and (d) t = t0 + 0.09 s. Larger
black particles (d = 0.2 mm) were used as markers at a proportion of 1–2% of the total particle weight. Snapshots of a flow of
particles of size d = 0.35 mm propagating on a 3 mm rough substrate at 25 cm from the gate. (e) Front arrival time at t0, (f)
t = t0 + 0.008 s, (g) t = t0 + 0.016 s, and (h) t = t0 + 0.065 s.
stage, the main body rapidly stopped moving but the head still propagated and stretched horizontally. The
amount of stretching increased with the substrate roughness, thus leading to thin (<1 cm) distal deposits
(Figure 8, see Movies S3 and S4 in the supporting information) with an almost horizontal (<1°) upper surface
and causing the long flow runout distances reported in Figures 5 and 6.
3.4. Flow-Substrate Interaction
The mechanisms of interaction between the granular flows and their substrate were investigated using high-
speed videos (Figure 9). The head of both initially fluidized and nonfluidized flows of fine particles
(d = 0.08 mm) slid over the substrate. Following this phase, dense clusters of flow particles dropped into the
substrate interstices. Settling velocities were measured either at the base of the clusters or by tracing the
markers (larger black particles) inside these clusters, with no significant difference between these two
methods. We obtained settling velocities of ~80–200 mm s1 for a 3 mm rough substrate, with the highest
values approaching the free-fall velocity for a height of 3 mm. Note that observations were made at the

Figure 10. Basal pore pressure P normalized to the lithostatic pressure PL = ρsεsgh, with ρsεs the bulk density of the flow
3
(~1450 kg/m ), g the gravitational acceleration, and h the flow thickness over the sensor, as a function of time. Time zero
corresponds to flow initiation at gate opening, and pore pressure was measured at various distances from the gate
according to methods 1 and 2 shown in Figure 4. Results are shown for flows propagating on smooth substrate and which
are (a) initially fluidized or (b) nonfluidized flows, flows propagating on 3 mm rough substrate and which are initially fluidized,
with (c) method 1 and (d) method 2, or nonfluidized, with (e) method 1 and (f) method 2. Note that the normalization of
negative pore pressure values to the lithostatic pressure has no physical meaning here; these values are only preserved to
show that underpressure occurs at the flow front.
channel side and that particle settling velocities could have been different for an unconfined substrate. The
duration of this phase of infilling of the interstices increased with the roughness. For example, it varied from
~0.06 s to ~0.11 s for a roughness of 0.7 mm to 3 mm, respectively, which corresponded to a travel distance of
~12.5 to 22 cm for initially fluidized flows and of ~7 to 15 cm for nonfluidized ones. For flows of particles of
size d = 0.35 mm, intense collisions with the substrate were observed at the flow base. For a substrate
roughness ≥ 1.5 mm, particles dropped as dilute clusters into the interstices immediately after the passage of
the flow front, at velocities of ~80–140 mm s1 (Figures 9e–9h).
3.5. Basal Pore Pressure

Results of pore pressure measurements at the base of flows of particle size d = 0.08 mm are presented in
Figure 10. The data for flows on a smooth substrate are similar to those reported in earlier studies [Roche et al.,
2010; Roche, 2012] and are shown here for comparison with cases involving rough substrates. The sliding
head of the flows generated under pressure proportional to the front velocity, whereas pore pressure
transmitted by the flow body depended on the initial conditions [Roche, 2012; Roche et al., 2013]. In the case
of initially fluidized flows, pore pressure was high during the early stages (i.e., close to the reservoir), because

Figure 11. Schematic representation of the autofluidization process for a flow of fine particles propagating on a rough sub-
strate. The air escapes from the substrate interstices, into which flow particles settle, and percolates upward through the flow.
it was advected rapidly from the reservoir and had not yet diffused much. This explains why pore pressure
could be higher than the lithostatic pressure for a given flow depth [Roche et al., 2010]. Pressure then
decreased through diffusion until the flow was totally defluidized. In contrast, initially nonfluidized flows did
not generate overpressure. An exception occurred at 10 cm from the reservoir where up to ~16% of the
particle weight was supported for a very short duration (Figure 10b), possibly due to some compaction
causing differential air-particle motion as the material was released from the reservoir [Roche et al., 2010].
On a 3 mm rough substrate, initially fluidized flows generated pressure signals similar to those on a smooth
base during the first stage (at 10–20 cm from the reservoir). However, during the intermediate stage,
overpressure supported ~70–90% of the particles’ weight at distances of up to 80 cm (Figures 10c and 10d). In
comparison, less than 50% of the particle weight was supported at that stage on a smooth substrate. Another
notable difference was that underpressure was small and of short duration for the rough substrate, particularly
when the sensors were located below the glued beads (see method 1 in Figure 4). For initially nonfluidized
flows (Figures 10e and 10f), the pressure signals differed fundamentally from those observed for a smooth
substrate. Overpressure was high at all stages of the flow, with at least ~70% of the particles’ weight supported.
For instance, nearly full bed support was detected even at a distance of 80 cm from the reservoir, whereas the
flow runout was 130 cm. Note that the duration of overpressure increased significantly between the early (at 10
and 20 cm) and intermediate to late stages (at 50 and 80 cm), showing that an increasing part of the flow had
high pore pressure. Furthermore, at late stages, this duration was almost equal to that of initially fluidized flows.
4. Discussion
4.1. Flow Mechanisms in Experiments
Our results showed that the runout of both initially fluidized and nonfluidized flows of group A fine particles
(d = 0.08 mm), propagating on a rough substrate, was increased by up to a factor of ~2 in comparison with a
smooth base. Kinematic and morphological data revealed that the increase of runout was acquired mostly during
the decelerating phase, as more material was transferred to the flow head, which then spread considerably to
form a thin and elongated deposit (Figures 7 and 8). High pore pressure at the base of initially nonfluidized flows
of fines on a rough substrate proved that autofluidization occurred during emplacement. We propose that this
was caused by flow particles dropping into the substrate interstices, causing an upward flux of escaping air
(Figure 11). Even if the flow particles did not fall entirely into the substrate interstices at the lowest roughness
investigated (as low as d0 = 0.08 mm), the small amount of air evacuated could have caused the slight runout
increase compared to a smooth substrate. This autofluidization mechanism is different from that proposed by
Bareschino et al. [2008] for flows generated in a rotating drum for which “plunging breakers” collapsed at the
front and forced the ambient air to percolate through the flow. In our experiments, the velocity of the upward
air flux can be estimated from volume conservation consideration. We consider the volume flux
Q ¼ SU; (5)
where U is the velocity and S the cross-sectional area. With Qp and Qia the volume flux of the settling particles
and of the interstitial air, respectively, and stating Qp = Qia then

εp Up ¼ 1 εp Uia ; (6)
with εp the particle volume fraction. The interstitial air velocity, however, has to be converted into a superficial
air velocity (Usa) in order to be compared with Umf, so that

Usa ¼ 1 εp Uia : (7)

Combining equations (6) and (7) leads to

εp ¼ Usa =Up : (8)
In this context, settling clusters (at velocities of

~80–200 mm s1) with particle volume fractions
as low as ~0.1 would cause air escape velocities
higher than the minimum fluidization velocity
(Umf ) of the fine flow particles (~ 8 mm s1).
Moreover, even upward air velocity less than Umf
can produce a drag force that reduces the internal
friction and thus increases the flow runout, albeit
less efficiently than Umf. Note that as Umf is length-
scale independent, air escape velocity in nature
similar to that estimated above is expected to
cause fluidization of pyroclastic flows containing
Figure 12. Schematic representation of the contacts (black circles) mainly ash (group A) particles with a mean grain
between the particles at the base of the flow head and a (a) smooth, size close to that in our experiments.
(b) moderately rough, and (c) very rough substrate.
According to this autofluidization mechanism,
the source of air in our experiments was available until the interstices were completely filled by the flow
particles. In that circumstance only the frontal part of the flow could be fluidized. As the duration of the
upward air flux increased with the volume of the substrate interstices, the length of this frontal part increased
with the roughness. This is consistent with the fact that the amount of stretching of the flow head increased
with the roughness, whereas the morphology of the flow body was almost unchanged. For a given
substrate roughness, rapid filling of the interstices close to the reservoir, due to some vertical velocity
component of the collapsing granular column, may explain the relatively short duration of overpressure
(Figure 10). On the other hand, slower filling at greater travel distances, where the material propagated
horizontally, is consistent with the observed longer overpressure phase. The autofluidization hypothesis is
also supported by the experiments with the substrate interstices partially filled with PEG or small beads
before the flow was generated, because volume reduction of the air available caused a significant decrease
(~20–35%) of runout for a given substrate roughness.
In the case of initially fluidized flows, air escape from the substrate interstices contributed to delayed diffusion
of pore pressure derived from the reservoir. The pressure measurements suggest that the effect of this
additional source of gas was negligible at early flow stages, when initial pore pressure had not decreased much,
but became very marked at longer travel distances because pore pressure was significantly higher than for a
smooth substrate (see data at 50 cm and 80 cm in Figure 10). For a given roughness, autofluidization increased
the flow runout, compared to a smooth substrate, in the same proportion as for initially nonfluidized flows.
The flow runout at the highest substrate roughness of 6 mm was shorter than at an intermediate roughness
of 1.5–3 mm (Figure 5) despite the fact that more fluidizing air was available from the interstices. Close
examination of the kinematic data revealed that both initially fluidized and nonfluidized flows on a 6 mm
rough substrate had a lower velocity than flows on smoother substrates, from the start of propagation
(Figure 6). This can be explained by the fact that, as the size of the interstices increased, the flow particles
were less able to move over the beads forming the substrate: first, the large glued beads could act as
obstacles to flow propagation as many flow particles collided with them, dissipating flow energy, and second,
as more flow particles fell into the interstices, the flow head lost its mass and thus spread less. These two
mechanisms could have caused the decrease of flow runout compared to intermediate substrate roughness.
Another mechanism may have contributed to the control of the flow propagation and runout. Experiments of
Géminard and Losert [2002], involving a horizontal plate sliding on a flat granular material, showed that the
dynamic friction coefficient decreased by up to ~40% when the roughness of the granular layer (i.e., the size of
the particles) increased by a factor of about 4. Pohlman et al. [2006] also observed a decrease in the friction
coefficient with increasing substrate roughness for their rotating drum experiments. Based on these studies and
considering the sliding head of fine particle flows in our experiments as a continuous medium (see Figures 9a
and 9b), the friction between the flow base and the substrate may have been reduced as the roughness

increased. This could be related to the fact that the contact area per unit length between the flow base and the
substrate decreased as the roughness (and thus the size of the interstices) increased (Figure 12), hence leading
to lower energy dissipation. This reduction of friction was probably negligible compared to the autofluidization
mechanism discussed above in accounting for the strong variation in flow runout. It may, however, explain the
increase of flow velocity with increased substrate roughness at early flow stages (Figure 6).
The flows of particles of size d = 0.35 mm are difficult to fluidize because Umf for these large particles is high
(99.5 mm s1) [after Roche et al., 2006]. Furthermore, rapid filling of the interstices (Figures 9e–9h) and thus short
duration of the associated air flux, coupled with the high permeability of these clusters of large particles (group
B) that cause rapid pore pressure diffusion, would have prevented autofluidization of the flow. Hence, the size of
the interstices, which controlled the duration of air escape, did not have any influence on the runout distance.
Note that Bareschino et al. [2008] reported the absence of autofluidization in their experiments involving large
beads of size d = 0.55 mm. The almost constant runout of flows of these coarse particles in our experiments
suggests that the amount of energy dissipation, through interactions between the flow particles and the
substrate, was about the same for the different roughnesses. This is consistent with the fact that the flow head
did not slide on the substrate as a continuous medium; instead, the flow particles collided with the substrate
beads and began to settle in the interstices almost immediately after the passage of the flow front. Longer
runout distances of flows on a smooth substrate may suggest lower energy dissipation. Rolling of flow particles
(particularly at the flow front) may have contributed to these longer runouts compared to rough substrates.
4.2. Implication for Pyroclastic Flows

Our results suggest that ash-rich pyroclastic flows, whose matrix consists of fine group A particles, can be
autofluidized when they propagate on a rough, horizontal substrate. In nature, it is likely that flow particles
will fall into the substrate interstices sufficiently rapidly to cause air escape at velocities greater than the Umf
of the pyroclastic material, which can be as low as 1 mm s1 [Druitt et al., 2007]. According to our experiments,
which show the highest runout values for flows of ~3 cm thick on substrates of roughness 1.5–3 mm, and
assuming the same values of Ro in experiments and nature, the autofluidization of 1 m to 10 m thick pyroclastic
flows is expected to cause maximum runout (other things being equal) at mean substrate roughness of
0.05–0.1 m to 0.5–1 m, respectively.
Although more air would be available to fluidize pyroclastic flows at greater roughness, large surface irregularities
would actually act as obstacles, and flows would lose much of their mass through sedimentation in the
interstices. In contrast to these fines-rich flows, the behavior of scoria or coarse-grained pumice flows would be
unaffected by air escape from the substrate and their runout is expected to be independent of the substrate
roughness. Irrespective of the flow type and nature of the substrate, propagation on steep slopes may occur in a
manner different from that described above and we acknowledge that this issue deserves further investigation.
The present study also has implications for models of pyroclastic flows because any source of fluidizing air at the
flow base will be essential in controlling the flow dynamics [e.g., Iverson et al., 2004; Denlinger and Iverson, 2004].
In particular, a basal air flux will increase the pressure diffusion time scale so that flow propagation will be favored.
5. Conclusions
Our experiments were carried out with fine (d = 0.08 mm) group A particles that are representative of the ash
matrix of many pyroclastic flows. They showed that flows propagating on horizontal rough substrates were
autofluidized by air escaping from the interstices during the particle settling, which caused longer runout
distances than in the case of smooth substrates. Such a counterintuitive relationship between flow runout
and substrate roughness was also reported by Andrews and Manga [2012] for dilute turbulent currents whose
physics are fundamentally different from those of the dense gas-particle mixtures we considered. In our
experiments, autofluidization was evidenced in initially nonfluidized flows with high pore fluid pressure
supporting at least ~70% of the particle weight. Other things being equal, the flow runout increased with the
substrate roughness because the volume of air available was proportional to the size of the interstices so that
the flows were fluidized for a longer duration. The runout increase was most marked at late stages of
emplacement. It occurred as autofluidization permitted more material to be transferred to the flow head,
which spread to form a thin (<1 mm), elongated deposit with a very small surface slope (<1°). The longest
runout, about twice that for a smooth substrate, was at a roughness of 1.5–3 mm. At the highest roughness

value of 6 mm, however, substrate irregularities acted as obstacles and the mass of the flow decreased
significantly through sedimentation into the substrate interstices, which caused a shorter runout even
though more air was potentially available in the interstices. In contrast, the flow runouts of large
(d = 0.35 mm) group B particles, which could not be autofluidized, were independent of substrate roughness.
This investigation suggests that air present in topographic lows of a natural substrate should be considered
as another external source of gas that can fluidize pyroclastic flows, along with air ingested at the front of the
flow or gas resulting from water vaporization or burnt vegetation [cf. Wilson, 1980; Bareschino et al., 2008].
Important implications of our study are that (1) the autofluidization mechanism can contribute to causing
long runout distances even on horizontal slopes and (2) ash-rich pyroclastic flows generated successively
under similar initial eruptive conditions (i.e., volume, mass flux, drop height, grain size distribution,
fluidization state, particle concentration…) during a volcanic crisis may have very different runout distances
around an edifice depending on the roughness of the substrate on which they propagate. Scaling of our
results to nature suggests that autofluidization can lead to the longest runout distances (other things being
equal) when the mean substrate roughness is a few centimeters to a few tens of centimeters for 1 m to 10 m
thick pyroclastic flows, respectively. Such roughness values are typical of the surface of many natural
substrates, including those consisting of pyroclastic flow or fall deposits, debris avalanche deposits (apart
from large hummocks), solidified lava flows, or even fluvial sediments. In this context, numerical simulations
of pyroclastic flows should take the nature of the topography and possible fluidization effects caused by air
escape from the substrate into careful consideration. The autofluidization mechanism evidenced by our
study might also occur in other types of fines-rich geophysical flows characterized by a long runout distance,
such as the Socompa debris avalanche [e.g., van Wyk de Vries et al., 2001]. We acknowledge, however, that this
Acknowledgments issue requires further investigation.
This work was supported by the Institut
de Recherche pour le Développement
(IRD, France). We thank Michael Barton
and an anonymous reviewer for con- References
structive comments. This is Laboratory Andrews, B. J., and M. Manga (2012), Counterintuitive effects of substrate roughness on PDCs, Abstract V11B-2753 presented at 2012 Fall
of Excellence ClerVolc contribution 96. Meeting, AGU, San Francisco, Calif., 3–7 Dec.
Bareschino, P., L. Lirer, A. Marzocchella, P. Petrosino, and P. Salatino (2008), Self-fluidization of subaerial rapid granular flows, Powder Technol.,
182(3), 323–333, doi:10.1016/j.powtec.2007.12.010.
Cas, R. A. F., and J. V. Wright (1987), Volcanic Successions, Modern and Ancient, Allen and Unwin, London, U. K.
Cole, P. D., E. S. Calder, T. H. Druitt, R. Hoblitt, R. Robertson, R. S. J. Sparks, and S. R. Young (1998), Pyroclastic flows generated by gravitational
instability of the 1996–1997 Lava Dome of Soufriere Hills Volcano, Montserrat, Geophys. Res. Lett., 25(18), 3425–3428.
Denlinger, R. P., and R. M. Iverson (2004), Granular avalanches across irregular three-dimensional terrain: 1. Theory and computation,
J. Geophys. Res., 109, F01014, doi:10.1029/2003JF000085.
Druitt, T. H. (1998), Pyroclastic density currents, in The Physics of Volcanic Eruptions, Special Publications, vol. 145, edited by J. S. Gilbert
and R. S. J. Sparks, pp. 145–182, Geological Society, London, U. K.
Druitt, T. H., E. S. Calder, P. D. Cole, R. P. Hoblitt, S. C. Loughlin, G. E. Norton, L. J. Ritchie, R. S. J. Sparks, and B. Voight (2002), Small-volume,
highly mobile pyroclastic flows formed by rapid sedimentation from pyroclastic surges at Soufrière Hills Volcano, Montserrat: An
important volcanic hazard, in The Eruption of Soufrière Hills Volcano, Montserrat, From 1995 to 1999, Geol. Soc. of London Mem., vol. 21,
edited by T. H. Druitt and B. P. Kokelaar, pp. 263–279, Geological Society, London.
Druitt, T. H., G. Avard, G. Bruni, P. Lettieri, and F. Maez (2007), Gas retention in fine-grained pyroclastic flow materials at high temperatures,
Bull. Volcanol., 69, 881–901.
Freundt, A., C. J. N. Wilson, and S. N. Carey (2000), Ignimbrites and block-and-ash flow deposits, in Encyclopedia of Volcanoes, edited by
H. Sigurdsson, pp. 581–599, Academic Press. San Diego, Calif.
Geldart, D. (1986), Gas Fluidization Technology, Wiley, Chichester, U. K.
Géminard, J.-C., and W. Losert (2002), Frictional properties of bidisperse granular matter: Effect of mixing ratio, Phys. Rev. E, 65, doi:10.1103/
PhysRevE.65.041301.
Girolami, L., T. H. Druitt, O. Roche, and Z. Khrabrykh (2008), Propagation and hindered settling of laboratory ash flows, J. Geophys. Res., 113,
B02202, doi:10.1029/2007JB005074.
Girolami, L., O. Roche, T. H. Druitt, and T. Corpetti (2010), Particle velocity fields in laboratory ash flows, Bull. Volcanol., 72(6), 747–759.
Goujon, C., N. Thomas, and B. Dalloz-Dubrujeaud (2003), Monodisperse dry granular flows on inclined planes: Role of roughness, Eur. Phys. J.
E Soft Matter, 11, 147–157.
Iverson, R. M. (1997), The physics of debris flows, Rev. Geophys., 35, 245–296.
Iverson, R. M., M. Logan, and R. P. Denlinger (2004), Granular avalanches across irregular three-dimensional terrain: 2. Experimental tests,
J. Geophys. Res., 109, F01015, doi:10.1029/2003JF000084.
Lajeunesse, E., A. Mangeney-Castelnau, and J. P. Vilotte (2004), Spreading of a granular mass on a horizontal plane, Phys. Fluids, 16, 2371–2381.
Lube, G., H. E. Huppert, R. S. J. Sparks, and A. Freundt (2011), Granular column collapses down rough, inclined channels, J. Fluid Mech., 675,
347–368.
Montserrat, S., A. Tamburrino, O. Roche, and Y. Niño (2012), Pore fluid pressure diffusion in defluidizing granular columns, J. Geophys. Res.,
117, F02034, doi:10.1029/2011JF002164.
Pohlman, N. A., B. L. Severson, J. M. Ottino, and R. M. Lueptow (2006), Surface roughness effects in granular matter: Influence on angle of
repose and the absence of segregation. Phys. Rev. E, 73, 031304, doi:10.1103/PhysRevE.73.031304.
Pouliquen, O. (1999), Scaling laws in granular flows down rough inclined planes, Phys. Fluids, 11, 542–548.

Rhodes, M. J. (1998), Introduction to Particle Technology, John Wiley, Chichester, U. K.

Roche, O. (2012), Depositional processes and gas pore pressure in pyroclastic flows: An experimental perspective, Bull. Volcanol., 74,
1807–1820.
Roche, O., M. A. Gilbertson, J. C. Phillips, and R. S. J. Sparks (2006), The influence of particle size on the flow of initially fluidized powders,
Powder Technol., 166, 167–174, doi:10.1016/j.powtec.2006.05.010.
Roche, O., S. Montserrat, Y. Niño, and A. Tamburrino (2008), Experimental observations of water-like behavior of initially fluidized, dam break
granular flows and their relevance for the propagation of ash-rich pyroclastic flows, J. Geophys. Res., 113, B12203, doi:10.1029/
2008JB005664.
Roche, O., S. Montserrat, Y. Niño, and A. Tamburrino (2010), Pore fluid pressure and internal kinematics of gravitational laboratory air-particle
flows: Insights into the emplacement dynamics of pyroclastic flows, J. Geophys. Res., 115, B09206, doi:10.1029/2009JB007133.
Roche, O., Y. Niño, A. Mangeney, B. Brand, N. Pollock, and G. Valentine (2013), Dynamic pore pressure variations induce substrate erosion by
pyroclastic flows, Geology, 41, 1107–1110, doi:10.1130/G34668.1.
Sparks, R. S. J. (1976), Grain size variations in ignimbrites and implications for the transport of pyroclastic flows, Sedimentology, 23, 147–188.
Sparks, R. S. J., and L. Wilson (1976), A model for the formation of ignimbrite by gravitational column collapse, J. Geol. Soc. London, 132,
441–451.
van Wyk de Vries, B., S. Self, P. W. Francis, and L. Keszthelyi (2001), A gravitational spreading origin for the Socompa debris avalanche,
J. Volcanol. Geotherm. Res., 105, 225–247.
Wilson, C. J. N. (1980), The role of fluidization in the emplacement of pyroclastic flows: An experimental approach, J. Volcanol. Geotherm. Res.,
8, 231–249.

&KDSLWUH,QIOXHQFHGHODSHQWHVXUO¶DXWRIOXLGLVDWLRQHWODG\QDPLTXHGHVpFRXOHPHQWVS\URFODVWLTXHV

;
<, +

,

9

!
"

"
-
%

%#

.

`%
"

%

" %
!
"

%

.
"

.

) 3"" :NÔJ 5
:NUO
+

%
#

9

%

K
&OQ

-
"2

"
-

"

%

K
"

"

%

`

"

:N
/O:M

%
0
V ".

2

&KDSLWUH,QIOXHQFHGHODSHQWHVXUO¶DXWRIOXLGLVDWLRQHWODG\QDPLTXHGHVpFRXOHPHQWVS\URFODVWLTXHV
³&KpGHYLOOH & 5RFKH 2 ,QIOXHQFH RI VORSH DQJOH RQ SRUH SUHVVXUH JHQHUDWLRQ DQG
NLQHPDWLFV RI S\URFODVWLF IORZV LQVLJKWV IURP ODERUDWRU\ H[SHULPHQWV %XOO 9ROFDQRO
GRLV´

Bull Volcanol (2015) 77:96
DOI 10.1007/s00445-015-0981-4
RESEARCH ARTICLE
Influence of slope angle on pore pressure generation

and kinematics of pyroclastic flows: insights
from laboratory experiments
Corentin Chédeville 1 & Olivier Roche 1
Received: 24 March 2015 / Accepted: 13 October 2015

# Springer-Verlag Berlin Heidelberg 2015
Abstract The influence of slope angle on pore pressure gen- and long runout distances of pyroclastic flows on moderate
eration and kinematics of fines-rich pyroclastic flows was in- topographies.
vestigated through laboratory experiments. Granular flows
were generated by the release of a column of fine glass beads Keywords Pyroclastic flows . Fluidization . Pore pressure .
(d=0.08 mm) in an inclined channel (0–30°). The granular Analog modelling . Inclined substrate . Substrate roughness
column could be fluidized while the channel base was either
smooth or made rough by glued beads of 3 mm diameter. Pore
pressure measurements reveal that the degree of Introduction
autofluidization, caused by air escaping from the substrate
interstices into which flow particles settled, was high at all Pyroclastic density currents are common in nature and can be
slope angles. Flow runout increase due to autofluidization, a major threat for populations as they are hot, fast, challenging
however, was reduced at slope angle higher than ∼12° because to predict, and have runout distances that can exceed 100 km
of the occurrence of a strong deceleration phase that limited (Valentine et al. 1989; Wilson et al. 1995; Druitt 1998;
the flow duration. This is probably caused by the combination Branney and Kokelaar 2002). Their dynamics can vary signif-
of flow head thinning at increased slope angle and settling of icantly depending on their triggering mechanism (eruption
particles into the substrate interstices until the flow ran out of column or dome collapse, boiling over, lateral explosion),
mass. Analysis of high-speed videos suggests that ingestion of volume, granulometry, sorting (Sparks 1976; Wilson 1980;
ambient air at the flow front did not occur, even on steep Druitt et al. 2007), and topography (Fisher et al. 1993;
slopes of 30°. Experiments at inclinations close to (25°) or Woods et al. 1998; Lube et al. 2007) as well as the nature of
slightly higher (30°) than the repose angle of the granular the substrate on which they propagate (Wilson 1980; Roche
material (28.5°) revealed the formation of a thin basal deposit et al. 2013; Chédeville and Roche 2014). The present study
that was then eroded as the flow thickness and velocity grad- addresses the dynamics of pyroclastic density currents with a
ually decreased. Our study suggests that air escape from sub- concentrated basal part, referred to as Bpyroclastic flows^
strate interstices in nature can be a significant external cause of hereafter.
pore pressure generation that favors low energy dissipation Pyroclastic flows propagate on slopes that range from more
than 30° near the summits of volcanoes down to ∼0° at more
distal areas. Several field studies have addressed the influence
of the slope angle on their deposition, segregation, and erosion
Editorial responsibility: G. Lube mechanisms (Sparks et al. 1997; Giordano 1998; Calder et al.
2000; Lube et al. 2007; Bernard et al. 2014; Brand et al. 2014).
* Corentin Chédeville Further insights were gained from analog and numerical
c.chedeville@opgc.univ-bpclermont.fr modelling of granular flows down inclined substrates that re-
vealed fundamental aspects of flow dynamics (e.g., GDR
1
Laboratoire Magmas et Volcans, Université Blaise Pascal, CNRS, MiDi 2004; Mangeney et al. 2007). Recent experimental
IRD, OPGC, 5 rue Kessler, 63038 Clermont-Ferrand, France works, in particular, have investigated the influence of a rough

96 Page 2 of 13 Bull Volcanol (2015) 77:96
erodible or rigid granular substrate on flow velocity and according to

runout distance (Mangeney et al. 2010; Lube et al. 2011;
Farin et al. 2014). These studies, however, involved flows of ∂P ∂2 P
¼D 2 ð1Þ
particles larger than 0.1–1 mm for which the influence of the ∂t ∂z
interstitial gas (air) is negligible, in contrast to fines-rich py- where P is the pore pressure, z is the bed height, t is the time,
roclastic flows (Roche et al. 2006; Druitt et al. 2007; Girolami and D is the hydraulic diffusion coefficient (Iverson 1997). D
et al. 2008; Rowley et al. 2014). Here, we define fines-rich is proportional to k/(1−εs)μβ, with k the bed hydraulic perme-
flows as concentrated granular mixtures with a high (typically ability, εs the particle volume fraction, μ the gas dynamic
>50 wt.%) proportion of ash and a low hydraulic permeability. viscosity, and β the gas compressibility. The pressure diffu-
Analog experiments involving fine particles (<0.1 mm) have sion timescale of a bed of thickness H increases proportionally
demonstrated that low material permeability greatly reduces to H2/D (Iverson 1997; Montserrat et al. 2012; Roche 2012).
diffusion of the gas pore pressure from within the granular At a laboratory scale, pore pressure diffuses almost instanta-
mixture, which favors flow propagation and increases runout neously (<<1 s) in beds of particles typically larger than
distances compared to equivalent but non-fluidized flows 100 μm, with permeability k>10−10 m2. In contrast, the low
(e.g., Roche 2012). These flows of fine particles consisted of permeability k<10−10 m2 of beds of finer particles (e.g., the
a sliding (i.e., non zero velocity at the base) head that gener- matrix of fines-rich ignimbrites) allows diffusion for longer
ated relative underpressure that scaled with the square of the durations, which can be up to several minutes or even hours
front velocity (Roche et al. 2013) while the flow body behind when scaled to the natural system (Druitt et al. 2007;
the head generated overpressure when pore pressure was pro- Montserrat et al. 2012; Roche 2012).
vided at source (Roche et al. 2010; Roche 2012). Recently,
Chédeville and Roche (2014) showed that flows of such fine
particles without initial pore pressure and propagating on a Experimental device and procedure
horizontal rough, rigid substrate could be autofluidized as
particles settled into the substrate interstices, which caused Dam break experiments were carried out in the same appara-
an upward flux of air that generated pore pressure in the tus as the one used by Chédeville and Roche (2014). The
granular mixture. The present study aims at extending the device consisted of a 20×10 cm reservoir connected to a
work of Chédeville and Roche (2014) by investigating the 300×10 cm tilting channel in which flows were generated
influence of the slope angle on the autofluidization mecha- (Fig. 1). Opening of the reservoir gate by means of a counter-
nism evidenced by their experiments. We report data on gas weight was fast enough to have a negligible influence on the
pore pressure at the flow base, flow runout distance, and front spreading of the granular material from the reservoir. The flow
kinematics as a function of the slope angle, varying from particles were sub-spherical glass beads of mean diameter d=
horizontal to 30°. 0.08 mm and with a density of 2500 kg m−3. To avoid possible
cohesion effects caused by ambient moisture, we fluidized the
particles with dry air in a fluidization rig for at least 10 min
Experimental methods before each experiment. For some experiments, a gas flux was
provided at the base of the reservoir through a porous plate in
Pore fluid pressure generation and diffusion order to fluidize the particles (i.e., generate pore pressure)
in gas-particle systems before they were released in the channel. The superficial gas
velocity was set above the minimum bubbling velocity (Umb)
Pore fluid pressure in a bed of particles can be generated when of beds of fine particles (∼13–16 mm s−1 >Umf) so that full bed
there is a relative vertical motion between the gas moving support was achieved and bed expansion (∼7–8 %) was max-
upward and the particles settling downward. The gas exerts imal. No gas flux was provided from the channel base in any
a drag force that increases with its velocity, which counterbal- experiment. The mass of particles initially placed in the reser-
ances the weight of the particles and reduces the interparticle voir was set to 3, 6, or 12±0.001 kg, which corresponded to
frictional forces (Rhodes 1998). The bed is said to be fluidized column heights of about 10, 20, and 40 cm, respectively. The
when the drag force equals the apparent weight of the parti- channel base was either smooth or covered by a layer of glued
cles, which occurs at superficial gas velocity called the mini- glass beads with a mean diameter d=3.0±0.1 mm in order to
mum fluidization velocity (Umf). This minimum velocity in- form a rough non-erodible substrate (see Chédeville and
creases with particle size and density, meaning that it is easier Roche (2014) for details). The choice of using a 3-mm rough
to fluidize beds of small and light particles. bed was motivated by the fact that this roughness had the
Once generated, the interstitial pore fluid pressure does not strongest influence on the runout of horizontal flows
vanish instantaneously even if there is no more relative gas- (Chédeville and Roche 2014). Moreover, complementary ex-
particle motion. The pore pressure decreases through diffusion periments with lower roughness (d=0.7±0.1 mm) showed

Bull Volcanol (2015) 77:96 Page 3 of 13 96
Fig. 1 Experimental device

pulleys
consisting of a (fluidization)
reservoir connected to a 3-m-long
inclinable channel whose base is
either smooth or made rough by
gluing 3-mm glass beads. When
the gate opens, the spreading of
the column of particles generates
a flow consisting of a sliding head
followed by a body at base of
sliding gate
which forms and aggrading
deposit
particles
air flux
porous plate glass beads
that the flows generated low pore pressure signals difficult to configuration represents a wide range of rough terrains that
analyze (see in the BResults^ section). could be commonly observed in natural configurations.
The channel was inclined at slope angles ranging from 0 to
30°, the latter being larger than the repose angle of the flow
particles (28.5±1°), while the reservoir base was always hor- Results
izontal. The flows generated had typical thicknesses of 0.5–
6 cm and maximum front velocities of 0.5–5 m/s depending Pore pressure data
on the initial conditions (i.e., presence or absence of pore
pressure in the reservoir, column height, and channel inclina- Pore pressure data were acquired at five different slope angles
tion). The experiments were filmed by means of a high-speed (0°, 8°, 16°, 25°, and 30°) for both initially fluidized and non-
video camera (Fastcam SA3) at rates of 250 to 3000 frames/s fluidized flows generated from a column of initial height of
with corresponding resolutions of 1024 × 1024 to 896 × 40 cm on both smooth and 3-mm rough substrates. For initial-
736 pixels, depending on the portion of the flow investigated. ly non-fluidized flows on a smooth substrate (Fig. 2), no pos-
To measure the gas pore pressure within the flow, we used itive pore pressure was measured at any slope angles investi-
piezoresistive pressure sensors at distances of 10, 50, 90, and gated except close to the reservoir (10 cm), in some experi-
150 cm from the gate. The sensors were covered by a 36-μm ments, where small positive pressure (<60 Pa) was recorded as
grid to prevent any contact with the particles and were placed in earlier similar works (Roche et al. 2010). The underpressure
at the same level as the top of the glued beads (Chédeville and phase, caused by the sliding head of the flows, had an ampli-
Roche 2014, their Fig. 4b). tude that increased with the slope angle and the distance in the
The experiments were scaled in order to ensure dynamic channel while its duration only increased with distance.
similarity with fines-rich pyroclastic flows according to the For flows propagating on a 3-mm rough substrate (Fig. 3),
dimensional analysis of Roche (2012) and Chédeville and pore pressure data were normalized with the lithostatic pres-
Roche (2014). Considering same values of the roughness sure PL =ρfgh, with ρf the bulk density of the flow (1450±
number R0 =h/D0 proposed by Chédeville and Roche (2014) 50 kg/m3), g the gravitational acceleration, and h the flow
in nature and experiments, with h the flow thickness and D0 height above the sensor. The flow height was measured from
the substrate roughness, and the fact that experimental flows the position of the flow upper surface along the channel
have a thickness h=1–6 cm, then the 3-mm rough substrate backwall on which a 2×2 cm grid was drawn. Error bars in
used in this study represents a natural roughness of ∼0.05–0.3 Fig. 3 correspond to the imprecision on the measurements of
to ∼0.5–3 m for 1-m-thick to 10-m-thick pyroclastic flows, the flow thickness and the possible variation of the flow den-
respectively. We thus believe that our experimental sity due to material expansion. The video frame rate was more

Fig. 2 Basal pore fluid pressure

as a function of time after the
release of a 40-cm high column of
particles from the reservoir, for
initially non-fluidized flows over
a smooth substrate, at inclinations
of a 0°, b 8°, c 16°, d 25°, and e
30°
(e)
than 4 times the sampling rate of the pressure sensors so that 150 cm), values were significantly higher (P/PL >70 %).
each pressure value could be precisely correlated to the instan- Note that there was no clear dependence of the pressure ratio
taneous flow thickness. The normalized pressure accounts for on the slope angle at a given distance. The flow length (x),
the degree of fluidization for positive pore pressure but is not corresponding to the duration of the positive pore pressure
relevant for the first underpressure phase, which is believed to peak (see below), increased with both the slope angle and
be a dynamic phenomenon (Roche et al. 2013). Positive pore the distance in the channel (Fig. 4a). Measurements of x were
pressure was measured for both initially fluidized and non- made either by correlating the peak duration with the flow
fluidized flows at all slope angles investigated. front position on the video or by multiplying the front velocity
For initially non-fluidized flows (Fig. 3a–e), underpressure by the duration of the overpressure phase if the front
was generated at the foremost part of the head but was of overpassed the channel length. Normalizing this length with
shorter duration and much lower amplitude (∼−100 Pa) than the total flow length (Fig. 4b, c) shows that a large proportion
for a smooth substrate (−200 to −500 Pa). Then the flow of the flow was affected by the overpressure, increasing from
generated a positive pore pressure that compensated 40– 20–60 % at the horizontal to 55–90 % at an inclination of 30°.
100 % of the lithostatic pressure. The ratio P/PL >1 measured Note that we considered that the length x affected by the over-
in some cases can be explained by rapid variation of the flow pressure included the foremost part of the flow at base of
thickness compared with the timescale of pressure diffusion which underpressure was measured instead (Fig. 4c). In fact,
once pressure was generated (see Roche et al. (2010), their this underpressure is believed to have a dynamical origin and
Fig. 12). At a slope angle larger than 8°, the lowest values of might be limited to the flow-substrate interface, possibly
the normalized pore pressure (P/PL ∼40–50 %) were measured masking overpressure within the flow above (Roche 2012;
close (10 cm) to the gate whereas at further distance (50– Roche et al. 2013).

Fig. 3 Basal pore fluid pressure,

P, normalized with the lithostatic
pressure, PL, as a function of time
for a–e initially non-fluidized and
f–j initially fluidized flows
propagating over a rough
substrate. Slope angle was 0–30°
and initial column height was
40 cm. High pore pressure
fluctuation for some data at 90–
150 cm is due to very low flow
thicknesses for which the sensor
noise becomes non-negligible.
Numbers above the curves
represent the peak raw pressure
values, in Pascal, for each sensor
location. Note that these values do
not necessarily correspond to the
highest raw pressure measured by
the sensors. Typical error bars are
represented for each sensor
location with the corresponding
color code and account for
imprecisions in measurements of
the flow thickness and possible
variation of bulk flow density due
to material expansion of 0–8 %.
Error bars smaller than the line
thickness are not represented. For
flows at inclinations of 0 and 8°,
positive pore pressure was also
measured at distances of 90 and
150 cm from the gate, but error
bars were so large that we chose
to not represent the corresponding
curves

Fig. 4 Initially non-fluidized flows on a rough substrate. a Flow length on the video or from the peak duration times the flow front velocity. The
affected by autofluidization (x) as a function of the slope angle, for sketch c (not to scale) illustrates the relations between the pore pressure
sensors located at various distances in the channel. b Ratio of x over the signal and the flow location when the front reaches the sensor (t1) and at
total flow length (xtotal) as a function of the slope angle, with same legend the end of the overpressure phase (t2). Measurements of x and xtotal
as in a. The two colors represent two experiments for each slope angle reported in a and b were both taken at time t2. Note that x also includes
investigated. Closed and open symbols represent, respectively, x obtained the first short underpressure phase at the flow front
from correlation of the pressure peak duration with the flow front position
In the case of initially fluidized flows over a rough sub- length. Experiments were carried out with initial particle col-
strate (Fig. 3f–j), the overpressure phase measured close to the umn heights of 10 or 20 cm, in both initially fluidized and
reservoir (10 cm) lasted longer than for initially non-fluidized non-fluidized conditions.
flows while its amplitude decreased with slope angle, from P/ For initially non-fluidized flows propagating on a smooth
PL∼100 % at 0° to P/PL∼40 % at 30°. Further downstream the substrate (Fig. 5a), the flow runout increased with the slope
influence of initial fluidization was not evident as initially angle and the data defined a trend similar to that reported by
non-fluidized flows generated similar pressure signals. There Mangeney et al. (2010) and Farin et al. (2014) for flows of
was no clear tendency of the variation of maximum pore pres- larger beads (0.6–0.8 mm). These authors found that
sure value with slope angle.
Results were comparable to those described above for r 2σ
¼ ; ð2Þ
flows propagating over a less rough substrate (d=0.7 mm) H tanδ−tanθ
except at a long distance from the gate (90–150 cm) for ini-
tially non-fluidized flows, where pore pressure signals were with r as the flow runout, H the initial column height, σ an
too low to be clearly distinguished from the sensor noise as the empirical coefficient that is equal to 0.5 (Mangeney et al.
flows became very thin (<0.2 cm). 2010), δ the effective friction angle of the material, and θ
the slope angle. For our experiments, the model fitted the
Flow runout data for a friction angle δ = 23°, lower than the material
repose angle (28.5±1°) but very close to the surface angle
Runout distances of flows propagating over smooth or 3-mm of the deposit left in the reservoir (∼22 ± 1°). At steeper
rough substrates were measured at slope angles ranging from slope angles, late stage motion of particles at the surface
0 to 20° with intervals of 2° (Fig. 5). Flow runout could not be of the deposit once the flow front stopped could cause
measured at higher inclinations as it exceeded the channel runout increase by up to 50 % at 20° (see error bars in

Fig. 5 Flow runout, r, normalized with the initial column height, H, as a represents the normalized runout as a function of the rescaled slope
function of the slope angle, θ, for a a smooth substrate and b a 3-mm angle. The purple diamond at 20° corresponds to the flow runout before
rough substrate (same legend for both graphs). In a, the solid line the front was overtaken by late-stage motion at the upper free surface.
corresponds to the best fit of the runout of non-fluidized flows, Error bars represent minimum and maximum runout distances measured
obtained using Eq. 2 for a friction angle of δ = 23° and the inset for 3–8 repeated experiments and can be smaller than symbols
Fig. 5a). For initially fluidized flows, the friction increased Flow front kinematics
with time as pore pressure diffused. Following Lucas and
Mangeney (2007) and Roche et al. (2011), who addressed Flow front kinematics and velocities are presented in Figs. 7
flows on horizontal slopes, we tried to define a mean effec- and 8 for different slope angles. The flows propagated in three
tive (reduced) friction angle so that Eq. 2 could fit our data. successive phases: acceleration, constant velocity, and decel-
We did not succeed as we found that the friction angle that eration, in agreement with trends reported in earlier works on
best fitted our data at θ=0° was δ=8.5° whereas it was as horizontal substrates (e.g., Chédeville and Roche 2014).
high as δ=20.1° at an inclination of 18°. When the channel base was rough, the relative duration of
The data for flows over a rough substrate reveal similar the final phase (deceleration) increased considerably with the
trends as observed for a smooth substrate, but with higher roughness, as the flows were (partially) autofluidized. In con-
normalized runouts at a given slope angle (Fig. 5b). At angles sequence, the frontal part of the deposit was greatly elongated
>12°, larger normalized runouts of initially non-fluidized compared to that of flows on a smooth substrate.
flows for column height H=20 cm compared with H=10 cm In the present study, both the flow duration and maximum
could be caused by a larger initial volume of material in the front velocity increased with the slope angle at given initial
reservoir (Farin et al. 2014). As for fluidized flows over a conditions (Figs. 7 and 8). For flows over a smooth substrate,
smooth substrate, we were not able to fit the data with incli- the last deceleration phase notably increased with inclination.
nation using a mean effective friction angle. Note that for When the slope was steep enough (∼12° for initially fluidized
slope angles of 0–4°, however, the Bbest^ mean friction angles flows and 16° for non-fluidized flows), late-stage mass move-
were δ=15° and 4° for initially non-fluidized and fluidized ment at the free surface of the flow overtook the distal limit of
flows, respectively. the already deposited material, thus increasing the runout by
The ratios of the runout of flows on rough (rr) and smooth up to several tens of centimeters. When the substrate was
(rs) substrates are represented in Fig. 6. The ratio rr/rs in- rough, the duration of the acceleration and constant velocity
creased with the initial column height (i.e., volume) and initial phases increased with the slope angle, whereas the duration of
fluidization. For all types of experiments, rr/rs increased up to the deceleration phase did not increase despite the higher max-
∼1.8–2.2 at low angles <6° then was approximately constant imum velocities. As a consequence, the total flow duration
at angles up to 12° and finally decreased down to <1 for a increased much less with slope angle compared to what was
slope of 20°. Data could not be reported at angles >14° for observed for a smooth substrate. Late-stage mass movement
initially fluidized flows as the runout exceeded the channel as described above was weak and occurred only near the
length. Note that for flows over a rough substrate, the runout reservoir.
considered was either (i) the well-defined distal limit of the
deposit determined from observation of flow particles in the Flow front morphology
substrate interstices (Fig. 6a) or (ii) the maximum extent of the
flows observed on videos, which could be several centimeters Detailed views of the flow front at different initial conditions
longer than the observed deposit and whose particles formed a are reported in Fig. 9. Flows over a smooth substrate had a thin
deposit that was too thin to be seen (Fig. 6b). However, both head whose free surface was at a shallow angle with respect to
sets of data show similar trends. the substrate and whose base was in direct contact with the

Fig. 6 Ratios of runout distance of flows on 3-mm rough substrate (rr) The purple diamond at 20° corresponds to the runout ratio where late-
over that of flows on a smooth substrate (rs) for initial column heights of stage motion overtaking the flow front on a smooth substrate was not
10 and 20 cm. Runout for a rough substrate was either the distal limit of a taken into account
the deposit observed directly in the channel or b the flow visible on video.
latter, except at high slope angles (25–30°) where a slight Sedimentation and deposit reworking
over-elevation of the foremost part could temporarily occur
(Fig. 9f). We did not observe any Bplunging breakers^ that At slope angles of 25 and 30°, the flows left no deposit in the
could entrain ambient air, as described by Bareschino et al. channel at any configuration. During propagation, however,
(2008) for instance. The head of flows over a rough substrate we observed the formation of a thin basal deposit (<5 mm at
slid over the glued beads at all angles investigated, and parti- 80 cm from the gate) that first aggraded and was subsequently
cles settled into the interstices (Fig. 9a–d; see Chédeville and eroded slowly by the material flowing above once both flow
Roche (2014) for a detailed description of flows over a hori- height (Fig. 10) and velocity had much decreased. This phe-
zontal substrate). Transient front instabilities observed at the nomenon was observed in all cases except for initially non-
foremost part of the flow probably resulted from collisions of fluidized flows over a smooth substrate. Note that for a rough
the particles with the substrate beads, but, as for a smooth substrate, deposition occurred sooner after passage of the flow
substrate, we did not observe any Bplunging breakers.^ front and was slower compared to a smooth substrate. At an
Fig. 7 Normalized flow runout,

x/H, as a function of normalized
time t/t0, with t0 =(H/g)1/2 at
various slope angles for initially
non-fluidized flows over a a
smooth substrate and b a rough
substrate, and initially fluidized
flows over c a smooth and d a
rough substrate. Measurements of
the front position were made at
time intervals of t/t0∼0.16 s.
Vertical black lines for some high
slope angle data indicate the front
being overtaken by a late-stage
motion at the upper free surface of
the flow. Some curves are
truncated as flows exceeded the
channel length

Fig. 8 Flow front velocity as a function of normalized time (t/t0, see all graphs). Vertical black lines for some high slope angle data indicate the
Fig. 7) for slope angles of 0 to 20° and for initially non-fluidized flows front being overtaken by a late-stage motion of particles at the upper free
propagating over a a smooth or b a rough substrate, and initially fluidized surface of the flow. Some curves are truncated because flows exceeded
flows propagating over c a smooth or d a rough substrate (same legend for the channel length
Fig. 9 Detailed views of the flow

front for various configurations. a (a) 0° -nf 3 mm (b) 30°- f 3 mm
Fluidized ( f ) flow, rough
substrate at 8°. b Fluidized flow,
rough substrate at 30°. c Non-
fluidized flow (nf), rough
substrate at 0°. d Non-fluidized
flow, rough substrate at 25°. e
Non-fluidized flow, smooth
substrate at 0°. f Fluidized flow,
smooth substrate at 30°.
Snapshots were taken at 80 cm (a,
b, d, and f), 45 cm (c), or 30 cm
(e) from the reservoir (c) 0°- nf 3 mm (d) 25°- nf 3 mm
(e) 0°- nf 3 mm (f) 30°- f 3 mm

Fig. 10 Thickness of the basal deposit, Td, and total thickness (flow + deposit), T, as a function of time at 80 cm from the gate and for inclinations of a
25° and b 30°. Some data are not represented for all flow durations because of difficulties in distinguishing the lower flow boundary
inclination of 30°, the maximum thickness of the deposit was (>2.5 m/s), efficient air percolation through the granular
smaller and its erosion was much faster than at 25°. mass caused the fine-grained material to acquire a friction-
less behavior. In our experiments, no similar front instabil-
ities were observed although the maximum front velocities
Discussion were higher (up to 5 m/s at an inclination of 30°) than those
in the rotating drum experiments. It is possible that the
Autofluidization plunging breakers reported by Bareschino et al. (2008)
were the consequence of their experimental configuration,
One of the most important results of our study is that, as the granular material was elevated at a flow rear by the
when initially non-fluidized flows propagated over a rotating drum before propagating on very steep flow sur-
rough substrate, high pore pressure (P/PL >0.75) was mea- faces towards the front.
sured at all slope angles (0–30°) and at most sensor loca-
tions (50–150 cm from the reservoir) investigated. This Influence of slope angle on flow dynamics
suggests that the autofluidization mechanism evidenced
by Chédeville and Roche (2014) for flows on a horizontal The variation of the flow runout with slope angle <16° for
rough substrate was also effective at high slope angles. our initially non-fluidized flows over a smooth substrate is
Though the amplitude of pore pressure signals on inclined well described by the scaling law of Mangeney et al.
surfaces was similar to that in the horizontal configuration, (2010) and Farin et al. (2014), which means that runout
the duration of the signals, which corresponded to the flow was essentially controlled by the initial column height. At
length affected by overpressure, increased with the slope angles >16°, this scaling law overestimated the runout,
angle for a given travel distance (Fig. 4). This could result particularly when late-stage motion was not considered
from a higher flow velocity and lower settling rate that (Fig. 5a). Farin et al. (2014) reported similar results for
increased the total filling time of the interstices by the flow experiments on steeply inclined rough substrates and pro-
particles, thus increasing the flow length affected by posed that runout was also controlled by the volume re-
autofluidization. This was confirmed by video measure- leased. The main difference with their study was the lower
ments that show that the filling time of the interstices in- effective friction angle in our experiments (23° against
creased from ∼0.11 to ∼0.21 s when the slope angle in- 27°), probably due to the different sizes of the flow parti-
creased from 0 to 30°. cles (d=0.08 mm against d=0.70 mm) and the nature of the
Another important result is the absence of overpressure substrate we used (smooth Plexiglas against glued glass
in initially non-fluidized flows propagating over a smooth beads), as the nature of the flow particles (glass beads)
substrate, even at a slope angle up to 30°. This implies that, and channel walls (Plexiglas) were identical for both stud-
in our configuration involving flows at almost maximum ies. Though scaling with a mean low effective friction an-
particle concentration, autofluidization was not possible or gle is possible for initially fluidized flows at a given slope
negligible. This contrasts with the experiments of angle (see Roche et al. (2011) for horizontal substrate), our
Bareschino et al. (2008) that involved a rotating drum results show that it is not applicable for a range of slope
and in which fine particles (glass beads or FCC) were angles. This conclusion also applies to our initially non-
autofluidized. In their study, autofluidization was the result fluidized flows on rough substrates that were partially
of successive Bplunging breakers^ that incorporated ambi- autofluidized during propagation and whose emplacement
ent air at the flow front. At sufficiently high flow velocity was controlled by a mean friction lower than that of the

non-fluidized material. The ratio of the flow runout on the Air ingestion at the flow front was proposed by
rough substrate over that on the smooth plane did not vary McTaggart (1960) and Wilson (1980) as a possible mecha-
much (∼1.8–2.2) at slope angles up to 10–12°, above nism for the fluidization of pyroclastic flows. This was
which this ratio decreased significantly. This shows that mainly because the flow head morphology was believed
the effect of runout increase due to substrate roughness to be similar to that of fluid gravity currents in
was reduced at high inclinations, which could be the con- Boussinesq conditions (i.e., with a low density difference
sequence of several mechanisms: (i) the autofluidization with the surrounding fluid) for which no-slip condition at
became less efficient at steeper slopes, but strong overpres- the lower boundary caused a convex-shaped head with an
sure signals suggest this was not the case; (ii) the continu- elevated nose beneath which the surrounding fluid could be
ous sedimentation of the flow particles in the substrate ingested (Allen 1971; Simpson 1972, 1986, 1997). In our
interstices reduced the mass of the flow whose head (the experiments, however, velocity at the lower boundary was
flow portion with a basal slip condition) ran out of materi- not zero as the flow base slid over the substrate, as com-
al; (iii) the increase of collision intensity between the flow monly observed for granular flows (Artoni et al. 2009).
particles and substrate beads, as the flow velocity increased This is probably the main reason why we did not observe
with the slope angle, could dissipate more flow energy than a long-lived elevated nose that would have favored basal air
for a smooth substrate. incorporation. Note also that mixing with ambient air at the
Kinematic data reveal that the presence of a rough sub- flow upper surface resulting from Kelvin-Helmholtz insta-
strate caused a larger increase of the maximum flow front bilities (e.g., Andrews and Manga 2012; Andrews 2014)
velocity and of the duration of the acceleration and constant was impossible because of the large density difference be-
velocity phases with an increasing slope angle compared to tween the flow and the surrounding medium, in contrast to
a smooth substrate (Fig. 8). Nevertheless, the deceleration Boussinesq gravity currents. In nature, however, pyroclas-
phase, which seemed to start only when the flow head tic flows may be less dense than in our experiments and are
thickness had much decreased (<2–3 mm), became abrupt commonly faster. For these reasons, we do not rule out that
(Fig. 8b, d) and its duration, notably for initially non- conditions of air incorporation at flow boundaries could be
fluidized flows, did not vary much (2.5<t/t0 <3.5) as the reached at least locally. Moreover, although we did not
slope angle increased, whereas it was drastically extended observe Bplunging breakers^ (c.f. Bareschino et al. 2008)
for flows on a smooth substrate (t/t0 varied from 2.5 to that would also have permitted air ingestion in our experi-
13.5 at inclinations of 0 and 20°, respectively). Rowley ments, such phenomena may occur in nature if a flow
et al. (2014) showed that experimental flows fluidized by passes a break in slope or an obstacle.
a continuous air flux along the flow base did not stop even Reworking of a freshly deposited material by the up-
on a horizontal substrate. In our experiments, however, on- per flowing part at a high inclination (25–30°) in our
ly the flow head was autofluidized through settling of the experiments may have implications for some pyroclastic
particles in the substrate interstices and the resulting mass flow deposit sequences. During some moderate to large
loss could have caused a rapid flow deceleration when, ignimbrite-forming eruptions, waxing and waning phases
eventually, the flow head became too thin to sustain are thought to alternate (Branney and Kokelaar 2002;
motion. Increasing the substrate roughness could enlarge Brown and Branney 2013; Williams et al. 2013; Brand
the flow portion affected by the fluidization mechanism et al. 2014), which could lead to successive episodes of
but, in the same time, could increase the volume of flow deposition and erosion (Brown and Branney 2004) and
particles lost by settling into the interstices. Furthermore, could render the deposit architecture difficult to decipher.
experiments of Chédeville and Roche (2014) on a horizon- In our experiments, the flows first formed a basal deposit
tal substrate showed that when the roughness was too high that was then progressively eroded as the flow thickness
(D0 > 0.3–0.6 mm), its positive influence on the runout and velocity decreased. The recent study of Lucas et al.
started to decrease. (2014) suggests that the basal friction of geophysical
granular flows can vary inversely with their velocity;
Implication for pyroclastic flows accordingly, onset of erosion at decreasing velocity in
our experiments is compatible with increasing friction
This work shows that the autofluidization mechanism at the base of the flowing mass. Another possibility is
found by Chédeville and Roche (2014) for horizontal that basal friction increased with time as the granular
substrates can also be efficient at high slope angles (at material defluidized progressively. We stress that our re-
least 30°). This suggests that, in nature, this mechanism sults are very preliminary and that further work is re-
could occur during the whole propagation of pyroclastic quired to better characterize conditions of deposit remo-
flows, from the steep flanks of volcanoes to the distal bilization, including particularly the critical flow velocity
more gentle slopes. and substrate slope angle.

Conclusion References
Our experiments on flows of fine particles over smooth or Allen JRL (1971) Mixing at turbidity current heads, and its geological
rough substrates inclined up to 30° have revealed that implications. J Sediment Petrol 41:97–113
Andrews BJ (2014) Dispersal and air entrainment in unconfined dilute
pyroclastic density currents. Bull Volcanol 76:852, doi: 8 10.1007/
(1) The autofluidization mechanism of fines-rich pyroclastic s00445-00014-00852-00444
flows evidenced by experiments of Chédeville and Andrews BJ, Manga M (2012) Experimental study of turbulence, sedi-
Roche (2014) and caused by air evacuation from the mentation, and coignimbrite mass partitioning in dilute pyroclastic
density currents. J Volcanol Geotherm Res 225–226:30–44. doi:10.
interstices of a rough substrate into which the flow par-
1016/j.jvolgeores.2012.1002.1011
ticles settled occurred at all slope angles investigated (0– Artoni R, Santomaso AC, Canu P (2009) Effective boundary conditions
30°). In contrast, autofluidization was not generated in for dense granular flows. Phys Rev E 79:031304. doi:10.1103/
case of a smooth substrate. PhysRevE.79.031304
(2) Flow front instabilities that may favor incorporation of Bareschino P, Lirer L, Marzocchella A, Petrosino P, Salatino P (2008)
Self-fluidization of subaerial rapid granular flows. Powder Technol
the ambient air such as plunging breakers or those typical 182:323–333. doi:10.1016/j.powtec.2007.12.010
of fluid gravity currents were never observed, even when Bernard J, Kelfoun K, Le Pennec J-L, Vallejo Vargas S (2014) Pyroclastic
the flows propagated on steep slopes (30°). We acknowl- flow erosion and bulking processes: comparing field-based vs.
edge, however, that such instabilities could occur in less modeling results at Tungurahua volcano, Ecuador. Bull Volcanol
76:858. doi:10.1007/s00445-014-0858-y
dense and/or faster pyroclastic flows or if these encoun- Brand BD, Mackaman-Lofland C, Pollock NM, Bendaña S, Dawson B,
ter a break in slope or an obstacle. Wichgers P (2014) Dynamics of pyroclastic density currents: con-
(3) The runout distance of flows on the rough substrate was ditions that promote substrate erosion and self-channelization—
about twice as that on the smooth incline for slope angles Mount St Helens, Washington (USA). J Volcanol Geotherm Res
276:189–214. doi:10.1016/j.jvolgeores.2014.01.007
up to 10–12°. For steeper slopes, strong deceleration of Branney MJ, Kokelaar BP (2002) Pyroclastic density currents and the
the flows on a rough substrate led to relative shorter sedimentation of ignimbrites. Geol Soc Lond Mem 27
runout distances, which became close to those for a Brown RJ, Branney MJ (2004) Bypassing and diachronous deposition
smooth substrate at a slope angle of 20°. This could result from density currents: evidence from a giant regressive bed form in
the Poris ignimbrite, Tenerife, Canary Islands. Geology 32:445–
from reduction of the flowing mass, as particles 448. doi:10.1130/G20188.1
sedimented into the interstices, and/or increase of the en- Brown RJ, Branney MJ (2013) Internal flow variations and diachronous
ergy dissipation through collisions on the rough substrate. sedimentation within extensive, sustained, density-stratified pyro-
(4) At a slope angle close to the repose angle of the flow clastic density currents flowing down gentle slopes, as revealed by
the internal architectures of ignimbrites on Tenerife. Bull Volcanol
material, a deposit first formed at the flow base and was
75:727. doi:10.1007/s00445-013-0727-0
then reworked by the material still in motion above it. Calder ES, Sparks RSJ, Gardeweg MC (2000) Erosion, transport and
The cause of this phenomenon is unclear. It appears that segregation of pumice and lithic clasts in pyroclastic flows inferred
erosion occurred once the flow thickness and/or velocity from ignimbrite at Lascar Volcano, Chile. J Volcanol Geotherm Res
104:201–235
were low enough, and in this regard, lower velocities of
Chédeville C, Roche O (2014) Autofluidization of pyroclastic flows
the flowing mass might have caused higher frictional propagating on rough substrates as shown by laboratory experi-
stresses on the basal deposit (Lucas et al. 2014). ments. J Geophys Res 119:1764–1776. doi:10.1002/2013JB010554
Further investigation is required to characterize in partic- Druitt TH (1998) Pyroclastic density currents. Geol Soc Lond Spec Publ
ular critical normal/shear stress ratio, flow velocity, and 145–182
Druitt TH, Avard G, Bruni G, Lettieri P, Maez F (2007) Gas retention in
slope angle at which erosion occurs. Such a reworking fine-grained pyroclastic flow materials at high temperatures. Bull
mechanism in pyroclastic flows could create deposits Volcanol 69:881–901. doi:10.1007/s00445-007-0116-7
that would not necessarily record the entire flow history Farin M, Mangeney A, Roche O (2014) Fundamental changes of granular
and that could locally display complex architecture with flow dynamics, deposition, and erosion processes at high slope an-
gles: insights from laboratory experiments. J Geophys Res 119:504–
remobilized material deposited downstream onto youn- 532. doi:10.1002/2013JF002750
ger products. Fisher R, Orsi G, Ort M, Heiken G (1993) Mobility of a large-volume
pyroclastic flow—emplacement of the Campanian ignimbrite, Italy.
J Volcanol Geotherm Res 56:205–220
Acknowledgments We are grateful to Greg Valentine and David Jessop GDR MiDi (2004) On dense granular flows. Eur Phys J E 14:341–365.
for the interesting discussions and comments that helped to improve the doi:10.1140/epje/i2003-10153-0
manuscript. We thank the associate editor Gert Lube, Ben Andrews, and Giordano G (1998) The effect of paleotopography on lithic distribution
Eric Bréard for the careful and constructive reviews. This work was and facies associations of small volume ignimbrites: the WTT Cupa
financed by the Laboratoire Mixte International SVAN of Institut de (Roccamonfina volcano, Italy). J Volcanol Geotherm Res 87:255–
Recherche pour le Développement (IRD, France) and by the French 273
Government Laboratory of Excellence initiative n°ANR-10-LABX- Girolami L, Druitt TH, Roche O, Khrabrykh Z (2008) Propagation and
0006, the Région Auvergne and the European Regional Development hindered settling of laboratory ash flows. J Geophys Res 113:
Fund. This is Laboratory of Excellence ClerVolc contribution number 169. B02202. doi:10.1029/2007JB005074

Iverson RM (1997) The physics of debris flows. Rev Geophys 35:245– insights into the emplacement dynamics of pyroclastic flows. J
296 Geophys Res 115:B09206. doi:10.1029/2009JB007133
Lube G, Cronin SJ, Platz T, Freundt A, Procter JN, Henderson C, Roche O, Attali M, Mangeney A, Lucas A (2011) On the run-out distance
Sheridan MF (2007) Flow and deposition of pyroclastic granular of geophysical gravitational flows: insight from fluidized granular
flows: a type example from the 1975 Ngauruhoe eruption, New collapse experiments. Earth Planet Sci Lett 311:375–385. doi:10.
Zealand. J Volcanol Geotherm Res 161:165–186. doi:10.1016/j. 1016/j.epsl.2011.09.023
jvolgeores.2006.12.003 Roche O, Niño Y, Mangeney A, Brand B, Pollock N, Valentine G (2013)
Lube G, Huppert HE, Sparks RSJ, Freundt A (2011) Granular column Dynamic pore pressure variations induce substrate erosion by pyro-
collapses down rough, inclined channels. J Fluid Mech 675:347– clastic flows. Geology 41:1107–1110. doi:10.1130/G34668.1
368. doi:10.1017/jfm.2011.21 Rowley PJ, Roche O, Druitt TH, Cas R (2014) Experimental study of
Lucas A, Mangeney A (2007) Mobility and topographic effects for large dense pyroclastic density currents using sustained, gas-fluidized
Valles Marineris landslides on Mars. Geophys Res Lett 34:L10201. granular flows. Bull Volcanol 76:855. doi:10.1007/s00445-014-
doi:10.1029/2007GL029835 0855-1
Lucas A, Mangeney A, Ampuero JP (2014) Frictional velocity- Simpson JE (1972) Effects of the lower boundary on the head of a gravity
weakening in landslides on Earth and on other planetary bodies. current. J Fluid Mech 53:759. doi:10.1017/S0022112072000461
Nat Commun 5:3417. doi:10.1038/ncomms4417 Simpson JE (1986) Mixing at the front of a gravity current. Acta Mech
Mangeney A, Bouchut F, Thomas N, Vilotte JP, Bristeau MO (2007) 63:245–253. doi:10.1007/BF01182551
Numerical modeling of self-channeling granular flows and of their Simpson JE (1997) Gravity currents in the environment and the labora-
levee-channel deposit. J Geophys Res 112:F02017. doi: tory. Cambridge University Press, Cambridge
02010.01029/02006JF000469 Sparks RSJ (1976) Grain size variations in ignimbrites and implications
Mangeney A, Roche O, Hungr O, Mangold N, Faccanoni G, Lucas A for the transport of pyroclastic flows. Sedimentology 23:147–188
(2010) Erosion and mobility in granular collapse over sloping beds. Sparks RSJ, Gardeweg M, Calder E, Matthews SJ (1997) Erosion by
J Geophys Res 115:1–21. doi:10.1029/2009JF001462 pyroclastic flows on Lascar Volcano. Chile. Bull Volcanol 58:557–
McTaggart K (1960) The mobility of nuées ardentes. Am J Sci 258:369– 565
382 Valentine G, Buesch D, Fisher R (1989) Basal layered deposits of the
Montserrat S, Tamburrino A, Roche O, Niño Y (2012) Pore fluid pressure peach springs tuff, northwestern Arizona, USA. Bull Volcanol 51:
diffusion in defluidizing granular columns. J Geophys Res 117: 395–414
F02034. doi:10.1029/2011JF002164 Williams R, Branney MJ, Barry TL (2013) Temporal and spatial evolu-
Rhodes MJ (1998) Introduction to particle technology. John Wiley, tion of a waxing then waning catastrophic density current revealed
Chichester by chemical mapping. Geology 42:107–110. doi:10.1130/G34830.1
Roche O (2012) Depositional processes and gas pore pressure in pyro- Wilson C (1980) The role of fluidization in the emplacement of pyroclas-
clastic flows: an experimental perspective. Bull Volcanol 74:1807– tic flows: an experimental approach. J Volcanol Geotherm Res 8:
1820. doi:10.1007/s00445-012-0639-4 231–249
Roche O, Gilbertson MA, Phillips JC, Sparks RSJ (2006) The influence Wilson C, Houghton B, Kamp PJJ, McWilliams MO (1995) An excep-
of particle size on the flow of initially fluidized powders. Powder tionally widespread ignimbrite with implications for pyroclastic
Technol 166:167–174. doi:10.1016/j.powtec.2006.05.010 flow emplacement. Nature 378:605–607
Roche O, Montserrat S, Niño Y, Tamburrino A (2010) Pore fluid pressure Woods AW, Bursik M, Kurbatov A (1998) The interaction of ash flows
and internal kinematics of gravitational laboratory air-particle flows: with ridges. Bull Volcanol 60:38–51

&KDSLWUH$XWRIOXLGLVDWLRQSDUFKXWHOLEUHGHSDUWLFXOHVGDQVXQHFRORQQHG¶DLU

=
+,

9&

!

+

'
M *E

.
%

%

9

/ONO

:O!Ô
"
O!OOOP

/O!/OOQ

"

"

9

%

!

-

"

%

.
"

`

)
W
9

.


͸ǤͳǤ

M
^ % %

"
-
%

#

"

%

9

]Me
:OOe
#

-

-

OM!^

" OOU!
L

"

"

"

%

<

%

"

1

%

%
-

%
#

.

%
%

"

.
%

%
:

"

%

!
%

%

G

"

"

@
:N]J )
X

:N]]J @
G
:N]UJ
G
@
:NUOJ )
:NU/J @
:NU^J X

+

:NNOJ
c"/OOJX
/OOUJ)

/O:/

%

"KH

" &
-
"K

%

W"
^: 0

"

-

%

-

%

-

"
-


&ŝŐƵƌĞϲ͘ϭ͗ĚŝƐƉŽƐŝƚŝĨĞǆƉĠƌŝŵĞŶƚĂůƵƚŝůŝƐĠƉŽƵƌĠƚƵĚŝĞƌůĂĐŚƵƚĞĚĞŵĠůĂŶŐĞƐŐƌĂŶƵůĂŝƌĞƐ;ǀŽŝƌĐŚĂƉŝƚƌĞϯƉŽƵƌ
ĚĠƚĂŝůƐͿ͘>ĂďĂƐĞĚƵůŝƚĚĞƉĂƌƚŝĐƵůĞĞƐƚŝŶŝƚŝĂůĞŵĞŶƚƉŽƐŝƚŝŽŶŶĠĞƵŶĞŚĂƵƚĞƵƌĚĞϮϬĐŵŽƵϵϬĐŵƐĞůŽŶůĞƐ
ĞǆƉĠƌŝĞŶĐĞƐ͘

%

-

% %

/O
NO

"

O!'MP

:O Ô

/O /OOQ 9

&

%

S:OO P
%
#
"

%

X

/OO'

-

%

` % "


"

3
^:9

W"
^:

-

dĂďůĞĂƵ ϲ͘ϭ͗ ŐĂŵŵĞ ĚĞ ƚĂŝůůĞ Ğƚ ĚŝĂŵğƚƌĞƐ ŵĠĚŝĂŶƐ ĐŽƌƌĞƐƉŽŶĚĂŶƚ ;ǀŽŝƌ ĐŚĂƉŝƚƌĞ ϯͿ͘ >Ğ ĚŝĂŵğƚƌĞ ŵĠĚŝĂŶ
ĐŽŶǀĞƌƚŝ ĞŶ ĠĐŚĞůůĞ ůŽŐĂƌŝƚŚŵŝƋƵĞ ĂŝŶƐŝ ƋƵĞ ůĞ ƚƌŝ ;ŽƵ ůĂ ĚŝƐƉĞƌƐŝŽŶͿ ŐƌĂŶƵůŽŵĠƚƌŝƋƵĞ ĐŽƌƌĞƐƉŽŶĚĂŶƚƐ ƐŽŶƚ
ĠŐĂůĞŵĞŶƚŝŶĚŝƋƵĠƐĞŶĐŽŵƉĂƌĂŝƐŽŶĂǀĞĐů͛ĠĐŚĂŶƚŝůůŽŶĚ͛ŝŐŶŝŵďƌŝƚĞ͘
ŝůůĞƐĚĞǀĞƌƌĞ /ŐŶŝŵďƌŝƚĞĂ
ŐĂŵŵĞĚĞ
ϬͲϰϱ ϰϱͲϵϬ ϱϯͲϭϬϲ ϭϬϲͲϮϭϮ ϮϱϬͲϰϮϱ ϲϬϬͲϴϬϬ ϯϬϬϬ ϬͲϰϬϬϬ
ƚĂŝůůĞƐ;ђŵͿ
ĚŵĠĚŝĂŶ;ђŵͿ ϯϳ͘ϰϮ ϳϮ͘ϱϰ ϴϴ͘ϭϲ ϭϲϬ͘ϯϰ ϯϯϲ͘ϵϱ ϳϬϬ ϯϬϬϬ ϲϯ
DĚ;૖Ϳ ϰ͘ϳϰ ϯ͘ϳϵ ϯ͘ϱϬ Ϯ͘ϲϰ ϭ͘ϱϳ Ϭ͘ϱϭ Ͳϭ͘ϱϴ ϰ
ʍ ૖ Ϭ͘ϯϭ Ϭ͘ϭϱ Ϭ͘ϭϯ Ϭ͘Ϯϴ Ϭ͘ϭϱ ǆ ǆ Ϯ͘Ϯϵ
Ă
ĚŽŶŶĠĞƐĚĞƌƵŝƚƚĞƚĂů͘ϮϬϬϳ͘
͸ǤʹǤ ±
͸ǤʹǤͳǤ ±±
9

-

'M!NOP

"

=>

%

-

-

9
%

&

%

%

OO]
O:/

"

VRXVSUHVVLRQ

"

"

%

% %

"

%
W"
^/ :

W"
^


&ŝŐƵƌĞϲ͘Ϯ͗;ĂͿWŚĂƐĞƐĚĞƉƌĞƐƐŝŽŶƉŽƵƌƵŶĞĞǆƉĠƌŝĞŶĐĞĚĞĐŚƵƚĞĚĞϮϬĐŵĚ͛ƵŶůŝƚĚĞƉĂƌƚŝĐƵůĞƐĚĞůĂŐĂŵŵĞϰϱͲ
ϵϬђŵĞƚĚĞϮϬĐŵĚ͛ĠƉĂŝƐƐĞƵƌ͘ƚϰĐŽƌƌĞƐƉŽŶĚăůĂĨŝŶĚĞůĂƉŚĂƐĞĚĞĐŚƵƚĞĞƚƚĚĐŽƌƌĞƐƉŽŶĚ ăůĂĚƵƌĠĞ ĚĞĚĞ
ĚĠĐƌŽŝƐƐĂŶĐĞ ĚĞ ůĂ ϮğŵĞ ƐƵƌƉƌĞƐƐŝŽŶ͘ ;ďͿ ĐĂƉƚĞƵƌƐ ŝŵƉůŝƋƵĠƐ ĚĂŶƐ ůĂ ŵĞƐƵƌĞ ĚĞƐ ƉƌĞƐƐŝŽŶƐ ƉŽƵƌ ůĞƐ ĚŝĨĨĠƌĞŶƚĞƐ
ƉŚĂƐĞƐĚĞů͛ĞǆƉĠƌŝĞŶĐĞ͘


&ŝŐƵƌĞϲ͘ϯ͗ƌĞůĂƚŝŽŶĞŶƚƌĞůĞƐŵĞƐƵƌĞƐĚĞƉƌĞƐƐŝŽŶĞƚůĂǀŝĚĠŽƉŽƵƌƵŶĞĞǆƉĠƌŝĞŶĐĞĚĞĐŚƵƚĞĚĞϮϬĐŵĚĞŚĂƵƚĞƵƌ
Ě͛ƵŶ ůŝƚ ĚĞ ƉĂƌƚŝĐƵůĞƐ ĚĞ ϮϬ Đŵ Ě͛ĠƉĂŝƐƐĞƵƌ ĐŽŶƐƚŝƚƵĠ ĚĞ ďŝůůĞƐ ĚĞ ǀĞƌƌĞ ĚĞ ϰϱͲϵϬ ђŵ͘ /ŵĂŐĞ ƉƌŝƐĞ ;ĂͿ ĂǀĂŶƚ
ů͛ĞǆƉĠƌŝĞŶĐĞ͕ ;ďͿ ƉĞŶĚĂŶƚ ůĞ ŵŝŶŝŵƵŵ ĚĞ ƐŽƵƐͲƉƌĞƐƐŝŽŶ͕ ;ĐͿ ƉĞŶĚĂŶƚ ůĂ ϭğƌĞ ƐƵƌƉƌĞƐƐŝŽŶ͕ ;ĚͿ ă ůĂ ĨŝŶ ĚĞ ůĂ ϭğƌĞ
ƐƵƌƉƌĞƐƐŝŽŶ͕;ĞͿăůĂĨŝŶĚĞůĂĐŚƵƚĞƋƵŝĐŽƌƌĞƐƉŽŶĚĠŐĂůĞŵĞŶƚĂƵŵĂǆŝŵƵŵĚĞůĂϮğŵĞƐƵƌƉƌĞƐƐŝŽŶ͘>ĞƐĐŽƵƌďĞƐ
ĚĞƉƌĞƐƐŝŽŶƐŽŶƚŝĚĞŶƚŝƋƵĞƐăĐĞůůĞƐĚĞůĂĨŝŐƵƌĞϲ͘Ϯ͘
+

"

"


W"
^//
W"
^

!

%
"

qUH VXUSUHVVLRQ 9

"

<

%

%:M H9

%

%

G

%
OM

"

"

"

%

% W"
^'9

D

:

W"
^ 9
"
"

O'M H

&ŝŐƵƌĞϲ͘ϰ͗ƉĂƐƐĂŐĞĚĞůĂďƵůůĞĚ͛ĂŝƌăƚƌĂǀĞƌƐůĞůŝƚĚĞƉĂƌƚŝĐƵůĞƐĞŶĨŽŶĐƚŝŽŶĚƵƚĞŵƉƐ͘>͛ĞǆƉĠƌŝĞŶĐĞĞƐƚůĂŵġŵĞ
ƋƵĞƉŽƵƌůĞƐĨŝŐƵƌĞƐϲ͘ϮĞƚϲ͘ϯ͘
?
"

%

qPH VXUSUHVVLRQ

< %

%
-


"
!

W"
^/WILQDOW"
^
9

< !

-

&

%
"

9

/

WG

,

%
3PD[ W"
^/
G

O!'MP'M!NOP
M!:O^P

<
%"

%

"

O!'MP
'M!NOP[൒O

W"
^M

'M!NOP

"
:O^!/:/ PJ /MO!'/M PJ ÔO!UOO P .

&

<

!
"

%
W"
^M
/MO!'/MP

&ŝŐƵƌĞϲ͘ϱ͗'ĂƵĐŚĞ͗&ŽƌŵĞĚĞůĂƐƵƌĨĂĐĞĚƵĚĠƉƀƚĂƉƌğƐůĂĐŚƵƚĞĚ͛ƵŶĞĐŽůŽŶŶĞĚĞϲϬĐŵĚĞŚĂƵƚĐŽŶƐƚŝƚƵĠĞĚĞ
ďŝůůĞƐ ĚĞ ϰϱͲϵϬ ђŵ ă ĚŝĨĨĠƌĞŶƚƐ ƚĞŵƉƐ ĂƉƌğƐ ů͛ŽƵǀĞƌƚƵƌĞ ĚĞ ůĂ ƉŽƌƚĞ͗ Ϳ ϯ͘ϰƐ͕ Ϳ ϰ͘ϭƐ͕ Ϳ ϰ͘ϮϵƐ Ğƚ Ϳ ϱƐ ;ĨŝŶĂůͿ͘
ƌŽŝƚĞ͗ĨŽƌŵĞĚĞůĂƐƵƌĨĂĐĞĨŝŶĂůĞĚƵĚĠƉƀƚĂƉƌğƐůĂĐŚƵƚĞĚ͛ƵŶĞĐŽůŽŶŶĞĚĞϰϬĐŵĚĞŚĂƵƚĐŽŶƐƚŝƚƵĠĞĚĞďŝůůĞƐĚĞ
ϮϱϬͲϰϮϱђŵ͘


͸ǤʹǤʹǤ
9

-

/O
NO

*"

- "

:

-

2
• 9
"-
:

/

%
#

/O

W"
^^1
%

"-

NO

&ŝŐƵƌĞ ϲ͘ϲ͗DĞƐƵƌĞĐŽŶƚĞŵƉŽƌĂŝŶĞ ĚĞƐϭğƌĞĞƚϮğŵĞƐƵƌƉƌĞƐƐŝŽŶƐƉĂƌůĞƐĐĂƉƚĞƵƌƐƉŽƐŝƚŝŽŶŶĠƐăůĂďĂƐĞ
;ĐŽƵƌďĞďůĞƵĞͿĞƚϯĐŵ;ĐŽƵƌďĞƌŽƵŐĞͿĚĞůĂĐŽůŽŶŶĞ͘;ZͿĐŽƌƌĞƐƉŽŶĚĂƵĐĂƉƚĞƵƌƉŽƐŝƚŝŽŶŶĠĂƵŶŝǀĞĂƵĚƵ
ƌĠƐĞƌǀŽŝƌ͘>͛ĞǆƉĠƌŝĞŶĐĞĐŽƌƌĞƐƉŽŶĚăůĂĐŚƵƚĞĚĞƉƵŝƐƵŶĞŚĂƵƚĞƵƌĚĞϮϬĐŵĚ͛ƵŶůŝƚĚĞƉĂƌƚŝĐƵůĞƐĚĞϯϬ
ĐŵĚ͛ĠƉĂŝƐƐĞƵƌĐŽŶƐƚŝƚƵĠĞĚĞďŝůůĞƐĚĞǀĞƌƌĞƐĚĞϰϱͲϵϬђŵ͘
• 9

/O ̱!:OOO+
̱!OO+
NO J%W"
^]
W"
Û
• G

NO
%

9
/

<

%%


%

/

%

/ W"
Û
dĂďůĞĂƵϲ͘Ϯ͗ĂƌĂĐƚĠƌŝƐƚŝƋƵĞƐĚĞƐĚŝĨĨĠƌĞŶƚƐůŝƚƐĚĞƉĂƌƚŝĐƵůĞƐƵƚŝůŝƐĠƐĚĂŶƐůĞƐĞǆƉĠƌŝĞŶĐĞƐĂǀĞĐ͗,Ϭů͛ĠƉĂŝƐƐĞƵƌ
ŝŶŝƚŝĂůĞ͕ߩŵůĂŵĂƐƐĞǀŽůƵŵŝƋƵĞ͕ߝůĂƉŽƌŽƐŝƚĠŬůĂƉĞƌŵĠĂďŝůŝƚĠĚƵůŝƚĚĞƉĂƌƚŝĐƵůĞƐǤ
+WKpRULTXH +UHHO
SDUWLFXOHV SRLGVNJ ȡ NJP İ NPð
FP PR\HQFP P
:/OOO /O /:N] :^^ O'M /N/1!:/

O!'M
/'OOO 'O '/^' :'O] O'' /'^1!:/
ÔOO :O :O/O :']O O': ]ON1!:/
:/OOO /O /O' :']M O': ^NM1!:/
:UOOO O O^^ :'Û O': ]:^1!:/
'M!NO
/'OOO 'O 'O/M :'N: O'O ^'N1!:/
OOOO MO MOUN :']' O': ^NU1!:/
ÔOO Ô ÔM :'N: O'O ^']1!:/
:/OOO /O /:O/ :'/] O' :/^1!::
M!:O^
/'OOO 'O ':N] :'O O' :/'1!::
:/OOO /O /O^' :'M' O'/ ]/1!::
:O^!/:/
/'OOO 'O 'O] :'] O': '/1!::
:/OOO /O /:'/ :'O: O'' /OM1!:O
/MO!'/M
/'OOO 'O '/U^ :'OO O'' /O^1!:O
ÔO!UOO :/OOO /O /O'N :'^' O': ^]U1!:O
OOO :/OOO /O :N/ :MÔ OU U:U1!ON
U' /O /OO :O/] - -
"

:/OOO /UM /U'^ :OM' - -
,ϬƚŚĠŽƌŝƋƵĞĐŽƌƌĞƐƉŽŶĚĂƵǆĠƉĂŝƐƐĞƵƌƐͨŐĠŶĠƌŝƋƵĞƐͩƵƚŝůŝƐĠĞƐƉŽƵƌĚĠĐƌŝƌĞůĞƐĞǆƉĠƌŝĞŶĐĞƐ
,ϬƌĠĞůĞƐƚů͛ĠƉĂŝƐƐĞƵƌŝŶŝƚŝĂůĞŵĞƐƵƌĠĞƉŽƵƌĐŚĂƋƵĞĞǆƉĠƌŝĞŶĐĞĞƚƵƚŝůŝƐĠĞƉŽƵƌůĞƐĐĂůĐƵůƐ
ʌŵсƉŽŝĚƐͬǀŽůƵŵĞĂǀĞĐůĞǀŽůƵŵĞĚƵůŝƚсϬ͘ϮΎϬ͘ϮΎ,ϬƌĠĞů
ߝсʌŵͬʌƉ;ʌƉсϮϱϬϬŬŐŵͲϯͿ
ŬĞƐƚĐĂůĐƵůĠŐƌąĐĞăů͛ĠƋƵĂƚŝŽŶϮ͘ϲ;ĐŚĂƉŝƚƌĞϮͿ


͸ǤʹǤ͵Ǥ ǯ±
+

"

G

2
3 = 3/ = ρP J+ ^:
%
3

3/

-

J
"%

ɏP
+

%

%

W

"

&

O S 3H3/ S :
&

" %

3H3/

-

%
#

-

%

3H3/

-

-

% %

:O
Ô ^^C"

:O
'O ^/'C"

%

/O
NO

+

-

#
"

%

"

%

(UUHXU 6RXUFH GX UHQYRL
LQWURXYDEOH
'M!NO P 9

"

%

="
> +

-

9
"-

/O
NO

%

"

^]
Û


&ŝŐƵƌĞϲ͘ϳ͗DĞƐƵƌĞƐĚĞƉƌĞƐƐŝŽŶŝŶƚĞƌƐƚŝƚŝĞůůĞĞŶĨŽŶĐƚŝŽŶĚƵƚĞŵƉƐƉŽƵƌƵŶĞŚĂƵƚĞƵƌĚĞĐŚƵƚĞĚĞϮϬĐŵĞƚĚĞƐ
ůŝƚƐĚĞƉĂƌƚŝĐƵůĞƐĚĞϭϬăϲϬĐŵĚ͛ĠƉĂŝƐƐĞƵƌƐŝŶŝƚŝĂůĞƐ͕,Ϭ͘>ĞƐĠĐŚĞůůĞƐĚĞƚĞŵƉƐƐŽŶƚĚŝĨĨĠƌĞŶƚĞƐĞŶĨŽŶĐƚŝŽŶĚĞƐ
ĠƉĂŝƐƐĞƵƌƐ͘>ĂƉƌĞƐƐŝŽŶĚƵƌĠƐĞƌǀŽŝƌĞƐƚŵĞƐƵƌĠĞƉĂƌůĞƐĐĂƉƚĞƵƌƐăĚĞƐŚĂƵƚĞƵƌƐĚĞϯϬĐŵĞƚϲϱĐŵ͘


&ŝŐƵƌĞϲ͘ϴ͗DĞƐƵƌĞƐĚĞůĂƉƌĞƐƐŝŽŶŝŶƚĞƌƐƚŝƚŝĞůůĞĞŶĨŽŶĐƚŝŽŶĚƵƚĞŵƉƐƉŽƵƌƵŶĞŚĂƵƚĞƵƌĚĞĐŚƵƚĞĚĞϵϬĐŵĞƚĚĞƐ
ůŝƚƐĚĞƉĂƌƚŝĐƵůĞƐĚĞϭϬăϰϬĐŵĚ͛ĠƉĂŝƐƐĞƵƌƐŝŶŝƚŝĂůĞƐ͘>ĞƐĠĐŚĞůůĞƐĚĞƚĞŵƉƐƐŽŶƚĚŝĨĨĠƌĞŶƚĞƐĞŶĨŽŶĐƚŝŽŶĚĞƐ
ĠƉĂŝƐƐĞƵƌƐ͘>ĂƉƌĞƐƐŝŽŶĚĂŶƐůĞƌĠƐĞƌǀŽŝƌĞƐƚŵĞƐƵƌĠĞƉĂƌůĞĐĂƉƚĞƵƌƉŽƐŝƚŝŽŶŶĠăƵŶĞŚĂƵƚĞƵƌĚĞϭϬϬĐŵ͘
9 !

1

%
:

-

3PD[

%:MOO++g:O
ÔOO++gÔ
W"
^] W"
^N 1
%
%

"

3PD[

W"

^N

,3PD[

&

G

%

+3PD[

W"
^N

% W"
^N

%

3PD[

3H3/aOU^9

:

"

%

W"
^:O
#
A

"

%
+

"


&ŝŐƵƌĞϲ͘ϵ͗sĂůĞƵƌƐŵĂǆŝŵĂůĞƐĚĞůĂƉƌĞŵŝğƌĞƐƵƌƉƌĞƐƐŝŽŶ͕WϭŵĂǆ͕ŵĞƐƵƌĠĞƐăůĂďĂƐĞĚĞůĂĐŽůŽŶŶĞ;ĂĞƚďͿĞƚă
ϭϬĐŵĂƵͲĚĞƐƐƵƐĚĞůĂƉŽƌƚĞ;ƌĠƐĞƌǀŽŝƌ͕ĐĞƚĚͿĞŶĨŽŶĐƚŝŽŶĚĞů͛ĠƉĂŝƐƐĞƵƌĚƵůŝƚĚĞƉĂƌƚŝĐƵůĞƐ;ďŝůůĞƐĚĞϰϱͲϵϬђŵͿ
ĂǀĞĐ;ĂͿĞƚ;ĐͿůĂƉŽƌƚĞƉůĂĐĠĞăϮϬĐŵĚĞŚĂƵƚĞƚ;ĐͿĞƚ;ĚͿăϵϬĐŵ͘>ĞƐůŽƐĂŶŐĞƐĞƚůĞƐĐƌŽŝǆƌĞƉƌĠƐĞŶƚĞŶƚůĞƐ
ŵġŵĞƐǀĂůĞƵƌƐĚĞƉƌĞƐƐŝŽŶĞŶĨŽŶĐƚŝŽŶ͕ƌĞƐƉĞĐƚŝǀĞŵĞŶƚ͕ĚĞů͛ĠƉĂŝƐƐĞƵƌŝŶŝƚŝĂůĞĚƵůŝƚĚĂŶƐůĞƌĠƐĞƌǀŽŝƌ;,ŝŶŝƚŝĂůͿ
ŽƵĚĞů͛ĠƉĂŝƐƐĞƵƌƌĞƐƚĂŶƚĞĂƵŵŽŵĞŶƚĚĞůĂŵĞƐƵƌĞĚĞWϭŵĂǆ;,WϭŵĂǆͿ͘WŽƵƌůĞĐĂƉƚĞƵƌĚƵƌĠƐĞƌǀŽŝƌ͕ůĞƐǀĂůĞƵƌƐĚĞ
,ϬĞƚ,WϭŵĂǆƐŽŶƚƌĠĚƵŝƚĞƐĚĞϭϬĐŵĚĞŵĂŶŝğƌĞăĐŽŶƐŝĚĠƌĞƌƐĞƵůĞŵĞŶƚů͛ĠƉĂŝƐƐĞƵƌĚƵůŝƚĂƵͲĚĞƐƐƵƐĚƵŶŝǀĞĂƵĚƵ
ĐĂƉƚĞƵƌ͘>ĂĚƌŽŝƚĞĐŽƌƌĞƐƉŽŶĚăůĂƉƌĞƐƐŝŽŶůŝƚŚŽƐƚĂƚŝƋƵĞƚŚĠŽƌŝƋƵĞ;W>сߩሻ ĂǀĞĐߩŵсϭϰϳϱŬŐŵͲϯ͘>ĂĨůğĐŚĞ
ĞŶ;ĂͿƌĞƉƌĠƐĞŶƚĞůĞƉĂƐƐĂŐĞĚĞ,Ϭă,WϭŵĂǆƉŽƵƌƵŶĞŵġŵĞĞǆƉĠƌŝĞŶĐĞ͘


&ŝŐƵƌĞϲ͘ϭϬ͗ĚƵƌĠĞĚĞůĂϭğƌĞƐƵƌƉƌĞƐƐŝŽŶƉŽƵƌĚĞƐďŝůůĞƐĚĞϰϱͲϵϬђŵĞŶĨŽŶĐƚŝŽŶĚĞů͛ĠƉĂŝƐƐĞƵƌŝŶŝƚŝĂůĞĚƵůŝƚĚĞ
ƉĂƌƚŝĐƵůĞƐ͕,Ϭ͘
9
-
/

"

"

%

%

:

W"
^:: 9

3/

9

/

WG "

" %

%

<

W"
^:/

&ŝŐƵƌĞϲ͘ϭϭ͗WƌĞƐƐŝŽŶŵĂǆŝŵĂůĞĚĞůĂϮğŵĞƐƵƌƉƌĞƐƐŝŽŶĞŶĨŽŶĐƚŝŽŶĚĞů͛ĠƉĂŝƐƐĞƵƌŝŶŝƚŝĂůĞĚƵůŝƚĚĞƉĂƌƚŝĐƵůĞƐ͕,Ϭ͘
>ĂĚƌŽŝƚĞƌĞƉƌĠƐĞŶƚĞůĂƉƌĞƐƐŝŽŶƚŚĠŽƌŝƋƵĞƉŽƵƌƋƵĞWͬW>сϭ


&ŝŐƵƌĞϲ͘ϭϮ͗ƵƌĠĞĚĞĚĠĐƌŽŝƐƐĂŶĐĞĚĞůĂϮğŵĞƐƵƌƉƌĞƐƐŝŽŶ͕ƚĚ͕ĞŶĨŽŶĐƚŝŽŶ;ĂͿĚĞů͛ĠƉĂŝƐƐĞƵƌŝŶŝƚŝĂůĞĚƵůŝƚĚĞ
ƉĂƌƚŝĐƵůĞ,Ϭ͕Ğƚ;ďͿĚĞ,Ϭϸ͘>ĞƐĚƌŽŝƚĞƐĚĞƚĞŶĚĂŶĐĞƉŽƵƌƵŶĞĐŚƵƚĞĚĞϮϬĐŵ;ďůĞƵͿĞƚϵϬĐŵ;ƌŽƵŐĞͿƐŽŶƚ
ƌĞƉƌĠƐĞŶƚĠĞƐĞŶƉŽŝŶƚŝůůĠĞƐ͕ĂǀĞĐůĞƐĠƋƵĂƚŝŽŶƐĞƚĐŽĞĨĨŝĐŝĞŶƚĚĞĐŽƌƌĠůĂƚŝŽŶƌĞƐƉĞĐƚŝĨƐ͘
͸ǤʹǤͶǤ
9

"
1

"

%

% "

@

/

%

-

"

O!'M P & 3
^:


%

"

@

f
% /O
'O 3
^/ 9

"

<`

S/MO!
'/M P

" &

.

'
M

&ŝŐƵƌĞϲ͘ϭϯ͗WƌĞƐƐŝŽŶĞŶĨŽŶĐƚŝŽŶĚƵƚĞŵƉƐƉŽƵƌĚŝĨĨĠƌĞŶƚĞƐŐĂŵŵĞƐĚĞƚĂŝůůĞĚĞƉĂƌƚŝĐƵůĞƐ͘,ĂƵƚĞƵƌĚĞĐŚƵƚĞĞƚ
ĠƉĂŝƐƐĞƵƌĚƵůŝƚĚĞƉĂƌƚŝĐƵůĞƐĚĞϮϬĐŵ͘>ĂůŝŐŶĞƉŽŝŶƚŝůůĠĞƌĞƉƌĠƐĞŶƚĞůĂƉƌĞƐƐŝŽŶůŝƚŚŽƐƚĂƚŝƋƵĞW>ăůĂďĂƐĞĚƵůŝƚ
ŝŶŝƚŝĂů;W>сϮϵϰϯWĂͿ͘EŽƚĞǌƋƵĞů͛ĠĐŚĞůůĞĚĞƚĞŵƉƐƉŽƵƌůĞƐƉĂƌƚŝĐƵůĞƐĚĞϬͲϰϱђŵĞƐƚƉůƵƐŐƌĂŶĚĞƋƵĞƉŽƵƌůĞƐ
ĂƵƚƌĞƐƚĂŝůůĞƐĚĞƉĂƌƚŝĐƵůĞ͘


&ŝŐƵƌĞ ϲ͘ϭϰ͗ DġŵĞ ůĠŐĞŶĚĞ ƋƵĞ ϲ͘ϭϯ ƉŽƵƌ ĚĞƐ ůŝƚƐ ĚĞ ƉĂƌƚŝĐƵůĞƐ Ě͛ĠƉĂŝƐƐĞƵƌ ,Ϭ ̱ ϰϬ Đŵ ;W> с ϱϴϴϲ WĂͿ͘ >ĞƐ
ƐŝŐŶĂƵǆĚĞƉƌĞƐƐŝŽŶĞŶƌĞŐŝƐƚƌĠƐƉĂƌůĞƐĐĂƉƚĞƵƌƐƉŽƐŝƚŝŽŶŶĠƐăůĂďĂƐĞĞƚͬŽƵăϯĐŵƉĞƵǀĞŶƚġƚƌĞĞŶƚŝğƌĞŵĞŶƚ
ŵĂƐƋƵĠƐƉĂƌůĞƐŝŐŶĂůĚƵĐĂƉƚĞƵƌăϭϱĐŵ;Ğǆ͗ϮϱϬͲϰϮϱђŵͿ͘

%

!

9 :

#

K

W"
^:W"
^:'

O^MOU

/O

-
3PD[H3/

OUM
:/
3/

+SPD[

O'
"

3PD[H3/ ̱O'/

O!'M

<


"
://]

G

O!'MP
.

]M
UOQ%

-

9/

<

1

:O^ P %

-

"

3/

%

&

"
W"
^:
W"
^:' 9

WG
"

+g'O
%:M

:O^!/:/
P
/O

O!'MP W"
^:M

%

"
Se

%

%

aU
#

'M!NO P

< "

%
+

W"
^:M

-

'M!NO P

&ŝŐƵƌĞϲ͘ϭϱ͗ĚƵƌĠĞĚĞůĂĚĠĐƌŽŝƐƐĂŶĐĞĚĞůĂƉƌĞƐƐŝŽŶĚĂŶƐůĞĚĠƉƀƚ;ϮğŵĞƐƵƌƉƌĞƐƐŝŽŶͿ͕ƚĚ͕͕ĞŶĨŽŶĐƚŝŽŶĚĞ,ϬϸƉŽƵƌ
ĚŝĨĨĠƌĞŶƚĞƐƚĂŝůůĞƐĚĞďŝůůĞƐĚĞǀĞƌƌĞ͘>ĞƐĚƌŽŝƚĞƐĞŶƉŽŝŶƚŝůůĠƐĐŽƌƌĞƐƉŽŶĚĞŶƚĂƵǆĚƌŽŝƚĞƐƉĂƐƐĂŶƚƉĂƌů͛ŽƌŝŐŝŶĞƋƵŝ
ƌĞĐŽƵƉĞŶƚůĞŵŝĞƵǆůĞƐĚŽŶŶĠĞƐƉŽƵƌĐŚĂƋƵĞƚĂŝůůĞĚĞƉĂƌƚŝĐƵůĞƐĞƚŚĂƵƚĞƵƌĚĞĐŚƵƚĞ;ĐŽƵůĞƵƌƐĐŽƌƌĞƐƉŽŶĚĂŶƚĞƐͿ͘


͸ǤʹǤͷǤ ±
9

%

'M!NOP
/O

-

%
"

9

:

"

%

"
W"
^:^

%

]MQ 'U\

3PD[H3/"

%

1

"K

"

2
39 = Q57 ^/
%
3

9
%
Q

5

%

"K 5̱ U:'$\!:!:
7

.

\
%
7
%

%

%

@.!9 2
3/ 3:
= ^
7/ 7:

:
/

%

-
%

"

9

%

2
7/
3/ = 3: ^'
7:
9
%

%

W"
^:^

3H3/ g :

!!

7g/OQ̱/N\

3ൌ
ൌ /N' + + g/O 9

%

%

7/
3/ = /N'; ^M
/N

%


7/
3/ = ^^
/N

&ŝŐƵƌĞ ϲ͘ϭϲ͗ ± ȋȀȌ
± ͺͻǦͿͶ υ ǯ± ǡ Ͷǡ ͸Ͷ Ǥ ȋȌ
ȋεɏͶȌȋȌ±ǯ±î
± >? ɏ@Ǥ ± ±
±ȋǤǤͷͶǦȌǤ±ǡǯ±±
ȋ Ȍ ǯ± ͷͶ ±
Ǥ
ȋ±ͼǤͼǡȌǤ


9
%

"

-

3PD[

-

K

W"
^:^

%

"

]MQ
9 /

%

9
%
-

%

W"
^:]

NO

K

-

:MOQ
/OOQ

&ŝŐƵƌĞ ϲ͘ϭϳ͗ ;ĂͿ sĂůĞƵƌ ŵĂǆŝŵĂůĞ ĚĞ ůĂ ϮğŵĞ ƐƵƌƉƌĞƐƐŝŽŶ ;ĚĠƉƀƚͿ ŶŽƌŵĂůŝƐĠ ƉĂƌ ůĂ ƉƌĞƐƐŝŽŶ ůŝƚŚŽƐƚĂƚŝƋƵĞ͕ ĞŶ
ĨŽŶĐƚŝŽŶĚĞůĂƚĞŵƉĠƌĂƚƵƌĞ͘;ďͿdĞŵƉƐĚĞĚŝĨĨƵƐŝŽŶĚĞůĂϮğŵĞƐƵƌƉƌĞƐƐŝŽŶĞŶĨŽŶĐƚŝŽŶĚĞůĂƚĞŵƉĠƌĂƚƵƌĞ͘WŽƵƌ;ĂͿ
Ğƚ;ďͿ͕ůŝƚƐĚĞϮϬĐŵĚ͛ĠƉĂŝƐƐĞƵƌĐŽŵƉŽƐĠƐĚĞďŝůůĞƐĚĞϰϱͲϵϬђŵ͘
͸ǤʹǤ͸Ǥ ±ǯ
9

"

]MQ %

-

%
5

-

9

"

:

%
#

-

/

UM

+g/UM W"
^:U

-

%

"

<
%
"

- W"
^:N9


< %

""

̱/!' H R

-

"

%

"

%

G

"

&

""

"

""

9
"

""

&

< &

!
%

%

W"
^M 9
<

K"

""

K

%

/O

NO

- ""
"B

"

<


&ŝŐƵƌĞϲ͘ϭϴ͗;ĂͲĚͿĠǀŽůƵƚŝŽŶĚĞůĂƉƌĞƐƐŝŽŶŝŶƚĞƌƐƚŝƚŝĞůůĞĞŶĨŽŶĐƚŝŽŶĚƵƚĞŵƉƐƉŽƵƌƵŶůŝƚĚ͛ŝŐŶŝŵďƌŝƚĞĚĞϮϴ͘ϱĐŵ
Ě͛ĠƉĂŝƐƐĞƵƌăĚŝĨĨĠƌĞŶƚĞƐƚĞŵƉĠƌĂƚƵƌĞƐŝŶŝƚŝĂůĞƐ͘>ĂŚĂƵƚĞƵƌĚĞĐŚƵƚĞĞƐƚĚĠƚĞƌŵŝŶĠĞƉĂƌůĂƉŽƐŝƚŝŽŶĚĞůĂƉŽƌƚĞ͘
;ĞͲĨͿ ŽŵƉĂƌĂŝƐŽŶ ĚĞƐ ƉƌŽĨŝůƐ ĚĞ ƉƌĞƐƐŝŽŶ ƉŽƵƌ ;ĞͿ ƵŶ ůŝƚ Ě͛ŝŐŶŝŵďƌŝƚĞ Ğƚ ;ĨͿ ƵŶ ůŝƚ ĚĞ ďŝůůĞƐ ĚĞ ϬͲϰϱ ђŵ ĚŽŶƚ
ů͛ĠƉĂŝƐƐĞƵƌ;ϮϬĐŵͿ͕ůĂƚĞŵƉĠƌĂƚƵƌĞ;ϴϬΣͿĞƚůĂŚĂƵƚĞƵƌĚĞĐŚƵƚĞ;ϮϬĐŵͿƐŽŶƚŝĚĞŶƚŝƋƵĞƐ͘ŶŽƚĞƌƋƵĞů͛ĠĐŚĞůůĞ
ĚĞƚĞŵƉƐŶ͛ĞƐƚƉĂƐŝĚĞŶƚŝƋƵĞĞŶƚƌĞĐĞƐĚĞƵǆŐƌĂƉŚŝƋƵĞƐ͘


&ŝŐƵƌĞ ϲ͘ϭϵ͗ ĨŽƌŵĂƚŝŽŶ Ě͛ƵŶ ĐŚĞŶĂů ƉĂƌ ůĞ

ƉĂƐƐĂŐĞ ĚĞ ů͛Ăŝƌ ĚĂŶƐ ƵŶ ůŝƚ Ě͛ŝŐŶŝŵďƌŝƚĞ ĚĞ
Ϯϴ͘ϱ Đŵ Ě͛ĠƉĂŝƐƐĞƵƌ ŝŶŝƚŝĂůĞ ƉŽƵƌ ƵŶĞ
ƚĞŵƉĠƌĂƚƵƌĞ ĚĞ ϴϱΣ Ğƚ ƵŶĞ ŚĂƵƚĞƵƌ ĚĞ ĐŚƵƚĞ
ĚĞϮϬĐŵ͘

9 :

"

%
- %

%
-

"

%

%

W"
^/O
9

"

%

%
.

#

!

% " %
%

1
%

3H3/
% "

%

%

̱O^ ̱ OUM

/O
NO G

-
-

NO

9

/

WG"

"

%

#

:^
^

+ g /UM J W"
^/:

%

%

%

-

"
<

:NOO
/NMO +

"

9

/

%

-
-


/O W"
^/:9

/

"
"

%

+

/O
-

"

%

O!'MP W"
^:U

O!'M P

"

<

-

%

-

U!:/e 9

"

<

a:

&ŝŐƵƌĞϲ͘ϮϬ͗ǀĂůĞƵƌŵĂǆŝŵĂůĞĚĞůĂϭğƌĞƐƵƌƉƌĞƐƐŝŽŶ͕WϭŵĂǆ͕ŶŽƌŵĂůŝƐĠĞƉĂƌůĂƉƌĞƐƐŝŽŶůŝƚŚŽƐƚĂƚŝƋƵĞĞŶ
ĨŽŶĐƚŝŽŶĚĞůĂƚĞŵƉĠƌĂƚƵƌĞƉŽƵƌƵŶůŝƚĚ͛ŝŐŶŝŵďƌŝƚĞĚ͛ĠƉĂŝƐƐĞƵƌ,Ϭ̱Ϯϴ͘ϱĐŵ͘>ĞůŽƐĂŶŐĞǀŝŽůĞƚĐŽƌƌĞƐƉŽŶĚă
ƵŶĞĞǆƉĠƌŝĞŶĐĞƉŽƵƌ,Ϭ̱ϮϬĐŵ͘

&ŝŐƵƌĞ ϲ͘Ϯϭ͗ ;ĂͿ ǀĂůĞƵƌ ŵĂǆŝŵĂůĞ ĚĞ ůĂ ϮğŵĞ ƐƵƌƉƌĞƐƐŝŽŶ͕ WϮŵĂǆ͕ ŶŽƌŵĂůŝƐĠĞ ƉĂƌ ůĂ ƉƌĞƐƐŝŽŶ ůŝƚŚŽƐƚĂƚŝƋƵĞ Ğƚ ;ďͿ
ĚƵƌĠĞ ĚĞ ĚĠĐƌŽŝƐƐĂŶĐĞ ĚĞ ůĂ ϮğŵĞ ƐƵƌƉƌĞƐƐŝŽŶ͕ ƚĚ͕ ĞŶ ĨŽŶĐƚŝŽŶ ĚĞ ůĂ ƚĞŵƉĠƌĂƚƵƌĞ ƉŽƵƌ ƵŶ ůŝƚ Ě͛ŝŐŶŝŵďƌŝƚĞ
Ě͛ĠƉĂŝƐƐĞƵƌ,Ϭ̱Ϯϴ͘ϱĐŵ͘>ĞůŽƐĂŶŐĞǀŝŽůĞƚĐŽƌƌĞƐƉŽŶĚăƵŶĞĞǆƉĠƌŝĞŶĐĞƉŽƵƌƵŶůŝƚĚĞϮϬĐŵĚ͛ĠƉĂŝƐƐĞƵƌ͘


͸Ǥ͵Ǥ
͸Ǥ͵ǤͳǤ Ǧ
9!

-

O:O/

%

!!

1

%

%

!!

%
!

"

%

S OM

9

!
"

%
%

%

̱ !Ô
!:]O +

%

O'N
O:]M

""

"
.

%

-

!OO !:MOO + 9 !

"
%

39

+
%

-

!

%

/O NO 9!

"

!

&

%
&

""

͸Ǥ͵ǤʹǤ ±ȋͳ°Ȍ
/D qUH VXUSUHVVLRQ "

%

!

"

W"
^/1

%
!!

"

%

)&/OOO1

!#

W"
^//

"

M
^9
&

%

"

#

"


%

" %

""

&

"
:

9

#

3H3/ ൌ:%

!

%

"
%

"

-

%

3H3/
"

:

""

1
%

OOOP
-

33/̱O'/
""

%

H

"
9
%

3H3/a:

-

S NO P
#

"

%

J @
:NU^ E
"

%

9%

#

9
%

<

%

%

/ H

O!'MP
:/M H

1

.

<

İ ൌ O'M İFRPSDFWp ̱ O'!O'

- <
Se

%
%

%

<

%

#

'

"

<

4DLU = 4GpS{W ^]
4DLU
4GpS{W

%

<
7

MU
'%

8 VD = 8 GpS{W : − ε Û
%
8VD %

8GpS{W %

"" < 9

%

%

: H
]O H

"

%

O!'M P
OOO P 1


%

-%

8PI

1" /:
/

3
^ 8PI/OQ
Δ3 8 UHO : − ε / ρ I 8 UHO / : − ε
= :MO / + :]M ^N
+ G ε G ε

%

8 PI : − ε / ρ I 8 PI / : − ε 3PI
:MO + :]M − = O ^:O
G/ ε G ε +
%
G

İ 8UHO %

%

ȡI

%

% .

%

3PI

"
%

3PI ̱3/
3PIH+

8PI

+9

%

% "

- %

:O^!/:/P

%

8PI

/MO!'/M P ̱:^ H +

%

%

]O H
:]O H

-

-

" %

3H3/ ̱
O'/
dĂďůĞĂƵ ϲ͘ϯ͗ sŝƚĞƐƐĞ ŵŝŶŝŵƵŵ ĚĞ ĨůƵŝĚŝƐĂƚŝŽŶ ă ϮϬΣ Ğƚ ϮϬϬΣ ĐĂůĐƵůĠĞƐ ;ĠƋƵĂƚŝŽŶ ϲ͘ϭϬͿ ƉŽƵƌ ůĞƐ ĚŝĨĨĠƌĞŶƚĞƐ
ŐĂŵŵĞƐĚĞƚĂŝůůĞƐĚĞďŝůůĞƐĚĞǀĞƌƌĞƵƚŝůŝƐĠĞƐĚĂŶƐůĞƐĞǆƉĠƌŝĞŶĐĞƐĞƚƉŽƵƌƵŶĞƉŽƌŽƐŝƚĠ͕ߝсϬ͘ϰϱ͘

ŝĂŵğƚƌĞĚĞƐ
ϬͲϰϱ ϰϱͲϵϬ ϱϯͲϭϬϲ ϭϬϲͲϮϭϮ ϮϱϬͲϰϮϱ ϲϬϬͲϴϬϬ ϯϬϬϬ
ƉĂƌƚŝĐƵůĞƐ;ђŵͿ
hŵĨϮϬΣ;ĐŵͬƐͿ Ϭ͘ϮϬ Ϭ͘ϳϳ ϭ͘ϭϯ ϯ͘ϴϴ ϭϱ͘ϱϭ ϰϴ͘ϴϬ ϭϲϳ͘ϳϲ
hŵĨϮϬϬΣ;ĐŵͬƐͿ Ϭ͘ϭϱ Ϭ͘ϱϱ Ϭ͘ϴϮ Ϯ͘ϴϭ ϭϭ͘ϲϴ ϰϯ͘ϰϲ ϮϬϬ͘ϲϵ

9
"

"

! 1

"

%

-

% W"
M
%

"
E

"

R

"

%

"
9


"

%

-

"

#

%

"KH

K

"

=VOXJJLQJ>
.

%

"

.

-

E

+ "

"
/
.

+H/ a / 5" /OO

L
-

"

%

+H/

OM

-

E

-

"

"

!"
-

9 %

:

h
%

%%

+

-

%
"

O!'M
P

%

"

%

-

&

"

-

-

#

%

#

% "

/OO] @
/OOU

%

"K!

%

"
% %

%

%

"

" %


͸Ǥ͵Ǥ͵Ǥ ±Øȋʹ°Ȍ
/D qPH VXUSUHVVLRQ

%

<

1

"

,
<

#

:

W"
^^

-

/

"

"

-

:

%

%

<
""

"
9"

#

"

%

H"
<

W"
^// ?

-

&
""

X
/O:/
+

:O^ P
"

-

<3PD[
"

%
3PD[H3/gOM!:

%
"

W"
^::

#
A

&

%

<

3PD[

+

<
"

"

:O^ P /

3 H3/ S O/M %

-
W"
^:W"
^:'

"

%%

<

"

9

D

<
#

9

"

-

%

%
%

!
"

<


&ŝŐƵƌĞϲ͘ϮϮ͗ŝŶƚĞƌƉƌĠƚĂƚŝŽŶĚĞƐĚĞƵǆĠƉŝƐŽĚĞƐĚĞƐƵƌƉƌĞƐƐŝŽŶƐĞŶƌĞŐŝƐƚƌĠƐƉĂƌůĞƐĐĂƉƚĞƵƌƐ͘;ĂͿϭğƌĞƐƵƌƉƌĞƐƐŝŽŶ
ŵĞƐƵƌĠĞĚĂŶƐůĞƌĠƐĞƌǀŽŝƌĞƚ ĂƉƉĂƌĞŶƚĠĂƵŵĠĐĂŶŝƐŵĞĚ͛ĂƵƚŽͲĨůƵŝĚŝƐĂƚŝŽŶƉŽƵƌƵŶĠĐŽƵůĞŵĞŶƚƐĞƉƌŽƉĂŐĞĂŶƚ
ƐƵƌƵŶĨŽŶĚƌƵŐƵĞƵǆ;ǀŽŝƌĐŚĂƉŝƚƌĞƐϰĞƚϱͿ͘;ďͿϮğŵĞƐƵƌƉƌĞƐƐŝŽŶŐĠŶĠƌĠĞƉĞŶĚĂŶƚůĂĐŽŵƉĂĐƚŝŽŶĚƵĚĠƉƀƚ͕ƋƵŝ
ƉĞƵƚġƚƌĞůŝĠăůĂĐŽŵƉĂĐƚŝŽŶĚĞů͛ĠĐŽƵůĞŵĞŶƚĂƉƌğƐƵŶĞĐŚƵƚĞ͘


͸Ǥ͵Ǥ͵ǤͳǤ ±Ø
9

<
"

O!'MP̱ :/

:O^!/:/P+"

%

:^
^

#
A

<
"
""

)"

%

" %

"

%

#

O!'MP

"

3
^: G

%

"

%

#

"

"
3
^:
/OO] %

Oe

S:^ P
"

:e

O!'MPE

"

"

O!
'MP

-"
&
< U!
:/e "

e

%

O!'M P 1

"

- <

"

+"
%

̱
O:H

#

̱:Oe

%

#

:H
/OO]

%

O!'MP%

%
SOOMH

-

9

-

"
-

%

#

!""

0
/OO]
-

"

"

-

-
#
"

"
ɏ ̱:OMOC"
"

:'OO!:MOO C" !

%

/OO]


-

<

WG

<
W"
^:/
W"
^:M

%

%-E%
:NN]

.

H

/OO]
X
/O:/
.
"KH

"
%
%
+ðH'

X
/O:/

%

'M!NO P

.
' /N
/
"

%

-
8VDS8PI

-

WGg+ðH'%

&G

-

G

.
'
-

%

%
.

F
$
"
:NMN
3 ° / 'W :H/ ∞ ° ª § Q+ · § ( Q + :) + · º °°½½

= ®: −
3O °
¦
+ Q =O °¯
® −: Q
«
«
LHUIF ¨ ¸
'W :H/ ¹
− LHUIF ¨ ¸ » ¾¾
'W :H/ ¹ ¼» °¿°
^::
¯ ¬ © © ¿
%
3

%
+
"

9

%#
-

9

2
3 § § π · ·
/
=
- ¨ − 'W ¨ ¸ ¸¸ ^:/
3O ¨ © / + ¹
© ¹
X
/O:/
F
$
"
:NMN.

'M!NOP

%

&
-

-

- 8VD ൑
8PI
+
-

,

<

W"
^/ 9

^:/

&-

-

%

'

`
%

-

%

H

W"

^/ !

W"
^/'R

O!'M


P
"
W"
^/!
%

'M!NO P + ൌ Ô
W"
^/' _

_&

'
%

'

!

'2

"
E

#
"

=
-#
> '
' & #

"

'

"

%

X
/O:/
)

/O:/

- 8VDa8PI

K
%
.

"

< +

'M!NO P

"

"

!#

"
%

!

E

-

X
/O:/

"

<
"

-

"K "

"

&

%

<

"
#

-

-

%

W"
^/'

<

%


/
&ŝŐƵƌĞϲ͘Ϯϯ͗ĚĠĐƌŽŝƐƐĂŶĐĞĚĞůĂƉƌĞƐƐŝŽŶĚĂŶƐůĞĚĠƉƀƚĞŶĨŽŶĐƚŝŽŶĚƵƚĞŵƉƐƉŽƵƌĚŝĨĨĠƌĞŶƚĞƐƚĂŝůůĞƐĞƚŶĂƚƵƌĞ
ĚĞƐ ƉĂƌƚŝĐƵůĞƐ͘ >ĞƐ ĞǆƉĠƌŝĞŶĐĞƐ ƐŽŶƚ ĂũƵƐƚĠĞƐ ƉĂƌ ƵŶ ŵŽĚğůĞ ƚŚĠŽƌŝƋƵĞĚĞ ĚŝĨĨƵƐŝŽŶ ĚŽŶƚ ůĞ ĂĠƚĠĐŚŽŝƐŝ ƉĂƌ
ĞƐƐĂŝͲĞƌƌĞƵƌ;ǀŽŝƌƚĞǆƚĞƉŽƵƌĚĠƚĂŝůͿ͘YƵĂŶĚůĞŵŽĚğůĞŶĞƌĞƉƌĠƐĞŶƚĞƉĂƐďŝĞŶůĞƐĚŽŶŶĠĞƐĞǆƉĠƌŝŵĞŶƚĂůĞƐ͕ĚĞƵǆ
ĐŽƵƌďĞƐƚŚĠŽƌŝƋƵĞƐ͕ĐŽƌƌĞƐƉŽŶĚĂŶƚăĚĞƵǆĐŽĞĨĨŝĐŝĞŶƚƐĚĞĚŝĨĨƵƐŝŽŶͨĞǆƚƌġŵĞƐͩϭĞƚϮƐŽŶƚƌĞƉƌĠƐĞŶƚĠĞƐ͘

/


&ŝŐƵƌĞϲ͘Ϯϰ͗DġŵĞůĠŐĞŶĚĞƋƵĞϲ͘Ϯϯ͕ƉŽƵƌĚĞƐůŝƚƐ
ĚĞ ďŝůůĞƐ ĚĞ ǀĞƌƌĞ ĚĞ ϰϱͲϵϬ ђŵ Ě͛ĠƉĂŝƐƐĞƵƌƐ
ĚŝĨĨĠƌĞŶƚĞƐ͘
͸Ǥ͵Ǥ͵ǤʹǤ ǯ±±Ø
`%
%

"

"

%

<

<
:M W"

^:U

%

#

-

"

%

/U

:M

%
-

%
.

:M

<

#

"

W"

^/MG

#
%

&

"

<

%

,

%

-
"

- 9


"
%

%

<

=
!

>

%

%

&
,

W"
^/^

%

"

%

O!'M P W"
^/M

"
'M!NOPW"
^/M

-

&ŝŐƵƌĞ ϲ͘Ϯϱ͗ǀƵĞ ƌĂƉƉƌŽĐŚĠĞƐƵƌƵŶĞƉĂƌƚŝĞ ĚĞƐĐŽƵƌďĞƐ ĚĞĚŝĨĨƵƐŝŽŶĚĞůĂƉƌĞƐƐŝŽŶĚĂŶƐůĞĚĠƉƀƚĞŶ

ĨŽŶĐƚŝŽŶĚƵƚĞŵƉƐƉŽƵƌĚŝĨĨĠƌĞŶƚĞƐ ƉŽƐŝƚŝŽŶƐĚĞĐĂƉƚĞƵƌƐĞƚƉŽƵƌ;ĂͿ ĚĞů͛ŝŐŶŝŵďƌŝƚĞ͕;ďͿĚĞƐďŝůůĞƐĚĞ
ǀĞƌƌĞĚĞϬͲϰϱђŵĞƚ;ĐͿĚĞƐďŝůůĞƐĚĞǀĞƌƌĞĚĞϰϱͲϵϬђŵ͘


&ŝŐƵƌĞϲ͘Ϯϲ͗ĠǀŽůƵƚŝŽŶƐƵƉƉŽƐĠĞĚĞůĂƉƌĞƐƐŝŽŶĞƚĚĞů͛ĠƚĂƚĚĞĐŽŵƉĂĐƚŝŽŶĚƵĚĠƉƀƚĞŶƚƌĞůĂďĂƐĞĞƚϭϱĐŵƉŽƵƌ
ů͛ŝŐŶŝŵďƌŝƚĞ͘W>ƌĞƉƌĠƐĞŶƚĞůĂƉƌĞƐƐŝŽŶůŝƚŚŽƐƚĂƚŝƋƵĞĞǆĞƌĐĠĞƉĂƌůĞĚĠƉƀƚĞƚĐƉƌĞƉƌĠƐĞŶƚĞůĂĐŽŵƉĂĐƚŝŽŶŵĂǆŝŵĂůĞ
ĚƵĚĠƉƀƚ͘
͸Ǥ͵ǤͶǤ
`%%

"

*&

:

!!

%

"

-

"

%

" %

33/ a :M 7 g /OOQ 1
%

-

<

% "

%

#

"K

%W"
^:^R
"K

-

%

%
%


%
"K 39gQ577
%

"K

%

"K "

%
7 ^^
9%
%
-

:

3PD[ %

7 a ]MQ %

%

%

W"
^:^
""

%

%

"K

-

"

"K

%

9

-

%

%

% ""

"

&

3H3/

:

R 7 S ]MQ
%

%
-

#

- %

"

R

%

"K

"

@
:NU^
9

%

-

".

-

-
-
"

-!

%

"

-

V7
:NÊ
V
%
C
/OO]

"

7

/OO]

"

-
""

"

@

K7
/O:O
9

"

+
-

/OQ /NC
/OOQ ']\%
"

:U:;:O!M+
/M];:O!M +

#

%

:/: C" !
O]MC"! W"
^/]1" ^N%
/

%

%

"

% "K %

"

%

"


X
.
5HS
//
R 5HS S :!:O "

"K
% RX

a :OOO!/OOO X"

"K

%

R:OS5HS/OOO
-

-
%

5H
%
:O
%

1

"

%

5H

"

8PI

"
5H

% :
MOO P

%

O'M
O^M 9
%

/OQ
/OOQ

"

"B
^:O

3
^

&ŝŐƵƌĞϲ͘Ϯϳ͗ĠǀŽůƵƚŝŽŶƚŚĠŽƌŝƋƵĞĚĞůĂǀŝƐĐŽƐŝƚĠĞƚĚĞůĂŵĂƐƐĞǀŽůƵŵŝƋƵĞĚĞů͛ĂŝƌĞŶĨŽŶĐƚŝŽŶĚĞƐĂƚĞŵƉĠƌĂƚƵƌĞ͘

-

%

&"

%

%

GS'/M!ÔOP
""

%
% "K%

""

1
%
#
%

%

"K!

-

"

%

""

%

"

%

:'

/OQ
/OOQ 3
^


͸ǤͶǤ
9

%

%
"

-
-
"
"
-
'
M2

%

""

!

"

'
M
K

^

-
-

.
9

"

%

9

" <

-

+

"
-

"

K

!

#

%

' M 9

"

WG
+ðH'1
%

"

-

%

8PI S

H

%

'M!NOP
:OÔ

'
M

S ^ ?

-

L

!

""

"

/OO'J["
/OONE

!
%

-

"

9/

"

<

-

%

"
%

+

:N]
`C
W& :NN

%

%

"
+

:NO/
)
?K

:NN:%%

"

.
&


!#
"

! G%

-
-

#

#

"

""

.

!

"

"

%
#

&

.

!

" 1 "

"
WG ̱O

/UM

-

%"

.

%

%
K

% "
/U
/
/OO]J X

/O:/
9
"
"

"

"

"
%
"
!
.

%

.

`C
W& :NN 7

%

! "

E

%

K

%

"

-

/OOQ

"

L

.

%

%
MOOQ 5":NU]J9

/O::

%

-

G

.

-

-

"

%

"

%
"K +
-

%

-

MOOQ /OQ W"

^/]

8PI
3

%


%"

"

'

-

&

#

%

/OO]
͸ǤͷǤ
͸ǤͷǤͳǤ ±±±
9

!

.

%

"

/O

NO 9

%

-
-

2
;ϭͿ ?

& "

%

%

:
!
+

.

"
-

'
M
;ϮͿ ?

"

<
"
%

-

.

%

"

9

-
!
2
7DLOOHGHVSDUWLFXOHV
9

%

O!'M P

.

SMOO P 9

"

%
%

"
%

8PI
"
"
+

S:OOP8PI

#

#

-

'
M 1
%


"

" 9

"

/MO!'/M P

%
!""

:OOP WGa:M
[ ൒/O +"

%

#

WG
"

O!
'MP+g/O
""

:

"

%

%

"

SOP

"

%

/

<

- &:/e
-

-

"

<
+DXWHXUGHFKXWH
` %
%

" %

/O
NO

-

<
/

%

%

+"

%

"

""

"

"
<+

%
#

""

"
#

(SDLVVHXUGXOLWGHSDUWLFXOHV
+

'M!NO P %
-

:

"

%

+

3H3/ a OU^
"

+ g :O!Ô

"
9

<

+ð

)&:NN^JE%
:NN]JX
/O:/


7HPSpUDWXUHGXOLWGHSDUWLFXOHV
9

"

%

"

"

#

%

"KH

8PI

"

+H+9a:
͸ǤͷǤʹǤ
+
-

!

.

-

#

2
E

-

"

%
.

'M!NO P +g:O!Ô

%

%

%
-

9
-
"
"

<

"

E %
"

"
-

3H3/

9

#

#

"

'M!NOPWG
S/

+g/O
WGa :O+gÔ

%

%

E

%

-

! %

%

%

"

%

--

%
"

H
-"
2
• 9
"

%

ɏ̱:'OO!:MOOC"
!
"
ɏ ̱:OOOC"!
#

%

%

̱/MOO C" ! "

ɏ൏:OOOC"!

- ɏ൐/OOOC"!


• "

"

"

O!'OOO
P
•

!

%

%

"

E

%

-

+
-

2
• ?

- .

-

#

• G"

"

%

.

%

"
.

• ?

E

<

-

"

C"

&

-

%

"

%

.


1RPHQFODWXUH

.

7E
W

WG

/

G

GPpGLDQ

+

+

+3PD[
3PD[

/ "

3

+
3/

+
3PD[ %
-

:

+
3PD[ %
-

/

+
7

\Q
İ
% .

+
ȡP
%

C"!
ȡS
%

C"!
ȡI
%

C"!
ȡD
%

C"!
J "%

!/
N

/
'

.
/!:
9 %

4DLU % !:
4GpS{W

< !:
8VD %

!:
8GpS{W %

< !:
8PI %

!:
8PE %

" !:
5HS

X
.

ߚ

+!:

&RQFOXVLRQVJpQpUDOHVHWSHUVSHFWLYHV

"

B
9

-

"

.
9

.
#

-

5 :NUOJ

/OO]J@
/OOU&
-
"K%

1
%

%-

%

"K

.

"
9

""

.

%

!

"

"
-

-

"B

-"

2
Ͳ :"

OMqM

S
"

"

"

'

M
Ͳ /

^
ͳǤ ±±±
ͳǤͳǤ

X

/OOUJ/O:O
X
/O:/

-

<


"

"

%- %

%

%

"

-

! E
"

"
%

"

"

#

%
#

X

/O:O% %
"

.

%

'M!NOP^

9

-

"

%

9

%

'M!NOP

-

%

-

%

%

f!

"
%

9

9

%

"B

3H3/ g O]!: %
+9

-

#

!

"

"
-

!

2
: 9

"
-

+"

"

"

!

"
:M

G

"
-

& '

!

R"

a

%

1
%

"
/MO!'/MP

"


/ 1

%

-

"

! "

%
"

%

%

&

8̱ /J+O:H/%
+O

%JX

/OOU1
%

"

"

H

%
"

"
- #

"

<

#

9 #

"

"

+ "

"

#

"

!

9

<

K
-

%

"

"

'G
%

#

/! %
%

#

"
!

"
-

&^
"

""

<

G
"

!

9

-

!

"
9

"

"
%

%

-

!

"
9"

%
"
"

#

-

#

!

"

%

%
"

.
#
%


±

-

%

!

-

"

%

OQOQ

"
-

9

.

"
2

% ,

%

O!
MQ&-

%
K

9

"

#

!
"
"
-

3H3/gO]!:

#

!

"

%

"

%"
#

K

"
-

"
-

UUHUV

%

Ʌa:O!:/Q

%

:^!/OQ

"

"

9

-

% +

-
"

"

"

"

%

"

"

"
-

#

%

.

G

#

#

#

"

E

"

"

-

%
"

%
.
!
"

-2) 3"":NÔJ
5 5C
:NU/J 7 /OOM 9


!
#

& OQ

"

""

%

%

-
-
%

S M H

&
K

?

%
<

"

/UMQ
<

"f

"
%

%

&

< SS :

ͳǤʹǤǦ

-

!

"

"

"

.
9

/O
NO

[g:O!Ô

O!'MP

/O/OOQ9

%

!

-

% "

"

-

%

2
¾
!
%
SO/

"

%
¾
:

&

%
!

¾
/

<

"

%
:

"

"
&
O


9!

"
.
1

#

:

7

%

%

%

%

9

-

%

%

#

%

-

"
- ?

#

/O :

+g:O!Ô

"
G S ÔO!UOO P

""

!

#
"

%

%

%

9/

%

<

%

%

"
<
?

< D %

/

1

#

%

%

S/MO!
'/M P <

%
""

R

%

"
Ga:OOP

Ca:O!::/

<

<

%

G

GS:OOPJCS:O!:: /

<

!

%

3H3/ g O'!ON

! "

WGLII ‫ ן‬+b +
"


O!'MP
<

-

%

&:/
e
e +

9
%

-

<

"

"

-

"

"

"

%

#

"

-

"
%

""

%

:

!!

= >

ʹǤ ±
9

-

""

"

.

%

.

`C
W& :NN

.

) ?K

:NN:
"

"

"

"

%
"

&
%

<

%

%

"

%

-!

-

%

#

<

%

"

9

.

#

K

%

%

S:O!://

"
%

"
%

G!
%

%

"

%


"

%

C
%
K

C

-

2
@ G"

/O::J \

E"
`%

d
5
:NNM+
7"3
V

:NUN
9

#

.

.
1

%

%

"

/OO]J X
/O:/ 9

""

!
#

%

S:OO!:MOP1
%

-

8PI "

-

!

#
"

!

%
"

`

!

-

.

%

%

9

.

"

a
/OOQ
""

%

-

#

#

%

-

-
9

"

%

-

E

K

#
"

%

"

9

%


%

͵Ǥ
+

"

!

"
-

"

"

!#
%"

"
""

"

9"
" "

-

"

!

9

<H

"

M "

#

.

%
<

<

"

E

"

"

"

<

%

E"

#

"

#

-

2
"

9

"
#

.

%

%

< "

!

-
&

%

<

+

"

-

-

-

%

*

-%

#

9


%

"

-
&

"

%

-

"

"

9

%

#

%-

&
"

.
G

"

!

)
/O:M

-

9

!

"

"K
B

+

:N]J7C:N]UJW
/OO
% 5 :NUO

.
?

%

G

-
#

%

-

"K
B

-

%

-

+%

%

%

%

%

%"

%

"K %

K %

9

%

#

-

=%
>

%

%

5pIpUHQFHVELEOLRJUDSKLTXHV

ZĠĨĠƌĞŶĐĞƐďŝďůŝŽŐƌĂƉŚŝƋƵĞƐ

G
GX @
:NUO 0
% "! K

F
E
["

-+F
3
/^2Mq'^
G% ) "F

0 :NN^ )" "
.

`

UO2:'^q:'U
G
$X9 :N]:)-".

"
" $7

+
':2N]q::
G
$X9 :NU'
%

". O27

.

. 1
%
G
M/U
G
F0$)") /O::1
"..
.

"
@
".N2:ONNq::O/
G
F 0$ )" ) /O:/ 1-

.

"
"
.
.
$ V
@

X
//M!//^2Oq''
G
F0$))" /O:/
%

"
+
G V::0!/]M

/O:/W)
"G@?7W !

]

GX5[0) :NUN0
+.7
@W
+

7

7

X$7

+
MN2:O^/q:O^N
G X 7 G + /OON 1

%
.

"
F+.X
%1]N2O:O'
G%GGc
$ :NU/0

-"%
. K+F
3

/2//q//


H
0 `$ \
F
XX /OOM @
F
$ W )

MU2NN!'/U
0
+99
G)K
++
+7 /OOU7
! K

" F+F
3
".:U/ /q
0
y @ W
@
/OOU 3

% "
K

F

1"X

U^2MNq^N
0
% G V" 0 0
% ) /OO]

!"

.

.
2
0
K..:NM^)7[

:NUO
7
[)
:NN]

0V ^N2]O:q]'O
0
&30 :NÛ@%.

$W)
:2/ONq/'U
0
$\
\9
+

$!9V
&V"7 /O:'+. F

C"

2 "
!
%
"
3"%
1 0V ]^2UMU
0C
W :N//3

C" F3GE
1":'2':M!''/
0
K
X3[
[19
$X :NN+
!%
"%.
$W)

/MO2Nq^N
0@@@"G9&
$ :NN3
:NU'
!

)

%
$%E
2".
-

0V MM2/]q'/
00) C!9 + C`)

/O:'. .
.

2

!
Kz)
7[

5" ?7G$V @

X
/]^2:UNq/:'
0
. )$ \C
+ :NN/ G
"

2 "
%

"" "

!
F"

"!"
"
0V M'2MO'qM/O
0
. )$ \C
+ /OO/ +.
.

"
)
/]@
7 9:'
0FX$0
.)$ /OO'0."

.
2
1%

"
"
%

+"
3

.
E@
"./2''Mq''U


0FX$0
.)$ /O:E
F%

F

-
%

.!
.
.
F" F
"

%

.

"
3

0
V ]M2]/]
0FX$G
F@) /O:M
+.
.
E27"[

1 .
V
/
G
+
^:!^'U
0 @ 9

+ `
F 0
/OO] G %
"

K
%
F
C
F

"

1"7 ^/2U]qN^
0
:NN/E
.
%
"
F
. F@
7 G0:O'2::Nq:/O]
0"
G 0
"K @5 /OO/ X
" . F "
2

. .
.
1+
7 9
/O/2'OMq':U
0"
G /O:/G
!

. "

.
K
.
$V @

X
/'!/''2N]q:O^
0C
7++
50 :N/U@WF"+ C
:E1"
/O2::N^!
:/OO
0C )E 5 G5 :NN^ 3

.
. "
F 0
V MU2:]Mq:N
0C)E+G+10
/OOMG%

% "
FX
+"+.Û2/]:qO:

17
+
50 :NNN). . F"

7

[V )
@
.X
9
/^ M2M]qM'O

177CX7$@
F
") /OOO1
"
"

. F

"
9 V

$V @

X
:O'2/O:q/M

)E 9
$X 0 7 @ )[ /OO] L

%
. "%.

$W)
MU^2:!N

)E7"
)0 7 /O:/3F%

.
F !
%

@
. X

9
N29:'Ô


.7`7"[ :NU]3
%
"" "

"
]NG
V
% E.@
7 G0NN2Oq:'

.7`7"[7CX7$ :NUU1-

!

$
@
.X
N 0:/2:M:'!:M/U

.7`7"[)
%
507 :NN^+. F"
%

F
2G
-

:UU\C
0V M]2'NqM::
+ :N]W F""

3E
E"9:M2:MO!:^^
!`jk
K@0
.)$ /OOM+"
%

. %
.
"

F !!
%

K"2 3
d"K "

)
- 0V Û2q/O
F[7$
"
$ :NMN
/
L- ?%
.+

`
FcC
XGW 5" $ V :NU] V

2

{ [
9M/U
XGW5"[)`WC
09
+
)@@V
$@ /O::
3
F.
-

."
%
. F
/OU!)

@i E"
`5 G"
F
F .
0
V ]2:MUq:ÔN
\V7
0 /O:M)" "
E27"[
1 .

V
/
G
+
'MNq']:

7$@

X /O::

.
/OO^ C!!
F )
V $%E
7

".MU2:M]q:^:/

%
X
L /O:' G K . F "" "

F..
-

$@
.X
::N2:]^'q:]]^

%
X
L /O:M E

"

"

C
. F2" .
-

0V
]]2N^
- ) 3[ /OO/ G"

.

"
" .
.

F

"
"

7

".'N2NO]qN/U

)[.` :N]U)
"F
+F
3
/O2/O:q/OM


+ :NN: )"
!F%

. !"

2

@
".:N2::OUq::::

+
17 3[ [ X X
X 7C X7$ c" 7X :NNU
+. F"

."%.
:NN^qN]9%

7

[V )
@
.X
9
/M :U2'/M!'/U
@7 )
[$ /OO G K

-. .
K$@
.X
:OU 0:O2/']
J
.[ :UM^9

%

&1-
%

.

V

50[
[1 :NN^1

3"
.

F
`
U:2MONqM:/

50[
[1 :NNU9"! C @
"./^ N2UOqUO^
%
\R
.)5070 :N]U@F"%

:N]'

% W
"@
@
7 G0UN2^N!U'

0WV1%
mG /OOUG

"
C
%
$@
.X
17 ::2WO/O:'

"
X + X ) E%
/OO' @ %

"
!

2:
. $@
.X
:ON2WO:O:'
3[7CX7$ :NU/G-"

@

$
V @

X
:2:']q:]:
3[0 X :NU^9

"

"

9C

L
"$V @

X
/N2:q/
3[ :NN/1

:U).:NUO

1`1 )7
[

5"0V M'2MM'qM]/
3[ :NNM 7

"
%

% "
$V @

X
^M2/]qN
3[ :NNU+.
.
E2$7@
{X7$7C
3

. V 1@
7 97
+:'M2:'Mq:U/


3[
17
+ [ X+ 9" 7 ` @1X
9$ 7C
X7$V"0 /OO/7!%
".
. F
.

. "
7

[ V )
|2
% K23[0+\C
13
1
7
[V )
:NNM:NNN@
7 9)

/:2/^q/]N
3[0@9

+c
$@ /OO'3
K
% "
"

F
"@
.X
9
:29O/Ô'
3[ G% @ 0 @ 9

+ )
K W /OO] @

!"

. F
"

0V ^N2UU:qNO:

C $ )" ) /OOU E . F $ @
. X

::20ON/O]

C $ 5
-
$ )" ) /OON 3 . .
.
2
1-

! F
$@
.X
::'20::/O

C$1L"3X
L /O:M+.
.
E27"[

1 .
V
/
G
+
^:]q^/N
K
1
E@
)G /OOOG

" F%
K"
$W
)
'/'2:^Nq:NM
1

"
$"X5
[X+
X[ :NU^`!
-%
%
`
/2MNUqÔ/
1"7 :NM/W F" C

1"+"'U2UN!N'
1
0@ :N L

" "

"% .
%9

@
)

^2'Mq'N
1L"3C
G0`
G
/OOUW.
:NN]0-".
% )
5
E
$@
.X
71::20O/::
1L"3C
G0V"0
/O:/)
F. .

.
"
).:U:NUO
)7[

$@
.
X
71::]20O^/OU


W ) )"

. G X
L /O:' W
"
" F .

"
"
2 " .

-

$@
.X
17 ::N2MO'qM/
W-@3 ` /OO'X

F
." F
"
".

2
%
.
1+
7 9
//:2:N]q
/:
W

` :N/3
"

"

V
.
3
37C
`@
"7 3
+\7
:2:q]'
W
XV :N^^)

. FG$7 /^'2MOq^
W
XV[
C
@ :NU/)+

)
2).U/O:NO/. F
"
$V @

X
:2Nq]:
W
XV[
C
@ :NU)+

)
2).U/O:NO/. F
"
zX
.$V @

X
:N2:UOq:U'
W
XV 7 C
[!? :NU' +. X C 7"
0
[

" 0

[

"']/
W
XVL@L)[[
C
@ :NN). "
!%
. Fz

"
E.$V @

X
M^2/OMq
//O
W
XV :NNM
" .
2 K

$ V
@

X
^^2/M]q/^
W
c 3[ /OO V

%. .

!
. F )
5
E
1
+
7 9
/:'2M^:qM]'
W0@
X+@ /OO/3

"

" K

+F
3
:/M2/UqU
W +? @ 9@ :NU' G .

%

F

""
"
K
1"7 N :/2:^^]q:^]M
W +5X)$5C
@+9+XF) :N]'3
7+
7
+%

%

@
X
^2M]qUU


W +5 :NNV
G
.

%
L- L- ?%
.''
W
G :NNU3
"!"
"
E21-

"F

.
" C.
0 V MN2':'q
'M
W
G :NNNW "!"
"
+EEG.

F F!"
"
0V Ô2M'MqM^]
W
G 5 $`
. 7` /OOO E"
C!! F
E2
7"[
1 .
V
G
+
MU:qMNN
N
@X) /OO'L

" F1+.$1:'2':q^M
@
:NU^@WK3
".$5
.{

@
:N]3.
" K+F
3
]2/UMq/N/
@
G
GX :N]U["

K
F
"%
"

+F
3
:N2:q:^
@
5"Gc :NUMWK F
F"
"

%

zEE
1-

!

1"7 'O '2^Mq^^:
@$!9
5 /OO/W

"
2

-"+.X
%1^M
@
K77
$ /O:O3

K"

E@

$:2/Nq'
@
)Gc
$@ :NN^3
" K

2
!
-+F
3

UN2/Nq^
@@ :NNU3

".

%
"
2
533 X % E. $
V @

X
U]2/MMq/]
@
XW0 /OO]3
1
3
.E
E

W

1-0"0
% WK
0
1" EWpEE:]]q
:U'
@ /OOU .

.

3

?%
0
+ /'/


@ 9 3[ X
L \.C d /OOU +"

"
. F$@
.X
::20O//O/
@9X
L3[
3 /O:O+
%
.

. FF

. F0V ]/2]']q]MN
@93[X
L /O:M3F%

" .
F21

-
. . K
" F$
V @

X
/N^2:qN
@
9 G% 1 /OO] 9"
2 3

"
. %.@
.X
9
'29O]O:
@& 3 ` K!&

0 /OO )

. " F

2
"
1+.$1::2:']q:M]
@.$)`3GX /OOMG
.
K

"
"F"
!

F+ X7 G'^:2:'']q:']
@""XW :N//3

%
.
C
5"
O
[
5)@G :NU^X"!

03 ?7G@
7
G0N]2N^q'O
[
$55
X1 :NN:)"
F

9F

)

+
7" 3 GK `
% $ @
. X

N^ 0]2:/''!:/'Ô
["Rd"[d$ /OON1-

. K
F
"
F%F
"
1"7 ^'2//'q
//''
[
[17$1 :NUO3
" "%.
$W)
NN2]UM!]NN
<
E%
X) :NN]3
.
FX
%@
.M2/'Mq/N^
E%
X)X
)19[
X@9"))$10
9 /OOOG

%.

.7

/NO2M:qM:^


E%
X)
"
X+ /OO: WF %. K
"

!

:-

.$@
.X
:O^ 0:2M]qMM/
E%
X) 9" )
"
X+ /OO' @ %

"
!

2 /
-

$ @
. X

:ON
WO:O:M
P
$ \V5

G5 :NU]["!

-"

F
GE1$2:^NUq:]O^
$"1
9)[d :NUOE@
$X)
$) 1WK
+
+
`
FcC!/'
$G) :N]O.

"
".W

{ 7W G
M]]
Q
\
\ 7
" + + + 0 /OON 3
"
.

% . F . F F
!

"% 1 0V ]:2:OM]q:O]M
\
+1 :N^^3

C $@
]'2]N!U
\

"@[ :NM]G
-

.

F
C

F

?7`07X
M:Û
\

757
%0 :NUU1 F
"

)7
[

).:U:NUO$@
.X
N2:']N!:'U:^
\

00
.)$ :NNM7
["!
.3.

3 C)%
77

".'/2Ô]q^:^
\FC$!$
C70-
+$+ G9%"
W
7$@

X
+

\ C ` 0!7 G 7 /O: +-.

-
"
)
/O:O
2+

"!

"..
.
$V @

X
/^:2/Ôq/N'
\K
.$ :N/]}
C
9
"
5
0
2 G
"V
C
"
GF
"
0
F~
"[
!+
!3
C.


9&

1)"

.!
GV
$+ /OO' 7
" "
K
+.W:^2/]:!/U:
9&

1)
$0[.@) /OOM@"K
+.
W:]2:OO/
9
7 9
+

$!9 /O:M 3F
.
%

.
%

$ V @

X

/N]2::q/]
9
+
)@@
/O::["!

@i
3 @"
+
`5G"

.3X)
.
0V ]2:MMq:M^M
9
F \5 @ 1X 0
5 :N'N
WK
+
E 1"

':2::O'q:::]
9
!@

1 @& d
X /OO] E

.

"
%
$W)
MN2']MqMO'
9F
X :NU/7

@%.WF2EE
)
F7
X

["!
.3.
$7

+
M/2/]Nq/N]
9
@[
[17CX7$[F)G /OO'G-.

"
$W)
MOU2:]Mq:NN
9
@[
[17CX7$W
G /OOM
F!
"
+.X
%1]/2O':O:
9
@7$+K3W
G+
$`[

7
)W /OO]WF

. " F2 G .

-

:N]M
`"

`
Fd
$V @

X
:^:2:^Mq:U^
9
@[
[17CX7$W
G /O::@
F"

$W)
^]M2']qÛ
9 G )"

. G /OO] ). "

"
V
)

)@
.X
9
'29:O/O:
9 G )"

. G G
$+ /O:' W %
.!F
C
"

1

.
`M2':]


&
)&$$ :NN^1-

.
F2

!

':
)&$$ /OOO@%.!%
"
!!
! F

$7

X
]O :2^'qU
)
%
5
. 7` 7" [ \" $ :NN' +
"

%

"!

:UU
. F \C
V E
$@
.X
NN2N'U]!NMO'
)
%
5
. 7` 7" [ :NN^ 7

".
\C :UU

.
0V M]2M:/qM/N
)" ) +

G
C $ /O:O X"
" 2

.
0 V ]2/:q )"

. G
0 W3`V
$+0
)L /OO]`

"
!

" " F

%
!

$ @
. X

::/2WO/O:]
)"

. G X
L [" L )" ` W @ 9 G /O:O 1
. "
%
"
$ @
. X
1 7 ::M2
WOO'O
) + :NM G C

3!X % 3 X. 7 `d
^'2/q^^
)"1[ G$
9+0) :NN'3

%
K

%

%
F
+F
3
]N2:q:O
)$ /O:M9

"%

% "
3

?%
0
+
) 3""\ :NÔ3
.

G$7 /MU2^NqU/
)
XG5X07FG :NUU"

. F"
.!
C %

) 7[

% ?7G 0 V
MO2:'q/M
)
[$ :N]NG K2G
F"
"
8$@
.X
71
U'2]M:!]M/O


)
[$ :NN^. F
C" . K`
]N2Ô:!Ô^
)
3+7X9 :N]]7
. FGC CW

GC@
".M2:]q:]^
)
L :NU/E

@
.
G0F
.C"

+F
3
2U:qU]
)
7 3 G X
L `k c /O:/ +

K"" $@
.X
::]2WO/O'
)
$@X
$ :NU'".

).:U:NUO
-%

)7[

E20.$WX
1-%
V 2E
1%
[K 5" ` X

` G
. +

:!:'/
)
7)+ X

\ :N]] 3

-
"

"! K

+F
3
:U2/Nq/'U
T
`C 7 W& 3 :NN +
.

%. ?K
V $
`%

:NNO!`%

:NN:2
%
. F$V
@

X
M'2:Nq
`
F@07GF.0
/OOO:OOOOc

-%

)

V
$%2
"
$V @

X
:OO2NqMO
U
+

WG :N]3

)+

:N/N!:N/
"
E5"+'MU2:!
:/^
+ $ ["" G$ \

F
XX 3 `[ /OO^ 1
. "
-

1+
7 9
/'^2'^^q'UO
+ `G 7
%
09 L $) 9
F X) /OO^ 7
"

"
2E

"

"
"+.
X
%
F1]O:O'
+
)7+)9 :NNWK FL/

""

.
1" 7
'U2^:]q^/:


+
L :NNN 7 " F " F F "
+. W
:: 2M'/qM'U
V
R
79X
$\ /OO'XC"F
"
..
0V
^]2:/Nq:'

X
.9 :NM'WK"
"

"

%
"
G$7 /M/2M]]q^:'
X
) /OOU E +
3
".2 7
1 5
. { 7

'MO
X

\+

[5 :NNO3

. "! K

zE1-

%

1"7 'M2:^/]q:^N
X

\ :NN:
.
F
1
%
G9`
FcC
X

\1$1+

[5 :NN3

.
"! K

zEE3

%

.
%

5

F
F

1"7 'U2:Û]q:^N]
X
G :UN/
W K"
5
F

d
V

E"

^2N']qNM'
X
L@
)G+$7CX7$ /OO/1-

""
F . F.@
.X
9
/N2'O!''
X
L@
)G+$7CX7$ /OO'1-

. "! K

" F F . F

$ @
. X

:ON20:O/O:
X
L @
)G + $ 7C X7$ /OOM E%
%
!
. F

-

"!
-
1+
7
9
/'O2'O:q':'
X
L@
)G+$7CX7$ /OO^3

K

F .
F
+F
3
:^^2:^]q:]'


X
L)
7`kc3G /OOU1-

% F
!C

% . K

C" F

%

" ! . F$@
.X
::20://O
X
L )
7 `k c 3 G /O:O +

C
"%.!
F2E"

. . F$@
.X
::M20ON/O^
X
L G ) )"

. G 9 G /O:: L
!
"
.
"% F2E" K
"

-

1+

7 9
::2]MqUM
X
L /O:/

"

. F2

-

%
0V ]'2:UO]q:U/O
X
L`kc)"

.G0 0+ C`V

@G /O:.
!

%

.. F@
".':2::O]q
:::O
X
L /O:M`
%
. .
.

1+
7 9
':U2::Mq:/M
X
L@
)G+$7CX7$ /OO/1-

""
F . F.@
.X
9
/N2'O!''
X $5 7 $1 :NU @%.

.

%.
"
$W)
:M2NMq::O
X
5E3+
70`

% F
7"V @
0V 'O2/!U
X)V
KK9 :NNN+

E"

+
"
W
E.$V @

X
N:2:]Nq:NU
X 7 7 X9 :N^: G!WF 3 |2 3

L" @
" X

E
d
?7@
" 7%+ +^^2:q]]
X
[ :NNG.

"

7

%7
%
X
?7G77!3+!/M
XF
.+$X
L3[X /O:'1-

.

.
.

"
"! K
" F0V ]^2UMM


W
7[W&3`C7 :NN/"

%"

.
F?K
% `
Ô2^^'q^^^
7%"
70 9 \\ :NUU +
K

"
"

F .

"$W)
:UN2::qM
7 C
[!?W
XV5
G :N]G

"

9
7

@
.7

"./O2MMqM]'
7 51[X+3
X7
7)
K)9`$3 :NN^+. F

$
:M :NN:
) + E2 `
F @
+". X
W

) +
+
) + +
2 +
E

V ".7
".?%
. 5"M'MqM]O
7
7X)X`
F)75 $G :NU'V " .
"
3

:U:M@
".:/2^MNq^^
7
%
$+\ 3K ? X :NN] +
" +
7 0 C
G

9
7
%
$+\ 5
\" + /OOO E

2 G
%
F
+F
3
::2/^:q/Û
7
%
X9 :N^^7

C7

:M'2:^Nq:^'
7
%
X9 :NÛ3
0 CFC
@
7 G!7
+:OU']
7

90
"
$1@ C
[ :NNM3
:UU
!

G"

% GC^^2^]qNM
7
7 [ + :NUO WK
F
F

"
"+F
3
/^2:^:q:^]
7$1 :N]/1

F
.

"%.
$W)

M2]MN!]Û
7$10
X1 :N]N3
.

"%.
% "%

K
$W)
N'2']]q'NM
7$1 :NU^)-"

"%.
G )
^2/'Mq/M
7 $1 :NN] @%.

%

. "

?%
.+
"


7X9 :NÔd
dV5

GWFd
dV
5

GWF?7@7+ +M'!W2:'Nq:MN
7CX7$5C
@+9 :N]3
""

2.
. C
`
/':2^/q^'
7CX7$7
75C
@+9 :N]+ "

@
".: 2::Mq
::U
7CX7$ :N]^@K

%"

. F7

"./2:']q:UU
7CX7$59 :N]^G

"
."%

$@
7 9:/2'':q'M:
7C X7$ :N]U @

. F2

0V ':2:!N
7CX7$59 [

@ :N]U3

"
"

%

. F.
$@
.X
U2:]/]q
:]N
7CX7$7"[

.7` :NUO3

. F

E

" "
"
%

$ V
@

X
]2U]qN^
7CX7$7"[

.7` :NUO3

. F

EE

F
" $ V
@

X
]2N]q:OM
7CX7$W +5[

X+CX$L_"9L3
X7+"
X :NUM
E"

@
`5 G"
$ V @

X

/'2/OMq/'U
7CX7$0
K
X3[
[19
$X[F)G)
[+$ :NN
7

!
"%.
F
%
". .::'2/'q/M]
7CX7$ :NN]

K%

1
+
7 9
:MO2:]]q:UN
7CX7$0C)E

.7`@
$@K
977"[5G5 :NN]
V +
$5
.7

7CX7$@
F
")
17)
F7$ :NN]1.. F
9 V
0% MU2MM]qM^M


7
C )$ @
G9 :NNM .K F
" % F

"

J
X
C
3

L
"?7G0V M]2:M:q
:^N
7
$9K+VCE /OO]3
!!

.
.""

"E$`
G)
@

:2M/q
M'O
X
3".$!X
7 G3&
&
57 :NNUV %
2
%

0V Ô /2:]q:''
3
K"\ :N'3
. $5
.`
FcC
3 ` $ V

"
7 :NN' L
%
.
$ @
. X

NN2:M^q:M^''
Y
V

@G :NU]7
F. "
0V 'N2^:^qÔ
V

@G 0
W
X V :NUN 0 .

+
7"
3 F

GK?7G0V M:2NMq':'
V

@G W
X V /OOO +. "
E2 7" [

1 .
V
G
+
M]:qMUO
V5.C
V
07
7W +5\
K
.9 /OO:G"%
""

7
%
$V @

X
:OM2//Mq/']
VK
K $G L )[ /OO^ W
%

.
"

"
. "

[ 0
%
?7G $ V
@

X
:M'2///q/^
V
%
C1VCE3@ /OO]3

"
2
3
:N^V

E.$@
.X
::/2WOO/^
Z
5C
@+9
X :N]:3F+!.

GK
$@
7 9
:/]2:]qMM


5C
@+9[
"XW5$` :NUO9F!
"
`
/U2/U^q
/U]
5C
@+97
7W""+ :NU:3
".

"
2GC"

-

. F$V @

X
:O2:q::
5C
@+9 :NU E"
.
"

$ V @

X

:]2^MqUU
5C
@+9 :NUML"

"

%
$V
@

X
/M2:M]q:]:
5@0 :N^NL

!
F!
F) @F![`
FcC'OU
5

$ /OOOWK
".
%

!3
.H

+F
3
::2/Oq/':
5 X /O:O 1

.
.
%

"
".2

.!K
F
!@
3 "
+

E.+
9

?\?%
. 9

//'
5X0
.)$0.39 /O:'3

% F-"

F"
.

%

.
"@
".'/2:O]q
::O
5 $` :NUO 3

K

. F2 G

-

$V @

X
U2/:q/'N
5$`5C
@+9 :NU/E"

2
. .
F$@
7 9:N2MU:qMN/
5 $` :NU' 3

K

. F /2

-

$V @

X
/O2MMqU'
5$` :NUM3

31`
Fd
EE3
3E"
+3
X7 G)+.1"7 :'2//Nq:O
5$`["0W\+$$) 5)L :NNMG

-
.F

"
F . F

`
]U2ÔMqÔ]
5$`["0W /OOO+.

E27"[

1 .
V
G
+
M'MqMM'
5$`[
5 /OOG
"E"
2)
3
0"
0 C
03 $@
:::2^Mq^]O


59 :N]^1-%
V 1!!EEE+1@
.$E
'M2M'qMM^
59[
$5 :NU:)".
". . F

"

%E23
:NUO
7[

5"
9+5)
-X?7@7+ +:/MOM:!M/'
5
K \[ 7
)W :N]N G
. "
@
7 G 7
+
:UO2:]]q:N'
5 3 :U]U

-

K

/^$:U]]`

$ )

@
+
::q:^]
5 $G5"$V :NU:X

F

$V @

X
:O2:q
'
5 G5 0C ) \% G :NNU 3

F F "
0
V Ô2UqM:
5"$V5C
@+9 :N]]3
"

27"
!"

"!
@
".M2]/Nq]/
5" $V :NU: 3

X

E"
|2 G. E
E"
)&9
R
.0V ''2:UNq/:/
5"$V5C
@+9 :NU:1
"
2G
.
)
- $V @

X
N2:::q::
[
p
[!c@
:NNMWK WF

!

"
2

.
"

+F
3
U/2/^Nq/]]
\
c33C777 :NN+. F
:NN:
?K

% $0V MM2:^^q:]M
c" 5! /OO [C K !

.
)

CC
E
`
FcC
c
$@ :NN^1

"! K
1"7
M:2:^]q/OM


c
)V
[
)G\
$G) /OOM3

"

K @

1"7 Ô2'M^]q'MUO
cC.7 :N]')
)%

1

E+. WF G

$7 X
E@
?%3C7
0@
7 :/2:]q^/

Les Despot Pyroclastique Et Leur Dynamisme

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Les Despot Pyroclastique Et Leur Dynamisme

Transféré par

Droits d'auteur :

Formats disponibles

Mécanismes d’auto-fluidisation des écoulements

pyroclastiques : approche expérimentale

To cite this version:

HAL Id: tel-01401494

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

  

!$%&' %$%(&' ' $)'$' %(*'(&'

    !"!

   => . >9

(  2 1 3 4 .,, 

   "

)   2 V "

Introduction gé né rale 19

1.1. Généralités ............................................................................................................ 26

1.2. Mécanismes de formation des courants de densité pyroclastiques 27

1.3. Mécanismes de transport des courants de densité pyroclastiques 32

1.3.1. Transport des particules par un gaz turbulent .................................. 34

1.3.2. Transport dans un mélange gaz-particules dense ............................ 38

1.4. Sédimentation des courants de densité pyroclastiques ........................ 43

1.4.1. Sédimentation à partir d’un courant dilué et turbulent ................. 43

1.4.2. Sédimentation à partir d’un écoulement dense ................................. 44

1.4.2.1. Sédimentation « en masse » ................................................................................... 44

1.4.2.2. Sédimentation par aggradation progressive .................................................. 46

1.5. Architecture des dépôts .................................................................................... 48

1.5.1. Dépôts de déferlantes ................................................................................. 48

1.5.2. Dépôts d’écoulements pyroclastiques de faible volume ................. 50

1.6. La problématique des ignimbrites ................................................................ 52

1.6.1. Caractéristiques des dépôts ..................................................................... 52

1.6.2. Rhéomorphisme ........................................................................................... 56

1.6.3. Rapport de forme des ignimbrites : HARI et LARI ............................. 56

1.6.4. Capacité à franchir des obstacles ............................................................ 59

1.7. Mobilité des courants pyroclastiques .......................................................... 60

2.1. Le processus de fluidisation : historique et applications ...................... 64

2.2. Fluidisation d’un milieu granulaire par un gaz ........................................ 65

2.2.1. Principe ........................................................................................................... 65

2.2.2. Vitesse caractéristiques de fluidisation (U mf, Umb) ........................... 67

2.2.3. Les différents groupes de particules ..................................................... 71

2.3. Défluidisation par diffusion de pression .................................................... 76

2.4. Quelques paramètres influant sur la fluidisation .................................... 77

2.4.1. Forces interparticulaires ........................................................................... 77

2.4.2. Autres paramètres affectant la fluidisation ........................................ 79

2.5. Fluidisation des écoulements pyroclastiques ........................................... 82

2.5.1. Indices de fluidisation des écoulements pyroclastiques ................ 82

2.5.2. Sources de fluidisation des écoulements pyroclastiques ............... 83

2.5.3. Implications pour la mise en place des écoulements

3.1. Démarche scientifique ...................................................................................... 92

3.2. Etudes d’écoulements générés par le principe de « rupture de

3.2.1. Travaux précédents ..................................................................................... 93

3.2.1.1. Courants gravitaires fluides ................................................................................... 93

3.2.1.2. Ecoulements granulaires secs (i.e. non-fluidisés) ......................................... 94

3.2.1.3. Ecoulements granulaires fluidisés ...................................................................... 98

3.2.2. Dispositif expérimental « rupture de barrage »

3.2.3. Les particules ................................................................................................. 102

3.2.4. Rugosité du fond du chenal ....................................................................... 102

3.2.5. Protocole expérimental ............................................................................. 104

3.2.6. Dimensionnement des écoulements expérimentaux ...................... 106

3.3. Expériences de chute de lit de particules (chapitre 6) ........................... 111

3.3.1. Dispositif expérimental ............................................................................. 111

3.3.2. Protocole expérimental ............................................................................. 115

3.4. Capteurs de pression de fluide ....................................................................... 115

3.4.1. Caractéristiques des capteurs ................................................................. 115

3.4.2. Calibration des capteurs ............................................................................ 116

3.5. Granulométrie des particules ......................................................................... 119

!$%&' %$%(&' ' $)'$' %(*'(&'

!"!

=> . >9

( 21 3 4 .,,

"

) 2 V "

(IIRQGUHPHQWGHFRORQQHpUXSWLYH)LJXUH

'pERUGHPHQWRX©ERLOLQJRYHUª)LJXUH

(IIRQGUHPHQWG¶XQG{PHGHODYH)LJXUH

'pFRPSUHVVLRQEUXWDOHG¶XQFRUSVPDJPDWLTXH)LJXUH

? "

+ "

E " "