Document Actions

IJEGE-11_BS-Sanvitale-et-alii

background image
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
375
DOI: 10.4408/IJEGE.2011-03.B-043
EXPERIMENTAL MEASUREMENTS OF VELOCITY
THROUGH GRANULAR-LIQUID FLOWS
N. SANVITALE
(*)
, E.T. BOWMAN
(**)
& R. GENEVOIS
(***)
(*)
University of Padua - IMAGE Dept. - via Ognissanti 39 - 35129 Padova, Italy - e-mail: nicoletta.sanvitale@unipd.it
(**)
University of Canterbury - Department of Civil and Natural Resources Engineering, Christchurch, New Zealand
(***)
University of Padua - Department of Geosciences, Italy
INTRODUCTION
The mechanics of concentrated granular-fluid
flows is characterized by momentum and energy
transfer caused by inertial grain collisions, grain
contact friction, viscous shear and solid-fluid inter-
actions (i
veRson
, 1997). Constitutive equations for
such flows are not completely defined and detailed
experimental observations represent a key tool to
understanding their kinematic behaviour. Physical
modelling of granular-fluid flows moving down an
inclined flume provides an important approach to
understanding geophysical granular flows like debris
flows and mud flows.
Usually in debris flow experiments it is possible
only to examine the behaviour at the flow margins
(the top and the sides) since the ability to probe the
internal behaviour of concentrated two-phase flows
is prevented by their high opacity. Experimental evi-
dence shows that the dynamics of the inner part of
the flow can be different from the motion close to the
sidewall (a
Rmanini
et alii, 2005). That is, the two di-
mensional nature of the sidewall measurements can
only partially represent the three dimensional fea-
tures of the flow inside the bulk. Past experimental
studies analyzed the internal characteristics of grain-
fluid mixtures. m
ainali
& R
aJaRatnam
(1994) meas-
ured velocity and concentration profiles for a highly
concentrated sand-water slurry, sampling the flow in
the centre of the flume. Velocity profiles of the entire
flow field of mudflows in a recirculatory flume were
ABSTRACT
This paper describes a series of “small scale” labo-
ratory flume tests designed to examine the motion and
arrest of concentrated granular-fluid flows, with a view
to understanding debris flows. A non intrusive optical
approach which relies on the matching of the refractive
index and the planar laser-induced fluorescence tech-
niques (PLIF), is adopted to analyse the flow behav-
iour of a granular medium in a ‘free surface’ condition
and in the context of relatively high-speed movement.
A PIV (Particle Image Velocimetry) approach, adapted
for geotechnical testing, is used to measure the veloc-
ity profiles as obtained from laser-illuminated images
through the transparent flowing granular material. PIV
analyses of experiments carried out using two different
mixtures, one characterized by well-graded and one by
uniform particle size distribution, are compared in or-
der to understand the effect of the granular composition
on the flow dynamics. The experimental results sug-
gest that the particle size distribution has an influence
on the mobility of flowing granular material at a given
moisture content. Flow of well-graded mixtures show
higher velocity, smaller flow thickness and a larger ru-
nout extent. The velocity profiles of the two mixtures,
measured at the control section of the flume slope, ex-
hibit different behaviour with a lower degree of shear-
ing for that with uniform particle size distribution.
K
ey
word
: debris flows, physical modelling, flume study,
PLIF, granular-fluid flows, PIV analyses
background image
N. SANVITALE , E.T. BOwMAN & R. GENEVOIS
376
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
METHODOLOGY
Matching the refractive indices of the components
of a solid-liquid system should render the mixture
transparent and the investigation of its inner parts pos-
sible. In order to distinguish the solid particles in the
fluid, a fluorescent dye is added to the fluid phase. The
dye is excited by a laser light with a wavelength tuned
to its absorption band. As soon as the dye de-excites
(on the order of nano to microseconds), it emits light
at a wavelength larger than that of excitation. This
light, known as fluorescence, enables the observa-
tions of both geometry and deformation through the
particle-fluid system: the particles in an image appear
dark against a bright background, as shown in Fig. 1.
EXPERIMENTAL APPARATUS
The experimental set up is shown in Fig. 2. The main
experimental device is a small scale flume with an alu-
minium head tank containing the flow material above
the top of the slope. A curved chute guides the flow
from the mouth of the tank to the top of the slope,
at which point the flow travels down a 150mm wide
by 2000mm long channel before being deposited on
a horizontal runout surface. The slope angle can be
varied from 0° to 40°. The bottom of the channel is
made of stainless steel which is artificially roughened
over the last 1m by means of a synthetic material with
a slip resistant surface (3M Safety-Walk 370). The
outward facing walls are made from Perspex except
for the part corresponding to the rough bed, which is
made of borosilicate glass. A high-speed digital cam-
era (SVSi Memview) records the downslope flow be-
haviour through the glass sidewalls. Before its release,
the prepared saturated material is stored in the head
tank and continually agitated to ensure that the least
possible segregation and consolidation takes place
before the release of the material. The pneumatically-
operated tank trapdoor is linked by a micro switch to
the high speed camera to ensure a time-delay between
the release of the material and recording of the flow.
The frame rate of the camera is set at 1100 fps.
LASER LIGHT SOURCE AND OPTICAL
DEVICES
The light source is a 800 mW solid state laser
(Changchun New Industries) at a wavelength of 532
nm. The laser beam is coupled into an optical fibre,
recollimated at the fibre output, and then sent through
obtained using a pulsed Doppler ultrasound tech-
nique by a
Rmanini
et alii (2003).
A well-known technique to achieve optical ac-
cessibility to the inner parts of particle-liquid me-
dium can be obtained by matching the refractive in-
dex of transparent solids and fluid. Optical matching
makes the entire mixture transparent allowing inves-
tigation of the internal points. Combining refractive
index matching with planar laser–induced fluores-
cence (PLIF) techniques enables the visualization
of deformation within a saturated granular flow and
enables the use of deformation measurement tech-
niques, such as Particle Image Velocimetry (w
Hite
et alii, 2003).
A number of studies have used the matching of
refractive indices to achieve optical accessibility to
liquid flow phenomena (b
udwiG
, 1994), and their
application to porous media using PLIF has been
undertaken by m
ontemaGno
& G
Ray
(1995); f
on
-
tenot
& v
iGil
(2002), s
toHR
et alii (2003); o
RPe
&
k
udRolli
(2007).
In this paper we present the preliminary results
obtained by applying the aforementioned techniques
to the analysis of fast moving granular-liquid flows
in a small scale flume. In the model described here,
data are extracted from 2-D vertical sections through
the flowing granular material. The physical and opti-
cal properties of the selected solid and liquid, and
the adopted methodology are presented. Further, the
results of the first experiments are discussed with
particular attention to the measurement of the veloc-
ity profiles.
Fig. 1 - Schematic layout of the adopted PLIF configura-
tion
background image
EXPERIMENTAL MEASUREMENTS OF VELOCITY THROUGH GRANULAR-LIQUID FLOWS
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
377
lar shape. The particles with size larger than 30mm
have been made by smashing a 28x15x6 cm block
of Pyrex® glass (Dow-Corning), a borosilicate glass
with the same optical characteristic of Duran but of a
different brand (no longer produced).
In order to create a distinction between fluid and
solid, the fluorescent dye Nile Red (Sigma-Aldrich)
is used. This dye is soluble in non-aqueous phase and
shows a meaningful quantum efficiency and absorp-
tion rate at the laser excitation wavelength. These
properties are necessary to achieve a high intensity of
emitted light, which is fundamental to capturing im-
a Powell lens generating a vertical sheet of light with a
thickness of around 2 mm. In some preliminary exper-
iments, the laser sheet illuminated the flowing mixture
from above, but the interaction between the incident
light and the surface of particles protruding from the
free surface of the flow was found to cause distortion
of the planar laser beam. This resulted in a broadening
of the laser sheet and in the illumination of particles
outside of the intended measurement plane focused on
by the camera (s
anvitale
et alii, 2010).
Modifications of the experimental apparatus were
made to apply the laser sheet from beneath the flow
through the bottom of the channel. The steel base in
the distal part of the channel (1 meter before the exit
point to the runout area) was replaced with a standard
glass plate. Then the upper glass surface was covered
by means of the same synthetic material to provide
the necessary roughness. A slit, 300 mm in length and
1.5-2 mm wide, cut through the rough material was
made to let the light pass through the base. With this
altered set-up the high-speed camera can record the
images of each flow along a laser illuminated longi-
tudinal plane passing through the slit. The illuminated
plane is located 35 mm from the sidewall and begins
approximately 110 mm before the exit point of the
flume channel to the horizontal deposition area. The
final set-up is shown in Fig. 2.
SOLID, FLUID AND DYE PROPERTIES
There are several key concerns in choosing ap-
propriate pairings of transparent particles and asso-
ciated fluids. They must be optically compatible in
terms of refractive index, reasonably economic and
non-hazardous to work with. Borosilicate glass (Schott
Duran
®
) is used as transparent material for the solid
particles. A hydrocarbon oil (Cargille Immersion Liq-
uid code 5095) is used as fluid phase.
This liquid closely matches the index of refraction
of Duran at the dye emission wavelength and at a tem-
perature of 22-24°C. The optical and physical proper-
ties of these two materials are reported in Table 1. The
borosilicate glass has been purchased from the manu-
facturer in rods with diameters from 4mm to 30 mm
and tubes with wall thickness ranging from 0.8 mm
to 2.0 mm. The tubes are crushed to obtain particles
smaller than 4.5 mm whereas the coarsest particles up
to 30 mm are cut from the rods in cylindrical pieces
and subsequently shaped in order to achieve an irregu-
Table 1 - Solid and fluid properties
Fig. 2
-
Experimental set up with laser
background image
N. SANVITALE , E.T. BOwMAN & R. GENEVOIS
378
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
rated with 3.94 litres of oil was used, for a resulting
overall sediment volume concentration of 0.57. The
concentration of the Nile Red in the oil was 2.5 mg/l,
the laser power was set around 600 mW and the frame
rate of the camera at 1100 fps. The flume slope angle
was 24.5°.
VELOCITY MEASUREMENTS
Measurements of the velocity profile inside the
flow were obtained using GeoPIV software (w
Hite
et
alii, 2003). GeoPIV operates by tracking the texture
(i.e. the spatial variation of brightness) of sub-regions
in an image, often called “patches”, in multiple imag-
es. The displacement vector of each patch in the time
interval occurring between two subsequent images is
found by locating the peak of the autocorrelation func-
tion of each patch.
The original GeoPIV algorithm has been modified
by W.A. Take (private communication) to the needs
of the present work, by supporting a static mesh, with
position and geometry fixed in the image and through
which particles flow. Each mesh consists of a single
column of patches. The patches overlap in the vertical
direction at a spacing of half of the patch size. The
displacements are calculated at two subsequent time
steps (i.e., two subsequent frames), providing an es-
timate of the instantaneous velocity field at the mesh
location.
For these experiments, velocity fields were esti-
mated at regular time intervals (60-90 frames, corre-
sponding to 0.054-0.082 s), by defining a static mesh
covering the entire height of the flow, as shown in Fig.
4. For all the tests, the first mesh was placed imme-
diately after the unsaturated coarse front, where fully
saturated conditions made the PLIF technique effec-
ages at a high frame rate during the experiment. Dur-
ing the tests a longpass filter (Schott OG550), with the
cut-point at 550 nm, is placed over the camera lens to
transmit the fluorescence signal and screen unwanted
scattered light.
TEST CONDITIONS
Debris flows are mixtures of granular sediment
and fluid, with both components influencing the re-
sulting flow and deposition behaviour. Field evidence
shows that virtually all debris flows contain a wide
particle size distribution (PSD), with sizes ranging
from boulders or gravel to silt and clay. The typical
mean particle size D
50
ranges between 2 mm and 200
mm while the values of the uniformity coefficient C
U
= D
60
/D
10
are of the order of 100-1000 (e.g. H
uRli
-
mann
et alii, 2003; P
ieRson
, 1980).
Usually, experimental flows are characterized by
a mean particle size which is necessarily smaller than
that found in the field, and use a relatively uniform
PSD (C
U
<5). Important aspects of their behaviour,
such as segregation of the particles and high mobility,
are unable to be replicated in such tests. Experimental
research (b
owman
& s
anvitale
, 2009) has shown that
the particle size distribution has a significant effect on
the mobility of flowing granular materials at a given
moisture content, and small laboratory flume tests can
reproduce key aspects of the debris flow behaviour if a
careful selection of the material, in terms of gradation
and particle size, is made. In order to analyze the influ-
ence of a change in the value of C
U
(at a given mean
particle size D
50
) on the kinematic behaviour of gran-
ular-liquid mixture, for the results reported herein,
two particle size distributions, namely PSD9-mod and
PSD16-up, were chosen (Fig. 3). Table 2 summarizes
the main parameters of the adopted solid material.
For both the PSDs two experiments were carried
out to ensure repeatability of results. In each run, a
mixture prepared with 12 kg of glass particles satu-
Table 2 - Properties of employed particle size distributions
Fig. 3 - Particle size distributions used for the experi-
ments
background image
EXPERIMENTAL MEASUREMENTS OF VELOCITY THROUGH GRANULAR-LIQUID FLOWS
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
379
well-graded materials. These tests also displayed
the greatest overall segregation during flow and
deposition. The deposits on the runout area exhibited
morphologic features common to many natural and
laboratory debris flow fans (m
aJoR
& i
veRson
, 1999)
with coarse particles concentrated at the front and at
the outermost margins, while the finer material was
emplaced at the rear of the fan. Figure 6 shows an
example of the resulting particle size distribution of
three samples collected at the front (“lobate” part of
sample) and in the middle (“centre” of sample) of
the runout zone for each PSD. We also collected a
sample at the distal part of the flume (“flume” sam-
ple), where some material deposited before reaching
the horizontal runout area. Both the replicates carried
out for each PSD showed essentially the same mor-
phologic characteristics in the deposit zone.
IMAGE ANALYSIS
Digital images of the flows are shown in Fig. 8
and Fig. 9. The flow proceeds from left to right and
the camera is angled so that the base of the channel is
parallel to the base of the camera in the images. Fig-
ure 8 shows images resulting from an experiment with
the well graded particle distribution (PSD9-mod). The
flows obtained using these gradations were thin and
elongated. A few bouncing grains preceded the arrival
of the front. The snout, consisting of coarse parti-
tive. Square patches with two different sizes, 16 or 32
pixels (1/16 or 1/8 of the image height), were used.
The former allowed a higher spatial resolution to be
obtained, and hence more detailed, although more
noisy, results. The latter led to more precise results
but at the expense of resolution. For each patch, the
velocity was calculated as the mean of the instanta-
neous velocities estimated over 30 successive frames
(corresponding to a time step of 0.027 s). Figure 5
shows an example of the distribution of the velocity
estimates obtained at each patch both for the x (hori-
zontal) and the y (vertical) components. This velocity
is an in-plane velocity, and the x and y axes are ori-
ented in directions parallel and perpendicular to the
mean flow direction, respectively. The vertical veloc-
ity was found to be at least one order of magnitude
smaller than the horizontal one and therefore, its con-
tribution was neglected in the analysis presented here.
The standard deviation was calculated for velocity at
each depth and values falling outside a proper confi-
dence interval were neglected and the mean recalcu-
lated accordingly. For patches of 32 pixels (hereafter
called 32pix), the confidence interval was set to three
standard deviations, while for patches of 16 pixels the
confidence interval was set equal to one standard de-
viation, in order to cope with the large scattering as-
sociated with the corresponding data.
RESULTS
DEPOSIT MORPHOLOGY
Figure 7 shows photos of the deposit fan for
both PSD. Both of the experiments carried out for
each PSD exhibited approximately the same shape
and runout, with differences in length smaller than
10%. The largest runout lengths were found for the
experiments with the highest values of C
U
, i.e. for
Fig. 4 - Example of static GeoPIV mesh
Fig. 5 - Example of measured velocity profiles for the body
of the flow. Each dot represents the velocity esti-
mate resulting from the comparison of two subse-
quent frames. The continuous lines are the veloci-
ties calculated as the mean of these instantaneous
velocities
background image
N. SANVITALE , E.T. BOwMAN & R. GENEVOIS
380
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
cles, had a velocity at the control section larger than
2 m/s. Both the front and the following body of the
flow had approximately the same height, around 25-
30 mm. The depth at the tail of the surge was around
10 mm. The flow segregated during motion with most
of the coarse particles concentrated towards the front.
Through the depth, the segregation was not so evident,
due the small thickness of the flow in comparison to
the average size of the grains. Nevertheless, the ten-
dency of the coarse particles to position close to the
free surface is visible in Fig. 8, many of them (mostly
the largest) appearing to “float” on it.
Figure 9 shows the images from a test using
PSD16-up. The flow behaviour was different from
that of the well-graded mixtures. The entire flow was
short and moved as a bulk. The front had a thickness
of around 70 mm, greater than the height of the cam-
era image, while the thickness of the tail was around
10-20 mm. From the images, the degree of segregation
appears to be lower, with coarse particles present also
in the body of the surge. The front was unsaturated
(individual particles are visible as illuminated block
and not as dark spot since PLIF needs fully saturated
condition to be effective) and highly concentrated.
Due to a significant number of air bubbles it was not
possible to obtain clear images of the front. The pen-
etration of the laser within the flow close to the front
was partially prevented by the high concentration of
the particles and the large thickness of the medium
(with the output laser power decreasing with the dis-
tance). The resulting images are blurred and not clear.
The level of detail improves significantly towards
the tail, where the solids concentration progressively
decreased. The average velocity of the snout was ap-
proximately 1.5 m/s.
MEAN VELOCITIES
Two experiments, one for each PSD, have been
analyzed. The analysis starts from the part of the
flow where fully saturated conditions make the PLF
technique effective, emphasizing the solid particles
as black spots against a bright background. Figure 10
shows the time evolution of the mean flow velocity.
This is calculated for each mesh, placed at different
positions in the flow, as the average of the mean ve-
locity estimates of its patches. In Fig. 11 the estimated
heights of the two flows obtained by plotting the po-
sition of the highest patch of each 16 pix mesh are
shown. Note that this is just an approximation of the
true height, since the meshes cannot cover the entire
depth of the flow. This is due to the restrictions posed
by the software, which requires that the top patch of
the mesh must be set just below the free surface. It
is useful for the following discussion, to distinguish
between two main regions for each flow, defined as
the “body” and the “tail”.
For PSD16-up, the body is defined as the region
showing the highest thickness and the largest mean
velocity: it occurs within the interval between 0 and
0.36 s (see Fig. 11). The remaining part of the flow,
with decreasing height and velocity, represents the
Fig.6 - Examples of the particle size distributions for the
lobate, centre and flume sample for each PSD
Fig. - Photos of runout zone after the test (a) PSD9-
mod; (b) PSD16-up
background image
EXPERIMENTAL MEASUREMENTS OF VELOCITY THROUGH GRANULAR-LIQUID FLOWS
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
381
to be related to the presence of large particles, whose
size is comparable with the flow thickness as shown
in Fig. 4. These particles, which move faster than the
surrounding matrix, preferentially at the surface of the
flow, occupy a significant part of the available space
and seem to push and accelerate the material immedi-
ately in front of them. Further analyses of other tests
with similar features (i.e. a well-graded mixture with
some large particles inside the flow) need to be per-
formed in order to elucidate this behaviour.
The height profiles confirm that PSD16-up moves
as a bulky surge, with a thickness significantly larger
than that of PSD9-mod. The heights of the flow drop
suddenly at the beginning of the tail, decreasing from
about 55 mm to less than 10 mm. The well graded
mixture (PSD9) has a reduced thickness which de-
creases smoothly toward the tail. Some examples of
the estimated velocity profiles are presented in Fig.
12. For both PSDs, two profiles from the body (a and
b) and two from the tail (c and d) are shown. The mean
values and the corresponding errors bars (standard de-
viation) are plotted for 16 pix meshes (blue lines). The
velocity obtained from 32 pix mesh is overlapped for
comparison. The two meshes present comparable re-
sults, with the 16pix giving more details and the 32
tail. For PSD9-mod, the body is identified between 0
and 0.45 s. This part of the flow exhibits a high (ex-
cept for the first two mesh positions) and rather con-
stant thickness (see Fig. 11). In the body the velocity
is high and, although fluctuating, slowly decreasing
(see Fig. 12). The tail is initially characterized by a
fast decrease of the velocity and height, which pro-
gressively becomes gentler.
Comparing the experiments, the most uniform
material (PSD16) moves with a lower velocity at any
time, even if at the end of the flow (in the tail) the ve-
locity of both the tests is comparable. A particular fea-
ture of the PSD9-mod velocity curve is represented by
large ripples at the initial part of the body. Such ripples
are not the result of numerical noise, but they seem
Fig. 8 - Digital images of the test using PSD9-mod. The
flow is proceeding from left to right. (a) During ar-
rival of the flow front; (b) In the body immediately
after the front; (c) In the core of the flow; (d) In the
tai
Fig. 9 - Digital images of the test using PSD16-up. The
flow is proceeding from left to right. (a) During
arrival of the flow front; (b) In the body immedi-
ately after the front; (c) In the core of the flow; (d)
In the tail
Fig. 10 - Mean velocity versus time
background image
N. SANVITALE , E.T. BOwMAN & R. GENEVOIS
382
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
pix providing smoother and more precise results
All the profiles show a non-zero slip velocity at
the bottom, of about 0.5 m/s, probably due to the rela-
tively low degree of roughness provided by the mate-
rial which covers the bed of the flume (with respect to
the size of the grains). In all cases, the flows appear to
show large shearing toward the base, while they tend
to move with a more constant velocity toward the sur-
face. This tendency is more pronounced for PSD16-
up-b, which moves with a plug in the upper part of
the flow.
DISCUSSION AND CONCLUDING RE-
MARKS
This paper presents the preliminary results of a
novel method for measuring the kinematic behaviour
of a transparent highly concentrated particle-fluid sys-
tem moving down an inclined channel, in a ‘free sur-
face’ condition and with relatively high-speed move-
ments. The use of a non intrusive optical approach
which relies on the matching of the refractive index of
the constituents and planar laser-induced fluorescence
(PLIF) techniques, makes the investigation of the field
velocity in the inner part of the flow possible.
Digital images recorded during a first series of
experiments were obtained with the laser plane posi-
tioned 35 mm from the outside face that is ¼ of the total
width of the employed flume apparatus. It is expected
that edge effects during shearing would be significantly
reduced at this distance but they would not be entirely
eliminated. However, in a narrow channel typical of de-
bris flows, some shearing will always be generated by
the side-walls of the channel and it is not the intention
of this work to completely eliminate all “edge effects”,
but only to render visible shearing within the flow. The
study of 2-D vertical sections through the interior of the
flow allows us to investigate the 3-D dimensional fea-
tures of the flow inside the bulk.
During an experiment, with a segregating system
of particles, the coarser material concentrates at the
front of the flow, resulting initially in an unsaturated
condition. For this part of the flow it was not possible
to use PLIF since the fluid phase is partially absent,
however in the body and tail of the flow where the
mixture was completely saturated and the solid con-
centration lower, the observation of the motion of in-
dividual particles within the fluid-particle system was
possible. A special PIV (Particle Image Velocimetry)
technique, GeoPIV, adapted for geotechnical testing
and modified to the needs of the present research, has
been used to measure the velocity profiles as obtained
from laser-illuminated images through the transparent
flowing mixture.
Two mixtures with the same sediment volume con-
centration and mean particle size D
50
, but consisting
different material i.e., a well-graded material with
C
U
=20.2 and a uniform one with C
U
=3.3, were inves-
tigated. The results show that the particle size distribu-
tion strongly affects the overall flow behaviour with
well-graded material exhibiting greater speed and
far greater runout in comparison to uniformly graded
mixtures. Also the heights of the flow and velocity
profiles measured at the control section of the flume
slope were quite different. The well-graded mixture
moved downslope the channel as an elongated and
thin flow, while the uniform mixture moved as a bulky
surge displaying a lower degree of shearing. Further
investigations are planned to analyze in detail the in-
fluence of the larger particles in the flow behaviour of
the well-graded mixtures and to examine the internal
shearing behaviour of the flows.
ACKNOWLEDGEMENTS
This research was financially supported through
the Royal Society of New Zealand Marsden Fund. The
authors would like to thank laboratory staff Messers
Siale Faitotonu, John Kooloos, Rob MacGregor, Peter
McGuigan, Ian Sheppard and Kevin Wines of the Uni-
versity of Canterbury for help in construction of the
flume apparatus and experimental testing. The authors
would also like to thank Dr W.A. Take for providing
a modified version of GeoPIV for use in the research.
Fig. 11 - Position of the highest patch of each mesh versus
time
background image
EXPERIMENTAL MEASUREMENTS OF VELOCITY THROUGH GRANULAR-LIQUID FLOWS
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
383
Fig. 12 - Typical velocity profiles for PSD9-mod and PSD16-up at different times for 16pix meshes (solid line) with the corre-
sponding error bars (estimated as standard deviation). The 32pix velocity (dashed line) is overlapped for comparison.
(a) and (b) velocity profiles in the body, (c) and (d) velocity profiles in the tai
background image
N. SANVITALE , E.T. BOwMAN & R. GENEVOIS
384
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
REFERENCES
a
Rmanini
a., d
alRì
’ C., f
RaCCaRollo
l., l
aRCHeR
m. & z
oRzin
E. (2003) - Experimental analysis of the general features of
uniform mud-flow. Proc. 3rd International Conference on Debris-Flow hazard mitigation: mechanics, prediction, and assess-
ment, Davos, 10-12 September 2003, Vol. 1: 423-434, edited by Millpress, Rottherdam 2003.
a
Rmanini
a., C
aPaRt
H., f
RaCCaRollo
l. & l
aRCHeR
m. (2005) - Rheological stratification in experimental free-surface flows of
granular-liquid mixtures. Journal of Fluid Mechanics, 532: 269-319.
b
owman
E.T. & s
anvitale
n. (2009) - The Role of Particle Size in the Flow Behaviour of Saturated Granular Materials. Proc.
17th International Conference on Soil Mechanics & Geotechnical Engineering (ICSMGE), 5-9 October, Alexandria, Egypt.
Edited by M.Hamza, S. Marawan and Y. El-Mossallamy. IOS Press, Amsterdam, the Netherlands, 1: 470-473.
b
udwiG
R. (1994) - Refractive index matching for liquid flow investigations. Experiments in Fluids, 17: 350-355.
f
ontenot
m.m. & v
iGil
R.d. (2002) - Pore-scale study of nonaqueous dissolution in porous media using Laser-Induced Fluo-
rescence. Journal of Colloid and Interface Science, 247: 481-489.
H
uRlimann
m., R
iCHenmann
d. & G
Raf
C. (2003) - Field and monitoring data of debris flow events in the Swiss Alps. Canadian
Geotechnical Journal, 40: 161-175.
i
veRson
R.M. (1997) - The physics of debris flows. Reviews of Geophysics, 35 (3): 245-296.
m
ainali
a. & R
aJaRatnam
n. (1994) - Experimental study of debris flow. Journal of Hydraulic Engineering, 120(1): 104-123.
m
aJoR
J.J. &. i
veRson
R.m (1999) - Debris-flow deposition - effects of pore-fluid pressure and friction concentrated at flow
margins. Geological Society of America Bulletin, 111: 1424-1434.
m
ontemaGno
C.d. & G
Ray
w.G. (1995) - Photoluminescent volumetric imaging:A technique for the exploration of multiphase
flow and transport in porous media. Geophysical Research Letters, 22 (4): 425-428.
o
RPe
v.o. & k
udRolli
a. (2007) - Velocity Correlations in Dense Granular Flows Observed with Internal Imaging. Physical
Review Letters, 98 (23): 238001
P
ieRson
T.C. (1980) - Erosion and deposition by debris flows at Mt. Thomas, North Canterbury, New Zealand. Earth Surface
Processes, 5: 227–247.
s
anvitale
n., b
owman
e.t. & G
enevois
R. (2010) - Optical investigation through a flowing saturated granular material. Zurich,
Switzerland: 7th International Conference on Physical Modelling in Geotechnics (ICPMG 2010), 28 Jun-1 Jul 2010.
s
toHR
m., R
otH
k. & J
aHne
b. (2003) - Measurement of 3D pore-scale flow in index-matched porous media. Experiments in
Fluids, 35: 159-166.
w
Hite
d.J., t
ake
w.a. & b
olton
m.d. (2003) - Soil deformation measurement using particle image velocimetry (PIV) and
photogrammetry. Geotechnique, 53 (7): 619–631.
Statistics