Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
in a Hele-Shaw cell
Giulia Spina
Relatrice: Prof.ssa Marina Serio
1
Purpose of the experiment
To study in detail the finger phenomena and the behavior of the finger at
different flow rates.
Fields of interest
CONTAMINANTS
• Unstable wetting front leads to much faster percolation of pollutants
• The MIM model explains the persistence of contaminants in ground
OIL ENGINEERING
• In order to avoid infiltration water-oil, gas-oil
2
Porous media
Porous soils are characterized by a strong irregular solid matrix, whose constituents got
dimensions that vary in a range of many orders of magnitude.
The complement of the solid matrix can be occupied by a liquid or a gas, or both.
Vw /V0
Va /V0
volumetric phase densities
porosity
a=air, w=water
3
Hele-Shaw cell
Hele-Shaw cell enables to 2D analysis of infiltration phenomena.
water light reference
Fine sand
homogeneous sand
LIGHT SOURCE
Air outflow
Non ponding water, two different flow rates (4.8 ml/min, 1.2 ml/min)
4
5
Water potential
In rigid, unsaturated soil wg m
Where:
g
wg(zz
0)
is the gravitational potential
pp
(
1/
r
m w a
1/
r ) is the matric potential, which accounts for
wa1 2
• wa surface tension
• r1 , r2 principal radii of curvature of the ellipsoid that represents the interface
• p pressure
• w water density
r>0 if it lies within the water phase
Often the equivalent height of a water column hww/wg is used instead of the
potential.
1 cmWC 1hPa=1mBar
6
z
meniscus z0
Example: in a capillary ( r1 r2 r )
wgz
( z
0)
2 /r
wa
The rise (or fall) in the capillary follows from surface forces.
7
Fluid-dynamics laws
• Conservation of mass t jw0
• volumetric water content
• jw water flux
Actually a conservation of volume since water may be considered as incompressible
jwK()
• Darcy-Buckingham law
p
The water flux is proportional to the pressure gradient. The proportionality constant is the
conductivity K .
This linear relation was found by Darcy in 1856; later improved by Buckingham by
substituting K with K ( )
In a porous media p is substituted by w , the water potential
• Richards equation
t
(
K
()
w)
0
Obtained by inserting the flux law in the conservation equation
8
Matric potential and hydraulic
conductivity
Hydraulic functions for various soil textures in the Mualem-van Genuchten (thick lines) and in the
Burdine-Brooks-Corey parametrization
• Non linear relations due to pore structure
• Hysteretic behaviour of the potential
As the sand refractive index always changes, this method leads to qualitative
results. In order to obtain the absolute water content, a calibration was
performed, by using X-Ray.
This is a very appropriate method in order to achieve a great spatial and temporal
resolution
10
Pressure measurements
• 12 tensiometers were employed to measure the water potential at a fixed point in the
cell.
• the presence of a air bubble in the tensiometer makes it less useful, because air can
grow and plug the porous membrane.
• every tensiometer has to be calibrated.
• Finger exactly reaching the sensor is a lucky case; the chance grows by increasing
the number of sensors, but this was not possible with our device (max tens. number
was 12).
11
Results: water content
using LTM
A C++ program calculates the mean light intensity in an area of about 15x15 pixels,
corresponding to the central part of the finger.
12
Results: pressure data
using Tensiometers
Results are more variegated, because the finger can reach the sensor in many different
ways, and what we obtain is always a mean. Here are showed only the best results.
13
Finger occurrence
After Saffman & Taylor (1958) for two fluids of different density and viscosity, driven by a
pressure gradient, one observes unstable displacement if:
kg
(1
)U
(1 )0
porosity
2 2
k permeability
density of replacing (1) and replaced (2) fluid viscosity
U interfacial velocity
Because both the viscosity and the density of water are much greater than that of air, the
equation simplifies to:
U
1/
kg 1Ks
Where K s is the saturated hydraulic conductivity. The first term is simply the flux through
the system.
So we obtain:
jw K s
This condition is achieved in our experiment by superimposing fine sand, with lower
saturated hydraulic conductivity, to the coarse sand.
14
Hysteresis and finger stability
Glass proposed in 1989 an explanation of the finger phenomena. Hysteresis in the
water retention curve has a great effect.
drainage
curve fringe
core
wetting
curve
15
Horizontal saturation
In order to check the explanation of Glass, we perform a horizontal analysis with the LTM.
As supposed, the water content is higher in the center of the finger. One can notice also
the presence of the peak due to finger tip.
16
Solute transport
After reaching a stable structure, we let a colored liquid, blue dye, flow in the cell.
One cannot see clearly the dye tip, because it mixes with water during transport from
pump to the top of the cell (this is an effect of convection and dispersion).
2,5 min
17
40 min
Convection-dispersion model (mention)
• breakthrough of solute step
• In a heterogeneous velocity field we expect that the variance of transport distances
increases with time.
• Each particle can change its velocity by moving from one streamline to another
through molecular diffusion.
• According to CLT, the distribution of travel distances approaches a Gaussian
distribution.
18
Convection-dispersion model (mention)
• One may consider two terms:
conv
js jwcs convective solute flux
jsdispD
effcs/
z dispersive component; described in analogy to a diffusion
process; z is the axis parallel to water flux.
cs , w is the solute concentration in the water phase, Deff is the effective diffusion
coefficient .
• The total solute flux law, according to conservation of mass, leads:
/
tcs
/z
j
wc
sD
c/
eff
s
z0
19
mobile-immobile model (mention)
• In many experiments the shape of the distribution is not Gaussian.
• This phenomena can be explained by separating the available into two
components, m , which is mobile, and im , which is not.
The total concentration is decomposed into the two parts of the water phase as:
s
c
mcs
,mc
ims
,im
• As only the mobile part is flowing, the convective dispersive solute flux states:
s
j j
wcs
,mDc
effs,
m /
z
• Inserting these two equations into the mass balance equation, and assuming the
fluid is well mixed, i.e. cm cim , leads to:
Rc
m/
t
js/
z0
Where R1im/m is the retardation factor. The solute is retarded with respect to
a hypothetical substance that is present in the mobile phase only.
In the experiment one can clearly see m and im , but the well mixed condition can be
achieved only on very longer time scales.
20
Future developments
• Numerical simulation: starting from the Richards equation one may add terms to copy
the behavior of fingers. For example Eliassi & Glass (Water Resources Research, vol
38, 2002) proposed a hold-back-pile-up term, related to a Nhd, hypodiffusion number,
that accounts for the material non-linearity
pile-up
hold-back
21
Acknowledgments
• Prof. Marina Serio, University of Turin
• Prof. Kurt Roth, Institute of environmental physics, University of Heidelberg
• D Sc Fereidoun Rezanezhad, Institute of environmental physics, University of
Heidelberg
• D Sc Marco Maccarini, Institute of physics and chemistry, University of Heidelberg
22
references
23