Reflection at an interface between a free fluid and a porous material:

theory versus experiment

Radim Ciz, Don Sherlock & Sam Howman

CSIRO Petroleum, ARRC, Kensington, Perth, Australia

Boris Gurevich
Curtin University of Technology, Department of Exploration Geophysics, Perth, Australia

ABSTRACT: According to the recent study by Denneman et al., (2002), the poroelastic reflection coefficient for
a plain interface between free fluid and an air-filled porous medium at low frequencies exhibits the significant
difference compared with the reflection from fluid/solid interface. In order to understand the nature of this
difference and to explore implications for reflections from an interface between two porous media, we derived
simple analytical expressions for poroelastic reflection coefficients at normal incidence. These expressions show
that the large difference between poroelastic and elastic reflection coefficients at low frequencies is due to the
large contrast between compressibilities of the pore fluid and the free fluid. In the low-frequency limit, the
poroelastic reflection coefficient reverts to the elastic value. However, this only occurs at frequencies as low
as 0.001 Hz. For the purpose of exploring whether the effects predicted by the theoretical analysis are real, we
conducted laboratory measurements of the reflection coefficient for an interface between a free fluid and an
air-saturated porous sandstone. The experimental results are consistent with the theoretical predictions within
experimental error and these measurements confirm the validity of theoretical expressions derived.

1 INTRODUCTION In particular, it was shown that stratification of porous

materials can produce significant additional attenua-
Theory of elastic wave propagation in porous saturated tion and dispersion of the propagating wave. These
media (Biot, 1962) predicts the effects of movement of phenomena are related to the fact that when a compres-
the pore fluid relative to the solid skeleton on propagat- sional or shear wave encounters an interface between
ing elastic waves. This opens potential opportunities two different porous saturated materials, a fluid flow
to estimate fluid and transport properties of porous across the interface may occur, which results in the loss
materials by measuring elastic waves. However, these of energy from the propagating wave. For a single inter-
opportunities are somewhat limited by the fact that face this phenomenon manifests itself in the fact that
at low frequencies relative fluid movement becomes the reflection coefficient from an interface between
negligible and the porous material behaves like an elas- two porous media is proportional to the square root
tic solid with the equivalent elastic moduli given by of frequency. Thus, fluid effects on reflection coeffi-
Gassmann (1951) equation. In particular, Biots theory cients may be observed at lower frequencies and recent
of poroelasticity predicts that elastic wave attenuation numerical calculations (Denneman, 2002) show that in
and dispersion only become significant at frequen- some cases the reflection coefficient exhibits strong
cies comparable to the so-called Biots characteristic deviations from its elastic value at frequencies much
frequency. For commonly encountered natural porous lower than Biots characteristic frequency. In particu-
materials such as sandstones or limestones saturated lar, numerical calculations have shown this behavior
with water, oil or gas, Biots characteristic frequency for an interface between a gas-saturated porous solid
is usually 0.1 MHz or higher. This is much higher than and a free liquid (Denneman et al., 2002). In the case of
the frequency of seismic exploration (2070 Hz) and a sealed-pore interface there is no fluid flow across the
well logging (550 kHz) typically used in exploration interface. Thus deformation of this system is equiva-
geophysics. lent to the deformation of an effective elastic medium.
Recently it was realized that dynamic poroelastic This corresponds to the classical elastic modelling
effects may be pronounced at lower frequencies when using velocities calculated by the static Gassmann
the porous medium is macroscopically heterogeneous. (1951) equation.

585
Copyright 2005 Taylor & Francis Group plc, London, UK
 In order to understand the nature of this differ-
ence and to explore implications for reflections from
an interface between two porous media, Gurevich
et al., (2004) derived simple analytical expressions for
poroelastic reflection coefficients at normal incidence.
These expressions show that the large difference
between poroelastic and elastic reflection coefficients
at low frequencies is due to the large contrast between
compressibilities of the pore fluid and the free fluid.
In the low-frequency limit, the poroelastic reflection
coefficient reverts to the elastic value. However, this
only occurs at frequencies as low as 0.001 Hz. Thus
we have a counter-intuitive situation that frequencies
Figure 1. Reflection and transmission from an interface of
of 10100 Hz are not low with respect to this problem. porous medium/fluid.
The main purpose of this paper is to explore whether
the effects predicted by the theoretical analysis are coefficients from an interface between a free fluid and
real. We conducted laboratory measurements of the a porous medium is
reflection coefficient for an interface between a free
fluid and an air-saturated porous sandstone. Piezo-
electric transducers were used to provide a pulsed
ultrasonic source at a controlled frequency. The out-
put from the transducers used for these experiments where the parameters Y is
have a dominant frequency of 50 kHz and a bandwidth
between around 10 kHz and 70 kHz. The transducers
were specially made for these experiments because the
frequency required is an order of magnitude lower
than that traditionally used in the laboratory. The
sandstones were made using a process known as the and v1a represents P-wave velocity in the poroelas-
Calcite In-situ Precipitation System (CIPS), which is tic half-space, v1b denotes P-wave velocity in the free
a method of cementing particulate soils through the fluid, is frequency. Poroelastic parameters Ca , Ha
injection of aqueous solutions and the precipitation and Na are related as
of calcite cement around sand grains and their con-
tacts. The technology was developed at the CSIRO to
improve the geotechnical properties of porous sedi-
ments and rocks (Kucharski et al., 1996). Laboratory
tests have shown that CIPS cemented sands closely
reproduce the acoustic and mechanical properties of
natural sandstones.
where is Biots effective stress coefficient. To ana-
lyze the parameter Y , which is the main difference
between elastic and poroelastic reflections coefficient,
2 OVERVIEW OF THEORETICAL ANALYSIS we follow further Gurevich et al., (2004).

We consider a porous medium with the uniformly

distributed porosity whose pores are filled with a 3 DISCUSSION OF THEORETICAL RESULTS
viscous fluid. The fluid is characterized by the bulk
modulus Kfa , density fa and steady-state shear viscos- The central result of the theoretical analysis is the
ity a . The grains of the solid are characterized by bulk expression for the reflection coefficient R11 for the
and shear moduli Kga , ga and density ga . The drained fast P-wave. Formula (1) shows that the expression for
solid matrix formed from grains and pore spaces are the reflection coefficient is the same as the familiar
characterized by bulk and shear moduli and density Ka , expression for two elastic media (e.g., Brekhovskikh
a , a , respectively, and steady-state permeability a . and Godin, 1990)
If the poroelastic half-space b in Figure 1 is a free
fluid with density b and bulk modulus Kb , we can
derive the expressions for the reflection and trans-
mission coefficients from an interface between a free
fluid and a porous medium. Following Gurevich et al., except for the factors Y . As previously shown (e.g.,
(2004), the closed-form expression for the reflection Bourbi et al., 1987), this factor is proportional to

586
Copyright 2005 Taylor & Francis Group plc, London, UK
 1 o Several key observations may be drawn from this
poroelastic plot and the analytical expressions (1) and (2):
0.9 elastic
The results at 10 Hz and at 10 kHz are in agree-
0.8 ment with the results of Denneman et al., (2002,
o Figure 8, p. 289) for normal incidence, and indeed
0.7
are very different from the elastic reflection coeffi-
|R11|

0.6 cient at frequencies as low as 10 Hz, as indicated by

the circles in Figure 2.
0.5
Nevertheless, this is not the result for the low-
0.4 frequency limit. Indeed, in the low-frequency limit
the poroelastic reflection coefficient reverts to the
0.3 elastic value. However, this only occurs at frequen-
0.2 cies as low as 103 Hz.
10-4 10-3 10-2 10-1 100 101 102 103 104 Thus 10 Hz is not in the low-frequency limit for
Frequency (Hz) equation (1).
Within the assumptions of low-frequency Biots the-
Figure 2. Reflection coefficient at an interface between a ory (and thus up
free fluid (water) and an air-saturated porous medium. Circles to at least 10 kHz), factor Y scales
linearly with . However, it translates into the
denote the exploration frequency range 10 Hz10 kHz. same linear dependency of the reflection coefficient
R only if Y = 1. If Y is notsmall, then R as given
by (1) is still a function of , but it is no longer a
Table 1. Mechanical properties of the porous matrix linear function of it, but a rational function.
(Denneman et al., 2002). Therefore, the large deviation of the poroelastic
reflection coefficient from the elastic one for an
Kg K g
interface between free fluid and air-saturated porous
(GPa) (GPa) (GPa)  (darcy) (kgm3 )
medium at very low frequencies is due to the fact
Sand 40 5.8 3.4 0.26 0.95 2760 that the proportionality constant between poroelas-
tic parameters and the square root of frequency is
very large.
To analyze this problem further, we assume both
Table 2. Mechanical properties of the sample pore and free moduli Kfa and Ka to be much smaller than the bulk
fluid (Denneman et al., 2002). modulus of solid grains Kga : Kfa = Kga and Ka = Kga .
Using this approximation, the poroelastic parameter
Kf (GPa) Kf (kgm3 ) (Pa s) Na is written as Na Kfa /a and Ca = Kfa /a = Ha .
Thus Y can be written as
Water 2.22 1000 0.001
Air 0.0001 1.2 1.82*105

the square root of frequency. The difference between

poroelastic and elastic reflection coefficients is con-
trolled mainly by ratios of poroelastic parameters
Ca /Ha , present in factor Y as given by equation (2).
The effect of the fluid flow is significant when this and c is Biots characteristic frequency. If the pore
difference is large. fluid is liquid (say, water), then b /fa , Kb /Kfa and Z
Since there is a significant difference between are of the order 1 (or smaller), and hence Y is small
normal incidence poroelastic and elastic reflection for frequencies = c . However, if the pore fluid is
coefficients at an interface between a free fluid and air, Y becomes finite (of order 1) at frequencies much
an air-filled porous medium (Figure 2), closed-from smaller than c Note that the first square root of equa-
expressions (1) and (2) would provide a necessary tion (5) increases, if one replaces water in the pores by
insight into these observations. To understand this air. More significantly, with air in the pores the second
effect, in Figure 2 we plot the reflection coefficient square root of equation (5) becomes very large due to
as given by equation (1) versus frequency for an inter- the small value of bulk modulus Kfa . It is this large
face between free water and an air-saturated porous contrast in fluid bulk moduli, that is the cause of the
sandstone. The parameters of the porous matrix, pore anomalous difference between elastic and poroelastic
fluid and free fluid are given in the Tables 1 and 2. reflection coefficients at the low frequencies.

587
Copyright 2005 Taylor & Francis Group plc, London, UK
4 LABORATORY EXPERIMENTS

The aim for the laboratory experiments was to com-

pare the predicted and measured reflection coeffi-
cients between a free fluid and air-saturated porous
sandstone. To verify the theoretical results obtained,
the special experimental configuration has been con-
structed for this purpose.

4.1 Equipment and experimental method

Figure 3. Schematic diagram of the experiment configu-
Piezoelectric transducers convert elastic or acousti- ration used to measure reflection amplitudes. The 50 kHz
cal energy into electric energy and vice versa. The transducers are submersed in a water tank with the sample
transducers used for these experiments were omni- placed on the top to form a reflection boundary between water
directional, and are suitable for measuring the reflec- and the sample. An incident angle of 20 degrees resulted
tion coefficients over a range of incident angles. The from the necessary addition of polystyrene foam between the
output from the transducers has a dominant frequency source and receiver to attenuate the direct arrivals.
of 50 kHz and a bandwidth between around 10 kHz and
70 kHz. The electrical input for the source transducer is
provided from a Panametrics 5055PR pulser/receiver. Table 3. Physical properties of the elastic materials used for
testing and calibration of the experiments.
These have a maximum pulse of 250 V and an out-
put adjustable in four steps to between 18 and 220 J, Material Density (g/cc) Velocity (m/s) Impedance
with a rise time of less than 10 ns. The received sig-
nals have a maximum frequency of 10 MHz with Acetal 1.41 2230 3140
the bandwidth controlled by an analogue high-pass Tivar 0.93 2120 1970
filter. The high-pass filter is applied before analogue- Acrylic 1.19 2670 3180
to-digital (A/D) conversion to prevent the very low
frequency amplitudes from saturating the A/D con-
verter. A preamplifier then boosts the analogue signal
with a voltage gain of 40 dB. verify that the methodology was valid and to calibrate
The signal is digitized using a Nicolet 430 digital the measurements to allow reflection coefficients to
oscilloscope, which is a two channel, 12 bit system be calculated. Three different elastic materials with
with a sample interval of 100 ns (10 Megasamples per known physical properties were placed on top of
second). The dynamic range of the recorded signal can the water filled container and the reflection ampli-
be adjusted between 30 mV and 120 V to optimize the tudes were measured. The materials used were Tivar,
signal-to-noise ratio. Triggering is controlled with a Acrylic and Acetal, with the physical properties listed
personal computer (PC) through the parallel printer in Table 3. The reflection coefficient was calculated
port. The oscilloscope is controlled through the IEEE- as the ratio of the measured amplitudes from the elas-
488 interface bus. tic materials relative to the amplitudes recorded from
The experiment was configured with the source and the water/air interface, which is assumed to have a
receiver transducer fixed in the base of a container reflection coefficient of unity.
at a depth of 100 mm (Figure 3). The container was Figure 4 shows the recorded waveforms from the
filled to the top with water such that the reflection water/air, water/Tivar and water/Acrylic interfaces.
ray path formed an incident angle of 20 degrees at The first arrival for the water/air interface is a peak,
the water/air interface. The transducers were separated representing a negative impedance contrast, while
by 20 mm of polystyrene foam which was necessary the first arrivals from the elastic material reflections
to attenuate the strong direct arrival and prevent it were a trough, representing positive impedance con-
from interfering with the reflected signal. Ideally the trasts. The amplitudes for a number of events were
experiments would be performed at normal incidence measured and compared to ensure that the appropri-
but this requires a single transducer to be used as ate events were being measured for valid results. It
both source and receiver which is not practical for was found that the first three events had consistently
such a low-frequency. In this case, the piezoelectric repeatable and comparable amplitudes that could be
element is still ringing from the source pulse when used for the calibration. Later arrivals had inconsis-
the reflected energy arrives and accurate reflection tent amplitudes as a result of interference from other
amplitudes cannot be measured. reflections from the walls of the container. Table 4
Before any measurements were made on porous shows the measured reflection amplitudes and a com-
samples, a number of elastic materials were tested to parison between the predicted and measured reflection

588
Copyright 2005 Taylor & Francis Group plc, London, UK
 Table 5. Physical properties of the two synthetic CIPS
sandstones used for reflection measurements.

Density Porosity Velocity

Sample (g/cc) (%) (m/s) Impedance

CIPS #1 1.83 29.1 2190 4000

CIPS #2 1.80 30.3 2640 4740

cement in a manner that mimics the cementation pro-

cess in nature where sediments are transformed into
calcarenites and limestone.
The CIPS process has recently been refined and
applied to sand packs with pre-determined grain size
distributions to produce sandstones with systemati-
cally varying properties such as porosity, permeability
and acoustic impedance (Sherlock and Siggins, 2003).
The aim is to provide a porous media representative
Figure 4. Reflection coefficients were measured by com-
paring amplitudes of early reflection arrivals relative to of natural sandstones but with physical properties that
amplitudes from the water/air interface. can be tailored for specific research applications where
precise control over individual parameters is desired.
The main reason for choosing CIPS sandstones for
Table 4. Measured reflection amplitudes, calculated reflec- these experiments was that they can be manufactured
tion coefficient (RC) and the predicted reflection coefficient as large homogeneous samples. It is not feasible to use
from theory for the four reflection experiments used for
verification and calibration of the experimental method.
standard core samples of natural sandstones because
of the low-frequency used. The wavelength of the sig-
Reflection Measured Theoretical nal from the 50 kHz transducers is 30 mm in water and
Interface amplitude RC RC around 5060 mm in a typical sandstone. To achieve a
valid reflection response it is necessary to use a sam-
Water/Air 11990 1.00 1.00 ple with thickness greater than one wavelength and
(assumed) a surface area significantly greater than the size of
Water/Tivar 1392 0.14 0.14 the first Fresnel zone, which in this case is around
Water/Acrylic 4175 0.40 0.36 75 mm. This meant that large homogeneous sandstone
Water/Acetal 4248 0.41 0.36 samples were required. An additional benefit of using
CIPS sandstones is that it will be possible in future to
fabricate a range of sandstones with pre-determined
coefficients from these tests. The results suggest that permeabilities to allow validation of the theoretical
measurement errors for small impedance contrasts are results over a wide range of experimental conditions
negligible, while for higher reflection coefficients the in a systematic and controlled manner. Table 5 list the
error increases to around 10%. For future experiments, physical properties of the two large CIPS samples used
the errors could be reduced through better isolation of for these experiments.
the source and receiver transducers and also the use of To create a sharp interface between a fluid and a dry
a larger water tank to prevent inference from spurious porous material it was necessary to find a non-wetting
reflections. fluid that would not get imbibed into the sample
through capillary forces. Most sandstones, including
the synthetic CIPS sandstone used for these experi-
ments, are water-wet. A range of fluids with varying
4.2 Samples and wettability
density, viscosity and surface tension were tested and
Once the methodology was established and verified, all were imbibed into the sandstone.
the reflection coefficients from two different syn- As it was not possible to find a fluid that was non-
thetic samples were measured. The artificial sandstone wetting relative to air, it was decided instead to change
samples were made using a process known as the the wetting properties of the CIPS sample. A chemical
Calcite In-situ Precipitation System (CIPS), the tech- traded under the name of StainProof (manufactured
nology was developed at the CSIRO (Kucharski et al., by Dry Treat) was found that disallows the passage of
1996). The process involves the application of a pro- water while still being permeable to air when coated on
prietary solution causing the crystallisation of calcite the surface of the sandstone. This was coated onto the

589
Copyright 2005 Taylor & Francis Group plc, London, UK
two CIPS samples and the experiments were successful experimentally obtained values of reflection coeffi-
using water as the non-wetting fluid. cients measured at particular frequencies is shown
in these figures. In both examples the theoretical
curves are in good agreement with the experimental
5 COMPARISON OF THEORETICAL AND observations.
EXPERIMENTAL RESULTS

Reflection coefficients obtained experimentally and 6 CONCLUSIONS

theoretically (using expression 1) are shown in Fig-
ures 5 and 6. Physical properties of the two synthetic Theoretical expressions for reflection coefficients for
CIPS sandstones are given in Table 5. Theoretical poroelastic interfaces show the large deviation of
curves for the poroelastic case (open pores) and the poroelastic coefficient from the elastic one for an
elastic reflection coefficient (sealed pores) are drawn interface between water and an air-saturated porous
as a function of frequency. The comparison with the medium at low frequencies. The analysis reveals that
this low frequency effect is due to the fact that pro-
portionality constant between poroelastic parameters
1 and square root of frequency is very large due to large
poroelastic contrast in compressibilities of the pore fluid and the
0.9 elastic free fluid.
measured
0.8 The final results for the sandstone blocks measured
0.7 are consistent with the theoretical predictions within
experimental error. It is possible to record higher fre-
0.6 quency measurements from the same samples using
|R 11|

0.5 standard laboratory piezoelectric transducers although

this would be of little benefit due to the asymptotic
0.4
nature of the reflection coefficient curve at higher fre-
0.3 quencies. Ideally, the experiments would be extended
0.2 to even lower frequencies than 50 kHz. It is possible
to build transducers specifically for lower frequency
0.1 measurements. However, it is not practical to attempt
0 measurements at frequencies lower than 50 kHz due to
10-4 10-3 10-2 10-1 100 101 102 103 104 the long wavelengths that would require prohibitively
Frequency (Hz) large samples to be used.
These measurements confirm the validity of theo-
Figure 5. Measured reflection coefficient at the interface retical expressions derived and demonstrate the pres-
between free water and air-saturated sample (CIPS #1) are ence of the large poroelastic effect at the interface
compared with theoretical models. between free fluid and fluid-saturated porous materials
well below Biots characteristic frequency.
1
0.9 poroelastic
elastic ACKNOWLEDGMENTS
0.8 measured
0.7 This work was supported by the Curtin University of
0.6 Technology small discovery/linkage grants scheme,
|R 11|

0.5 Centre of Excellence for Exploration and Production

Geophysics, Curtin Reservoir Geophysics Consortium
0.4 and CSIRO Postdoctoral Fellowship Program.
0.3
0.2
REFERENCES
0.1
0 Biot, M.A. 1962. Mechanics of deformation and acoustic
10-4 10-3 10-2 10-1 100 101 102 103 104 propagation in porous media. Journal of Applied Physics,
Frequency (Hz) 33: 14821498.
Bourbi, T., Coussy, O. & Zinszner, B. 1987. Acoustics of
Figure 6. Measured reflection coefficient at the interface porous media. Paris: Technip,.
between free water and air-saturated sample (CIPS #2) are Brekhovskikh, L.M. & Godin, O.A. 1990. Acoustics of
compared with theoretical models. layered media. Vol. 1, Heiledberg: Springer-Verlag.

590
Copyright 2005 Taylor & Francis Group plc, London, UK
Denneman,A.I.M., Drijkoningen, G.G., Smuelders, D.M.J. & Kucharski, E., Price, G., Li, H. & Joer, H.A. 1996. Laboratory
Wapenaar, K. 2002. Reflection and transmission of waves evaluation of CIPS cemented calcareous and silica sands.
at a fluid/porous-medium interface. Geophysics, 67: 7th Australia New Zealand Conference on Geomechanics:
282291. 102107.
Gassmann, F. 1951. Elastic waves through a packing of Sherlock, D. H. & Siggins A. F. 2003. The development
spheres. Geophysics, 16: 673685. of synthetic CIPS sandstones for geophysical research.
Gurevich, B., Ciz, R. & Denneman, A.I.M 2004. Sim- ASEG Conference and Exhibition, Adelaide, Feb 18.
ple expressions for normal incidence reflection coeffi-
cients from an interface between fluid-saturated porous
materials. Geophysics, 69: 13721377.

591
Copyright 2005 Taylor & Francis Group plc, London, UK