Schmitt CMC-NCE Contributions May2013

Please cite this article in press as: Njiekak, G., et al.
, CO
2
rock physics as part of the Weyburn-Midale geological storage project. Int. J.
Greenhouse Gas Control (2013), http://dx.doi.org/10.1016/j.ijggc.2013.02.007
ARTICLE IN PRESS
G Model
IJGGC-867; No. of Pages 16
International Journal of Greenhouse Gas Control xxx (2013) xxxxxx
Contents lists available at SciVerse ScienceDirect
International Journal of Greenhouse Gas Control
j our nal homepage: www. el sevi er . com/ l ocat e/ i j ggc
CO
2
rock physics as part of the Weyburn-Midale geological storage project
Gautier Njiekak
, Douglas R. Schmitt, Helen Yam, Randolf S. Kofman

Department of Physics, Institute for Geophysical Research, CCIS 4-183, University of Alberta, Edmonton, AB, Canada T6G 2E1
a r t i c l e i n f o
Article history:
Received 17 August 2012
Received in revised form7 February 2013
Accepted 11 February 2013
Available online xxx
Keywords:
Seismic monitoring
Ultrasonic velocity
CO
2
rock physics
Carbonates
Weyburn reservoir
a b s t r a c t
To develop condence in the seismic techniques that are used to (i) qualitatively locate and track the
movement of the CO
2
plume, and(ii) quantitatively determine the amount of CO
2
in place in the Weyburn
pool, a good understanding of the effects of CO
2
onseismic waves is critical. For this purpose, an extensive
series of ultrasonic measurements were performed on over twenty core samples from the Weyburn-
Midale carbonates (Marly and Vuggy units) and the overlying and underlying formations. Care was taken
to separate pore uid effects from pore pressure build-up effects during the experiments. This allowed for
the sampling of the effects of the CO
2
s varying phase states (gasliquidsupercritical uid) on the overall
rock seismic response. The current paper provides a subset of measurements conducted on four samples
fromthe Marly and Vuggy units. Of the observations arising from the measurements, there are two that
are of particular note. First, both the P- and S-wave speeds decrease substantially as the CO
2
transforms
from gas to either liquid or supercritical phase. This observation is consistent with the increase of CO
2
uid density across these phase boundaries. Second, across the gasliquid phase transition both wave
speeds drop abruptly as would be expected for the change in the physical properties of the CO
2
across
this rst order phase boundary. In contrast, across the gassupercritical phase boundary the velocities
change more gradually. This suggests that it may be difcult to distinguish the gassupercritical boundary
using seismic reection techniques. Illustrative modelling of seismic reectivities within a hypothetical
geological formation with physical properties equal to that of one of the measured samples, however,
suggests that a CO
2
liquidwater contact is a good seismic reector.
2013 Elsevier Ltd. All rights reserved.
1. Introduction
The Weyburn-Midale geological storage project is a research
component of EnCanas CO
2
EnhancedOil Recovery project that has
been underway since 2000 in the Weyburn pool, Saskatchewan,
Canada. As is well known, a major concern with geological CO
2
sequestration is the potential for CO
2
leakage. Social acceptance of
such projects depend critically on the ability to monitor movement
of CO
2
in the subsurface. Geophysical remote sensing techniques,
particularly the use of 4-D reection seismology, are regarded as
a promising tool to monitor the ow of CO
2
underground. The
effectiveness of seismic monitoring will depend upon the abil-
ity to acquire high-resolution seismic data, since seismic methods
have shown to be very effective in detecting time-lapse changes
(Tura and Lumley, 1999; Carcione et al., 2006; Kazemeini et al.,
2010; Robinson and Davis, 2011; Ivanova et al., 2012). There is
currently a great need for a better understanding of how injec-
tion of CO
2
will inuence the seismic properties of rocks and
also of how seismic methods can help get reliable quantitative
Corresponding author. Tel.: +1 780 710 9304; fax: +1 780 492 0714.
E-mail addresses: njiekak@ualberta.ca, gnjiekak@yahoo.com(G. Njiekak).
estimates of subsurface CO
2
. To better dene the dependence of
seismic properties on the characteristic porosity, mineralogy, and
uid saturation in the Weyburn reservoir, we have performed an
extensive series of petrophysical and rock physics testing on over
twenty carbonate reservoir and cap-rock core samples from the
Weyburn oileld. This was done under a large variety of conning
and pore pressures, temperatures and saturation states. Saturating
uids included nitrogen gas, CO
2
, and water.
The current paper focusses only on the effects of the different
phases of CO
2
on the propagation of longitudinal (P) and trans-
verse (S) elastic waves through candidate Weyburn carbonates.
CO
2
at depths intended for geological sequestration can be in gas,
liquid or supercritical phase states. It is well known that such
pressuretemperature dependent variations will inuence seis-
mic wave speeds and reectivities through changes in the uid
compressibility and density. It is these changes that will ulti-
mately be sensed using surface seismic techniques; and numerous
authors have carried out theoretical studies to model the expected
responses (e.g., Ravazzoli and Gmez, 2011; Khatiwada et al., 2012;
Ghosh and Sen, 2012; Guan and Tutuncu, 2012). However, to our
knowledge, there is a general lack of experimental tests of such
effects. It is well knownthat CO
2
s seismic properties, i.e., bulkmod-
ulus andbulk density, are considerably different thanthose of other
1750-5836/$ see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijggc.2013.02.007
Please cite this article in press as: Njiekak, G., et al., CO
2
ARTICLE IN PRESS
G Model
2 G. Njiekak et al. / International Journal of Greenhouse Gas Control xxx (2013) xxxxxx
Fig. 1. Location of the Weyburn eld within the Williston basin.
in situ pore uids such as oil and brine. Therefore introduction of
CO
2
into the pore space would yield a change in the overall seis-
mic signature allowing the movement of the subsurface CO
2
to be
qualitatively tracked from seismic surveys. Quantitative determi-
nation of the level of CO
2
saturation from the changes in seismic
signals however is muchmore difcult. Furthermore, unlike typical
pore uids encountered in situ and as mentioned above, CO
2
can
exist in three different phase states and this adds complexity to the
accurate interpretation of seismic reectivity.
Most CO
2
ooding experiments to date involved the injection of
CO
2
into a porous media pre-saturated with another in situ uid
and the acoustic variations observed are usually from a combi-
nation of pore pressure and uid substitution effect. The results
of these studies are difcult to apply to seismic interpretations
because the partial saturation of CO
2
is usually unknown and the
observed changes solely due to CO
2
cannot be quantitatively deter-
mined. To differ from previous works (e.g., Purcell et al., 2010;
Lei and Xue, 2009; Shi et al., 2007; Kummerow and Spangenberg,
2011), this work involves full CO
2
saturation and the separation
of seismic effects due to changes in the pore uid properties from
thoseinducedbyporepressurebuild-up. It is important topoint out
that full CO
2
saturation is unlikely in actual geological sequestra-
tion projects. However, before attempting to understand the wave
behaviours through more realistic mixed brineCO
2
saturations,
it is instructive to have an end member study in order to con-
straint themorecomplexandmoreprobablecases of partial/patchy
saturation.
In this paper, we carry out a series of measurements of the
ultrasonic P and S-wave velocities through rocks fully saturated
with CO
2
under a variety of pressure and temperature conditions.
A brief review of the Weyburn-Midale geology sets the stage for
the characterization of the core materials studied. The experimen-
tal set-up and the protocols developed to make measurements that
isolate as much as possible the effects of the CO
2
uid properties
are described. The results of a representative subset of samples are
then given. The paper concludes with a simple illustrative case that
studies the changes in the seismic reectivity with pore pressure
(and hence phase state) for a waterCO
2
liquid contact lying within
a geological formation that has the measured physical properties
of one of the samples.
2. Weyburn reservoir properties
The Weyburn unit, located in the southeast corner of
Saskatchewan (western Canada), is a 180km
2
oileld discovered
in 1954 (Fig. 1) The eld produces 2534 degree API mediumgrav-
ity sour crude from the Midale beds of the Mississippian Charles
Formation (e.g., Way and Hancock, 1999). The reservoir, located at
an average depth of 1.5km, is subdivided into an upper dolostone
unit: the Marly, with an average thickness of 6m, and a lower lime-
stone unit: the Vuggy, which averages around 15m in thickness
(Burrowes, 2001), Fig. 2.
The Vuggy contains porous coated-grain wackestones to grain-
stones developed along a carbonate shoal that forms good-quality
reservoir, and low porosity mudstones; these are interpreted to
represent intershoal deposits of poor reservoir quality (Burrowes,
2001; Whittaker et al., 2004). The porosity of the shoal deposits
can reach 20% and the permeability is ranging from10mD to over
500mD, with an average of 50mD. In the intershoal sediments
the porosity is up to 12% and the matrix permeability varies from
0.1mD to 25mD, averaging 3mD (e.g., Brown, 2002 and references
therein).
The Marlyis a microsucrosic dolostone havingporosities of upto
38%. Permeabilityvaries from1mD toover 100mD, withanaverage
of 10mD. The Marly is more homogeneous than the Vuggy, and
2
ARTICLE IN PRESS
G Model
G. Njiekak et al. / International Journal of Greenhouse Gas Control xxx (2013) xxxxxx 3
Fig. 2. A schematic NESWcross-section through the Weyburn oileld illustrating truncation of inclined Mississippian strata at the Sub-Mesozoic Unconformity (Whittaker
and Rostron, 2001).
Table 1
Rock physical properties.
Sample nr. Formation Length
a
(cm) Grain density
(g/cm
3
)
Porosity based on
He-pycnometry (%)
Porosity based on Hg
porosimetry (%)
Porosity based on
micro-CT images (%)
Air permeability
b
(mD)
M1 Marly 5.62 2.67 8 7 8 <2
M2 Marly 4.80 2.83 17 14 12 <2
V1 Vuggy 4.68 2.67 8 9 9 H: 13
V: 30
V2 Vuggy 5.63 2.71 11 9 5 H: 16
V: 35
a
Refers to the length of the plugs that were used for ultrasonic measurements. Each plug had a diameter of 3.81cm.
b
H and V describe the permeability measured parallel and normal to the plugs axis, respectively.
deposition of its sediments, prior to dolomitization, took place in
relatively quiet waters.
The Midale Evaporite, a competent anhydrite layer, provides
the upper seal to the reservoir. It is a diagenetically altered zone
that has very lowpermeability and occurs at the up-dip portion of
the Midale beds subjacent to the regional Sub-Mesozoic Unconfor-
mity, which represents a depositional hiatus of approximately 60
million years. The relatively impermeable Triassic Lower Watrous
Member extends across much of southern Saskatchewan (e.g.,
Whittaker et al., 2004, and references therein) lying above the
Sub-Mesozoic Unconformity. The Frobisher unit, a succession of
upwards shoaling, shallowmarine carbonate-evaporite sediments,
forms the bottomseal to the Midale beds.
3. Sample characterization
Standard petrophysics tests and imaging analysis were per-
formed on fourteen samples representative of the Weyburn
reservoir layers (Marly and Vuggy units) and six samples from
the overlying and underlying formations. Later in this paper we
focus, however, on the results from a smaller representative sub-
set. The petrophysics tests included (i) He pycnometry for the
determination of the samples grain density and estimation of
bulk porosity using a Quantachrome
TM
Multipycnometer (model
MVP-D160-E), (ii) Hg injection porosimetry for the estimation of
the pore size distribution using a Micromeritics AutoPore IV 9500
(maximum capillary pressure of 413.6MPa in principle allowing
penetration of Hg through 3nm pore throats), (iii) permeability
measurement using an air permeameter (the NERs Tiny Perm II),
(iv) X-ray diffraction (XRD) analysis carried out using a Rigaku
Geigerex Power Diffractometer in the Department of Earth and
Atmospheric Science at the University of Alberta and (v) X-ray u-
orescence (XRF) whole rock chemistry provided commercially.
Imaging analysis of the rock fabric was carried out at a
variety of different scales. Textures were analysed in two dimen-
sions using thin sections. Scanning electron microscopy (SEM)
was used in assisting the examination of microstructure. In
addition to this, X-ray computed tomography (CT resolution:
0.5mm0.35mm0.35mm voxels) and micro-computed X-ray
tomography (micro-CT resolution: 5m5m5m vox-
els) analyses were performed to quantify the 3-dimensional
characteristics of the rock fabrics.
This paper describes only those measurements on four sam-
ples fromthe Marly and Vuggy units, which are the targets for the
Table 2
XRF whole rock analysis.
Sample nr. SiO
2
(%) Al
2
O
3
(%) Fe
2
O
3
(T) (%) MnO (%) MgO (%) CaO (%) Na
2
O (%) K
2
O (%) TiO
2
(%) P
2
O5 (%) Cr
2
O
3
(%) V
2
O5 (%) LOI (%) Total (%)
M1 5.35 0.66 1.57 0.026 2.82 47.34 <0.01 0.01 0.03 0.03 <0.01 <0.003 40.83 98.66
M2 8.98 1.43 0.67 0.015 14.36 32.84 0.01 0.01 0.08 0.03 <0.01 0.003 40.63 99.06
V1 0.98 0.32 0.24 0.009 1.23 53.47 <0.01 <0.01 0.01 0.01 <0.01 <0.003 42.82 99
V2 2.87 0.75 0.47 0.009 6.24 45.52 0.1 0.02 0.02 0.01 <0.01 0.004 43.31 99.32
2
ARTICLE IN PRESS
G Model
Fig. 3. Thin sections (plane-polarized light), left column, and SEM images, right column, of the four selected samples. (a) and (b) Marly samples. Note the well cemented
dolomite grains of sample M1 compared to those of sample M2 fromthe SEMimages. Dol =Dolomite, Cal =Calcite. (c) and (d) Vuggy samples. Note the abundance of fossil
fragments in the thin sections; red arrowin (d) indicates intraparticle porosity in foraminifera. Red arrows in (c) indicate (cement) crystals that are likely lling vugs.
storage of CO
2
. The general behaviour of these four samples is sim-
ilar to what was recorded on the other Marly and Vuggy samples
investigatedduringour research. Table1shows thedensity, various
measures of the porosity, and the permeability of the four samples.
The XRF whole rock analyses are presented in Table 2.
3.1. The Marly unit
The Marly unit is represented by two dolomite samples (M1 and
M2). Dolomite is the dominant mineral in these samples, form-
ing more than 90% of the rock mineralogical composition. Calcite,
2
ARTICLE IN PRESS
G Model
Fig. 4. Log differential intrusion vs. pore size for the matrix of the four studied samples. Dominant pore throat sizes are 0.65m in samples M1 and M2 and 0.35m and
0.55 min samples V1 and V2, respectively.
ankerite, quartz, and feldspars are the other constituent minerals.
Sample M1 is more compact than sample M2, with tightly packed
dolomite crystals making most of its matrix (Fig. 3). In both sam-
ples, the porosity is generally intercrystalline and the main pore
throat size is around 0.65m(Fig. 4).
3.2. The Vuggy unit
The two studied Vuggy samples (V1 and V2) display a mud-
dominated fabric typical of wackestone to packstone facies. Calcite
is the main mineral in sample V1, forming more than 90% of its
matrix. Other minerals present in sample V1 are dolomite and
quartz. In sample V2, calcite and dolomite are the main miner-
als, both making more than 95% of the rock (in an approximately
2 to 1 ratio). Ankerite, quartz, and feldspars are the other minerals
observed in sample V2. Porosity is vuggy, moldic, intergranular and
intragranular inbothVuggysamples (Fig. 3). Hginjectionporosime-
try indicates dominant pore throat sizes are 0.35mand 0.55m
in samples V1 and V2, respectively (Fig. 4).
3.3. Porosity estimation
Different methods were used to estimate the samples porosity
(Table 1). He pycnometry was used to determine the porosity of
the plugs that were later usedfor the ultrasonic measurements; the
bulk volumes of the four plugs range from53 to 64cm
3
. Smaller
sample fragments (bulk volumes lower than 5cm
3
) were used for
the Hg porosimetry test.
An attempt was made to estimate porosity based on micro-CT
scans completed for 5mm cubic fragments from each sample.
The scan resolution was set to 5m for each sample. Each scan
produces a sequential stack of 8-bit grey-scale images, providing
a detailed view of the internal structure of each sample. Porosity
determination was based on binary images created by threshold-
ing a region of interest (ROI) within the original stack of grey-scale
images. The threshold pixel brightness, between 0 (black) and 255
(white), was selected manually and represents the transition from
pore to mineral.
Samples M1 and V1 showa relative good match between poros-
ity values based on the three methods (micro-CT image analysis,
He pycnometry and Hg injection porosimetry; Table 1). For sam-
ples M2 and V2, the micro-CT porosity values are the lowest; the
rock fragments that were used to generate the micro-CT images
may represent local textural variations (with a very dense matrix)
in the two samples. Such textural patterns have been observed in
carbonates rocks, where heterogeneous pore-size distributions and
permeabilities canrange inscale over at least 6orders of magnitude
(e.g., Baechle et al., 2008).
In the following sections of this paper, porosity obtained by
means of He pycnometry will be used in the analysis and inter-
pretation of the ultrasonic velocities.
4. Ultrasonic measurements
4.1. Experimental apparatus
The pulse transmission technique was the underlying method-
ology used for the ultrasonic measurements in our study. The
experimental setupincludedseveral functional units suchas apres-
sure vessel to encase and apply an hydrostatic conning pressure
2
ARTICLE IN PRESS
G Model
Fig. 5. The experimental setup for the ultrasonic measurement.
to the core sample; a heat tape wrapped around the pressure vessel
to control the temperature; a pulse generator (JSR-PR35); a set of
specially constructed transmitting and receiving transducers (res-
onant frequency=1MHz) to allow for the external excitation and
recording of the pulsed signal; tanks as pore uid sources; inde-
pendent pumps (a Quizix
TM
dual-cylinder pumping system) for
regulating conning and pore pressures and a digital oscilloscope
(NI USB-5133 100MS/s) for recording waveforms (Fig. 5, see also
Yam, 2011).
4.2. Experimental procedure
The measurements here were adapted from those developed
in Yam (2011). A given test began with the coring of a cylinder
of the sample of interest. The ends of this cylinder were ground
as parallel to one another as possible and the sample was then
dried under vacuum at modest temperatures (70
C). The sam-

ple was then prepared for the high-pressure measurement suite by
attaching the set of ultrasonic transducers to its ends. The sample-
transducers assemblage was then hermetically sealed to prevent
any contamination by the hydraulic uid once in the pressure ves-
sel. The assembly was subsequently placed in the pressure vessel.
The transmitted signal was generated by triggering the transmit-
ting transducer witha fast-rising 200Vsquare wave using the pulse
generator. The generated elastic wave after propagating through
the sample was recorded by the digital oscilloscope at a sampling
interval of 10ns. To reduce random noise effects, the nal wave-
formrecorded was a stack of at least 100 traces. For uid-saturated
experiments, the desired pore uid was introduced into the sample
via stainless steel tubing that connected the pore space of the sam-
ple to the pore uid reservoir located outside of the vessel through
the vessel lid.
4.3. Testing sequence
The testing sequence consisted of:
(1) Obtaining the P- and S-waveforms under a series of increasing
conning pressures (upto 35MPa) but withthe pore space sub-
ject to vacuumto provide the dry properties, this was repeated
at two different temperatures (23
C and 50
C) to evaluate the
effect of temperature on the rock frame.
(2) Performing measurements with the pore space saturated
with inert nitrogen gas. These tests carried out at constant
differential pressure (difference between the conning and the
pore pressure) allowed for assessment of any potential effec-
tive stress variations that could occur in the sample. This was
critical because the rock elastic properties themselves depend
nonlinearly on conning pressure.
(3) Putting the pore space of sample under vacuum for 68h and
then repeating the measurements with full CO
2
saturation
under a variety of pore pressure and temperature conditions
in order to sample the full range of CO
2
phase states. These
measurements are all carried out at constant differential pres-
sure. This involves at least 8 different suites of measurements
carried out under conditions of either constant temperature or
constant pore pressure.
(4) Repeating the dry measurements once the suite of CO
2
tests
were completed. This allowed for the assessment of any
mechanical change that could have altered the rock dry prop-
erties.
(5) Finally carrying out the measurements under full saturation
with distilled water.
For all the dry and saturated measurements, the measurements
were takenafter incremental pressure change of upto3MPa during
pressurization and/or depressurization cycles. The samples were
allowedtoequilibrate at constant pressure for about 10min prior to
acquisitionof thewaveforms. All theuid-saturatedmeasurements
were carried out at constant differential pressure of 15MPa. About
one week was needed to complete a full suite of measurements on
a given sample. In this paper, the results fromonly a representative
subset of the four selected samples are presented.
4.4. Velocity calculation and error analysis
The pulse-transmission technique used here involves generat-
ing a compressional or a shear wave from one end cap (source
transducer). The generated wave propagates through the sample
andis receivedontheoppositeend(receiver transducer). Thetrans-
mit time of this pulse through the sample is measured and the
velocity is simply the ratio of the samples length to this time. The
time is estimated in two steps.
First, we measured the travel time (t
b
) through the aluminium
buffer in the absence of any sample by picking the rst extremum
of the recorded signal. This buffer transit time t
b
depends weakly
on the conning pressure; and this pressure dependence was
determined by cap-to-cap direct measurements under pressure.
Similarly, then the travel time (t
bs
) through the aluminium buffer
and the sample is measured and picked. The travel time through
the sample (t
s
) is simply the difference t
bs
t
b
. Consequently, the
wave velocity v of the sample can be calculated using the simple
relation:
v =
L
s
t
s
=
L
s
t
bs
t
b
(1)
There are sources of errors associated with the determination of
the velocity: the sample length and the travel time picking for both
the buffer calibration measurements and sample measurements.
The samples were prepared with their end faces parallel to within
210
5
m; therefore 210
5
mis the error in sample length. The
travel time error depends on the quality of the signal and it varies
between 1010
9
s for high quality signals (high pressure mea-
surements) and 3010
9
s for low quality signals (low pressure
measurements).
Other less apparent factors and errors considered in the veloc-
ity calculation are (i) sample shortening under pressure, and (ii)
thermal expansion of the aluminium buffer caps and the sam-
ple for measurements done at 50
C. Considering all the different

2
ARTICLE IN PRESS
G Model
Fig. 6. Dry velocities at T =23
C and at T =50
C. (a) Samples M1 and M2 fromthe Marly unit, (b) samples V1 and V2 fromthe Vuggy unit. Black arrows show the evolution
of measurements (pressurization and depressurization cycles).
possibilities of errors, thelargest percentageerrors inP- andS-wave
velocity in this study are estimated at 0.7% and 0.4%, respectively.
4.5. Experimental results
4.5.1. Under dry (vacuum) conditions
On dry samples, and as expected, the compressional (P) and
shear (S) wave velocities increase with conning pressure and
decrease with temperature (Fig. 6). Figs. 7 and 8 showthe P- and S-
waveforms recorded on samples M1 and V1. Waveforms recorded
onthe two other samples (M2 andV2) display similar trends. The P-
andS-wave velocities will generally decrease withporosity. Ineach
sample group (more in the Vuggy than in the Marly), the higher the
porosity, the higher the magnitude of change of wave velocities
with increasing conning pressure (Table 3).
4.5.2. Under uid-saturated conditions
As mentioned before, for the CO
2
-saturated experiments, mea-
surements were done under different conditions of temperature
and pore pressure in order to inspect the pore uid effects on the
Table 3
Percentage change of P- and S-wave velocities with increasing conning pressure
from1 to 35MPa and at roomtemperature (T 23
C).
Sample nr. and porosity Dry rock at T =23
C
P-wave S-wave
M1 ( =8%) 8% 4%
M2 ( =17%) 9% 4%
V1 ( =7%) 2% 1%
V2 ( =11%) 4% 4%
2
ARTICLE IN PRESS
G Model
1 35 1
25
26
27
28
29
30
T
)
c
e
s
(

e
m
i
25 12 7 1
40
41
42
43
44
45
46
T
)
c
e
s
(

e
m
i
25 12 7 1
40
41
42
43
44
45
46
T
)
c
e
s
(

e
m
i
1 35 1
40
41
42
43
44
45
T
)
c
e
s
(

e
m
i
25 12 7 1
24
25
26
27
28
29
30
31
T
)
c
e
s
(

e
m
i
25 12 7 1
24
25
26
27
28
29
30
31
T
)
c
e
s
(

e
m
i
(c) (a)
(b) (d) (f)
(e)
Up-pressurization
Cycle
Down-pressurization
Cycle
Up-pressurization
Cycle
Down-pressurization
Cycle
Fig. 7. Dry and CO
2
-saturated waveforms on sample M1. Top and bottom left: P- and S-waveforms on the dry sample. Measurements were done during up- and down-
pressurization cycles. Note how P- and S-wave travel times decrease with increasing conning pressure. Top and bottom centre: Waveforms were acquired on the CO
2
-
saturated sample at T =23
C; CO
2
was turning from gas to liquid in the pore space. Top and bottom right: CO
2
was turning from gas to a supercritical uid in the pores,
T =50
C. Note the decrease of the S-wave travel times and the different behaviours of the P-wave speeds (decrease at lowpore pressure and increase at higher pore pressure)
from the CO
2
-saturated waveforms.
seismic wave velocities. The key here was to make sure that any
variation in the waveforms was due to change in the pore uid
physical properties. Thus, for each sample, a constant differential
pressure of 15MPa was maintained by varying the conning pres-
sure accordingly to the pore pressure. Assuming that the equal
amount of pore pressure increase cancelled the equal amount of
conning pressure increase, any waveformvariation observed was
solely caused by pore uid effects.
4.5.3. Nitrogen-saturated measurements
Prior to CO
2
-saturated measurements, each sample was sub-
jected to constant differential pressure measurement under inert
nitrogen gas saturation. As change in the nitrogen properties (den-
sity, bulk modulus) only slightly affects sound waves travel times
under therangeof pressureandtemperatureconditions usedinthis
study (Fig. 9), no substantial change in the P- and S-wave velocities
meant that the applied differential pressure (15MPa) was the main
factor controlling the variationof the velocity. This behaviour of the
wave velocities did lend condence to selecting the set of high-
quality measurements that were suitable for the purpose of our
study. Fig. 10 shows that P- and S-wave velocities remained con-
stant on all the four studied samples during the nitrogen-saturated
measurements.
4.5.4. CO
2
-saturated measurements
Results presented here are from measurements made at con-
stant temperature (23
C and 50
C) with the pore pressure varying

from1MPa to25MPa ineachcase. CO
2
is agas at lowporepressures
and is a liquid or a supercritical uid at higher pore pressures
depending on the temperature (Fig. 11). The wave velocities of the
two constant temperature runs for the four samples are plotted in
Fig. 12. On these plots, velocities of the dry sample at a differen-
tial pressure of 15MPa are also shown. P- and S-waveforms of the
CO
2
-saturated samples M1 and V1 are shown in Figs. 7 and 8.
Once CO
2
was injected into the sample, irrespective of the
temperatures, both P- and S-wave velocities decreased with the
pore pressure. Values of CO
2
velocity (i.e. wave velocity on CO
2
-
saturated sample) remained almost identical to the dry velocities
at low pore pressure (1MPa) (Fig. 12). With increasing pore pres-
sure, the drop of the CO
2
velocities was more immediate for the
low temperature run; the large drop generally occurred at a pore
pressure of about 57MPa and 1015MPa for the low tempera-
ture (T =23
C) and high temperature (T =50
C) runs, respectively.
Overall, the decrease of the P- andS-waves travel times right before
the gasliquid and gassupercritical uid transitions were up to
4% and 2%, respectively. Over the phase transition, P-wave veloc-
ity recovered while S-wave speed kept decreasing. The recovery of
2
ARTICLE IN PRESS
G Model
1 35
21
22
23
24
25
26
T
)
c
e
s
(

e
m
i
25 12 7 1
35
36
37
38
39
40
41
Pore Pressure (MPa)
T
)
c
e
s
(

e
m
i
25 12 7 1
35
36
37
38
39
40
41
T
)
c
e
s
(

e
m
i
1 35 1
35
36
37
38
39
40
T
)
c
e
s
(

e
m
i
25 12 7 1
21
22
23
24
25
26
27
T
)
c
e
s
(

e
m
i
25 12 7 1
21
22
23
24
25
26
T
)
c
e
s
(

e
m
i
c) ( (a)
(b) (d) (f)
(e)
Up-pressurization
Cycle
Down-pressurization
Cycle
Up-pressurization
Cycle
Down-pressurization
Cycle
Fig. 8. Dry and CO
2
-saturated waveforms on Sample V1. Top and bottomleft: P- and S-waveforms on the dry sample (up- and down-pressurization measurements); waves
speeds increase with pressure. Top and bottomcentre: Waveforms were acquired on the CO
2
-saturated sample at T =23
C; CO
2
was turning fromgas to liquid in the pore
space. Top and bottomright: CO
2
was turning fromgas to a supercritical uid in the pores, T =50
C. The CO
2
-saturated waveforms showsame trends as on sample M1 (see
Fig. 7).
Fig. 9. P-wave velocity (sound speed) in pure N
2
(curve 1) and pure CO
2
(curves 2
and 3) as a function of conning pressure. The ranges of pressure are those applied
in our study. Curves 1 and 2 obtained at roomtemperature (23
C) while the curve

3 is obtained at 50
C. For each curve, blue open symbols represent velocities in

gas phases. Pink, red, and green solid symbols represent velocities in supercritical
N
2
, liquid CO
2
and supercritical CO
2
, respectively. Note the slight and linear vari-
ation (9%) of the P-wave velocity in N
2
under the conditions applied in our study.
P-wave velocity data at different pressures are based on thermodynamic proper-
ties obtained from the National Institute of Standards and Technology chemistry
webBook (Lemmon et al., 2011).
the P-wave velocity is gradual and is up to 1% at a pore pressure of
25MPa. The total drop of the S-wave travel times is up to 4% over
the entire 125MPa pore pressure interval. Included in Fig. 9 are
the behaviours of P-wave velocity in pure CO
2
based on thermody-
namic properties obtained fromthe National Institute of Standards
and Technology (NIST) chemistry webBook (Lemmon et al., 2011).
The similarities between the behaviour of the P-wave velocity in
pure CO
2
and in saturated rock illustrate the sensitivity of the over-
all seismic response of the rock to the changing properties of the
pore uid during the experiments.
5. Discussion
5.1. Observed CO
2
-saturated wave velocity
Of the observations arising fromthe measurements done in this
study, the following are of particular note:
- First, both the P- and S-wave speeds decrease substantially as
the CO
2
gas is injected into the sample, i.e., when the pore pres-
sure is increased but remains below 78MPa (e.g., Fig. 12).
2
ARTICLE IN PRESS
G Model
Fig. 10. Dry (vacuum) and nitrogen-saturated velocities at T =23
C. The dry samples have no pore pressure by denition. A constant differential pressure of 15MPa was used
for the nitrogen-saturated experiments (a) samples M1 and M2 fromthe Marly unit, (b) samples V1 and V2 fromthe Vuggy unit.
Considering the equations of P- and S-wave velocity in a homo-
geneous isotropic medium(e.g., Geertsma, 1961),
V
P
=
K +(4/3)
and (2)
V
S
=
, (3)
where V
P
and V
S
are P- and S-wave velocities, respectively, and ,
K, andare the bulk density, bulk modulus, andshear modulus of
the elastic medium, respectively, the drop of the wave velocities
are consistent with the increase of the CO
2
gas density.
- Second, the pore pressure at which the largest drop occurred
depends on the CO
2
s temperature. At 50
C, the largest drop

occurs above the CO
2
s critical pressure at about 1015MPa as
changes in CO
2
s physical properties remain gradual over the
gassupercritical uid boundary (Fig. 12). At 23
C, the immedi-
ate nature of the wave responses is due to the sharp contrast in
the bulk modulus and density of the CO
2
when the gasliquid
boundary is traversed; the largest drop usually occurred around
57MPa. The recovery of the P-wave velocity after the phase
transition suggests that CO
2
s bulk modulus dominates over its
density for the control of the P-wave behaviour at this stage (see
P-wave velocity curves in Fig. 12).
- Third, using seismic reection techniques to distinguish the
gassupercritical uid boundary in the eld will be difcult, the
2
ARTICLE IN PRESS
G Model
Fig. 11. CO
2
properties (bulk modulus and density) as a function of pressure and temperature based on the thermodynamic model of Span and Wagner (1996). Arrows (i)
and (ii) refer to the conditions under which CO
2
-saturated rock velocities were measured. Measurements were done under constant differential pressure of 15MPa.
mainreasonbeing the gradual change of the P- and S-wave veloc-
ities on the CO
2
-saturated samples across this phase boundary.
5.2. Modelled CO
2
-saturated wave velocity
The wave frequencies applied in our laboratory work are much
higher than those employed in seismic surveys over geological
sequestration projects. Applying the laboratory results to seis-
mic monitoring will require the high frequency data to be scaled
down to low frequencies by adequate rock physics models. In this
paper, the expected velocities for the studied samples were cal-
culated by applying the Gassmanns equation (Gassmann, 1951).
The Gassmanns model (Gassmann, 1951) provides an estimate of
relaxed velocity at zero frequency, which is a lower bound of the
uid-saturation effect (e.g., Han and Batzle, 2004; Mavko, 2005). It
is widely used to predict the effect of uids on both log and seis-
mic velocity data. By Gassmanns formulation, the bulk modulus
of a saturated rock, K
sat
, is related to the rock frames bulk modu-
lus (frame modulus), K
dry
, the mineral grains bulk modulus (solid
constituent of the rock), K
s
, the pore uids bulk modulus (uid
constituent of the rock), K
f
, and the porosity, , of the rock through:
K
sat
= K
dry
+
(1 (K
dry
/K
s
))
2
(/K
f
) +((1 )/K
s
) (K
dry
/K
2
s
)
(4)
The pressure and temperature dependent bulk modulus and the
density of the pore uid (CO
2
) were determined from the ther-
modynamic properties obtained from NISTs online chemistry
webBook (Lemmon et al., 2011), the bulk and the shear moduli
of the rock frame were determined from ultrasonic pulse trans-
mission measurements on the dry sample and it was estimated
as a function of temperature to account for the softening of the
sample frame with heat, the bulk density of the sample is sim-
ply the ratio of the samples dry mass to the samples envelope
volume, the porosity of the sample was determined by means of
He pycnometry and the bulk modulus of the mineral grains was
obtained from Carmichael (1989). For sample V2, whose mineral
composition includes two main minerals, calcite and dolomite, a
more accurate value of the mineral grains bulk modulus, K
s
, was
determined using Hills average equation (Hill, 1952):
K
s
=
1
2
(K
V
+K
R
) (5)
where K
V
is Voigts average (Voigt, 1928) and K
R
is Reusss average
(Reuss, 1929). These two averages respectively give the maximum
and minimumvalues possible for a mixture:
K
V
=
n
i=1
f
i
K
i
(6)
1
K
R
=
n
i=1
f
i
K
i
(7)
where f
i
is the fraction amount of the ith mineral and k
i
is the bulk
modulus of that corresponding mineral.
In Gassmanns formulation, the shear modulus is regarded to be
independent of the pore uid because uids are incapable of sus-
taining shear forces. Therefore, the shear modulus of the saturated
rock used in the calculation was the same as the shear modulus of
the dry rock. Fromthe determined saturated bulk or shear moduli,
the corresponding saturated P- or S-wave velocity was found from
using Eqs. (2) and (3) where the bulk density of the saturated rock,
sat
, was given by:
sat
= (1 )
s
+
f
(8)
s
and
f
are the mineral grains and pore uid densities, respec-
tively. The density of the mineral grains was determined by means
of He pycnometry.
Figs. 13 and 14 compare the modelled with the observed P- and
S-wave velocities under CO
2
saturation for the four samples. For
both P- and S-waves, the observed laboratory and the Gassmanns
modelled velocities behave similarly over the pore pressure range
(125MPa). Overall, except for sample M1, Gassmanns equa-
tion generally underpredicts the measured velocities by up to 3%.
For each lithological group, in most cases the accuracy in the
Gassmanns prediction decreases with the porosity. On sample M1,
the lowest porous Marly sample, there is a relatively good match
between the measured and the modelled data fromaround 7MPa
upward; for lower pore pressures (i.e., whenCO
2
is inthe gas state),
Gassmanns model overpredicts the observedvelocities at lowtem-
perature andunderpredicts themat higher temperature (50
C). For
vuggy sample V1, which has the same porosity as sample M1, the
divergence between the modelled and the measured velocities is
even greater.
A possible explanation for this could be that Gassmanns model
assumes complete interconnection of the rock pore spaces and
also the possibility for the uid to move freely through the pores.
2
ARTICLE IN PRESS
G Model
Fig. 12. CO
2
-saturated velocities at T =23
C and at T =50
C. Dry wave velocities obtained under a conning pressure of 15MPa are plotted at pore pressure =0MPa to provide
a reference for the changes seen (see the solid star symbols). Black solid square symbols and black open square symbols represent wave velocities when CO
2
was gas and
liquid in the rock pore spaces, respectively; the experiment was performed at 23
C. Orange solid circle symbols and orange open circle symbols represent wave velocities
when CO
2
was gas and a supercritical uid (Sc CO
2
) in the rock pore spaces, respectively; the experiment was performed at 50
C.
Unfortunately, the small pore dimensions (main pore throat size
is less than 1m in both samples) make the visual assessment
of the degree of connectivity difcult. One possible reason for
the poorer performance of Gassmanns formulation for sample
V1 could be that the connectivity assumption is violated. If one
takes into consideration that the dominant pore throat dimen-
sion in sample V1 (0.35m) is small relative to that of sample
M1 (0.7m), it could be easier to have the Gassmanns assump-
tion on pore connectivity violated in sample V1 than in sample
M1.
In addition to the variation in the rock textural pat-
terns, other factors which may have limited the success of
the Gassmanns predictions on some of the tested samples
include:
- the high frequency applied in our studies; the Gassmanns equa-
tion is only valid for describing wave behaviour pertaining to low
frequencies and strictly only at zero-frequency (static deforma-
tion), and
- the lack of pressure equilibrium in the sample when the
pore/conning pressures were changed. For example, in some
carbonates, discrepancies (up to 100m/s and 80m/s for P- and
S-wave velocities, respectively) were observed between the val-
ues of the dry velocity acquired during the rst dry run of the
testing sequence andthose measuredjust after the CO
2
-saturated
measurements. These discrepancies may have beencaused either
by the fact that the tested samples needed much more time to
recover or by the alteration of the rock solid frame by the satu-
rating uid.
2
ARTICLE IN PRESS
G Model
Fig. 13. Observed laboratory data and Gassmanns modelled velocities for samples from the Marly unit under a constant differential pressure of 15MPa at T =23
C and
T =50
C: (a) sample M1, (b) sample M2.

5.3. Implications for seismic monitoring
The results here are not intended to fully describe the real reser-
voir conditions where one will expect partial saturations of the CO
2
,
whatever pressure and temperature dependent phase it may be
in, with other reservoir uids that will include brine and hydro-
carbons. This could be further complicated by heterogeneous, or
patchy, saturation at a variety of scales throughout the reservoir.
The experiments discussed here provide the end-member, fully
saturated case and are intended to illustrate how the variations
of CO
2
phase will inuence the acoustic properties of the rock. A
full modelling of the expected seismic responses based on these
laboratory results is beyond the scope of this paper.
However, at face value the results do provide some implica-
tions towards seismic monitoring of CO
2
uids during geological
sequestration in carbonates. In an ideal situation, we would hope
that we could assess movement of uids remotely by monitoring
changes intheseismic reectivitywithtime, andthat thesechanges
Fig. 14. Observed laboratory data and Gassmanns modelled velocities for samples from the Vuggy unit under a constant differential pressure of 15MPa at T =23
C and
T =50
C: (a) sample V1, (b) sample V2.

2
ARTICLE IN PRESS
G Model
Fig. 15. Water-saturated velocities under a constant differential pressure of 15MPa for sample V2. P-wave velocities slightly increase with pore pressure while S-wave
speeds barely change.
Fig. 16. P- and S-waves acoustic impedances for sample V2 under a constant differential pressure of 15MPa. Black solid square symbols and black open square symbols
represent acoustic impedances when CO
2
is gas and liquid in the rock pore spaces, respectively. Orange solid circle symbols and orange open circle symbols represent acoustic
impedances when CO
2
is gas and a supercritical uid (Sc CO
2
) in the rock pore spaces at T =50
C, respectively. Note that units of impedance are here reported in Rayls with
1 MRayl =10
6
kg/m
2
s.
could allow us to indicate how the uid saturation (including the
phase state of the CO
2
) would vary. For the simplest case of nor-
mal incidence seismic reections from the interface between two
elastic materials, the reection co-efcient R of the reected wave
normalized with respect to the amplitude of the incident wave is
R =
Z
2
Z
1
Z
2
+Z
1
(9)
where Z
i
=V
i
i
is the acoustic impedance that is the simple product
of the wave speed and the material density and medium 1 and 2
are the upper and lower material layers, respectively.
For purposes of demonstration, the P- and S-wave impedances
are calculated from the measured wave speeds and calculated
density for sample V2 (Figs. 15 and 16). When water-saturated
the P-wave acoustic impedances slightly increase with pore pres-
sure due to a combination of increasing density and water bulk
modulus. In contrast, the water-saturated S-wave impedance
remains essentially constant. Under CO
2
-saturated conditions, the
acoustic impedance decreases for both waves when CO
2
is gas in
the pore space and recovers when CO
2
turns into liquid or a super-
critical uid (Fig. 16).
These varying acoustic impedances can be used to calculate (Eq.
(9)) reection co-efcients for a hypothetical geological structure
consisting of a formation with the acoustic properties of sample
V2 that has a fully CO
2
-saturated zone immediately overlying a
fully water-saturated zone (Fig. 17). Such a CO
2
uid over water
geometry is not unexpected in the subsurface as the CO
2
is always
less dense than water at the corresponding in situ conditions and
would be expected to buoyantly migrate upwards until the seal
rocks are encountered. This may, for example, be the case elegantly
detected at the Sleipner project (e.g., Arts et al., 2008; Chadwick
et al., 2006).
Fig. 17. Hypothetical normal incidence P- and S-waves reection co-efcients in a formation with the acoustic properties of sample V2 assuming a CO
2
over water geometry.
In (a), solid and open symbols represent reection co-efcients when CO
2
is gas and liquid in the rock pore spaces, respectively. In (b), solid and open symbols represent
reection co-efcients when CO
2
is gas and a supercritical uid in the rock pore spaces, respectively; the CO
2
velocity data used in the calculation (see Fig. 16) were acquired
at T =50
C.
2
ARTICLE IN PRESS
G Model
The room temperature case might represent a relatively cold
reservoir. In this situation, both R
P
and R
S
rst increase up to the
gasliquid phase transition pressure after which they both rapidly
decrease (Fig. 17a). This suggests that the seismic reectivity would
brighten as the phase transformation pressure is approached but
then nearly discontinuously dimonce the transformation had been
effected. The change in R
S
is even more dramatic. Indeed, once the
CO
2
is liquidR
S
nearlyvanishes. Continuedpressure increase would
lead to a change in the sign of R
S
that would be manifest as a change
in the polarity of the observed shear wave seismic arrival. How-
ever, above the gasliquid phase transition the reectivities do not
vary rapidly and this may make further monitoring of reservoir
evolution difcult.
At the higher temperature of 50
C (Fig. 17b), both R

P
and R
S
again increase with pressure to the gassupercritical transition,
and once the transition is crossed they decrease. In contrast to the
colder 23
C example (Fig. 17a), both reection co-efcients retain

the same sign and, hence, there is no ip of the reections polar-
ity. Further, the values of both remain relatively large throughout
the range of pressures. The decrease of the reection co-efcients
is more sudden over the CO
2
gasliquid transition than over the
gassupercritical uid transition.
Fig. 17 only shows one relatively simple case of the expected
changes in reectivity anticipated for a zone fully saturated with
CO
2
overlying one that is fully water saturated. As such, one should
not assume that they provide a general example of what might be
expected in all cases. In a real situation, one would need further to
consider partial saturations, temperature, and changes in effective
stress. Further, there is no rm criteria to suggest what values of
seismic reectivity are sufciently strong for detection either as
this will depend critically on the actual geological structure, the
dimensions of the zone of interest, its distance from the seismic
source, the seismic acquisition geometry, and the ambient seismic
noise levels during recording. Application of these experimental
results through modelling of expected seismic responses must take
all of these other factors into account.
6. Conclusions
Ultrasonic measurements have been performed on carbonates
fromthe Weyburnreservoir (Marly andVuggy units) to identify the
variationof their seismic properties due toCO
2
saturation. Care was
taken to separate seismic effects due to changes in the pore uid
properties from those induced by pore pressure build-up during
the experiments. This allowed for the investigation of the effects
of CO
2
s varying phase states on the rock seismic properties. From
the analysis of the recorded waveforms, CO
2
phase changes (gas
to liquid and gas to supercritical uid) are marked by a drop in P-
and S-wave velocities of up to 4%. This decrease is controlled by the
increasing density of the CO
2
, and hence the density of the satu-
rated rock, across the phase boundary. However, P-wave velocities
will usually recover after the phase transformation. The gradual
variations of the P- and S-wave velocities over the gas to supercrit-
ical uid transition suggest that this will be difcult to detect using
seismic reection methods.
Full CO
2
saturation was applied in our study, this may not
reect real-world situations where CO
2
may coexist with other
in situ uids such as oil and/or brine in rock pore spaces and
also where patchy saturation is more likely to occur. However,
the current study provides an end-member understanding of what
happens with CO
2
as a pore uid. To better represent the condi-
tions encountered in geological sequestration, future work should
consider adding other possible in situ pore uids such as brine and
oil. Various saturations of CO
2
in various phase states should be
attempted to determine (i) if the different phases of CO
2
or phase
transition can be identied and (ii) the resulting implications for
quantitative assessment of CO
2
saturation state.
An attempt was made to model the core-measured velocities
using Gassmanns formulation. The Gassmann-modelled velocities
show the same trends as the measured CO
2
-saturated veloci-
ties. The modelled data can match the measured velocities but
can also be higher or lower by up to 3% over the range of pore
pressures (125MPa) used in this study. Based on P- and S-wave
impedances for one of the studied carbonate samples under CO
2
-
and water-saturation, we investigated the expected changes in
seismic reectivity for a zone fully saturated with CO
2
overlying
a fully water saturated zone. The results suggest that the seismic
reectivity wouldbrighten as the phase transformation(gas toliq-
uid or gas to a supercritical uid) pressure is approached but then
dimonce the transformation had been effected.
Acknowledgments
The development of the CO
2
measurement system was sup-
ported by NSERC and the Canada Research Chair program. Yamwas
supported in part by an NSERC CGS Scholarship and Njiekak and
Kofman in part by a research contract from the Weyburn-Midale
CO
2
Monitoring andStorage Project as well as a researchgrant from
Carbon Management Canada. The technical assistance of Len Tober
and Lucas Duerksen was invaluable.
References
Arts, R., Chadwick, A., Eiken, O., Thibeau, S., Nooner, S., 2008. 10 years experience of
monitoring CO
2
injection in the Utsira sand at Sleipner, offshore Norway. First
Break 26, 2936.
Baechle, G.T., Colpaert, A., Eberli, G.P., Weger, R.J., 2008. Effects of microporosity on
sonic velocity in carbonate rocks. The Leading Edge 27, 10121018.
Brown, L.T., 2002. Integration of rock physics and reservoir simulation for the inter-
pretation of time-lapse seismic data at Weyburn Field, Saskatchewan. M.Sc.
Thesis, Colorado School of Mines, Golden, CO, 207 pp.
Burrowes, G., 2001. Investigating CO
2
storage potential of carbonate rocks during
tertiary recovery from a billion barrel oil eld, Weyburn, Saskatchewan: Part
2 reservoir geology (IEA Weyburn CO
2
Monitoring and Storage Project); In:
Summary of Investigations 2001. Volume 1, Saskatchewan Geological Survey,
Sask. Energy Mines, Misc. Rep. 2001-4.1, pp. 6471.
Carcione, J.M., Picotti, S., Gei, D., Rossi, G., 2006. Physics and seismic modeling for
monitoring CO
2
storage. Pure and Applied Geophysics 163, 175207.
Carmichael, R.S., 1989. Practical Handbook of Physical Properties of Rocks and Min-
erals. CRC Press, Inc., Boca Raton, FL, 741 p.
Chadwick, R.A., Noy, D., Lindeberg, E., Arts, R., Eiken, O., Williams, G., 2006. Cal-
ibrating reservoir performance with time-lapse seismic monitoring and ow
simulations of the Sleipner CO
2
plume. In: 8th International Conference on
Greenhouse Gas Control Technologies (GHGT-8), Trondheim, Norway.
Gassmann, F., 1951. Elasticity of porous media: Uber die Elastizitat poroser Medien.
Vierteljahrsschrift der Naturforschenden Gesselschaft in Zurich 1, 121.
Geertsma, J., 1961. Velocity-log interpretation: the effect of rock bulk compressibil-
ity. Society of PetroleumEngineers Journal 1, 235248.
Ghosh, R., Sen, M.K., 2012. Predicting subsurface CO
2
movement: from laboratory
to eld scale. Geophysics 77, 2737.
Guan, S., Tutuncu, A.N., 2012. Constraining elastic rock properties in a CO
2
enhanced
oil recovery reservoir using sonic log data. In: Proceedings of the 46th US Rock
Mechanics/Geomechanics Symposium, Chicago, IL, USA, p. 13.
Han, D., Batzle, M.L., 2004. Gassmanns equation and uid-saturation effects on
seismic velocities. Geophysics 69, 398405.
Hill, R., 1952. The elastic behaviour of crystalline aggregate. Proceedings of the
Physical Society 65, 349354.
Ivanova, A., Kashubin, A., Juhojuntti, N., Kummerow, J., Henninges, J., Juhlin, C., Lueth,
S., Ivandic, M., 2012. Monitoring andvolumetric estimationof injectedCO
2
using
4D seismic, petrophysical data, core measurements and well logging: a case
study at Ketzin, Germany. Geophysical Prospecting 60, 957973.
Kazemeini, S.H., Juhlin, C., Fomel, S., 2010. Monitoring CO
2
response on surface
seismic data; a rock physics and seismic modeling feasibility study at the CO
2
sequestrationsite, Ketzin, Germany. Journal of Applied Geophysics 71, 109124.
Khatiwada, M., Adam, L., Morrison, M., van Wijk, K., 2012. A feasibility study of
time-lapse seismic monitoring of CO
2
sequestrationina layeredbasalt reservoir.
Journal of Applied Geophysics 82, 145152.
Kummerow, J., Spangenberg, E., 2011. Experimental evaluation of the impact of
the interactions of CO
2
SO
2
, brine, and reservoir rock on petrophysical proper-
ties: a case study fromthe Ketzin test site, Germany. Geochemistry, Geophysics,
Geosystems 12, 110.
2
ARTICLE IN PRESS
G Model
Lei, X., Xue, Z., 2009. Ultrasonic velocity and attenuation during CO
2
injection into
water-saturated porous sandstones: measurements using difference seismic
tomography. Physics of the Earth and Planetary Interiors 176, 224234.
Lemmon, E.W., McLinden, M.O., Friend, D.G., 2011. In: Linstrom, P.J., Mallard, W.G.
(Eds.), Thermophysical Properties of Fluid Systems in NIST Chemistry WebBook,
NIST Standard Reference Database Number 69, National Institute of Standards
and Technology, Gaithersburg MD, 20899, http://webbook.nist.gov (retrieved
20082011).
Mavko, G., 2005. Expert answers. CSEG Recorder 30 (5), 812.
Purcell, C., Mur, A., Soong, Y., Mclendon, R., Haljasmaa, I.V., Harbert, W., 2010. Inte-
grating velocity measurements in a reservoir rock sample from the SACROC
unit with an AVO proxy for subsurface supercritical CO
2
. The Leading Edge 29,
192195.
Ravazzoli, C.L., Gmez, J.L., 2011. AVA seismic reectivity analysis in carbon diox-
ide accumulations: sensitivity to CO
2
phase and saturation. Journal of Applied
Geophysics 73, 93100.
Reuss, A., 1929. Berechnung der iessgrense von mischkristallen auf grund der plas-
tizitatbedingung fur einkristalle. Zeitschrift fur Angewandte Mathematik aus
Mechnik 9, 4958.
Robinson, A., Davis, T.L., 2011. Improvingefciencyof CO
2
sequestrationusingshort-
term seismic monitoring: applications from the Postle eld EOR programme.
First Break 29 (1), 7578.
Shi, J., Xue, Z., Durucan, S., 2007. Seismic monitoring and modelling of super-
critical CO
2
injection into a water-saturated sandstone: Interpretation of
P-wave velocity data. International Journal of Greenhouse Gas Control 1,
473480.
Span, R., Wagner, W., 1996. A new equation of state for carbon dioxide covering
the uid region fromthe triple-point temperature to 1100K at pressures up to
800MPa. Journal of Physical Chemistry 25, 15091596.
Tura, A., Lumley, D., 1999. Estimating pressure and saturation changes from time-
lapse AVOdata. In: 69thAnnual International Meeting, SEG, ExpandedAbstracts,
pp. 16551658.
Voigt, W., 1928. Lehrbuch der Kristallphysik: Teubner.
Way, W.P., Hancock, S.K.,1999. Weyburn CO
2
miscible ood project eld pro-
duction and pipeline operation impacts. In: 8th Petroleum Conference of the
South Saskatchewan Section. The PetroleumSociety of the Canadian Institute of
Mining, Metallurgy and Petroleum, Regina, Canada.
Whittaker, S.G., Rostron, B., 2001. Geologic storage of CO
2
in a carbonate reservoir
withinthe WillistonBasin, Canada: anupdate. In: FifthInternational Conference
on Greenhouse Gas Control Technologies, Cairns, Queensland, Australia, August
2000. Pergamon, Oxford, pp. 385390.
Whittaker, S., Rostron, B., Khan, D., Hajnal, Z., Qing, H., Penner, L., Maathuis, H., Gous-
sev, S., 2004. Theme 1: geological characterization. In: Wilson, M., Monea, M.
(Eds.), IEA GHG Weyburn CO
2
Monitoring and Storage Project Summary Report
20002004, Proceedings of the 7thInternational Conference onGreenhouse Gas
Control Technologies, vol. III. , pp. 1569.
Yam, H., 2011. CO
2
rock physics: a laboratory study. M.Sc. Thesis, Univ. of Alberta,
Edmonton, 286 pp.
1. INTRODUCTION
Generally, there will be a difference between static and
dynamic bulk moduli and this is usually ascribed to the
difference in strain amplitude. Another potential cause
for the difference between the two moduli is the
presence of cracks because the difference appears to
diminish at high confining pressure [1, 2]. Attempts to
assess effects of pore fluid on the relationships between
static and dynamic moduli have not been considered
useful as there is a significant difference between static
and dynamic moduli even in carefully dried rocks [3].
Today, increasing interest for CO
2
sequestration will
require further understanding of the effects of pore fluid
on elastic properties. This is quite critical as free CO
2
at
depths intended for CO
2
sequestration can be in gas,
liquid or supercritical phase states. Injection of CO
2

could affect rock solid frame and result in changes in the
rock physical properties. Proper knowledge about the
relationships between static and dynamic moduli might
be useful to optimize the characterization of CO
2

saturated geological formations. The current paper
describes laboratory measurements of ultrasonic P- and
S- waves velocities and strain on a dolostone under dry
and CO
2
saturated conditions at different stress levels.
The variations in the strain magnitude of the rock sample
and in the derived static and dynamic bulk moduli are
discussed in relation to the CO
2
's varying phase state
(gas-liquid-supercritical fluid).
2. METHODOLOGY
The studied sample is a dolostone with a (mercury
injection) porosity of 21%. The rock matrix consists of
relatively loosely packed crystals (largely dolomite).
Very small calcite grains and feldspar flakes are
disseminated within the matrix. Quartz is sometimes
visible. The sample grain density is 2.79 g/cm
3
. The test
was performed on a cylindrical plug, 3.81 cm in
diameter and 3.9 cm long. The plug axis was parallel to
the weakly layered fabric of the rock.
Sample preparation: P-wave and S-wave velocities
were measured in the axial direction by a set of specially
constructed ultrasonic transducers mounted on each end
of the cylindrical plug The piezoelectric material used to
build the transducers were made with resonant
frequencies centered around 1 MHz. Before mounting
the transducers, the plug end faces were ground as
parallel to one another as possible and the sample was
dried under vacuum at temperatures around 70 C.

ARMA 12-582

Effect of CO
2
's varying phase on static and dynamic moduli of a fully
saturated dolostone

Njiekak, G.
Department of Physics, Institute for Geophysical Research, Univ. of Alberta, Edmonton, AB, Canada
Yam, H., Kofman, R.S. and Schmitt, D.R.
Department of Physics, Institute for Geophysical Research, Univ. of Alberta, Edmonton, AB, Canada

Copyright 2012 ARMA, American Rock Mechanics Association
This paper was prepared for presentation at the 46
th
US Rock Mechanics / Geomechanics Symposium held in Chicago, IL, USA, 24-27 June
2012.
This paper was selected for presentation at the symposium by an ARMA Technical Program Committee based on a technical and critical review of
the paper by a minimum of two technical reviewers. The material, as presented, does not necessarily reflect any position of ARMA, its officers, or
members. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of ARMA
is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The
abstract must contain conspicuous acknowledgement of where and by whom the paper was presented.

ABSTRACT: Laboratory measurements of ultrasonic P- and S- waves velocities and strain have been performed on a dolostone
under dry and CO
2
saturated conditions at different stress levels. The current work attempts to assess the effects of the CO
2
's
varying phase state on the static and dynamic bulk moduli. Experiments were conducted under a constant differential pressure of
15 MPa (difference between the confining and the pore pressure), so the seismic effects recorded would be reflective of changes in
the pore fluid properties. Results show that P- and S- waves are sensitive to changes in the pore space and CO
2
phase changes (gas
to liquid and gas to supercritical fluid) are evident from signal variations. Stress-strain curves of the CO
2
saturated sample indicate
an increase of the strain with pore pressure. This is likely attributed to the closure of isolated pores (that means pore not connected
to the inter-crystalline pore space available for fluid flow) with increasing confining stress. The static bulk modulus of the
saturated sample decreases with pore pressure. The dynamic bulk modulus also decreases with pore pressure but recovers when
CO
2
turns from gas to liquid or to supercritical fluid in the rock pore spaces.

For strain measurement, two strain gauges were glued
directly on the sample surface, near the middle of the
plug (Fig. 1 ). One gauge was gl ued parallel to the
pl ug axi s (vertical) and the other perpendicular
(horizontal) to it, using the M-bond 200

catalyst and
M-Bond 200

adhesive. The sample with the affixed
strain gauges was then pushed into a Klearon

tubing and a small window cut into the tubing to allow
access to the soldering area of the strain gauges. The
assembly (sample + strain gages) was then sealed to
prevent the leakage of the hydraulic oil from reaching it
while being in the pressure vessel. The same procedure
was used to prepare the fused quartz standard. In
f act , i n order to correct for pressure effects, fused
quartz, whose pressure dependent properties are well
known, was also measured. The fused quartz and
dolostone strains were measured simultaneously within
the pressure vessel. Then the observed strain values for
the standard (fused quartz) were corrected by using the
calculated strain values obtained from its known
compressibility. In the meantime, these correction
values, which were pressure effects, were used to correct
the sample strain values.

Fig. 1. A sample with the affixed strain gauges.

Velocity measurement: The two prepared jacketed
samples (dolostone and fused quartz) were then placed
in a pressure vessel. Aside from the source/receiver
transducers, the experimental setup for the measurement
of the velocity includes several functional units such as a
pulse generator, a digital oscilloscope (NI USB- 5133
100 MS/s), a cylindrical pressure vessel, a CO
2
tank, a
thermocouple, and an electrical resistance tape (Fig. 2).
The transducer is exited by a fast-rising, 200V square
wave, sent by the pulse generator (Panametrics, model
5800 PR). The generated elastic wave after propagating
through the sample is recorded by a digital oscilloscope
at a sampling interval of 10 nanoseconds. The recorded
waveform is the stack of at least 100 traces to reduce the
noise level in the signal. Arrival times were determined
by picking the first extremum of the recorded signal. Net
traveltimes through the sample were found by correcting
the arrival times for delay (determined by calibration) in
the transducers. Velocities were found as samples
length divided by net traveltime. The velocity
measurements were performed under dry conditions at
room temperature and under CO
2
saturation conditions at
different temperatures (~ 22C and 50C). The dry
measurements were done with the dolostone pore space
under vacuum (pore pressure = 0 MPa) while the
confining pressure could vary from 1 MPa upward. For
the CO
2
saturated measurements, CO
2
was introduced
into the dolostone sample via stainless steel tubing that
connects the pore space of the sample to the CO
2
tank
located outside of the vessel through the vessel lid. The
confining and pore pressure systems are independent of
each other such that different pressure conditions can be
applied in irrespective of each other by using different
pumps located outside of the pressure vessel. During the
CO
2
saturated measurements, the differential pressure
(difference between the confining and the pore pressure)
was kept constant, so seismic effects that are solely
reflective of changes in the pore fluid properties could be
recorded (Fig. 3). The sample was flooded with CO2 for
at least 3 hours before the measurements to allow its full
saturation. The measured ultrasonic velocities were
converted into dynamic bulk moduli by using the
mercury injection porosity (21%) of the dolostone and
the density appropriate to the situation.


Fig. 3. CO2 properties (bulk modulus and density) as a
function of pressure and temperature based on the
thermodynamic model of Span and Wagner (1996). Arrows (i)
and (ii) refer to the conditions under which CO
2
-saturated rock
velocities were measured. Measurements were done under
constant differential pressure of 15 MPa.

Strain measurement: the experimental setup includes the
following components:
(i)- a multichannel Wheatstone bridge circuit as four
separate strain gauges wer e used for each
measurement;
(ii)- a constant voltage power supply exciting the bridge
at V
in
= 2.5 volt;
(iii)- a digital voltmeter measuring the output voltage
V
out ;
(iv)- a digital data logger (USB-5133) was used to
digitally read the voltage across the bridge and the
output of the pressure transducer and to record them in
the computer using LabView software. The later was
coded to initiate the experiment and record the voltmeter
readings in both unstrained and strained conditions.

Since the value of the input voltage (Vin =2.5 volt) and
the gauge factor (usually provided by strain gauge
vendors) were known, the strain of the sample was easily
calculated using the following equation:
( )
r f
r
V G
V
2 1
4
+
=
(1)
where is the strain, V
r
the difference of the ratio of V
out

(output voltage) to V
in
from unstrained to strained
condition, and G
f
the gage factor. In this experiment, V
out

was not recorded under unstrained conditions; V
r
represents the difference of the ratio from the least
strained condition to each higher strained condition.
Thus, the strain values obtained in this study represent
the difference between the minimal strain (recorded
when the confining pressure was equal to 1 MPa and the
pore pressure equal to 0 MPa) and any higher strains
resulting from the increase of the pressure.
There are two sources of error inherent in the observed
strain value indicated by the strain gauge that need to be
considered in the strain calculation. The first is that due
to the applied pressure and the second is temperature on
the gauge elements (Jansen, 1997). Temperature effect
was corrected by subtracting the thermal output value
obtained from the graph of thermal output (supplied by
the vendor) from the strain measurements at the test
temperature. As mentioned above, calibration
measurements on fused quartz were used to correct the
pressure effects. The observed strain values for the
standard (fused quartz) were corrected by using the
calculated strain values obtained from its known
compressibility [e.g., 4, 5]. The correction values were
then used to correct the observed sample strain data.
Static bulk moduli of the studied samples were
determined from the obtained strain values. The
measured strain of a sample can be represented as
follows:
L
L
=
(2)
where L/L is the relative change of the sample length.
Since bulk modulus is related to the volume change, the
static bulk modulus of the sample is given by:
3 ) / ( 3 /
P
L L
P
V V
P
K

=
=
(3)
where P is the change in the pressure and V/V the
resulting change of the sample's volume.
3. RESULTS
The dry experiments were performed at room (~22C)
and at high temperatures (~50C), with the confining
pressure varying from 1 MPa to 35 MPa. Both
pressurization and depressurization were applied to the
sample over this pressure interval. Both P-and S- wave
velocities exhibit an increase with pressure, likely
caused by the progressive closure of compliant pores
(Fig. 4).

Fig. 4. Ultrasonic P- and S- wave velocities on dry dolostone
at room (green symbols) and at high (50 C) temperature (red
symbols). Black arrows show the evolution of the
measurements (pressurization and depressurization cycles).

Figure 5 shows variations of the P- and S-waves
velocities (Vp and Vs) on the dolostone sample as a
function of pore pressure. As pore pressure increases, for
the lower temperature run (T = 22C), CO
2
changes from
a gas phase to a liquid phase in the pore spaces while for
the higher temperature run (T = 50C), CO
2
changes
from a gas phase to a supercritical fluid phase. Vp and
Vs decrease with increasing pore pressure but Vp
slightly recovers after the phase transformation (gas to
liquid or to supercritical fluid). For the lower
temperature run (T = 22C), the elastic wave response to
CO
2
's phase transition is sharp and significant. In
contrast, the elastic wave responses of the higher
temperature run (T = 50C) to CO
2
's phase transition are
gentle and lack any abrupt change.

Fig. 5. P- and S-waves velocities on the CO
2
saturated
dolostone. Results of low and high temperature measurements
are shown (see Fig. 3).

The stress-strain curves of the dry sample at room
(22C) and high (50C) temperatures are shown in figure
6 . Strains from both vertical and horizontal strain
gauges are almost identical, indicating a lack of
anisotropy in the studied sample. Under high confining
pressure (35 MPa), a difference of 8 to 10 % was
recorded between the horizontal and vertical strain
components on the dry sample. Figure 7 shows the
stress-strain curves of the CO
2
saturated sample; a
difference of 11 to 14 % was recorded between the two
strain components. Either on dry or CO
2
saturated
dolostone, the difference between the horizontal and
vertical components slightly increases with the
temperature, likely due to thermal expansion of the
tested sample. The strain magnitude of the sample when
flooded with CO
2
will increase with pore pressure.
The static and dynamic bulk moduli of the dry sample
usually show an increase with confining pressure at
room (22C) and high (50C ) temperatures (Fig. 8),
with the static to dynamic ratio changing from 0.3 to 0.7.
The static bulk modulus used in the static to dynamic
ratio is calculated from the average of the vertical and
horizontal moduli. Under CO
2
saturation conditions,
both moduli decrease with pore pressure but the 'satura-

Fig. 6. Stress-strain relations for the tested sample showing the
vertical strain (blue) and the horizontal strain (pink). Both
were measured during pressurization (solid symbols) and
depressurization (open symbols) under dry conditions at
(a).room temperature and at (b) 50 C. The uncertainty for the
strain values is estimated at 0.1%.

ted' dynamic modulus will recover after the phase
transformation (Fig. 9). The decrease of the 'saturated'
dynamic bulk modulus is more gradual over the gas-
supercritical fluid transition than over the gas-liquid
transition (Fig. 10a). The static modulus seems to behave
in the opposite way, with the decrease being smoother
over the gas-liquid transition. The 'saturated' static to
dynamic ratio is around 0.4 for the differential pressure
(15 MPa) applied during the CO
2
saturated
measurements.

Fig. 7: Stress-strain relations for the CO
2
-saturated sample
showing the vertical (blue) and the horizontal strain (pink). (a)
Experimental results at room temperature. CO
2
was turning
from a gas to liquid in the rock pore space. (b) Experimental
results at 50 C. CO
2
gas transformed into a supercritical fluid
in the pore space. The red and yellow symbols refer to the
strain on the dry sample (pore pressure = 0 MPa) under a
confining stress of 15 MPa .

Fig. 8: Dynamic and static bulk modulus of the dry sample.
The static bulk modulus is calculated from the average of the
vertical and horizontal moduli. Black arrows show the
evolution of measurements (pressurization and
depressurization cycles).

Fig. 9. Relationships between the static and dynamic bulk
moduli. The static bulk modulus (open pink symbols) for the
CO
2
-saturated sample decreases with pore pressure. The
dynamic bulk modulus (open black symbols) also decreases
with pore pressure but recovers when CO
2
turns from gas to
liquid in the rock pore space. The black and pink solid
symbols refer to the moduli of the dry sample (pore pressure =
0 MPa) under a confining stress of 15 MPa.
4. DISCUSSION AND CONCLUSIONS
A good amount of compliant pores might be present in
the sample as indicated by the increase of the velocities
on the dry sample. However, the measured strains of the
dry sample from both vertical and horizontal strain
gauges under a pressure range of 1 to 35 MPa are almost
similar, suggesting a lack of anisotropy in the sample.
Results show that elastic waves are sensitive to changes
in the pore space and CO
2
phase changes (gas to liquid
and gas to supercritical fluid) are evident from signal
variations. The drop of the ultrasonic P- and S- waves
velocities once the dolostone sample is saturated with
CO
2
is likely due to the increase of the density of the
CO
2
supercritical fluid relative to the gas phase.

Fig. 10. Relationships between the dynamic (a) and static (b)
bulk moduli for the low (22 C) and high (50 C) temperature
measurements. The black and pink solid symbols refer to the
moduli of the dry sample (pore pressure = 0 MPa) under a
confining stress of 15 MPa.

Another observation was the increase of the strain
magnitude of the saturated sample with pore pressure.
Considering the higher compressibility of gas compared
to liquid and supercritical fluid, the increase of the pore
pressure (which causes CO
2
to turn from gas to liquid or
supercritical fluid in the pore space) is not expected to
induce further volume decrease of the sample. Possible
physical reason for the observed behavior is the closure
of isolated pores present in the tested sample. The
isolated pores are not considered to be connected to the
inter-crystalline pore space available for fluid flow. It is
worth reminding here that, during the two CO
2
saturated
experiments, the differential pressure was kept constant
at 15 MPa by varying the confining pressure accordingly
to the pore pressure. Assuming that the equal amount of
pore pressure increase cancelled the equal amount of
confining pressure increase, any strain increase at an
advanced stage of the saturation was likely caused by the
closure of isolated pores.
Generally, the static and dynamic bulk moduli of the dry
sample increase with confining pressure. The bulk
moduli of the saturated sample decrease with pore
pressure. The dynamic modulus seems more sensitive to
the CO
2
's varying phase state than the static modulus as
it slightly recovers after the phase transformation.
Overall, the static to dynamic ratio vary between 0.3 to
0.7 under the range of effective pressures (up to 35 MPa)
applied in this study.
ACKNOWLEDGMENTS
2
measurement system was
supported by NSERC and the Canada Research Chair
program. Yam was supported in part by an NSERC CGS
Scholarship and Njiekak and Kofman in part by a
research contract from the Weyburn-Midale CO
2

Monitoring and Storage Project. The technical assistance
of Len Tober and Lucas Duerksen was invaluable.

REFERENCES
1. Walsh, J.B., and W.F. Brace. 1966. Elasticity of rock:
A review of some recent theoretical studies: Rock
Mechanics and Engineering Geology 4: 283297.
2. Jizba, D., and A. Nur. 1990. Static and dynamic moduli
of tight gas sandstones and their relation to formation
properties. 31st Annual Logging Symposium, Society of
Professional Well Log Analysts, Paper BB.
3. King, M.S. 1970. Static and dynamic elastic moduli of
rocks under pressure. 11th U. S. Symposium on Rock
Mechanics, American Rock Mechanics Association,
Proceedings, 329351.
4. Bass, J.D. 1995. Elasticity of minerals, glasses, and
melts. In T.J. Ahrens, ed. Mineral Physics and
Crystallography: A Handbook of Physical Constants:
American Geophysical Union, Washington, D.C., 45-
63.
5. Ohno, I., M. Abe, M. Kimura, Y. Hanayama, H. Oda,
and I. Suzuki. 2000. Elasticity measurement of silica
glass under gas pressure. American Mineralogist 85:
288291.

Laboratory Investigations of Seismic Signatures of CO
2
Saturation
for Geological Sequestration
Gautier Njiekak
Department of Physics, Institute for Geophysical Research, University of Alberta
Helen Yam
Randolf S. Kofman
Douglas R. Schmitt
ABSTRACT: The current study provides an experimental approach that aims at documenting the effects
of different phases of CO
2
(gas, liquid, supercritical fluid) on rock seismic responses. This has
implications on obtaining quantitative estimates of fluid saturations for geological storage. The ultrasonic
pulse transmission technique is the underlying methodology used in this work. Unlike previous rock
physics studies involving CO
2
as pore fluid, care is taken to fully saturate the studied samples and also to
separate pore fluid effects from pore pressure build-up effects during the experiments. This allows to
record seismic effects that are solely reflective of changes in the pore fluid properties. Measurements
were done on a CO
2
saturated Berea sandstone. Laboratory results show P- and S- waveform variations
(velocity, amplitude) in response to CO
2
's varying phase state. Overall, CO
2
phase changes (gas to liquid
and gas to supercritical fluid) are marked by a drop in velocities, with the drop being smaller and less
immediate over the gas-supercritical fluid transition than over the gas-liquid transition.
1. INTRODUCTION
Due to the sensitivity of seismic waves to pore fluid contents, time lapse seismology is regarded as a
promising monitoring method for geological CO
2
sequestration projects and is employed in all industrial
scale projects (Sleipner, Weyburn, In Salah). Therefore, understanding the effect of CO
2
as a pore fluid on
the overall rock seismic response is critical. There exist a very limited number of laboratory studies on
examining the elastic effects of CO
2
on pore fluid seismic responses. All laboratory work to date has used
the ultrasonic pulse transmission technique and demonstrated that the presence of CO
2
in the pore space
will lower seismic velocities due to its higher compressibility compared to other possible in-situ fluids
such as brine or oil. Today, one of the main challenges for seismic monitoring surveys is the quantitative
assessment of the injected CO
2
. This is quite crucial as possible alteration of the reservoir rock frame
and/or changes in the fluid physical properties following the injection of CO
2
could affect the overall
seismic behaviour of the rock. To address this issue, an experimental procedure was set up to inspect the
seismic effects from (i) CO
2
as a lone pore fluid, and (ii) the different phase states of CO
2
. This is
particularly critical as CO
2
can be in gas, liquid, or supercritical phases even at the relatively modest pore
pressures and temperatures in the uppermost kilometer of the earth's crust. Real situations sometimes
differ from the experimental conditions set up in this study as CO
2
may coexist with other in-situ fluids in
the pore space. However, this study provides an end member understanding of the possibilities with CO
2
in the pore space, which is pivotal when quantifying CO
2
using seismic techniques. Moreover, unlike
previous studies involving CO
2
flooding in the laboratory, care was taken to separate seismic effects due
to changes in the pore fluid properties from those induced by pore pressure build-up during such
experiments.
ROCKENG12 - ROCK ENGINEERING FOR NATURAL RESOURCES 529
2. EXPERIMENTAL SETUP AND PROCEDURE
The experiments conducted in this study were based on the ultrasonic pulse transmission technique. The
application of this method allowed the recording of an ultrasonic P- or S- wave pulse after propagation
through a sample of known length. From the final waveforms the transit time of the pulse was deduced,
and with the length of the sample the elastic wave velocity was calculated. The measurements were
carried out in the laboratory of the Experimental Geophysics Group at the University of Alberta.
2.1 Experimental Setup
The experimental setup includes several functional units such as a pulse generator, a digital oscilloscope
(NI USB- 5133 100 MS/s), source/receiver transducers, a cylindrical pressure vessel, fluid tanks, a
thermocouple, and an electrical resistance tape (Fig. 1). The transducer is exited by a fast-rising, 200V
square wave, sent by the pulse generator (Panametrics, model 5800 PR). The generated elastic wave after
propagating through the sample is recorded by a digital oscilloscope at a sampling interval of 10
nanoseconds. The recorded waveform is the stack of at least 500 traces to reduce the noise level in the
signal. The desired pore fluid is introduced into the sample via stainless steel tubing that connects the pore
space of the sample to the pore fluid reservoir located outside of the vessel through the vessel lid. The
confining and pore pressure systems are independent of each other such that different pressure conditions
can be applied in irrespective of each other by using different pumps located outside of the pressure
vessel. Both the confining pressure (pressure vessel) and the pore pressure systems are capable of
reaching a maximum pressure of 70 MPa from the pumps. Higher confining pressures, up to 200 MPa,
could also be achieved by using a separate air pump connected to the pressure vessel.
Figure 1 - The experimental setup for the ultrasonic measurement
2.2 Experimental Procedure
Ultrasonic measurements were made on samples under a variety of dry and saturated conditions. The
dry measurements were done with the pore space under vacuum (pore pressure =0 MPa) while the
confining pressure could vary from 1 MPa upward.
530 21ST CANADIAN ROCK MECHANICS SYMPOSIUM
For the CO
2
saturated measurements, measurements were done under different conditions of temperature
and pore pressure in order to inspect the pore fluid effects on the seismic waves velocities. The key here
was to make sure that any variation in the waveforms was due to change in the pore fluid physical
properties. Thus, for each sample, a constant differential pressure was maintained by varying the
confining pressure accordingly to the pore pressure. Assuming that the equal amount of pore pressure
increase cancelled the equal amount of confining pressure increase, any waveform variation observed
was solely caused by effects of the fluid(s) present in the pore space.
Running experiments under constant differential pressure could also be used to test the full saturation of
the studied sample. Generally, it is difficult to know if a rock is fully saturated, and most authors simply
make this assumption after taking precautions during sample preparation. Rarely, however, is any
additional evidence given to support this contention. The same problem occurred in this study and in
order to satisfy this assumption a quality control procedure was used. Prior to CO
2
saturated
measurements, the sample was subjected to constant differential pressure measurement under inert
nitrogen gas saturation. As nitrogen gas, unlike CO
2
, does not undergo a phase transformation under the
range of pressure and temperature conditions used in this study, no substantial change in the P- and S-
wave velocities with constant effective pressure was taken to indicate full saturation of the sample. While
this latter behaviour may also be indicative of other processes than partial water vapour-nitrogen
saturation (e.g., chemical reactions at the pore surfaces), it does lend confidence to selecting the set of
-
2.3. Sample Preparation
For the measurement, cylindrically shaped plugs of up to 3.81cm in diameter and of up to ~6 cm in
length were used. The plug end faces were ground to parallel (within 0.02 mm) in order to enhance the
signal transmission and to avoid errors in the velocity measurements as much as possible. Then the
samples were dried under vacuum at 70C temperature for 48 hours and kept in a desiccator jar
afterwards until in use. The sample to be measured was pushed into an impermeable Klearon
TM
tube. An
ultrasonic transducer was placed on each end of the sample and the two transducers were aligned so that
S-wave ceramic were properly polarized (See Bakhorji, 2009 for details on the building of the
source/receiver transducers). To enhance the sealing mechanism and prevent the leakage of the hydraulic
oil from reaching the sample while being in the pressure vessel, steel hose clamps were used so that they
could be tightened around the Klearon
TM
tube just above the aluminum part of the transducers. The
jacketed assembly of the sample was then placed inside the confining pressure vessel.
3. CASE STUDY: BEREA SANDSTONE
3.1. Sample Description
Berea sandstone is predominantly composed of quartz and minor traces of clay minerals (Fig. 2). The
porosity and permeability of Berea sandstone are known to be relatively high, making it a good candidate
for petrophysical testing (e.g., Shankland and J ohnson, 1993; Hart and Wang, 1995; Zhan et al, 2010).
The studied specimen has a porosity of 19%, with an air permeability of about 238 mD. The main pore
throat size is around 11m (Yam, 2011).
a) b)
Figure 2 - (a) Photograph of the Berea sandstone's plug and (b) a scanning electron microscope
image of the sample
a)
P-waves
3800
3850
3900
3950
4000
4050
4100
4150
0 10 20 30 40 50
Confining Pressure (MPa)
Pressure Up
Pressure Down
b)
S-waves
2350
2400
2450
2500
2550
2600
0 10 20 30 40 50
Pressure Up
Pressure Down
Figure 3 - Dry P-and S-wave velocities with their respective errors as a function of confining
pressure for the Berea sandstone
Figure 4 - Phase diagram of CO
2
. C =critical point. Arrows (i) and (ii) refer to the conditions
under which CO
2
-saturated rock velocities were measured
a) b)
c) d)
Figure 5 - Normalized P- and S-wave waveforms as a function of pore pressures collected during
the constant temperature runs at T =23C (a and b) and T =55C (c and d). A constant
differential pressure of 15 MPa was maintained throughout each measurement run
a) b)
Figure 6 - CO
2
saturated P-and S-wave velocities of the Berea sandstone for the two constant
temperature runs at T =23C and T =55C while under a constant differential pressure (Pd) of
15 MPa. Error bars are shown
3.2. Ultrasonic Measurements
Measurements were made on the sandstone sample under dry and saturated conditions. The dry
measurements were done with the pore space under vacuum (pore pressure =0 MPa) while the confining
pressure varied from 5 MPa to 40 MPa. Both pressurization and depressurization were applied to the
sample over this pressure interval. Both P-and S- wave velocities exhibit a nonlinear increase with
pressure (Fig. 3). The wave velocities increase rapidly at low pressures that taper off with higher
pressures. This behavioural increase is caused by the progressive closure of compliant pores where most
of the compliant pores will close at low pressures, therefore causing a rapid velocity increase.
For the CO
2
saturated measurements, two different constant temperature runs were conducted while pore
pressure varied from 2 MPa to 25 MPa. The constant temperature runs were done at 23 C and 55 C (See
arrows (i) and (ii) in figure 4). For each measurement run, a constant differential pressure of 15 MPa was
maintained. As pore pressure increases, for the lower temperature runs (23C), CO
2
changed from a gas
phase to a liquid phase in the pore spaces while for the higher temperature runs (55C), CO
2
changed
from a gas phase to the supercritical fluid phase. The results of the measurements show waveform
variations (velocity, amplitude) in response to CO
2
's varying phase state. For the lower temperature run (T
=23C), the elastic wave response to CO
2
's phase transition is sharp and significant. In contrast, the
elastic wave responses of the higher temperature run (T =55C) to CO
2
's phase transition are gentle and
lack any abrupt change (Figs 5 and 6). Overall, CO
2
phase changes (gas to liquid and gas to supercritical
fluid) are marked by a drop in P- and S- wave velocities of up to 3.9 and 4.4%, respectively, over the
entire investigated pore pressure interval (=25 MPa). This is likely due to the increased density of the
liquid or the supercritical fluid relative to the gas (Fig. 6). The P-wave velocity reduction (3.6-3.9%)
recorded in this study is lower than the P-wave velocity decrease (7.5-14.5%) observed by Lei and Xue
(2009) when injecting different phases of CO
2
in water-saturated sandstone samples. However, the
differences between these values should be interpreted with caution as experimental conditions (e.g.,
investigated pore pressure interval, nature of the pore fluids) are not similar to the ones set up in the
current study.
4. CONCLUSIONS
An experimental procedure has been set up to inspect the variation of P- and S- waves velocities as a
function of CO
2
saturation. The work focuses on inspecting the effects of different CO
2
phases (gas,
liquid and supercritical fluid) on the overall rock seismic behaviour. Results show that elastic waves are
sensitive to changes in the pore space and CO
2
phase change is evident from signal variations. CO
2
phase
changes (gas to liquid and gas to supercritical fluid) are marked by a drop in velocities of up to 4.4 %.
The abruptness in the velocity change differs significantly between the gas-liquid and gas-supercritical
fluid transitions; and the gradual variations seen for the latter suggest that this will be difficult to detect
using seismic reflection techniques.
5. ACKNOWLEDGEMENTS
2
measurement system was supported by NSERC and the Canada Research
Chair program. Yam was supported in part by an NSERC CGS Scholarship and Njiekak and Kofman in
part by a research contract from the Weyburn-Midale CO
2
Monitoring and Storage Project. The technical
assistance of Len Tober and Lucas Duerksen was invaluable.
6. REFERENCES
BAKHORJ I, A., 2009. Laboratory measurements of static and dynamic elastic properties in carbonate,
Thesis Ph.D., Univ. of Alberta, Edmonton, 333 p.
HART, D. J ., and WANG, H. F., 1995. Laboratory measurements of a complete set of poroelastic moduli
for Berea sandstone and Indiana limestone, Journal of Geophysical Research, 17, 741-751.
LEI, X., and XUE, Z., 2009. Ultrasonic velocity and attenuation during CO2 injection into water-
saturated porous sandstones: Measurements using difference seismic tomography, Physics of the
Earth and Planetary Interiors, 176, 224-234.
SHANKLAND, T. J., and J OHNSON, P. A., 1993. Elastic wave attenuation and velocity of Berea
sandstone measured in the frequency domain: Geophysical Research Letters, 20, 391-394.
YAM, H., 2011. CO
2
Rock Physics: A laboratory study. MSc. Thesis, Univ. of Alberta, Edmonton, 286 p.
ZHAN, X., SCHWARTZ, L. M., TOKSOZ, M. N., SMITH, W. C., and MORGAN, F. D., 2010. Pore-
scale modeling of electrical and fluid transport in Berea sandstone: Geophysics, 75, F135.
Rock Physics Facilities and Research in the
Experimental Geophysics Group at the
University of Alberta
Douglas R. Schmitt
Sanna Aqil, Aiman Bakhoriji
1
, Youcef Bouzidi
2
, Mizan H. Chowdhury, Randolph Kofman,
Jaime Melendez, Gautier Njiekak, Miryam Ortiz-Osornio, Heather Schijns, Brendan Snow,
Helen Yam
3
Institute for Geophysical Research, Department of Physics, University of Alberta, Edmonton, Alberta, Canada
Introduction
The Experimental Geophysics Group (EGG) carries out a
wide variety of field and laboratory studies. These studies
focus on understanding what controls the physical properties
of rocks and how these relate to geophysical surface observa-
tions. Here, we hope to give some overview of the laboratory
facilities that EGG has developed over the years for charac-
terizing porous rock samples and for making ultrasonic P and
S wave measurements for both conventional rock property
evaluation and for more fundamental studies of wave propa-
gation in porous and anisotropic media. This leads to a
summary of some of the recent laboratory results obtained.
The motivations for much of this research vary and often it is
in support of our research activities in scientific drilling,
geomechanics, and stress determination. One portion of our
work focuses on conventional laboratory rock property meas-
urements, in these we gain knowledge of the properties of the
rock under conditions of pressure, temperature, and satura-
tion expected in situ. The results in such cases provide
constraints on geophysical observations, such as providing
the difficult-to-obtain dry or drained bulk modulus necessary
for Gassmann fluid substitution calculations. The second
portion concentrates on measurements that allow us to better
understand how waves propagate through complex mate-
rials. This work is of a more fundamental nature and it seeks
to better constrain our interpretations of geophysical data.
Experimental Facilities
Measurements under In Situ Conditions: Rock is a notoriously
complex material. One aspect of this is that the P and S wave
speeds are highly dependent on in situ conditions, and in
particular, the velocities of nearly all rocks depend on the
effective pressure they are subject to. As such, making a
measurement of a rock property at atmospheric pressure will
mislead an interpretation of field data if it is assumed this is
representative of the same property at depth in the earth.
Consequently, the measurements must usually be made on
rocks subject to confining and pore pressures. This signifi-
cantly complicates carrying out such tests as now one must
adequately seal a rock sample to protect it from the fluid
transmitting confining pressure. Further, one must be able to
send and receive signals from the sensors inside the pressure
vessel to the data acquisition systems outside.
To carry out such experiments, EGG has constructed two
main pressure vessel systems here informally called Blue and
Red. In part for reasons of safety, these vessels are capable of
reaching relatively high confining pressures of 300 MPa
(~45,000 psi or about 12km depth) and 200 MPa (~30,000 psi
about 8 km depth), respectively. Red (Fig. 1) is most often
used on smaller diameter (~2.54 cm) cylindrical plug samples
in simpler cases that do not require the sample to contain
pore fluids. The high confining pressure that can be achieved
is much greater than necessary for most sedimentary rock
studies (< 50 MPa), but has been used on studies of the
40 CSEG RECORDER January 2012
Continued on Page 41
C
o
o
r
d
i
n
a
t
e
d

b
y

S
a
t
i
n
d
e
r

C
h
o
p
r
a
A
R
T
I
C
L
E
1
Now at Saudi Aramco, Dharan
2
Now at the Petroleum Institute, Abu Dhabi
3
Now at Imperial Oil, Calgary
Figure 1. Red pressure vessel with capacity to 300 MPa total pressure with
undergraduate student Brendan Snow making P and S wave measurements on
shocked materials.
January 2012 CSEG RECORDER 41
anisotropy of metamorphic and igneous rock whose elastic prop-
erties are strongly controlled by microcracks. Such high pres-
sures are necessary to close most of the microcracks in order to
obtain values representative of the intrinsic pore-free material.
Blue (Fig. 2) is substantially more complicated as it allows for
independent controls on pore fluid pressure, confining pressure,
and temperature. This machine is currently extremely busy in
rock physics studies related to CO
2
sequestration and recently
completed an extensive study of carbonates from the Arab D
formation [Bakhorji, 2009]. The vessel can accommodate samples
with diameters nearly up to 10 cm; and this added real estate
allows for additional experimental flexibility. Pore pressures are
varied and maintained by a Quizix dual-cylinder pumping
system. Temperatures up to ~100C are varied using a ribbon
heater wrapped around the pressure vessel. The entire control
and acquisition system is operated by a specially written
Labview program.
Our current data acquisition systems are based on racked modules
that include high resolution loggers (to continuously record pres-
sures, temperatures, and strain) as well as a 12-bit high speed
digital oscilloscope for acquiring the ultrasonic waveforms (typi-
cally sampling at rates ~10 ns/sample). This information is all
displayed on a large screen from which the experiment can be
monitored (Fig. 3), all of the acquisition takes place using the same
specialized Labview program that control the sample conditions.
Ultrasonic Transducers: EGG has developed considerable experience
in the construction of differing kinds of ultrasonic transducers from
piezo-electric ceramics. This allows for a great deal of flexibility in
the design of measurements, which would not be possible using
commercially available sensors. Our current generation of time-of-
flight transducers allows for the simultaneous recording of P and
S waveforms (see Fig. 4), and we are currently working on a further
stacked system that will have a second S wave ceramic in order to
simplify measurements of elastic wave anisotropy (where two split
shear waves need to be measured[Njiekak et al., 2011]. We are also
working on hardening our transducers so that they can survive
longer the severe conditions of pressure they experience while in
the pressure vessel.
Technical concerns of sealing a sample for pressure and routing
information into and out of the pressure vessel aside, an ultrasonic
velocity measurement is otherwise conceptually a simple affair. In
most applications, a cylindrical rock sample is first cored and its
ends are ground to be as parallel and flat as possible. The cylinder
is then pushed into a flexible tube that is necessary to exclude the
confining pressure oil from the rocks pore space. Two dual P-S
wave transducers are then mounted on either end of the cylinder
taking care to ensure the S-wave particle polarizations are aligned
together. In this configuration, one transducer assembly transmits
while the other receives. The transmitter is activated, crudely, by
applying a fast-rising high voltage step to the piezo-electric
ceramic producing a mechanical disturbance. This disturbance
propagates as an elastic wave through the transducer assembly
Article Contd
Rock Physics Facilities and Research
Continued from Page 40
Figure 2. Blue pressure vessel with confining pressure capacity to 200 MPa and
controlled pore pressure capacity to 70 MPa with the Quizix and 150 MPa manu-
ally. Dr. Gautier Njiekak at the controls.
Figure 3. Mizan H. Chowdhury at the controls of Blue during a CO
2
measurement.
buffers and the sample to be detected by the receiver that trans-
duces it back to a voltage that is recorded by the high speed digi-
tizer to produce the waveform. The wave velocity is obtained by
dividing the sample length (sometimes corrected using strain) by
the wave transit time (corrected for the transducer delay) as
picked from the observed waveform.
Reflectivity Experiments: Our ultrasonic goniometer is a unique
system for carrying out fundamental studies of wave reflectivity
(Fig. 5). This essentially consists of a large aperture transmitter
and a small near-point receiver both aligned to the reflection
point on the surface of a block of the material under study. This
combination differs from all other similar reflectivity systems
that employ two identical transducers. The advantage of our
system is that the large transmitter mimics to a degree (although
not perfectly) a plane wave and it also transmits a large energy
acoustic pulse. In other systems, the effects of transducer diffrac-
tion must be considered for both the transmitter and the receiver
and the subsequent corrections required add additional error to
the measurements. The goniometer can also be configured for
material transmission tests.
Dielectric Properties: A collaboration with the Potash Corporation
of Saskatchewan has allowed us access to an advanced instru-
ment (Agilent E4991A) radio frequency (10 MHz to 3 GHz)
impedance/material analyzer (Fig. 6a). This instrument allows
one to measure the dielectric properties of materials over the
frequencies typically used in ground penetrating radar (GPR).
Briefly, for low loss materials, the relative dielectric permittivity
krelates to the speed of an electromagnetic wave n through
n =
c
k
Knowledge of k allows calculation of the reflection co-efficient R
from the interface between two materials of differing k
1
and k
2
using the Fresnel equations that at normal incidence reduce to
R =
k
1
k
2
k
1
+ k
2
This is being used to study GPR wave speed dispersion and
attenuation in the potash and surrounding formations, this is
Article Contd
Figure 4. An example of recorded P and S waveforms through a carbonate obtained
as a function of confining pressure. Note the decrease in the travel time with
increasing pressure as it cycles first up then down.
Figure 5. Description of the reflectivity measurement system. a) photograph of the
experimental set up that shows the large piezoelectric source and the small receiver
at equal angles along the vertical goniometer. The horizontal goniometer allows for
the sample to be rotated to allow for studies of anisotropic reflectivity. b) definitions
of the geometry of the measurement system inclues the angles of incidence q, of
material principal symmetry axis horizontal rotation f, and material symmetry
axis tilt y.
primarily done as an aid to mine development and safety as well
as providing a better understanding of clay containing materials
[Aqil and Schmitt, 2009].
Material Characterization: In many rock physics studies in the
literature, the materials under study are often poorly character-
ized; and this limits the ultimate utility of the work. As such,
obtaining as much information about the materials composition
and structure has been a priority in most of our projects, and a
number of differing techniques are employed to obtain such
information. The Rock Physics Laboratory operates the
following pieces of equipment:
Helium Pyncnometer: This device uses Boyles Law for the
expansion of He gas to accurately determine the volume of the
solid (mineral) portion V
m
of a given rock sample or a powder.
Since the corresponding mass of the sample is easily meas-
ured, the pyncnometer (Quantachrome Multipyncnometer)
(Fig. 6a) most directly provides a measure of the samples
grain density r
g
; and as such it can provide constraints on the
mineralogical composition. Further, if knowledge of the bulk
density r
b
or the envelope volume V
e
is available then this
information can be used to calculate the porosity via:
f =
V
e
V
m
V
e
Mercury Injection Porosimetry: This is a substantially more
sophisticated device (Micromeritics AutoPore IV) Fig. 6b
that measures the samples porosity but also gives a wealth
of additional data on the structure of the porosity, and most
particularly some idea of the distribution of the pore dimen-
sions (or more strictly the pore throat dimensions).
Petroleum engineers refer to this data as capillary pressure
curves. Although the operation of this instrument is beyond
the space necessary here, the basic concept relies on the
nonwetting properties of liquid Hg. Unlike most other
fluids such as water that will be sucked by capillary forces
into the rock, pressure must be applied to Hg to force it to
enter the rocks pores. The higher the applied pressure the
smaller the pore throat the Hg will breach according; and
therefore carefully monitoring the Hg pressure and the
volume injected allows for constraint on the proportions of
the porosity at different scales. An example of one
porosimetry curve is shown in Fig. 7d.
3D Scanner: Porosity f is defined simply as the ratio of the pore
volume V
f
to the envelope volume V
e
. While this is a rather
trivial relation, actually determining V
e
in reality can be quite
challenging particularly if the sample dimensions are irregular.
Aside from its utility in making images of the samples Aside
from its utility in making images of the samples.
The laboratory also makes use of additional geological characteri-
zation tools available elsewhere on campus for making geological
thin sections, carrying out X-ray diffraction mineral identification,
scanning electron microscopy (SEM), and 3D micro-CT x-ray
Article Contd
Figure 6. Various laboratory instruments. a) 10 MHz to 3 GHz dielectric imped-
ance measurement system, b) He pyncnometer, c) Hg porosimeter, d) 3D image
scanner.
Figure 7. Examples images as part of the characterization of a vuggy carbonate
core sample a) small cube for micro-CT analysis, b) thin section,c) scanning elec-
tron microscope image, , d) Hg injection porosimetry curves, e) portion of the
micro-CT (micron resolution x-ray tomography) scan showing the output
greyscale image (top), the threshold solid portion (left corner) and the threshold
pore section (right corner).
imaging (see Fig. 6). If necessary, full X-ray CT scanning of cores
can be carried out at the Edmonton facilities of Alberta Innovates
Technology Futures. Full chemical analysis using XRF tech-
niques are usually sent out to commercial providers.
Some Current Measurement Programs
CO
2
Sequestration: As noted above, Blue has been configured to
make measurements on CO
2
saturated rock samples. These are
particularly interesting as the physical properties of CO
2
itself
varies substantially over the range of relatively modest pressures
and temperatures that are expected to be encountered during
sequestration. Indeed, depending on the temperature gradients
in an area, one might expect to encounter CO
2
as either a gas, a
liquid, or a supercritical fluid within the uppermost km of the
earths crust. The density and bulk modulus (and hence sound
speed) of CO
2
will consequently vary dramatically and this influ-
ences the seismic properties of the material [Yam and Schmitt,
2011]. An example (Fig. 8) of the variation in P and S wave
velocity with pore pressure as measured at two temperatures of
28C and 40C, where CO
2
gas transforms to either a liquid or a
supercritical fluid, respectively, illustrates the differences that
can be observed. It is interesting to note that the gas-liquid phase
transformation is abrupt but that from gas-supercritical fluid is
much more gradual; and the latter phase transformation will
likely be difficult to detect with seismic monitoring.
Anisotropy: In sedimentary basins seismic anisotropy is produced
at many scales be it from preferred mineralogical alignment or
layering. While anisotropy appears to be nearly ubiquitous, there
is actually very little hard data with regards to anisotropy in the
earth or in core samples. Core scale anisotropy is interesting as it
might provide useful attributes as to the rock structure and
composition. This scale is also important as it can be considered
as it must also affect the overall anisotropy of a formation. Three
different studies of anisotropy are currently underway:
Shales: In part, the vigorous interest in unconventional
resources is stimulating our work in the measurement of the
anisotropy of shales. Building on previous measurements
carried [Hemsing and Schmitt, 2006; Wong et al., 2008] we are
measuring the elastic anisotropy of a series of shales from
the Western Canadian Sedimentary Basin. Under the
assumption that the material is transversely isotropic, three
plugs are cut at directions perpendicular, parallel, and
oblique to the materials layering (for sediments). The P-
wave and the two orthogonal S-wave velocities are then
measured as a function of confining pressure. An example of
a relatively anisotropic Mannville shale from southern
Alberta (Fig.9a) displays P- and S-wave anisotropies of d ~
0.08 and g = 0.12, respectively, with e = 0.1750.02.
Metamorphic Schists: This work is motivated by the need to
understand the off-axis anisotropy of a thick schist section as
measured in a deep borehole [Schijns, 2009; Schijns et al.,
2011] drilled into the Fennoscandian shield near the
Outokumpu, Finland mining district. In this configuration,
small P- and S-wave piezoelectric ceramics [after Mah and
Schmitt, 2001] are mounted directly and at strategic posi-
tions relative to the foliation plane and lineation direction on
the surface of cylinders of the schist.
Reflectivity from an anisotropic material: The use of AVO
methods to look for fracture orientations has taken off in
recent years. However, these method almost always assume
a set of vertically oriented parallel fractures that results in the
very specific geometry with a horizontal transversely
isotropic (HTI) symmetry. In the real earth, however, fracture
and joint systems are not so constrained by theoretical
simplifications; and consequently it is interesting to examine
the amplitude versus angle of incidence and azimuth (AVA-
Az) response of such formations. Fig. 7 shows a set of reflec-
tivity curves obtained from a nearly TI medium tilted at 45
with respect to the horizontal [Ortiz-Osornio and Schmitt,
2011] It is interesting to note that the responses show the
most variation only past the P-wave critical angle.
Article Contd
Figure 8. Variations in the ultrasonic P- and S-wave velocities with pore pressure
at temperatures of 28C and 40C. All these measurements were taken at constant
differential pressure of 15 MPa.
Future Directions
With regards to the materials studied, a good deal of the work in
the immediate future will likely focus on continued studies of
CO
2
saturation on the seismic properties of rocks. We have
completed a large study of carbonates associated with the
Weyburn CO
2
project under various temperatures and pressures.
We are currently working on sandstone materials for Carbon
Management Canada with the eventual goal of developing
seismic models for purposes of time lapse monitoring. We will be
broadening this work to better understand the physical proper-
ties of CO
2
saturated brines themselves and carry this into the
rock samples. The system and experimental protocols we have
set up to carry out the CO
2
studies is somewhat unique and has
attracted interest from additional collaborators; so CO
2
studies
will likely continue for some time.
Beginning with Bakhorji [2009] leading from the much earlier
work of Schmitt and Li [1995] we have also begun including
pseudo-static strain measurements simultaneously to the
dynamic ultrasonic velocity measurements. The strains versus
pressure plots yield the static bulk moduli which is useful
geomechanical information. This will further be supplemented
by strength determinations that are necessary in the design of
fracture operations. [Melendez and Schmitt, 2011]
Of most interest, however, is our work towards the construction of
an instrument to measure the moduli of rocks at seismic frequen-
cies. Currently, we are working in collaboration with the Australian
National University (Canberra) to study the effects of fluid satura-
tion in cracked rocks. High frequency measurements are made in
the Rock Physics Laboratory while low-frequency forced-oscilla-
tion tests are carried out at ANU on the same quartzite rock
samples. We are currently designing a new system to measure the
seismic band bulk moduli of the material directly. R
Acknowledgements
The laboratory was put together using grants from a variety of
sources but most crucially by NSERC through equipment grants
and, in earlier times, the Alberta Oil Sands Technology and
Research Authority. The most recent support of the Potash
Corporation of Saskatchewan for the purchase of the dielectric
material analyzer has been most appreciated. Similarly, most of
our high pressure systems have been build from scratch as we
were fortunate to be able to scavenge vessels, expensive pressure
gauges, and costly high pressure fittings from a variety the labs of
retired professors in condensed matter physics. In such a case,
access to skilled technicians and machinists has been key and none
of this work would have been possible without the past assistance
of L. Tober, J. Haverstock, G. Lachat and current support from L.
Duerksen, P. Zimmerman, T. Paget, and T. Walford.
References
Aqil, S., and D. R. Schmitt (2009), Dielectric permittivity of natural salt rocks contami-
nated with clay, paper presented at Soc. Exploration Geophysicists 79th Annual Mtg.,
Houston, Oct. 25-30.
Bakhorji, A. (2009), Laboratory measurements of static and dynamic elastic properties in
carbonate, 333 pp, University of Alberta, Edmonton.
Hemsing, D. B. (2007), Laboratory determination of seismic anisotropy in sedimentary rock
from the Western Canadian Sedimentary Basin, 208 pp, University of Alberta, Edmonton.
Hemsing, D. B., and D. R. Schmitt (2006), Laboratory determination of elastic anisotropy
of shales from Alberta, paper presented at SEG 76th Annual Meeting, New Orleans.
Mah, M., and D. R. Schmitt (2001), Experimental determination of the elastic coefficients
of an orthorhombic material, Geophysics, 66(4), 1217-1225.
Melendez, J., and D. R. Schmitt (2011), Experimental measurements of seismic
anisotropy in rocks, paper presented at 1st Int. Workshop on Rock Physics, Golden,
August 7-12.
Article Contd
Figure 9. a) Ultrasonic P- and S-wave anisotropies observed from a sample of
Mannville shale [see Hemsing, 2007; Hemsing and Schmitt, 2006]. b) S-wave
velocities observed at different angles through a metamorphic schist. c) Variation in
the reflectivity with angle of incidence and azimuth from a nearly TI medium with
the symmetry axis tilted at 45 to the horizontal.
Njiekak, G., R. Kofman, and D. R. Schmitt (2011), Simultaneous ultrasonic measure-
ment of compressional and two directional shear wave velocities with a single pair of trans-
ducers, paper presented at Recovery 2011 CSPG CSEG CWLS Convention,
Calgary, May 9-14.
Ortiz-Osornio, M., and D. R. Schmitt (2011), Physical Modeling of the Reflectivity and
Transmissivity of Anisotropic Materials, Abstract DO27, paper presented at 73rd
EAGE Conference & Exhibition Vienna, June 23-26.
Schijns, H. (2009), Seismic Anisotropy Measurements of a Fractured Crystalline
Formation in Outokumpu, Finland Using High Resolution VSP, 124 pp, University of
Alberta, Edmonton.
Schijns, H., D. R. Schmitt, P. J. Heikkinen, and I. T. Kukkonen (2011), Seismic
Anisotropy Observations in the Crystalline Upper Crust: Observations and Modeling from
the Outokumpu Scientific Borehole, Finland, Geophysical Journal International, in
press, 44 ms.
Schmitt, D. R., and Y. Y. Li (1995), A High-Pressure Technique for Determining the
Microcrack Porosities of Damaged Brittle Materials, Can. J. Phys., 73(5-6), 330-337.
Wong, R. C. K., D. R. Schmitt, D. Collis, and R. Gautam (2008), Inherent transversely
isotropic elastic parameters of over-consolidated shale measured by ultrasonic waves and
their comparison with static and acoustic in situ log measurements, Journal of
Geophysics and Engineering, 5(1), 103-117.
Yam, H., and D. R. Schmitt (2011), CO
2
rock physics: a laboratory study, Canadian Well
Logging Society Insite Magazine, 30, 13-16.
Additional resource:
1
Now at Saudi Aramco, Dharan
2
Now at the Petroleum Institute, Abu Dhabi
3
Now at Imperial Oil, Calgary
Article Contd
Douglas R. Schmitt is a Tier 1 Canada Research Chair in Rock Physics (since 2002) and a Professor of
Geophysics and Physics at the University of Alberta. His research group carries out a unique blend of field and
laboratory experiments that focus primarily on understanding of rock physics and mechanics and how they
influence geophysical observations. He has and is playing active roles in international scientific drillling proj-
ects on 5 continents. This research focus carries over to his teaching of an annual geophysics field school and
graduate and undergraduate level courses in rock physics. He is a graduate of the University of Lethbridge
(BSc in Physics, 1980) and the Seismological Laboratory of the California Institute of Technology (PhD, (1981-
87)). Prior to joining the University of Alberta in 1989, he carried out postdoctoral research in the geomechanics
of hydraulic fracturing in the Dept. of Geophysics at Stanford University (1987-88). He is a A. vonHumboldt Research Fellow
(Geophysikalische Insitute, Uni. Karlsruhe, 1996-97) and a visiting scientist in the Research School of Earth Sciences at the
Australian National University, Canberra (2005). He worked as an exploration geophysicist at Texaco Canada Resources Ltd.,
Calgary (1980-81). He has received the A. vonHumboldt fellowship, the University of Alberta Faculty of Science Research Award
(1999), the Distinguished Alumnus of the Year for the University of Lethbridge (2008), and the Canadian Society of Exploration
Geoiphysicists Meritorious Service Award (2009). He currently servies as vice-chair of the IODP Science and Technology Panel
and as a member of the ICDP Scientific Advisory Group. He is the only registered Professional Geophysicist at the Unviversity
of Alberta and serves (since 2005) on the Board of Examiners for the Association of Professional Engineers, Geologists, and
Geophysicists, of Alberta.
CWLS Magazine
June 2011 Issue 1 Volume 30
L
O
G
G
I
N
G
SO
C
I
E
T
Y

Rt
Ro Rw F
Sw
C
A
N
A
D
I AN W
E
L
L

13 CO
2
Rock Physics:
A Laboratory Study
18 Drill Cuttings and Rock Properties for the
Monteith Formation, Nikanassin Group
CANADIAN WELL LOGGING SOCIETY
11
L
O
G
G
I NG SOC
I E
T
Y

Rt
Ro Rw F
Sw
C
A
N
A
DI AN W
E
L
L

13
L
O
G
G
I NG SOC
I E
T
Y

Rt
Ro Rw F
Sw
C
A
N
A
DI AN W
E
L
L

CO
2
Rock Physics: A Laboratory Study
Helen Yam and Douglas R. Schmitt
Summary
CO
2
in the upper sections of a sedimentary basin can exist in a
gas, a liquid or a supercritical fluid phase state. Pore fluids can
have a substantial effect on the behavior of seismic waves where
seismic methods are widely used for monitoring geological
CO
2
sequestration projects. To observe the seismic effects as-
sociated with the different phase states of CO
2
, ultrasonic pulse
transmission experiments were conducted on a Berea sandstone
subjected to full CO
2
saturation over a range of pressures and
temperatures. Over CO
2
phase transitions the acquired wave-
forms show significant changes in the arrival time and strength
variations.
Introduction
Geological CO
2
sequestration is regarded as an effective
method in mitigating the amount of CO
2
released into our at-
mosphere (Benson and Cole, 2008; Chadwick et al, 2009).
Depleted oil and gas reservoirs, saline aquifers, and deep un-
mineable coal beds together have an estimated storage capacity
up to 11,000 gigatons of carbon (IPCC, 2005). Concerns about
the safety and the integrity of underground CO
2
containment
have enlisted various methodologies for monitoring the behav-
iour of the injected CO
2
, where seismic methods are highly
favourable. The presence of a pore fluid has been shown to have
a strong impact on the elastic properties of a porous medium
(Batzle and Wang, 1992). Seismic methods, particularly time-
lapse seismic, are employed in all the current large scale se-
questration projects (Sleipner, Weyburn, and In Salah) and in
many smaller scale or pilot projects (Frio, SACROC,
Nagaoka). Given the possible pressure and temperature condi-
tions at the upper depths of a sedimentary basin, CO
2
can be
in a gas, liquid, or supercritical fluid phase state, where each will
yield a different effect on seismic waves from the consequent
variations in density and fluid compressibility. Therefore to ef-
fectively utilize seismic methods for monitoring sequestration
projects, the thorough understanding of the elastic properties of
CO
2
is essential. Ultrasonic pulse transmission experiments
have been conducted on a fully CO
2
saturated Berea sandstone
for a variety of temperatures and pressures that are reflective of
subsurface conditions for which CO
2
can be a gas, liquid and
supercritical fluid. Although realistically the pore space may
also contain brine or hydrocarbons, the purpose of full CO
2
saturation is to provide an end member understanding of what
extreme behaviours can be expected with CO
2
in the pore
space.
CO
2
Phase States and Physical Properties
Based on the empirical thermodynamic model of Span and
Wagner (1996), the phase diagrams of CO
2
bulk modulus and
density over a temperature range of 0-60 C and for pressures
up to 30 MPa are shown in figure 1a, and 1b. In general within
a fluid, the bulk modulus and density are the physical proper-
ties that will influence wave propagation; bulk modulus has a
direct proportional effect while density has an inverse propor-
tional effect on wave velocities. The accepted critical point of
CO
2
is at 30.9782 C and 7.3773 MPa (Span and Wagner,
1996), and is marked by a red dot in the phase diagrams. Below
the critical temperature and pressure, CO
2
can be either a gas
or liquid with respect to the possible subsurface conditions. The
vapour-liquid boundary is clearly discernible by an abrupt
change in all the physical properties. As the critical point is ap-
proached the properties of the gas and liquid phases converge
and no distinction is apparent thereafter, yielding the supercrit-
ical fluid phase. A unique characteristic of supercritical fluid is
they have physical behaviors of both a gas and a liquid. As a
consequence of this characteristic, any gas or liquid phase tran-
sitions gradually into the supercritical phase, in contrast to the
gas-liquid transition. Another unique feature of supercritical
fluid is in the region close to the critical point, large density
variations occur over a small range of pressure and temperature.
Figure 1: Phase diagrams of CO
2
s: (a) bulk modulus (b) and density. The
red dot on both phase diagrams marks the critical point of CO
2
. The
vapor-liquid boundary is clearly discernible from the sudden change in
physical properties. The boundaries of the supercritical fluid phase state are
marked by the white dotted line. The temperature and pressure conditions
applied to the CO
2
during ultrasonic pulse transmission measurements are
shown by the white arrows.
14
L
O
G
G
I NG SOC
I E
T
Y

Rt
Ro Rw F
Sw
C
A
N
A
DI AN W
E
L
L

Porous Medium Sample
A portion of Berea sandstone, predominantly composed of
quartz grains, is used as the host sample in this study. From
helium porosimetry measurements, the sample has a porosity of
19.0% and from air permeability measurements it has a
permeability of ~240 millidarcies. A photograph and a
scanning electron microscopy (SEM) image of the sample are
shown in figure 2. From mercury porosimetry tests, the pores
are mainly 11 microns in size, with a minor component of pores
smaller than this.
Experiment
To determine the elastic properties of the sample, the ultrasonic
pulse transmission technique is applied. A cylindrical sample
size of 50 mm in length and 25.4 mm in diameter is placed in
between a transmitting and a receiving transducer made of P-
and S- piezoelectric ceramics with a center frequency of 1
MHz that is mounted on an aluminum buffer cap. The sample-
transducer assembly is placed inside a pressure vessel that is
filled with hydraulic oil, where the hydraulic oil serves as the
hydrostatic pressurizing medium and provides the confining
pressure. The sample is placed in a flexible, clear, impervious
Tygon tube to seal it from hydraulic oil contamination. CO
2
is introduced into the vacuumed sample via stainless steel tub-
ing that connects the sample to a CO
2
tank of 99.9% purity, lo-
cated on the outside of the vessel. The confining and pore fluid
systems are independent of each other such that different pres-
sure conditions can be applied irrespectively of each other by
using different pumps. The temperature of the experiment is
controlled by an electrical resistance tape that is wrapped
around the vessel and is measured by a K-type thermocouple
located next to the sample inside the vessel.
The transmitted signal is generated by exciting the transmit-
ting transducer with a fast-rising 200V square wave from a
Figure 2: Berea Sandstone a) photograph b) SEM image
JSR-PR35 pulse generator. The propagated signal is recorded
by a digital oscilloscope made by National Instrument, at a
sampling rate of 10 nanoseconds. The final waveform is a stack
of over 500 traces to reduce random noise effects. From the fi-
nal waveforms the transit time of the signals is deduced, and
with the length of the sample, the elastic wave velocities conse-
quently calculated. The general scheme of the experiment is
shown in figure 3.
Ultrasonic measurements are conducted on the sample under
dry and CO
2
saturated conditions. The dry measurements are
completed with the pore space under vacuum (Pp =0 MPa)
while the confining pressure (i.e. differential pressure when Pp
=0 MPa) varies from 2 MPa to 40 MPa. The resulting P-and
S-wave velocities show a significant nonlinear dependence on
confining pressure as was expected for the Berea sandstone.
This pressure dependency of the elastic wave velocities is at-
tributed to the closing of micro-cracks in the sandstone under
pressure. For the CO
2
saturated measurements, five different
constant temperature runs are conducted while pore pressure
varies from 2 MPa to 25 MPa. The constant temperature runs
are completed at 23 C, 28 C, 40 C, 45 C, and 55 C. In fig-
ure 1, superimposed on the phase diagrams of CO
2
are the dif-
ferent temperatures and pore pressures conditions explored
during the series of CO
2
saturated measurements. As pore
pressure increases, for the lower temperature runs (23 C and
28 C) CO
2
changes from a gas phase to a liquid phase, while
for the higher temperature runs (40 C, 45 C, and 55 C) CO
2
changes from a gas phase to the supercritical fluid phase. For
each CO
2
measurement run, a constant differential pressure of
Figure 3: Experimental setup for ultrasonic pulse transmission for CO
2
saturated measurements.
CO
2
Rock Physics continued
15
L
O
G
G
I NG SOC
I E
T
Y

Rt
Ro Rw F
Sw
C
A
N
A
DI AN W
E
L
L

Continued on next page
15 MPa is maintained by varying the confining pressure ac-
cordingly to the pore pressure. As mentioned earlier, the elastic
wave velocities of the Berea sandstone exhibits a pressure de-
pendency. By maintaining a constant differential pressure, any
resulting variations in elastic wave velocities are solely caused
by the pore fluids behaviour, if it can be assumed that the in-
tegrity of the sample has not been altered.
Laboratory results
Figure 4 shows the recorded P- and S-wave waveforms with
their amplitudes displayed in color as a function of pore pres-
sure for only the lowest and highest temperature run, T = 23 C
and T = 55 C respectively. In general, as pore pressure increases
the signals arrive later and in a given plot, two arrival trends can
be seen. Furthermore, the waveforms signal strength shows
changes near the anticipated phase transition. For the lower
temperature run, the elastic wave response to the CO
2
phase
transition is sharp and is related to the distinct contrast be-
tween the gas to liquid physical properties seen in the bulk
modulus and the density phase diagrams of CO
2
. The sharp
change in the waveforms occurs at a pore pressure close to a
CO
2
vapour-pressure of 6.1 MPa at 23 C. In contrast, the
elastic wave responses of the higher temperature run to CO
2
s
phase transition is gentle and less obvious. This is a conse-
quence of the subtle differences between the gas to supercriti-
cal fluid physical properties. Furthermore, the change in wave-
forms trend does not occur until well past the critical pressure
Figure 4: P- and S-wave (left and right, respectively) waveforms collected
for T = 23 C and T = 55 C (top and bottom, respectively) constant
temperature runs.
of CO
2
. Referring to the density phase diagram of CO
2
, large
changes in density do not occur till about 10 MPa at 55 C.
The P- and S-wave velocities of all the waveforms were deter-
mined and are displayed in figure 5. On these plots, the veloc-
ity of the dry sample at a differential pressure of 15 MPa is
shown by the black dot. From these velocity plots the following
are observed:
1. Once CO
2
is introduced into the pore space, irrespective of
the phase state of the CO
2
, the elastic waves traveling
through the sample are slower than when the pore space is
empty;
2. As pore pressure increases, the elastic wave velocities gener-
ally decreases with the exception of the T = 23 C P-waves
at high pore pressures;
3. Over the entire 25 MPa pore pressure interval, P-wave ve-
locities changed between 3.6-3.9% while S-wave velocities
changed between 3.9-4.4%. The change in wave velocity is
greater and sharper for the lower temperature runs than the
higher temperature runs;
4. Wave velocity variations with pore pressure are greater
within the gaseous phase state than within the liquid or su-
percritical fluid phase state.
For all S-waves of a given temperature run, as pore pressure in-
creases the wave velocity decreases. This behavior is expected
since theoretically changes to S-wave velocities are only due to
bulk density changes, which increases with pore pressure.
Between temperature runs, the decrease in S-wave velocity
with increasing density is less consistent. The integrity of the
sample may have changed between the measurement run. For
P-waves, as pore pressure increases the wave velocities generally
decreases except for the T = 23 C run at high pore pressures.
Figure 5: Ultrasonic a) P- wave b) S-wave velocities measured of the
sample saturated with CO
2
under various temperatures and pore pressures
and when the sample is not saturated (black dot at Pp = 0 MPa). All
measurements on the two plots were completed with a differential pressure
of 15 MPa.
16
L
O
G
G
I NG SOC
I E
T
Y

Rt
Ro Rw F
Sw
C
A
N
A
DI AN W
E
L
L

CO
2
Rock Physics continued
The decrease in P-wave velocities suggests that the effect of the
change in bulk density is greater than the effect of the change
in bulk modulus; the two have opposing effects on P-wave ve-
locity and they both increase as pore pressure increases. For the
T = 23 C run at high pore pressures, the increase observed sug-
gests that the change in bulk modulus dominates. Referring to
figure 1, in the liquid phase state the T = 23 C run do undergo
a greater change in bulk modulus than the other four tempera-
ture runs; therefore the unique behavior observed for the
T = 23 C run at high pore pressures is reasonable.
Conclusions
P- and S-wave laboratory data were acquired for various pres-
sure and temperature conditions on a fully CO
2
saturated Berea
sandstone. The results show that elastic waves are sensitive to
changes pertaining to the pore space content and CO
2
phase
change may be evident from signal variations. The elastic wave
velocities show a sharp change when CO
2
transitions from a
gas to liquid. However, the elastic wave velocities show a
smoother and a more subtle change when CO
2
transitions from
a gas to a supercritical fluid. The change observed here is for an
extreme end member, and this does not imply that the same re-
sults can be attained when there is more than just CO
2
in the
pore fluid.
Acknowledgements
HY would like to thank the NSERC-CGS M.Sc. award for
support, and the NSERC Discovery Program for support of
the laboratory work and purchase of some of the equipment
that allowed us to appropriately control fluid pressures in the
samples. Also, a big thanks to UofA Physics Dept. technicians,
Len Tober and Lucas Duerksen for their technical support in
making these experiments possible.
References
Batzle, M. L., and Wang, Z., 1992, Seismic properties of pore
fluids, Geophysics, 57, 1396-1408.
Benson, S. M., and Cole, D. R., 2008, CO
2
Sequestration in
Deep Sedimentary Formations, Elements, 4, 325-331.
Chadwick, R. A., Arts, R., Bentham, M., Eiken, O., Holloway,
S., Kirby, G. A., Pearce, J.M., Williamson, J. P., Zweigel, P.,
2009, Review of monitoring issues and technologies associated
with the long-term underground storage of carbon dioxide,
Geological Society of London Special Publications, 313, 257-
275.
IPCC, 2005, Underground geological storage, In: Metz B.,
Davidson O., de Coninck H. C., Loos M., Meyer L. A. (eds),
IPCC Special Report on Carbon dioxide Capture and Storage,
prepared by Working Group III of the Intergovernmental
Panel on Climate Change, Cambridge University Press.
Span, R., and Wagner, W., 1996, A new equation of state for
carbon dioxide covering the fluid region from the triple-point
temperature to 1100 K at pressures up to 800 MPa, J. Phys.
Chem. Ref. Data, 25, 1509-1596.

Page 1 of 9

Seismic behavi our of CO
2
saturated Fontainebleau
sandstone under in situ conditions
Md Mizanul Huq Chowdhury
a
, Douglas R. Schmitt
b
and Randolf Kofman
c

a
University of Alberta, Edmonton, AB, Canada, mhchowdh@ualberta.ca;
b
University of Alberta,
Edmonton, AB, Canada, dschmitt@ualberta.ca;
c
University of Alberta, Edmonton, AB,
Canada, rkofman@ualberta.ca.

Summary
Understanding the seismic response of a rock during saturated with CO
2
is an important
component for the monitoring of geological greenhouse gas sequestration and of volcanic
hazards. Additionally, the study of the effect of CO
2
on seismic wave propagation is
scientifically interesting because CO
2
can exist in gas, liquid, and supercritical fluid phases
over the modest temperature and pressure ranges typically accessible in the upper 2 km of
the earths crust, CO
2
s critical point lies near 31C and 7.4 MPa. We have carried out a
series of ultrasonic pulse transmission experiments on several samples of fully CO
2

saturated Fontainebleau sandstone over pore fluid pressure ranges of 1 MPa to 20 MPa and
at two constant temperatures below (21C) and above (50C) the critical temperature, these
ranges were chosen to cross the gas-liquid and gas-supercritical transitions, respectively.
Introduction
In geological sequestration CO
2
leakage is a vital concern; monitoring and verifying the
subsurface movement and phase behaviour of the injected CO
2
is very important to ensure
the integrity of the reservoir. Seismic methods are thought to be one way to monitor the
changes in subsurface because the seismic velocities are sensitive to a rocks mineralogical
composition, porosity and pore fluid contents. As such, it is necessary to understand the
degree to which changes in the CO
2
saturation state will influence the seismic properties.
This work gives simultaneous measurements of ultrasonic compressional and shear wave
velocities on Fontainebleau sandstones as functions of pore pressure and temperature.
Fontainebleau sandstone is collected from the Paris region, France. It shows a large porosity
variation from 2%-28% and consists of nearly pure quartz (99.8%). Its pore geometry shows
significant variation with porosity and has a wide spectrum of microstructure. The main
motivation of this work is to obtain a good understanding of the rock physics involved with
CO
2
as pore fluid.
Physical and elastic properties of CO
2
Second International Workshop on Rock Physics, Southampton, 4 9 August 2013
Page 2 of 9

The bulk modulus and density phase diagrams of CO
2
are shown in the Fig. 1 as function of
pressure and temperature on the basis of Span and Wagners (1996) thermodynamic model.
The critical point of CO
2
according to this model is at 31
0
C and at 7.4 MPa. Depending on
the subsurface situations, CO
2
can be either gas or liquid under this temperature and
pressure. A sudden change in physical properties of CO
2
clearly indicates the gas-liquid
boundary in the diagram, which gradually vanishes as the red point (critical point) reached.
The supercritical fluid phase state begins just after that point. The supercritical fluid phase is
no longer a simple vapour or a liquid, but has some of the properties of of gas and liquid in
the same time. One result of this is that the transition from liquid or gas to supercritical is
smooth. This differs significantly from the abrupt vapour-liquid transition that appears as a
sharp discontinuity in Fig. 1.

Fig 1: CO
2
phase diagrams as function of pressure and temperature according to the
thermodynamic model of Span and Wagner (1996). Fig 1(a) gives the bulk modulus and the
fig 1(b) is for density phase diagram of CO
2
.The critical point of CO
2
is clearly indicated by
the red dot in both phase diagrams. The gas-liquid boundary is easily noticeable because of
the sudden change in physical properties. The white dotted line gives the boundaries for the
supercritical fluid phase. White arrows shows three cases of our measurements as (i)
gaseous to liquid, (ii) gaseous to supercritical fluid and (iii) liquid to supercritical fluid
transitions.
Page 3 of 9

Gassmanns Rock-fluid interaction Model:
Gassmanns equation is a simple and widely used rock-fluid interaction models applied in
modelling responses inthe seismic frequency region ~100 Hz. Gassmann considered the
elementary elasticity of the pore fluid, the mineral grains, and the rock frame of the saturated
sample under a set of assumptions that include:
1) A microscopically homogenous and isotropic medium,
2) Similar bulk and shear moduli for all minerals that constitute the rock sample,
3) Vanishing viscosity and a free movement of the fluids in pore regions which are
interconnected (i.e. making this problem essentially static).
4) Completely saturated pore space all times,
5) No chemical interaction between pore fluid and rock minerals.
6) Quasi-static conditions are maintained to have frequencies low enough.
In this Gassmann's formulation the medium's saturated bulk modulus, K
sat
, depends on the
rocks frame modulus, K
dry
; the bulk modulus of the mineral grains, K
s
; the bulk modulus of
the fluid, K
f
; and the porosity of the rock medium, , through the following:

2
2
(1 )
1
dry
s
sat dry
dry
f s s
K
K
K K
K
K K K

= +
+ +
,
sat dry
= (1.1)
Here
sat
&
dry
are the saturated and dry shear modulus of the rock frame. P- and S-wave
velocities than can be calculated using the following formulas:

4
3
sat sat
p
sat
K
V
+
= ,
sat
s
sat
V

= , (1 )
sat s f
= + . (1.2)
Here
sat
is the saturated density and
f
is the fluid density.
Sample Properties
Initial measurements have been made on three samples. The table summarizes their
properties and the Fig 2 shows the samples:

Properties Fontainebleau
sample 1
Fontainebleau
sample 2
Fontainebleau
sample 3
Mass (gm) 117.1 167.2 130.4
Bulk Volume (cm
3
)
45.16 65.59 44.74
Grain Density (kg/m
3
)
2650 2650 2509
Bulk Density (kg/m
3
)
2593 2549.16 2914
Porosity (%) 10 12.5 6.28
Modal Pore size (m) 20 20 2.1
Page 4 of 9

Fig 2: Fontainebleau samples.

Fig 3: Schematic diagram of the set-up.
Experimental setup & protocol
The ultrasonic pulse transmission method was used to determine P- and S-wave velocities. This
method conceptually involves measuring the travel time of the ultrasonic wave travelling
Page 5 of 9

Attaching PZTs and
putting in the vessel
through the sample. The experimental set up consists of several functional units such as a
pulse generator, source/receiving transducers, a digital oscilloscope, a pressure vessel that
can apply confining pressure up to 200 MPa, a fluid reservoir, and a thermocouple. Fig 3
shows simplified schematic set-up of the experiment.
A series of measurements are done in our experiment including dry runs where the sample is
free from any liquid or gas to obtain the dry frame moduli, Nitrogen runs to test the effective
pressure, than various CO
2
saturated runs and at the end H
2
O runs. The whole experimental
protocol is shown in the following flow chart:

In the flow chart PZT stands for Piezoelectric Transducer, P
d
for differential pressure and P
p

for pore pressure.
Examples:
Dry & nitrogen runs: In the dry measurement, there is no saturating fluid and the pore
space of the samples are subject to a vacuum. Two series of dry run measurements, one at
21
0
C and another at 50
0
C with increasing confining pressure were made. Once these are
completed the sample is saturated with N
2
gas in order to test the effective pressure
response of the material. Fig 4 shows the P-wave velocity vs time plots of the dry
measurements and the N
2
run as a function of the confining pressure. The N
2
run was
carried out at a constant differential pressure of 15 MPa.
Vacuumed at 70
0
C
Vacuumed
overnight
CO
2
saturated runs:
1) Liquid to Gas (T=21
0
C and
P
d
=15 MPa)
2) Gas to Supercritical fluid
(T=50
0
C & P
d
=15 MPa).
Vacuumed
overnight
H
2
O

Nitrogen runs at
21
0
C and P
d
=15
MPa
Dry runs at 21
0

and 50
0
C
Sample preparation
Page 6 of 9

Fig 4. Dry measurements with nitrogen run for 21
0
C. In the dry case as the confining
pressure increases the wave velocities also increases. At a given confining pressure the
wave velocities during up-cycle (pressurizing) is always lower than down-cycle
(depressurizing) showing wave velocity hysteresis. The wave velocities increase rapidly at
lower pressure but this increment diminishes at higher pressure because of number of
closure of compliant pores reduces at high pressure.
CO
2
runs: Three different CO
2
saturated runs were carried out to see the transition of 1)
liquid to gas, 2) gas to supercritical fluid and 3) liquid to supercritical fluid. Fig 5 and 6 shows
the three transitions respectively:
Fig 5: Liquid to Gas transition measurements. For S-wave we see a significant velocity drop
of ~3% on that pressure range(P
p
=5-8 MPa) . In this case the temperature was constant (T~
21
0
C) and the differential pressure 15 MPa.

Fig 6: In the vapour to supercritical transition measurement the P-wave (left) velocities show
a gradual change near transition pressure (P
p
~710 MPa). In the liquid to supercritical fluid
transition measurements (right) the transition is gradual around the transition temperature
~31
0
C for S-wave (right) case.
Page 7 of 9

Comparison between observation and Gassmanns model prediction:
Gassmanns equation is used to predict the P- and S-wave velocities in the same range of
temperatures and pressures and Fig. 7 shows the comparison for liquid to gas and gas to
supercritical cases. We can see that the trend for each case is similar but there is a
difference in the velocities. Gassmanns equations are applicable for low frequency (~100
Hz) measurement but we did our runs at high frequency (~1 MHz) and all the assumptions of
the model cannot be fulfil for a real measurements.
Fig. 7: Comparison of the observed and the
Gassmanns model prediction for liquid to gas and gas to supercritical fluid cases of CO
2

saturated runs.
Conclusions
We have measured P- and S-wave velocities for various pressure and temperature
conditions in two Fontainebleau sandstones to check the change in velocities during the
phase transitions. The velocities change gradually across the gas-supercritical and liquid
supercritical transitions in agreement with the nature of these second order phase transitions.
The gas-liquid transition is, conversely, a first order transition discontinuous in CO
2
density
and bulk modulus and explains the abrupt changes in a wave speeds across the phase
boundary. However, in the real situation the rock may not only saturate with CO
2
, there may
be other pore fluid present there too to consider. The comparison between the observed data
and Gassmanns model predictions are also shown where we see a similar trend for the two
cases for CO
2
saturation measurements although there is a large dispersion between the
observed and calculated values.
Acknowledgements
This research is funded by the NCE-Carbon Management Canada project grant to DRS.
Equipment used in these tests were supported from NSERC grants.
References
Yam, Helen, CO
2
rock physics: a laboratory study, 2011,MSc thesis, Department of Physics,
University of Alberta.
Span, R., and Wagner, W., 1996, A new equation of state for carbon dioxide covering the
fluid region from triple-point temperature to 1100K at pressure up to 800 MPa, J . Phys.
Chem. Ref. Data, 25, 1509-1596.
Gassmann, F., 1951,Uber die elastizitat poroser medien: Vierteljahrsschrift der
Page 8 of 9

Naturforschenden Gesellschaft in Zurich, 96, 123.

Page 9 of 9

Use of CO2 as a Fluid in Fundamental Studies
of Wave Propagation through Porous Media

D.R. Schmitt
1
, G. Njiekak
1
, M. Chowdhury
1
, R. Kofman
1
, A. Rabbani
1
, and H. Yam
1

1
Institute for Geophysical Research, Department of Physics, University of Alberta,
CCIS 4-138, Edmonton, Alberta T6G 2E1; PH (780) 492-3985; FAX (780) 492-0714;
email: dschmitt@ualberta.ca

ABSTRACT
The compressional and shear wave speeds of a fully CO
2
saturated porous
ceramic are measured as functions of pore fluid pressure and temperature. The wave
speeds show considerable variation dependent upon the phase state of the CO
2
. The
observations are further in good agreement with Biots model of wave propagation.
The ability to change the CO
2
s compressibility, density, and viscosity by relatively
modest changes in temperature and pressure provides for a somewhat tunable fluid
that can be used to study fluid induced variations of wave propagation behavior.
INTRODUCTION
The rapid increase in the concentration of CO
2
into the atmosphere from
various anthropogenic sources motivates the search for ways to both reduce the
amount of CO
2
produced and to otherwise prevent it from being released. In
achieving the latter, the geological sequestration of greenhouse gases emitted from
large sources, such as coal-fired electrical power plants, has been proposed. Remote
tracking of the motion and possible escape of such sequestered fluids will require that
some kind of remote monitoring of the reservoir into which the CO
2
has been injected
be carried out. This has motivated numerous studies [e.g., Khatiwada et al., 2012] in
which the seismic responses have been modeled. Despite this large amount of more
theoretical work, there are few experimental studies of the acoustic properties of CO
2

saturated rocks under expected in situ conditions [e.g., Vialle and Vanorio, 2011]
This lack of experimental work motivated the development of a system and
methodology to carry out P- and S-wave speed measurements in porous media
saturated with CO
2
. A major focus of this research obviously examines the effects in
natural porous rocks from which actual candidate reservoirs are made. Real rocks,
however, are notoriously complex with disperse pore shapes and sizes, physical
properties that depend nonlinearly with confining stress that are further hysteretic.
Consequently, the first measurements were carried out on a synthetic sintered
alumina in order to better understand wave propagation through a simpler porous
media. The observed wave speeds agree well with theoretical predictions of the Biot
model [Biot, 1956a; 1956b; Johnson et al., 1987] when saturated with gaseous N
2
,
liquid H
2
O, or the gas, liquid, and supercritical phase states of CO
2
. Although
perhaps obvious in hindsight, CO
2
s ability to pass between these three different
phase states at relatively modest temperature and pressures allow for tuning of the
saturating fluids viscosity, density, and compressibility. This permits a wide range of
acoustic tests to be performed in which the fluid properties can be varied without
having to physically replace the saturant.
Overall we are motivated to carry out this line of investigation in order to
answer questions related to the seismological tracking of sequestered CO
2
. This
contribution, however, focuses more on our measurements and observations in
simpler and more easily characterized synthetic porous alumina. We first briefly
review the relevant pressure and temperature dependent physical properties of CO
2
,
this review is necessary in order to develop the rationale for our experimental
protocols. The newly developed experimental apparatus is described and some
relevant observations provided. The wave speeds observed are compared to those
predicted using Biots model. The observed and predicted wave speeds mostly agree
well. However, under certain situations they unexpectedly diverge, and this
divergence led to an important modification of our experimental protocol.
BACKGROUND
The critical point of CO
2
is at only 30.9782 C and 7.3773 MPa (Figure 1). These
conditions are easily achieved at relatively shallow depths that depend on the local
geothermal gradient and the degree to which the fluids in the reservoir are
pressurized. Regardless, in the shallow earth CO
2
can exist in vapour, liquid, or
supercritical phases. A typical situation could have CO
2
injected into the reservoir at
2 km depth with temperatures of 50 C or more and pore fluid pressures that could be
as high as 20 MPa. Under these conditions any free CO
2
within the pore space will
be supercritical. Buoyant leakage of CO
2
escaping from the intended reservoir can
revert back to vapour or liquid as it migrates upwards through the geological column.
In Figure 1, the liquid to vapour (i.e. boiling) phase transition locus is shown as a
dark line that terminates at the critical point CP. This is a first order phase transition
which physically means that a latent heat of transition is involved and mathematically
that the first derivative of the property with respect to a thermodynamic variable (e.g.
pressure or temperature) is discontinuous. This is obvious in the 3D surface
rendering of a number of CO
2
s properties in Figure 2 where the latent heat is
represented by the jump in the enthalpy across the transition. The density, bulk
modulus, and viscosity are all correspondingly discontinuous across the transition.
The transformations from liquid or vapour to supercritical fluid are continuous or
higher order, phase transitions. These are not associated with a latent heat as may
also be seen in the smooth variation in the enthalpy across the boundaries.
Correspondingly, the physical properties vary smoothly although the change across
the vapour supercritical transition is more transilient.

Figure 1. Phase diagram for CO2 in the vicinity of the critical point CP.

In the Biot model, even though the fluid phase is neglected the fluids density,
viscosity, and bulk modulus all play a role in determining the frequency dependence
of wave speeds and attenuation. Further, as shown in Figure 2 all of these properties
depend on both the fluid pressure and the temperature of the system. Conversely, in
CO
2
, the pressure and temperature strongly control the fluids properties and hence
the wave propagation in the saturated porous media.

EXPERIMENTAL METHOD
The experimental method and particularly the protocols developed for making these
measurements are presented in more detail elsewhere [Njiekak et al., 2013; Yam,
2011] and only a brief overview is necessary here.
The porous sintered alumina employed in these measurements is commercially
obtained and it is a filter media employed in soil fluid pressure tensiometry. The
parameters needed to carry out predictions using the Biot model [Biot, 1956a; 1956b;
Johnson et al., 1987] were measured using protocols developed in Bouzidi and
Schmitt [2009] and are rehearsed in Table 1. The high porosity of this medium results
from the sintering of highly irregular shards of alumina. Unexpectedly and despite its
high porosity this ceramic displayed little pressure dependence and undetected
hysteresis to over 40 MPa of total confining stress when the pore space was subject to
vacuum. Further, under conditions of constant differential (or effective) pressure of
15 MPa the observed N
2
saturated waves speeds also showed little variation. Hence,
all of the experimental results presented here were carried out at a constant
differential pressure in order to reduce any changes in the ceramics frame moduli
thus enhancing those changes due to variations in the fluid properties. An added
feature to this material is that its dry frame is exceedingly stiff for a porous material
and this too assists in studying the influence of the CO
2
fluids on the wave
propagation. Of course, the fluid properties must be determined for each fluid
pressure and temperature and the model developed by Span and Wagner [Span and
Wagner, 1996] was used to provide the necessary values.
Both P- and S-waves through the sample were simultaneously determined using the
pulse transmission technique from ultrasonic waveforms acquired at a range of pore
fluid pressures and temperatures. The samples were jacketed and subject to both
controlled confining and pore pressures, and as noted most of the CO
2
measurements
were carried out at a constant differential pressure of 15 MPa. Temperature control
was more problematic due to the large thermal inertia of the pressure system. Transit
times were converted to wave speeds using the known length of the sample after
accounting for the transducer/end cap delays that were calibrated as functions of
confining pressure and temperature.

Figure 2. Three dimensional representations of fluid CO
2
s pressure and
temperature dependencies for the density, the viscosity, the bulk modulus, and
the enthalpy. Underlying data obtained using Span and Wagners [1996] model.

A major objective of the experiments was to map out the effects of the changes in the
fluid properties, and particularly the phase state, on the ultrasonic wave speeds. As
such, the experimental protocol proceeded by acquiring appropriate waveforms under
sample conditions of i) varying confining pressure with the pore space under vacuum
(dry), ii) under constant effective pressures at different temperatures with the pore
space saturated with gaseous N
2
, iii) a series of different runs at constant effective
pressure with medical grade (99.9% pure) CO
2
saturant and at either constant pore
fluid pressure or constant temperature (see Figure 1), and, finally, iv) at constant
effective pressure at different temperatures with the pore space saturated with liquid
deionized water.

Table 1. Measured Physical Properties of the Sintered Alumina Sample
Property Symbol Value
Porosity 0.584
Tortuosity 1.73
Permeability k 97 mD
Dry Frame Bulk Modulus K
d
106.5 GPa
Solid Bulk Modulus K
s
251.7 GPa
Dry Frame Shear Modulus
d
88.9 GPa
Solid Shear Modulus
s
162.5 GPa
Pore Diameter d 2.06 m
Dry Frame Density
d
1700 kg/m
3
Solid Density
s
3950 kg/m
3

RESULTS AND DISCUSSION

A full presentation of the experimental results cannot be made in the space available
and the reader is referred to Yam [2011] for the summaries of all of the data obtained.
Here we focus on those changes only across the gas-liquid and gas-supercritical
transitions for variable differential pressure tests held at constant temperatures of
28C and 40C respectively, and at constant differential pressure (i.e., the difference
between the confining and the fluid pressure).

The raw sets of longitudinal ultrasonic waveforms for the 28C (Figure 3a) and 40C
(Figure 3b) runs display the lengthening of the pulse transit times with increasing
fluid pressure. A lengthening of the transit time equates to a decreased wave speed.
At 28C the gas-liquid transition occurs at 6.92 MPa. Below this the waveforms
have higher amplitudes and the transit time lengthens gradually towards the phase
transition pressure. The transit time abruptly changes across the first-order gas-liquid
transition. Similar behaviour is seen at 40 C. For this run the gas-supercritical fluid
transition is at 7.38 MPa, and again below the transition the pulses arrive earlier and
are stronger in amplitude. However, one important difference between the 28C and
the 40C waveform sets is that the waveforms evolve from gas-saturated to
supercritical fluid-saturated more gradually in concert with the more muted higher-
order gas-supercritical phase transition.

Figure 3 | Observed longitudinal ultrasonic waveforms at varying pore fluid
pressures along the 28C and 40C trajectories. a. Variation in the ultrasonic
waveform character and traveltime for the 28C trajectory. The gas to liquid
transition occurs at 6.8918 MPa as delineated by the vertical dashed line. b.
Variation in the ultrasonic waveform character and traveltime for the 28C
trajectory. The gas to liquid transition occurs at 7.3850 MPa as delineated by
the vertical dashed line. Note that the horizontal axis denotes the pressures at
which the measurements are made and is not linearly increasing.
The observed longitudinal and transverse wave speeds extracted from the waveforms
(Figure 3) are plotted against the CO
2
pore fluid pressure in Figure 4. At 28C the
wave speeds drop drastically across the phase transition while, at 40C the speeds
vary more slowly, these directly reflect the differences in the fluid properties through
the gas-liquid and gas-supercritical fluid transitions, respectively. In all cases the
observed wave speeds decrease by as much as 5%. This behaviour results primarily
from the increase in the density of the saturating CO
2
fluid (and hence the overall
saturated bulk media density) upon the transformation from gas to liquid or to
supercritical fluid.

According to the Biot model, the speeds and attenuations of mechanical waves
propagating through fluid-saturated porous media depend on frequency. This
dispersion arises from the competition between the fluids inertia and viscosity as it
moves relative to the solid frame as the wave passes. The frequencies (~1 MHz) of
the ultrasonic measurements substantially exceed those typically encountered in
active seismic measurements in the uppermost crust for monitoring purposes (~100
Hz). As noted, the wave speeds disperse with frequency in fluid-saturated porous and
permeable media; and independently of the measurements values of the wave speeds
were calculated at frequencies of 1 MHz and 100 Hz, these bounds represent the top
and bottom boundaries, respectively, of the shaded dipper-like shapes in Figure 4.
The observed values are in generally good agreement with the theoretical predictions
calculated using the high frequency model of Biot by less than 0.5% and 0.4% for the
V
P
and V
S
, respectively.

Figure 3 | Comparison of observed and calculated wave speeds, the observed
data points are represented by connected down-pointing or up-pointing triangles
along the 28C and 40C trajectories, respectively. The light and darker filled
irregular shapes delimit the calculated high and low (LF) frequency Biot model
limits for the 28C and 40C trajectories, respectively. The observed data lie
close to the high frequency calculations and the lower bounds are close to the
letters LF.

Since the wave speeds at observed at high frequency are adequately matched by Biot
model calculations, we use the low frequency extensions of this theory to predict the
seismic waves speeds at 100 Hz. It is worth noting that these values differ
insignificantly from the zero frequency limit [Gassmann, 1951] that does not
incorporate viscous effects. The low frequency bounds, denoted by LF in Figure 4,
are of most concern to those seismological transit-time dependent methods that are
sensitive variations in seismic wave speeds. The predicted wave speeds at low
frequency display significantly greater variations with pore fluid pressure than the
high frequency observations and predictions. The abrupt nature of the 28C gas to
liquid transition would be most easily distinguished by wave travel-time variations.
At 40C, however, although a large over-all drop is seen, the gradual nature of the
transition makes tracking of this phase variation difficult over small pressure
variations. Further, once a phase transition has occurred, V
P
and V
S
evolve more
slowly and changes will be more difficult to detect.

CONCLUSION
An experimental apparatus has been developed to make ultrasonic P- and S-wave
speed measurements on CO
2
saturated porous media over a range of pressure and
temperature conditions expected for in situ geological sequestration. CO
2
can exist as
a vapour, a liquid, or a supercritical fluid at these ranges. The initial tests were
carried out using a highly porous sintered alumina ceramic, and through this material
the observed wave speeds were found to be in good agreement with those predicted
using Biots model for wave propagation. Further, although this is perhaps obvious in
hindsight, the relatively modest temperatures and pressures required to change the
phase state of CO2 allow its density, viscosity, and bulk moduli to be easily tuned
during an experiment without need to re-saturating the material with a different fluid.
REFERENCES
Biot, M. A. (1956a), Theory of propagation of elastic waves in a fluid-saturated porous
solid .2. Higher frequency range, J. Acoust. Soc. Am., 28(2), 179-191.
Biot, M. A. (1956b), Theory of propagation of elastic waves in a fluid-saturated porous
solid .1. Low-frequency range, J. Acoust. Soc. Am., 28(2), 168-178.
Bouzidi, Y. S., and D. R. Schmitt (2009), Measurement of the speed and attenuation of
the Biot slow wave using a large ultrasonic transmitter, J. Geophys. Res., 114.
Gassmann, F. (1951), U ber die elastizitat poroser medien., Vierteljahrsschr.
Naturforsch. Ges. Zuerich, 96, 1-23.
J ohnson, D. L., J . Koplik, and R. Dashen (1987), Theory of dynamic permeability and
tortuosity in fluid-saturated porous-media, J. Fluid Mech., 176, 379-402.
Khatiwada, M., L. Adam, M. Morrison, and K. van Wijk (2012), A feasibility study of
time-lapse seismic monitoring of CO
2
sequestration in a layered basalt reservoir, J. Appl.
Geophys., 82, 145-152.
Njiekak, G., D. R. Schmitt, H. Yam, and S. R. Kofman (2013), CO
2
rock physics as part
of the Weyburn-Midale geological storage project, Int. J. Greenhouse Gas Control, in
press.
Span, R., and W. Wagner (1996), A new equation of state for carbon dioxide covering the
fluid region from the triple-point temperature to 1100 k at pressures up to 800 mpa, J.
Phys. Chem. Ref. Data, 25(6), 1509-1596.
Vialle, S., and T. Vanorio (2011), Laboratory measurements of elastic properties of
carbonate rocks during injection of reactive CO
2
-saturated water, GRL, 38.
Yam, H. (2011), CO
2
rock physics: A laboratory study, 285 pp, University of Alberta,
Edmonton.
Franais Home
3B6.4 ID:6762 11:30
Influence of rate of temperature variation on CO2 phase change in saturated synthetic rock observed with
ultrasonic measurements.
Arif Rabbani , Randolf Kofman , Douglas R. Schmitt
Institute for Geophysical Research, Department of Physics, University of Alberta, Edmonton, Alberta
Contact: rabbani@ualberta.ca
Ultrasonic measurements using a fully CO2 saturated synthetic sample have been conducted over a variety of
temperatures and pressures representative of subsurface conditions aimed to monitor the evolution of seismic
responses resulting from changes in the gas, liquid, or supercritical phase state of CO2. We have observed that during
isobaric (31 MPa of confining pressure and 6 MPa of pore pressure) heating and cooling experiments the rate of
temperature change has a strong effect on the specific temperature at which the expected phase change would occur.
It is also noticed that the gas-to-liquid phase change (cooling) is not as strongly influenced by the cooling rate as the
liquid-to-gas phase change (heating). The high rate runs of heating and cooling have significant effects on both slope
and hysteresis in velocity versus temperature plots. The phase transition of pure, isolated CO2 under 6 MPa of pore
pressure is expected at ~22 C of temperature (NIST website) but in our experiments faster rate of heating of ~2.5
min/C shifts the phase change to ~34C from ~24C at slower rate of ~20 min/C. The time-dependent heat transfer
out of and into the sample during condensation and evaporation respectively may be the possible explanations for this
disparity. In particular, the temperature is measured in the confining fluid within ~5 cm of the top of the sample
assembly which will not necessarily represent the temperature throughout the sample; unless it is given time to
equilibrate. The time to achieve equilibrium suggested that thermal responses could in part be responsible for the
delays. To further investigate this observation a simplified model, the lumped capacity model, was implemented which
estimated the time more than hour to reach the equilibrium invariant with the temperature difference.
Abstract https://www1.cmos.ca/agenda/abstract.aspx?AID=6762
1 of 1 15/05/2013 7:46 AM
Franais Home
3B6.3 ID:6767 11:15
Ultrasonic wave speeds in CO2 saturated Fontainebleau sandstone under in situ conditions
Md Mizanul Huq Chowdhury , Douglas Rr Schmitt , Randolf Kofman
Contact: mhchowdh@ualberta.ca
In a CO2 geological sequestration project, potential CO2 leakage is one of the vital concerns. Monitoring the
subsurface movements and the phase state of the injected CO2 is therefore very important. Understanding the seismic
response of the subsurface rock in the sequestration is also crucial for the societal acceptance of this process.
Additionally, the study of the effect of CO2 on seismic wave propagation is scientifically interesting because CO2 can
exist in gas, liquid, and supercritical fluid phases over the modest temperature and pressure ranges typically
accessible in the upper 2 km of the earths crust, CO2s critical point lies near 31C and 7.4 MPa. We have carried out
a series of ultrasonic pulse transmission experiments on several samples of fully CO2 saturated Fontainebleau
sandstone over pore fluid pressure ranges of 1 MPa to 20 MPa and at two constant temperatures below (21C) and
above (50C) the critical temperature, these ranges were chosen to cross the gas-liquid and gas- supercritical
transitions, respectively. The porosity of the Fontainebleau samples is found to be in the range of 10-13% based on He
pyncnometry and Hg intrusion porosimetry. Across the liquid to gas transition we observed a 1.9- 3.9% and a
3.14-3.5% velocity drop in P- and S-waves, respectively. A more subtle (<1.2%) drop in the P-wave velocity occurs
across the gas-supercritical transition, but no discontinuity appears in the S-wave. The velocities change gradually
across the gas-supercritical and liquid supercritical transitions in agreement with the nature of these second order
phase transitions. The gas- liquid transition is, conversely, a first order transition discontinuous in CO2 density and
bulk modulus and explains the abrupt changes in a wave speeds across the phase boundary. The comparison
between the observed data and Gassmanns model predictions shows a similar trend for the all three cases for CO2
saturation measurements.
Abstract https://www1.cmos.ca/agenda/abstract.aspx?AID=6767
1 of 1 15/05/2013 7:45 AM

GeoConvention 2013: Integration 1
Investigations of Seismic Signatures of CO
2
Saturation as Part of
a Geological Storage Project
Gautier Njiekak*, Department of Physics, Institute for Geophysical Research, Univ. of Alberta, Edmonton, AB,
Canada
njiekak@ualberta.ca
Douglas R. Schmitt, Department of Physics, Institute for Geophysical Research, Univ. of Alberta, Edmonton, AB,
Canada
and
Randolf S. Kofman, Department of Physics, Institute for Geophysical Research, Univ. of Alberta, Edmonton, AB,
Canada

Summary
The current study provides an extensive series of measurements that aims at predicting the seismic
behaviour of CO
2
-saturated rocks. This has implications on monitoring CO
2
geological storage using
seismic methods. The ultrasonic pulse transmission technique is the underlying methodology used in
this work. Unlike previous rock physics studies involving CO
2
as pore fluid, care is taken to separate
pore fluid effects from pore pressure build-up effects during the experiments. This allows for the
sampling of the effects of the CO
2
s varying phase states (gas liquid - supercritical fluid) on the overall
rock seismic response. Measurements were done on core samples from the Weyburn oilfield,
Saskatchewan. Of the observations arising from the measurements done in this study, there are two
that are of particular note. First, both the P- and S-wave speeds decrease substantially as the CO
2
transforms from gas to either liquid or supercritical phase. This observation is consistent with the
increase of CO
2
fluid density across these phase boundaries. Second, across the gas-liquid phase
transition both wave speeds drop abruptly as would be expected for the change in the physical
properties of the CO
2
across this first order phase boundary. In contrast, across the gas-supercritical
phase boundary the velocities change more gradually. This suggests that it may be difficult to
distinguish the gas-supercritical boundary using seismic reflection techniques.

Keywords: CO
2
sequestration; seismic monitoring; rock physics; ultrasonic velocity; Weyburn reservoir.

1. Introduction
As is well known, societal acceptance of CO
2

geological sequestration depends critically on the
ability to monitor movement of CO
2

particularly the use of 4-D reflection seismology, will be a key component in monitoring motions of CO
2

underground (e.g., Carcione et al., 2006; Ivanova et al., 2012). As such, there is a great need for a
good understanding of how CO
2

will influence the seismic properties of rocks. This is doubly
interesting in that free CO
2

at depths intended for CO
2
sequestration can be in gas, liquid or
supercritical phase states. However, to our knowledge, there is a general lack of experimental
tests of such effects. Here we propose a laboratory protocol for the measurement of P- and S-
wave speeds and attenuations through natural rocks saturated with CO
2
. This protocol was
developed on the basis of experiences gained during an extensive series of laboratory

measurements of P- and S-wave velocities of carbonate reservoir and cap-rock core samples from
the Weyburn, Saskatchewan, project and some test sandstones.
2. Experimental Protocol
The protocol includes two important components. First, the samples chosen undergo a rigorous
material characterization that includes He and Hg porosimetry, XRF whole rock chemistry, thin
sectioning, SEM scanning, and micro-CT imagery. This allows for an understanding of the mineralogy,
the porosity, and the architecture of the pore space within the rock. The more crucial set of tests,
however, are those in which the rock sample is subject to a large variety of confining and pore
pressures, saturation states, and temperatures. A given test begins with coring of a cylinder of
the given sample. The ends of this cylinder are ground as parallel to one another as possible and
the sample is then dried under vacuum at modest temperatures (~70 C). The sample is then
prepared for the high-pressure measurement suite by attaching a set of specially constructed
ultrasonic transducers to its ends and then hermetically sealing this. This assembly is placed in a
pressure vessel. The P- and S-wave measurement protocol then follows by
i) Obtaining the P- and S-waveforms under a series of increasing confining pressures but with the
pore space subject to vacuum to provide the dry properties, this is carried out at different
temperatures to evaluate the effect of temperature on the rock frame.
ii) Repeating these measurements with the pore space saturated with inert nitrogen gas. An
additional suite here tests carried out at constant differential pressure (difference between the
confining and the pore pressure) allows for assessment of any potential effective stress variations
that could occur in the sample. This is critical because the rock elastic properties themselves depend
nonlinearly on confining pressure.
iii) Repeating these measurements with full CO
2

gas saturation under a variety of pore pressure and
temperature in order to sample the full range of CO
2
phase states. These measurements are all
carried out at constant differential pressure. This involves at least 5 different suites of measurements
carried out under conditions of constant temperature or constant pore pressure (Fig. 1).
iv) Repeating the dry measurements once the suite of CO
2

tests are completed, and
v) Finally carrying out the measurements under full saturation with liquid water.
These series of measurements provides a great deal of information that allows for prediction of the
seismic behaviour of the given rock, but at the cost that a given run requires upwards of a week of
dedicated time in the measurement system.
2.1. Ultrasonic measurements: experimental setup
The experimental setup for the ultrasonic measurements includes several functional units such as a
pulse generator, a digital oscilloscope (NI USB- 5133 100 MS/s), source/receiver transducers, a
cylindrical pressure vessel, fluid tanks, a thermocouple, and an electrical resistance tape (Fig. 2, see
also Yam, 2011). The transducer is exited by a fast-rising, 200V square wave, sent by the pulse
generator (Panametrics, model 5800 PR). The generated elastic wave after propagating through the
sample is recorded by a digital oscilloscope at a sampling interval of 10 nanoseconds. The recorded
waveform is the stack of at least 500 traces to reduce the noise level in the signal. The desired pore
fluid is introduced into the sample via stainless steel tubing that connects the pore space of the sample
to the pore fluid reservoir located outside of the vessel through the vessel lid. The confining and pore
pressure systems are independent of each other such that different pressure conditions can be applied
in irrespective of each other by using different pumps located outside of the pressure vessel. Both the
confining pressure (pressure vessel) and the pore pressure systems are capable of reaching a
maximum pressure of 70 MPa from the pumps. Higher confining pressures, up to 200 MPa, could also
be achieved by using a separate air pump connected to the pressure vessel.


Fig. 1. CO
2
properties (bulk modulus and density) as a function of pressure and temperature based on the
thermodynamic model of Span and Wagner (1996). C = critical point. Arrows (i), (ii), (iii), (iv), and (v) refer to the
conditions under which CO
2
-saturated rock velocities are measured. For (i), (ii), (iii) and (iv), measurements are
done under constant temperature. (v)* refers to measurements done under constant pore pressure (about 11
MPa in this case) and T = 50 C.


3. Case study
3.1. Sample description
The studied sample is a dolostone from the Weyburn area, Saskatchewan, with a (mercury injection)
porosity of 14% and air permeability lower than 2 mD. The rock matrix consists of relatively loosely
packed crystals (largely dolomite). Very small calcite grains and feldspar flakes are disseminated within

the matrix. Quartz is sometimes visible. The porosity is generally intercrystalline and the main pore
throat size is around 0.65 m. The sample grain density is 2.83 g/cm
3
. The ultrasonic measurements
were performed on a cylindrical plug, 3.81 cm in diameter and 4.8 cm long (Fig. 3).

Fig. 3. (a) Photograph of the dolostone's plug and (b) a scanning electron microscope (SEM) image of the
sample. Note the loosely compacted dolomite crystals in the SEM image.

a)
P-wave
4000
4100
4200
4300
4400
4500
4600
4700
4800
0 10 20 30 40
V
e
l
o
c
i
t
y

(
m
/
s
)
23 C, Dry Sample
50C, Dry Sample
S-wave
2550
2600
2650
2700
2750
2800
0 10 20 30 40
V
e
l
o
c
i
t
y

(
m
/
s
)
23 C, Dry Sample
50 C, Dry Sample

b)
P-wave
4100
4200
4300
4400
4500
4600
4700
4800
0 10 20 30 40
V
e
l
o
c
i
t
y

(
m
/
s
)
Dry Sample at 23 C
N2-Saturated Sample,
Pd = 15 MPa
S-wave
2600
2650
2700
2750
2800
0 10 20 30 40
V
e
l
o
c
i
t
y

(
m
/
s
)
Dry Sample at 23 C
Pd = 15 MPa

Fig. 4. (a) Dry and (b) nitrogen-saturated P-and S-wave velocities as a function of confining pressure for the
dolostone sample. Error bars are shown. In (a), velocities recorded at 23C and at 50C are shown. The
hysteresis of the dry velocity curves can be explained by the difference between the closing and opening rate of

the crack-like pores at a given pressure during pressurization and depressurization cycle, respectively (Sharma
and Tutuncu, 1994). Pd = differential pressure.

Fig. 5. Dry and CO
2
-saturated waveforms on the dolostone sample. (a): P- and S-waveforms on the dry sample.
Measurements were done during up- and down pressurization cycles. Note how P- and S- wave travel times
become faster with increasing confining pressure. (b): Waveforms that were acquired on the CO
2
-saturated
sample at T = 23 C; CO
2
was turning from gas to liquid in the pore space. (c): CO
2
was turning from gas to a
supercritical fluid in the pores; T = 50 C. Note the decrease of the S-wave travel times and the different

behaviours of the P-wave speeds (decrease at low pore pressure and slight increase at higher pore pressure)
from the CO
2

Fig. 6. CO
2
saturated P-and S-wave velocities of the dolostone sample for the two constant temperature runs at T
= 23C and T = 50C while under a constant differential pressure of 15 MPa. Error bars are shown. Dry wave
velocities obtained under a confining pressure of 15 MPa are plotted at pore pressure = 0 MPa to provide a
reference for the changes seen. Black arrows indicate velocities when CO2 is gas, liquid and a supercritical fluid
in the pore spaces.

3.2. Ultrasonic measurements
Under dry (vacuum) conditions:
On the dry sample, and as expected, the compressional (P) and shear (S) wave velocities increase with
confining pressure and decrease with temperature (Fig. 4a). Figure 5a shows the P- and S-waveforms
recorded on the dry sample.

Under fluid-saturated conditions:
Only nitrogen- and CO
2
-saturated measurements are provided in this sub-section. The measurements
conditions were set up so that any variation in the recorded waveforms was due to change in the pore
fluid physical properties. Thus, for each sample, a constant differential pressure of 15 MPa was
maintained by varying the confining pressure accordingly to the pore pressure. Assuming that the equal
amount of pore pressure increase cancelled the equal amount of confining pressure increase, any
waveform variation observed was solely caused by pore fluid effects.
Nitrogen-saturated measurements: Prior to CO
2
-saturated measurements, each sample was subjected
to constant differential pressure measurement of 15 MPa under inert nitrogen gas saturation. In general
within a fluid, the bulk modulus and density are the physical properties that will influence wave
propagation; bulk modulus has a direct proportional effect while density has an inverse proportional
effect on wave velocities. As change in the nitrogen properties (density, bulk modulus) only slightly
affects sound waves travel times under the range of pressure and temperature conditions used in this
study (Lemmon et al., 2011), no substantial change in the P- and S- wave velocities meant that the
applied differential pressure was the main factor controlling the variation of the velocity. This behaviour
of the wave velocities did lend confidence to selecting the set of high-quality measurements that were
suitable for the purpose of our study. Figure 4b shows that P- and S-wave velocities barely changed
when both confining and pore pressures were increased at the same rate. This suggests that the
effective stress that controlled the elastic response of the rock remained almost constant during the
nitrogen-saturated measurements.

CO
2
-saturated measurements: Results presented here are from measurements made at constant
temperature (23 C and 50 C) with the pore pressure varying from 1 MPa to 25 MPa in each case. As
pore pressure increases, for the lower temperature runs (23C), CO
2
changed from a gas phase to a
liquid phase in the pore spaces while for the higher temperature runs (50C), CO
2
changed from a gas
phase to the supercritical fluid phase (See arrows (i) and (ii) in Fig. 1). The wave velocities of the two
constant temperature runs are plotted in figure 6. On these plots, velocities of the dry sample at a
differential pressure of 15 MPa are also shown. P- and S-waveforms of the CO
2
-saturated sample are
shown in figures 5b and 5c.
Once CO
2
gas was injected into the sample, irrespective of the temperatures, both P-and S-wave
velocities decreased with the pore pressure. The drop of the P- and S-wave velocities was more
immediate for the low temperature run; the large drop occurred at a pore pressure of about 5 to 7 MPa
and 10 to 15 MPa for the low temperature (T= 23 C) and high temperature (T = 50 C) runs,
respectively. At 23 C, the immediate nature of the wave responses was likely due to the sharp contrast
in the density of the CO
2
when the gas-liquid boundary was traversed. Overall, the decrease of the P-
and S- waves travel times right before the gas-liquid and gas-supercritical fluid transitions were up to
3%. Over the phase transition, P wave velocity recovered while S-wave speed kept decreasing. The
recovery of the P-wave velocity was gradual and was up to 1 % at a pore pressure of 25 MPa. The total
drop of the S-wave travel times was up to 4 % over the entire 1 to 25 MPa pore pressure interval.

4. Conclusion
An experimental procedure has been set up to inspect the effects of CO
2
on the seismic properties of
rocks. The series of measurements allows for the assessment of the effects of the CO
2
phase changes
(gas to liquid and gas to supercritical fluid) on the overall rock seismic behaviour. CO
2
phase changes
(gas to liquid and gas to supercritical fluid) are marked by a drop in velocities of up to 4% over the entire 1
to 25 MPa pore pressure interval used in this study. The abruptness in the velocity change differs
significantly between the gas-liquid and gas-supercritical fluid transitions; and the gradual variations seen
for the latter suggest that this will be difficult to detect using seismic reflection methods. Real situations
sometimes differ from the experimental conditions set up in this study as CO
2
may coexist with other in-
situ fluids in the pore space. However, this study provides an end member understanding of the
possibilities with CO
2
in the pore space, which is critical for the setup of an accurate monitoring of CO
2

plume in the subsurface. In fact, any change in the conditions of pressure and temperature of a CO
2

storage area or any upward migration of the CO
2
towards the surface could lead to the transformation of
CO
2
into either a gas or a liquid or a supercritical fluid.
Acknowledgements
2
measurement system was supported by NSERC and the Canada
Research Chair program. Njiekak and Kofman were supported in part by a research contract from the
Weyburn-Midale CO
2
Monitoring and Storage Project as well as a research grant from Carbon
Management Canada. The technical assistance of Len Tober and Lucas Duerksen was invaluable.

References
Carcione, J.M, Picotti, S., Gei, D., and Rossi, G., 2006, Physics and seismic modeling for monitoring CO2 storage. Pure Appl.
Geophys., 163,175-207.
Ivanova, A., Kashubin, A., Juhojuntti, N., Kummerow, J., Henninges, J., Juhlin, C., Lueth, S., and Ivandic., M., 2012,
Monitoring and volumetric estimation of injected CO2 using 4D seismic, petrophysical data, core measurements and well
logging: a case study at Ketzin, Germany. Geophysical Prospecting, 60, 957973.
Lemmon, E.W., McLinden, M.O., and Friend, D.G., 2011, Thermophysical Properties of Fluid Systems in NIST Chemistry
WebBook, NIST Standard Reference Database Number 69, Eds. Linstrom, P. J. and Mallard, W. G., National Institute of
Standards and Technology, Gaithersburg MD, 20899, http://webbook.nist.gov, (retrieved 2008-2011).

Sharma, M.M., and Tutuncu, A.N., 1994, Grain contact adhesion hysteresis: A mechanism for attenuation of seismic waves.
Geophysical Research Letters, 21, 2323-2326.
Span, R., and Wagner, W., 1996, A new equation of state for carbon dioxide covering the fluid region from the triple-point
temperature to 1100 K at pressures up to 800 MPa, J. Phys. Chem., 25, 1509-1596.
Yam, H., 2011, CO2 Rock Physics: A laboratory study. MSc. Thesis, Univ. of Alberta, Edmonton, 286 p.

Influence of cooling and heating rate on CO
2
condensation and
evaporation observed in a saturated synthetic rock sample
Kofman R.*, A. Rabbani, G. Njiekak, Department of Physics, University of Alberta
rkofman@ualberta.ca
and
D. R. Schmitt, Department of Physics, University of Alberta

Summary
The possibility of reducing carbon dioxide (CO
2
) emissions into atmosphere by capture and storage into
subsurface reservoirs has led to many different studies related to seismic monitoring of geologically
stored carbon to aid in the interpretation of field seismic observations, simultaneous ultrasonic
measurements of wave velocities are being conducted on a variety of porous materials. These
materials include synthetic sintered alumina, shales, sandstones, and carbonates. The results of such
measurements will improve the models obtained from related seismic surveys.

This paper focuses on the effects of CO
2
as a pore fluid on the overall rock seismic response.
Laboratory measurements using a fully CO
2
saturated synthetic sample have been conducted over a
variety of temperatures and pressures representative of subsurface conditions in which CO
2
can exist
in a gas, liquid, or supercritical state. Although full CO
2
saturation is not expected in most in situ cases,
where pore space will likely be shared with brine or hydrocarbons, this work provides an end member
understanding of the evolution of seismic responses resulting from changes in the phase state of CO
2
.
However, we have observed in these experiments that there is a delay in the temperature at which the
gas-to-liquid and liquid-to-gas phase transition occurs during isobaric heating and cooling experiments,
which appear controlled, in large part, by the rate of change of temperature and by whether the phase
transition is crossed from low to high temperatures or vice versa. The time to achieve equilibrium
suggested that thermal responses could in part be responsible for the delays. To further investigate this
observation a simplified model, the lumped capacity model, was implemented to estimate the time
required to reach a desired temperature

Introduction
Seismic velocities within a target formation change in response to the injection of CO
2
. These changes
reflect the influence of the injected CO
2
fluid on the related elastic moduli of the saturated target rock at
various pressures and temperatures. The velocity changes are evident in time-lapse vertical seismic
profiles (VSP) as strong reflectors appearing within the injection region (e.g. Zhou et al., 2010). To
better understand these changes, investigators measure the elastic properties of the typical target
rocks expected in such surveys using a variety of temperatures, heating and cooling rates, confining
and pore pressures, and assorted pore fluids. At present, this study builds on previous work done by
Yam (2011) and Njiekak et al (2013) where a large quantity of dynamic and static measurements have
been recorded using both natural and synthetic samples. In particular, recent modifications to vital
pieces of laboratory equipment have allowed for more thorough and efficient measurements of both
heating and cooling cycles, in addition extending CO
2
pore fluid pressurization experiments to lower
pressures and temperatures. During recent runs, it has become obvious that there are a number of
items that workers must be aware of when conducting these types of experiments. The remainder of
this abstract will focus primarily on the results of measuring a highly porous high-fired alumina sample
saturated with CO
2
.

Theory and Method

To investigate the effects of the CO
2
fluid, both strain and ultrasonic measurements have been
conducted on the synthetic sample. Strain is measured using a pair of single element 350 gauges.
Ultrasonic travel times are measured using 1 MHz compression-wave (P-wave) and shear-wave (S-
wave) transducers mounted on aluminum end caps; typically there is one P-wave and one S-wave
transducer stacked on each end cap. The sample is a 5.08 cm segment of porous ceramic (high fired
alumina) rod. The sample is sealed between the end caps and placed in pressure vessel filled with
hydraulic fluid. An ultrasonic pulser/receiver is used to send pulses to the set of transducers on the
pulser end cap, and receives the transmitted signal through the transducers on the receiver end cap,
under a variety of pressures and temperatures. These end caps are calibrated by performing
pressurization/depressurization runs at specified temperatures with no sample placed between them.
The recorded travel times are later subtracted from the actual sample data to remove the effects of the
end caps. Once assembled and placed in the pressure vessel, the sample is vacuumed for at least 12
hours after which several 'dry' pressurization/depressurization runs are completed at various
temperatures. The dry runs are completed to measure the response of the unsaturated rock.
Following the dry runs, several similar runs are completed while the sample was saturated with nitrogen

and later CO
2
and finally water, using an effective pressure of 25 MPa. The sample was vacuumed for
at least 12 hours prior to changing the fluid saturant. Experiments were set up such that velocities and
strains could be measure during the various CO
2
phase changes possible during well-site injection (e.g.
liquid to gas, liquid to supercritical) by varying related pressures and temperatures appropriately. A
more detailed explanation of the experimental procedure and equipment can be found in Yam (2011).
During the heating and cooling isobaric CO
2
run (Confining pressure = 31 MPa; Pore pressure = 6
MPa), it was observed that the rate of temperature change has a strong effect on the specific
temperature at which the expected phase change would occur. As the density of the saturant has a
strong effect on the elastic properties and, therefore, the observed velocities in the sample it is useful to
consider the properties of CO
2
for this particular run (Figure 1). Two possible explanations for this
disparity include: 1) the transport of CO
2
into (during condensation) and out of (during precipitation) the
sample during the phase transition; and 2) the time-dependent heat transfer out of (during
condensation) and into (during evaporation) the sample. As the pore-pressure is maintained, the heat
transfer within the sample, and between the sample and the surrounding fluid, seems most critical. In
particular, the temperature measured during each run is the temperature of the confining fluid,
measured within ~5 cm of the top of the sample assembly. As such, the recorded temperature will not
necessarily represent the temperature throughout the sample assembly; unless it is given time to
equilibrate. To determine an approximate equilibration time at each temperature, and help explain the
observed phase transition lag, a lumped capacitance method was applied (Wendl, 2005) to model the
approximate thermal transient state within the material upon an otherwise rapid external temperature
change in the confining fluid. This model assumes that temperature gradients within the materials are
negligible.


Figure 1: A diagram presenting the density of CO
2
as a function of temperature at constant pressure (6 MPa).
The phase change is clearly evident as a sharp drop in density. The liquid to vapor phase change occurs at ~22
0
C (NIST website).

Lumped capacitance method:

It is a relatively simple and special case where the temperature in the solid is spatially invariant, i.e.
temperature gradients are negligible, ( ) () ()
. This analysis, known as the

lumped capacitance model, assumes that the convective resistance between the boundary of the solid
and its surroundings is larger than the conductive resistance of energy transfer in the solid. The energy
balance equation can be written by considering only the convective energy transfer at the boundary of
the conductor.
The overall energy balance is:
(1)
where
(()
) is the rate of heat transfer into solid and the rate of increase of heat
energy of the solid is
. This yields the following:
()) (2)

and

Maximum temperature
Time
() Time dependent temperature
Density of the sample in
Specific heat capacity of the sample in

Heat transfer coefficient of fluid (oil) to solid in

Volume of the sample
Area of the sample

Radius of the cylindrical sample in meters (m)
Height of the sample in meters (m)
Equation (2) can be rewritten as:
()
) (3)
where
is the initial temperature of the conductor.

Therefore, temperature changes take the form of a simple exponential as spatial gradients are small
enough to be neglected.
()
(4)
Where:
Time constant
Thermal capacitance

Convection thermal resistance

From equation (4), the temperature of a solid as a function of time becomes:
()
(5)
Therefore, the time needed to heat a sample from
to a target temperature is:

(6)

and the total energy transferred over some period of time can be written as:

()
which gives

) (7)
where

The model holds if the Biot number is
( Thermal conductivity).

Sample calculation:
Here a sample, having a temperature of T
0
= 15 C, is placed in a bath having a temperature of 45 C.
Using this model we can calculate the time, t, required for the sample to reach a temperature T = 25 C
and determine the approximate amount of time required to reach equilibrium with the bath.
Given that:

where
is the characteristic length. The thermal resistance to heat transfer from the sample is due to
the dielectric layer (the tubing isolating the sample from the hydraulic fluid) and the convection
coefficient. That is

So the convection coefficient for the sample and tubing is

and the effective Biot number is

.
Since , the lumped capacitance model is applicable. Replacing h in equation (6) with , yields:

where:
V

Resulting in:

So the time required for the sample to reach 25 C is ~9.7 minutes. The time required for the sample to
reach equilibrium with the bath is roughly 2 hours (Figure 2).

Figure 2: A plot of the time required for a sample at 15 C to equilibrate with a surrounding fluid at 45 C based on
the lumped capacity model. The black asterisk represents the results of the sample calculation. The time required to
reach equilibrium is not dependant on the temperature differential. Note that this calculation is only intended to be a
rough estimation.

Examples
The following plots are included to provide some details observed with the synthetic sample (Figure 3
and 4). They also provide clear evidence of the lag observed during the liquid-to-gas and gas-to-liquid
CO
2
phase transition during a series of isobaric heating and cooling runs (Figure 4). For the sake of
brevity, only the P-wave data are provided. The first plot (Figure 3) illustrates the stiffness of the dry
sample. Essentially, there is very little change in the measured P-wave velocities or hysteresis outside
experimental uncertainty, clearly emphasizing the effect of CO
2
saturation. It is worth noting that in
both the heating and cooling runs, the flow-rate of the CO
2
into (during cooling) and out of (during
heating) increased significantly immediately prior to the observed velocity changes, but appeared to
remain relatively low and stable during the velocity changes.


Figure 3: The room-temperature dry run reveals that P-wave velocities do not change significantly with varying
confining pressure and constant temperature. The S-wave data provides similar results. The hysteresis between
the pressurization and depressurization cycles reflects the effect of pressure on the sample porosity.


Figure 4: The P-wave velocities for the CO
2
saturated sample as a function of temperature show the influence of
the rate of heating and cooling on the temperature at which the phase change occurs. It is clear that the gas-to-
liquid phase change (cooling) is not as strongly influenced by the cooling rate as the liquid-to-gas phase change
(heating). The heating runs show a significant variation in both slope and temperature. The approximate rates
are as follows: High Rate Heating: ~2.5 min/C; High Rate Cooling: ~5 min/C; Low Rate Heating: ~20 min/C;
Low Rate Cooling: ~20 min/C. The expected phase transition of pure, isolated CO
2
under these conditions is
~22 C (NIST website).

Conclusions
As efforts to sequester CO
2
in subsurface reservoirs increases, the ability to effectively monitor storage
sites during and following injection should be a high priority. In addition to other techniques, the ease
and affordability of deploying seismic arrays to monitor such sites over extended time-frames make it
an attractive option. To improve the results of seismic surveys, knowledge of the physical properties of
the reservoir rocks under in situ conditions is essential. Workers must continue to improve their ability
to measure these properties. These recent results suggest that care must be taken when conducting
ultrasonic measurements of CO
2
samples saturated when phase transitions are involved. Detailed
modeling can be an effective means to mitigate these issues and determine the optimum parameters at

which experiments should be conducted. And, while this paper deals specifically with a synthetic
sample, the principals can be instructive for future work with natural reservoir rocks.

Acknowledgements
This work was supported by the Carbon Management Canada Network Centre of Excellence and
NSERC.

References
The National Institute of Standards and Technology (NIST), 2011, Isobaric properties for carbon dioxide, U.S. Department of
Commerce, http://www.nist.gov/index.html
Njiekak, G., Schmitt, D. R., Yam, H. and Kofman, R.S., 2012, CO2 rock physics as part of the Weyburn-Midale geological
storage project: International Journal of Greenhouse Gas Control, accepted pending revision.
Wendl M. C., Fundamentals of Heat transfer theory and applications, class notes for ME 371, Departments of Mechanical
Engineering and School of Medicine, Washington University, Saint Louis, U.S.A., 2005.
Yam, H., CO2 Rock Physics: A laboratory Study: M.Sc. thesis, University of Alberta.
Zhou, R., L. Huang, J. T. Rutledge, M. Fehler, T. M. Daley, and E. L. Majer, 2010, Coda-wave interferometry analysis of time-
lapse VSP data for monitoring geological carbon sequestration: International Journal of Greenhouse Gas Control (4), 679686.

Seismic behaviour of CO
2
saturated Fontainebleau sandstone
under in situ conditions
Md Mizanul Huq Chowdhury*, University of Alberta, Edmonton, AB, Canada,
mhchowdh@ualberta.ca
and
Douglas R. Schmitt, University of Alberta, Edmonton, AB, Canada,
dschmitt@ualberta.ca

Summary
Understanding the seismic response of a rock in the CO
2
sequestration is important for the societal
acceptance of geological greenhouse gas sequestration and for monitoring of volcanic hazards.
Additionally, the study of the effect of CO
2
on seismic wave propagation is scientifically interesting
because CO
2
can exist in gas, liquid, and supercritical fluid phases over the modest temperature and
pressure ranges typically accessible in the upper 2 km of the earths crust, CO
2
s critical point lies near
31C and 7.4 MPa. We have carried out a series of ultrasonic pulse transmission experiments on
several samples of fully CO
2
saturated Fontainebleau sandstone over pore fluid pressure ranges of 1
MPa to 20 MPa and at two constant temperatures below (21C) and above (50C) the critical
temperature, these ranges were chosen to cross the gas-liquid and gas-supercritical transitions,
respectively.

Introduction
In a geological CO
2
sequestration project CO
2
leakage is a vital concern for which monitoring
and verifying the subsurface movement and phase behavior of the injected CO
2
is important to
ensure the storage integrity. Seismic methods are seemed to be a convenient way to monitor the
changes in subsurface in a CO
2
sequestration as seismic velocities are equally sensitive to a
rocks mineralogical composition, porosity and pore fluid contents. However, proper
interpretation of such field observations requires appropriate knowledge of the materials seismic
properties. Here we report on the simultaneous measurement of ultrasonic compressional and
shears wave velocities through saturated Fontainebleau sandstones. Fontainebleau sandstone
is collected from the Paris region, France. It shows a large porosity variation from 2%-28% and
consists of pure quartz (99.8%). Its pore geometry shows significant variation with porosity and
has a wide spectrum of microstructure. The main motivation of this work is to obtain an
understanding on the rock physics involved with CO
2
as pore fluid.

Physical and elastic properties of CO
2
2
are shown in the Fig. 1 as function of pressure
and temperature on the basis of Span and Wagners (1996) thermodynamic model. The critical point of
CO
2
0
C and at 7.4 MPa. Depending on the subsurface situations CO
2

can be either gas or liquid under this temperature and pressure. A sudden change in physical
properties of CO
2
clearly indicates the gas-liquid boundary in the diagram, which gradually vanishes as

the red point (critical point) reached. The supercritical fluid phase state starts just after that point. The
supercritical fluid phase has distinct property-it shows the physical behavior of gas and liquid in the
same time. This results a smooth transition of liquid-supercritical or gas-supercritical phases.

Fig 1: CO
2
phase diagrams as function of pressure and temperature according to the thermodynamic
model of Span and Wagner (1996). Fig 1(a) gives the bulk modulus and the fig 1(b) is for density phase
diagram of CO
2
.The critical point of CO
2
is clearly indicated by the red dot in both phase diagrams. The
gas-liquid boundary is easily noticeable because of the sudden change in physical properties. The
white dotted line gives the boundaries for the supercritical fluid phase. White arrows shows three cases
of our measurements as (i) gaseous to liquid, (ii) gaseous to supercritical fluid and (iii) liquid to
supercritical fluid transitions.

Gassmanns Rock-fluid interaction Model:
Gassmanns equation is one of the simple and widely used rock-fluid interaction models applicable for
seismic frequency region ~ 100 Hz. He considered the elementary elasticity of the pore fluid and
mineral grains of the sample that is saturated.
He also considered some assumptions to formulate his equation that are:
1) A microscopically homogenous and isotropic medium,
2) Similar bulk and shear moduli for all minerals that constitute the rock sample,
3) A zero fluid viscosity and a free movement of the fluids in pore regions which are interconnected,
4) Completely saturated pore space all times,
5) No interaction between pore fluid and rock minerals i.e. no change in rock's stiffness,
6) Quasi-static conditions are maintained to have frequencies low enough.
In this Gassmann's formulation the medium's saturated bulk modulus, K
sat
, has relation with frame
modulus, K
dry
, bulk modulus of the mineral grains, K
s
, bulk modulus of the fluid, K
f
, and the porosity of
the rock medium, , through the following:

2
2
(1 )
1
dry
s
sat dry
dry
f s s
K
K
K K
K
K K K

,
sat dry
(1.1)
Here
sat
&
dry
are the saturated and dry shear modulus of the rock frame. P- and S-wave velocities
than can be calculated using the following formulas:

4
3
sat sat
p
sat
K
V

,
sat
s
sat
V

, (1 )
sat s f
. (1.2)

Here
sat
is the saturated density and
f
is the fluid density.

Sample Properties
At this writing, two Fontainebleau samples have been used in our measurements. The table
summarized their properties and the Fig 2 shows the samples:

Fig 2:
Fontainebleau
samples.

Fig 3: left panel: SEM image. Right panel: micro CT image of the Fontainebleau sample.

Experimental setup & protocol
The ultrasonic pulse transmission method was used to determine
P- and S-wave velocities. This method conceptually involves
measuring the travel time of the ultrasonic wave travelling through
the sample. The experimental set up consists of several functional
units such as pulse generator, source/receiving transducers, a
digital oscilloscope, a pressure vessel that can apply confining
pressure up to 200 MPa, a fluid reservoir, and a thermocouple.
Fig 4 shows simplified schematic set-up of the experiment.

Fig 4: Schematic diagram of the set-up.

Properties Fontainebleau
1
Fontainebleau 2
Mass (gm) 117.1 167.2
Bulk Volume (cm
3
)
45.16 65.59
Grain Density (kg/m
3
)
2650 2650
Bulk Density (kg/m
3
)
2593 2549.16
Porosity (%) 10 12.5
Modal Pore size (m) 20 20
Quartz
Pores

Attaching PZTs and
putting in the vessel
A series of measurements are done in our experiment including dry runs where the sample is free from
any liquid or gas to obtain the dry frame moduli, Nitrogen runs to test the effective pressure, than
various CO
2
saturated runs and at the end H
2
O runs. The whole experimental protocol is shown in the
following flow chart:

In the flow chart PZT stands for Piezoelectric Transducer, P
d
for differential pressure and P
p
for pore
pressure.

Examples:
Dry runs: In these measurements the sample is free from any fluid. We did two dry run measurements
one in 21
0
C and another one with 50
0
C. Fig 4 shows the P-and S-wave velocity vs time plots of the dry
measurements.

Fig 4. Dry measurements for 21
0
C (left) and 50
0
C (right). As confining pressure increases the wave velocities
also increases. At a given confining pressure the wave velocities during up-cycle (pressurizing) is always lower
than down-cycle (depressurizing) showing wave velocity hysteresis. The wave velocities increase rapidly at lower
pressure but this increment diminishes at higher pressure because of number of closure of compliant pores
reduces at high pressure.
Nitrogen runs: After the dry runs, the Nitrogen runs are done to check the effective pressure. In these
measurements we noted a very less variations in velocities but still we saw some hysteresis in the data Fig 5
shows the nitrogen run measurements.

Fig 5. Nitrogen runs- left one
is for P-wave velocity and the
right one shows the S-wave
velocity plot.
Nitrogen runs
at 21
0
C and
P
d
=15 MPa
Dry runs at 21
0

and 50
0
C
Sample preparation
Vacuumed at 70
0
C
Vacuumed
overnight
CO
2
saturated runs:
1) Liquid to Gas (T=21
0
C and
P
d
=15 MPa)
2) Gas to Supercritical fluid
(T=50
0
C & P
d
=15 MPa).
3) Liquid to supercritical fluid
(P
p
= 8MPa & P
d
=15 MPa).

Vacuumed
overnight
H
2
O
Run

CO
2
runs: Three different CO
2
saturated runs were done to see the transition of 1) liquid to gas, 2) gas to
supercritical fluid and 3) liquid to supercritical fluid. Fig 6, 7 and 8 shows the three transitions respectively:

Fig 6: Liquid to Gas transition measurements. For P-wave (left) we see a clear 1.96% velocity drop around
their vapour pressure (P
p
=5-6 MPa) when the transition happens while for S-wave (right) we see a
significant velocity drop of ~3% on that pressure range. In the both cases the temperature was constant (T~ 21
0

C).

Fig 7: Gas to supercritical transition measurement. Both P-wave (left) and S-wave (right) velocities show a
gradual change near transition pressure (P
p
~78 MPa).

Fig 8: Liquid to supercritical fluid transition measurements. The transition is gradual around the
transition temperature ~31
0
C for both P-wave (left) and S-wave (right) case.

Comparison between observation and Gassmanns model prediction: Gassmanns equation is used
to predict the P- and S-wave velocities in the same range of temperatures and pressures and Fig. 9 shows the
comparison for all three cases. We can see that the trend for each case is similar but there is a difference in the

velocities. Gassmanns equations are applicable for low frequency (~100 Hz) measurement but we did our runs at
high frequency (~1 MHz) and all the assumptions of the model cannot be fulfill for a real measurements.

Fig. 9 : Comparison of the observed and the Gassmanns model prediction for all three cases of CO
2
saturated
runs.

Conclusions
We have measured P- and S-wave velocities for various pressure and temperature conditions in two
Fontainebleau sandstones to check the change in velocities during the phase transition. The velocities
change gradually across the gas-supercritical and liquid supercritical transitions in agreement with the
nature of these second order phase transitions. The gas-liquid transition is, conversely, a first order
transition discontinuous in CO
2
density and bulk modulus and explains the abrupt changes in a wave
speeds across the phase boundary. However in the real situation the rock may not only saturated with
CO
2
, there may be other pore fluid present there too to consider. The comparison between the
observed data and Gassmanns model predictions are also shown where we see a similar trend for the
all three cases for CO
2
saturation measurements.
Acknowledgements
Md Mizanul Huq Chowdhury would like to thank professor Schmitt groups research professional Randy
Kofman and the postdoctoral fellow Gautier Njiekak for their technical support in making this
experiment.

References
Yam, Helen, CO2 rock physics: a laboratory study, 2011,MSc thesis, Department of Physics, University of Alberta.
Span, R., and Wagner, W., 1996, A new equation of state for carbon dioxide covering the fluid region from triple-point
temperature to 1100K at pressure up to 800 MPa, J. Phys. Chem. Ref. Data, 25, 1509-1596.
Gassmann, F., 1951,Uber die elastizitat poroser medien:Vierteljahrsschrift der Naturforschenden Gesellschaft in Zurich,
96, 123.

Final ID: MR33A-2427
Investigations of Seismic Signatures of CO2 Saturation
for Geological Sequestration
G. Njiekak;
1
; R. S. Kofman;
1
; M. H. Chowdhury;
1
; D. R. Schmitt;
1
;
1. Physics, Institute for Geophysics, Univ. of Alberta, Edmonton, AB, Canada.

Body: The effectiveness of seismic monitoring will depend upon the ability to collect high-resolution seismic data,
since seismic methods have shown to be very effective in detecting time-lapse changes. As is well known, a typical
geological storage area for CO2 should be located at a depth greater than 1000 m where CO2 is in its supercritical
state (ScCO2). However, any change in the conditions of pressure and temperature of the storage area or any upward
migration of the ScCO2 towards the surface could lead to its transformation into either a gas or a liquid. To accurately
monitor the motion of the CO2 plume in such case or in situations where CO2 could be stored in shallowly deep areas
(e.g., the Missouri Carbon Sequestration Project), an understanding of the effects of the CO2s varying phase states
on the seismic waves will be required. However, to our knowledge, there is a general lack of experimental tests of
such effects. To address this issue, the current paper is focussing on petrophysical and rock physics testing on
carbonate samples with various textural patterns and porosity ranging from ~ 4% to 19%. An experimental procedure
that allowed for the deconvolution of the effects of CO2 phase states on the seismic responses was set up; it included
ultrasonic measurements of P- and S- waves under (i) vacuum conditions for the determination of 'dry' properties of
the studied samples, and (ii) saturated conditions; saturating fluids included nitrogen gas, CO2 and water. All the fluid-
saturated measurements were performed at a constant differential pressure of 15 MPa (difference between the
confining and the pore pressure) and were done as follows:
-first, measurements were conducted under inert nitrogen gas saturation; this allowed for assessment of any potential
effective stress variations that could occur in the studied sample. This was critical because the rock elastic properties
themselves depend nonlinearly on confining pressure.
-Second, measurements were performed with CO2 in the rock pore space under a variety of pore pressure and
temperature in order to sample the full range of CO2 phase states.
-Third, measurements were repeated with full water saturation.
It is worth noting that samples were put under vacuum for 6 to 8 hours before starting any set of saturated
measurement.
Of the numerous observations arising from these measurements, the followings are of particular note. First, both the
P- and S-wave speeds on the studied samples decrease substantially as the CO2 transforms from gas to either liquid
or supercritical phase in the rock pore spaces. This observation is consistent with the increase of CO2 fluid density
across these phase boundaries. Second, across the gas-liquid phase transition both wave speeds drop abruptly as
would be expected for the change in the physical properties of the CO2 across this first order phase boundary; the
gas-supercritical fluid transition is characterized by a more gradual variation of the wave velocities. Third, P-wave
velocity usually recovers after the phase transition; this suggests that CO2's bulk modulus dominates over its density
for the control of the P-wave behaviour at this stage.
Seismic behaviour of CO2 saturated Fontainebleau sandstones under in situ conditions.
M. H. Chowdhury;
1
; D. R. Schmitt;
1
; G. Njiekak;
1
; R. S. Kofman;
1
; H. Yam;
1
;
1. University, edmonton, AB, Canada.

Body: Understanding the seismic response of a rock in the CO2 sequestration is important for the societal acceptance
of geological greenhouse gas sequestration and for monitoring of volcanic hazards. Additionally, the study of the
effect of CO2 on seismic wave propagation is scientifically interesting because CO2 can exist in gas, liquid, and
supercritical fluid phases over the modest temperature and pressure ranges typically accessible in the upper 2 km of
the earths crust, CO2s critical point lies near 31' C and 7.4 MPa. We have carried out a series of ultrasonic pulse
transmission experiments on several samples of fully CO2 saturated Fontainebleau sandstone over pore fluid
pressure ranges of 1 MPa to 20 MPa and at two constant temperatures below (21' C) and above (50' C) the critical
temperature, these ranges were chosen to cross the gas-liquid and gas-supercritical transitions, respectively. The
porosity of the Fontainebleau samples is found to be in the range of 10-13% based on He pyncnometry and Hg
intrusion porosimetry. P- and S-wave velocities were determined from laboratory data and were plotted against the
different pore fluid pressures and temperatures to check the behaviour of the sandstones in those situations. The
measurements were all acquired at a constant effective (differential) pressure of 15 MPa in order to minimize the
otherwise significant pressure dependent velocity changes. Across the liquid to gas transition (at 21' C) between 5-6
MPa we observed a 1.9-3.9% and a 3.14-3.5% velocity drop in P- and S-waves, respectively. A more subtle (<1.2%)
drop in the P-wave velocity occurs across the gas-supercritical transition (at 50' C), but no discontinuity appears in the
S-wave. Other tests conducted at constant pore fluid pressure with changing temperatures crossed the liquid-
supercritical boundary. In general, the velocities change gradually across the gas-supercritical and the liquid-
supercritical transitions in agreement with the nature of these second-order phase transitions. The gas-liquid transition
is, conversely, a first order transition discontinuous in CO2 density and bulk modulus and explains the abrupt changes
in wave speeds across this phase boundary. When considered with the fluid density, these changed amplify the
differences in the saturated materials acoustic impedance and suggest that the saturation state will be an important
aspect to seismic monitoring.
A Proposed Protocol for Evaluating the Seismic Properties of CO
2
Saturated
Rocks: Experiences Gained from the Weyburn, Saskatchewan Sequestration
Project

Douglas R. Schmitt
1
,
Gautier Njiekak
1
, Helen Yam
1
, Randolph Kofman
1
, Mizan Chowdhury
1

1
Institute for Geophysical Research, Dept. of Physics, University of
Alberta, Edmonton, Alberta, Canada

Summary

The current study provides an extensive series of measurements that aims at predicting the seismic
behaviour of CO
2
-saturated rocks. This has implications on monitoring CO
2
geological storage using
seismic methods. The ultrasonic pulse transmission technique is the underlying methodology used in this
work. Unlike previous rock physics studies involving CO
2
as pore fluid, care is taken to separate pore
fluid effects from pore pressure build-up effects during the experiments. This allows for the sampling of
the effects of the CO
2
s varying phase states (gas liquid - supercritical fluid) on the overall rock seismic
response. Measurements were carried out on core samples from the Weyburn-Midale oilfield,
Saskatchewan. Of the observations arising from the measurements done in this study, there are two that
are of particular note. First, both the P- and S-wave speeds decrease substantially as the CO
2
transforms
from gas to either liquid or supercritical phase. This observation is consistent with the increase of CO
2

fluid density across these phase boundaries. Second, across the gas-liquid phase transition both wave
speeds drop abruptly as would be expected for the change in the physical properties of the CO
2
across this
first order phase boundary. In contrast, across the gas-supercritical phase boundary the velocities change
more gradually. This suggests that it may be difficult to distinguish the gas-supercritical boundary using
seismic reflection techniques.

Keywords: CO
2
sequestration; seismic monitoring; rock physics; ultrasonic velocity; Weyburn reservoir.

1. INTRODUCTION

As is well known, societal acceptance of CO
2

geological sequestration depends critically on the
ability to monitor movement of CO
2

particularly the use of 4-D reflection seismology, will be a key component in monitoring motions of CO
2

underground (e.g., Carcione et al., 2006; Ivanova et al., 2012). As such, there is a great need for a
good understanding of how CO
2

will influence the seismic properties of rocks. This is doubly
interesting in that free CO
2

at depths intended for CO
2
sequestration can be in gas, liquid or
supercritical phase states. However, to our knowledge, there is a general lack of experimental tests
of such effects. Here we propose a laboratory protocol for the measurement of P- and S-wave speeds
and attenuations through natural rocks saturated with CO
2
. This protocol was developed on the basis of
experiences gained during an extensive series of laboratory measurements of P- and S-wave velocities
of carbonate reservoir and cap-rock core samples from the Weyburn, Saskatchewan, project and
some test sandstones.

2. EXPERIMENTAL PROTOCOL

The protocol includes two important components. First, the samples chosen undergo a rigorous
material characterization that includes He and Hg porosimetry, XRF whole rock chemistry, thin
sectioning, SEM scanning, and micro-CT imagery. This allows for an understanding of the mineralogy,
the porosity, and the architecture of the pore space within the rock. The more crucial set of tests,
however, are those in which the rock sample is subject to a large variety of confining and pore
pressures, saturation states, and temperatures. A given test begins with coring of a cylinder of the
given sample. The ends of this cylinder are ground as parallel to one another as possible and the
sample is then dried under vacuum at modest temperatures (~70 C). The sample is then prepared for
the high-pressure measurement suite by attaching a set of specially constructed ultrasonic transducers
to its ends and then hermetically sealing this. This assembly is placed in a pressure vessel. The P-
and S-wave measurement protocol then follows by
i) Obtaining the P- and S-waveforms under a series of increasing confining pressures but with the pore
space subject to vacuum to provide the dry properties, this is carried out at different temperatures
to evaluate the effect of temperature on the rock frame.
ii) Repeating these measurements with the pore space saturated with inert nitrogen gas. An
additional suite here tests carried out at constant differential pressure (difference between the
confining and the pore pressure) allows for assessment of any potential effective stress variations that
could occur in the sample. This is critical because the rock elastic properties themselves depend
nonlinearly on confining pressure.
iii) Repeating these measurements with full CO
2

gas saturation under a variety of pore pressures and
temperatures in order to sample the full range of CO
2
phase states. These measurements are all carried
out at constant differential pressure. This involves at least 5 different suites of measurements carried out
under conditions of constant temperature or constant pore pressure (Fig. 1).
iv) Repeating the dry measurements once the suite of CO
2

tests are completed, and
v) Finally carrying out the measurements under full saturation with liquid water.
These series of measurements provides a great deal of information that allows for prediction of the
seismic behaviour of the given rock, but at the cost that a given run requires upwards of a week of
dedicated time in the measurement system.

2.1. Ultrasonic measurements: experimental setup

The experimental setup for the ultrasonic measurements includes several functional units such as a pulse
generator, a digital oscilloscope (NI USB- 5133 100 MS/s), source/receiver transducers, a cylindrical
pressure vessel, fluid tanks, a thermocouple, and an electrical resistance tape (Fig. 2, see also Yam,
2011). The transducer is exited by a fast-rising, 200V square wave, sent by the pulse generator
(Panametrics, model 5800 PR). The generated elastic wave after propagating through the sample is
recorded by a digital oscilloscope at a sampling interval of 10 nanoseconds. The recorded waveform is the
stack of at least 500 traces to reduce the noise level in the signal. The desired pore fluid is introduced into
the sample via stainless steel tubing that connects the pore space of the sample to the pore fluid reservoir
located outside of the vessel through the vessel lid. The confining and pore pressure systems are
independent of each other such that different pressure conditions can be applied in irrespective of each
other by using different pumps located outside of the pressure vessel. Both the confining pressure
(pressure vessel) and the pore pressure systems are capable of reaching a maximum pressure of 70 MPa
from the pumps. Higher confining pressures, up to 200 MPa, could also be achieved by using a separate
air pump connected to the pressure vessel.

Fig. 1. CO
2
properties (bulk modulus and density) as a function of pressure and temperature based on the thermodynamic model
of Span and Wagner (1996). C =critical point. Arrows (i), (ii), (iii), (iv), and (v) refer to the conditions under which CO
2
-
saturated rock velocities are measured. For (i), (ii), (iii) and (iv), measurements are done under constant temperature. (v)* refers
to measurements done under constant pore pressure (about 11 MPa in this case) and T =50 C.


3. CASE STUDY: DOLOSTONE

3.1. Sample description

The studied sample is a dolostone from the Weyburn area, Saskatchewan, with a (mercury injection)
porosity of 14% and air permeability lower than 2 mD. The rock matrix consists of relatively loosely
packed crystals (largely dolomite). Small calcite grains and feldspar flakes are disseminated within the
matrix. Quartz is sometimes visible. The porosity is generally intercrystalline and the main pore throat
size is around 0.65 m. The sample grain density is 2.83 g/cm
3
. The ultrasonic measurements were
performed on a cylindrical plug, 3.81 cmin diameter and 4.8 cmlong (Fig. 3).

a)

b)

Fig. 3. (a) Photograph of the dolostone's plug and (b) a scanning electron microscope (SEM) image of the sample. Note the
loosely compacted dolomite crystals in the SEM image.

a)
P-wave
4000
4100
4200
4300
4400
4500
4600
4700
4800
0 10 20 30 40
Confi ni ng Pressure (MPa)
V
e
l
o
c
i
t
y

(
m
/
s
)
23 C, Dry Sample
50C, Dry Sample
S-wave
2550
2600
2650
2700
2750
2800
0 10 20 30 40
V
e
l
o
c
i
t
y

(
m
/
s
)
23 C, Dry Sample
50 C, Dry Sample

b)
P-wave
4100
4200
4300
4400
4500
4600
4700
4800
0 10 20 30 40
V
e
l
o
c
i
t
y

(
m
/
s
)
Dry Sample at 23 C
Pd = 15 MPa
S-wave
2600
2650
2700
2750
2800
0 10 20 30 40
V
e
l
o
c
i
t
y

(
m
/
s
)
Dry Sample at 23 C
Pd = 15 MPa

Fig. 4. (a) Dry and (b) nitrogen-saturated P-and S-wave velocities as a function of confining pressure for the dolostone
sample. Error bars are shown. In (a), velocities recorded at 23C and at 50C are shown. The hysteresis of the dry velocity
curves can be explained by the difference between the closing and opening rate of the crack-like pores at a given pressure
during pressurization and depressurization cycle, respectively (Sharma and Tutuncu, 1994). Pd =differential pressure.

Fig. 5. Dry and CO2-saturated waveforms on the dolostone sample. (a): P- and S-waveforms on the dry sample. Measurements
were done during up- and down pressurization cycles. Note how P- and S- wave travel times become faster with increasing
confining pressure. (b): Waveforms that were acquired on the CO
2
-saturated sample at T =23 C; CO
2
was turning fromgas to
liquid in the pore space. (c): CO
2
was turning fromgas to a supercritical fluid in the pores; T =50 C. Note the decrease of the S-
wave travel times and the different behaviours of the P-wave speeds (decrease at low pore pressureand slight increase at higher
pore pressure) fromthe CO
2

Fig. 6. CO
2
saturated P-and S-wave velocities of the dolostone sample for the two constant temperature runs at T =23C and T =
50C while under a constant differential pressure of 15 MPa. Error bars are shown. Dry wave velocities obtained under a
confining pressure of 15 MPa are plotted at pore pressure =0 MPa to provide a reference for the changes seen. Black arrows
indicate velocities when CO2 is gas, liquid and a supercritical fluid in the pore spaces.

3.2. Ultrasonic measurements

Under dry (vacuum) conditions:
On the dry sample, and as expected, the compressional (P) and shear (S) wave velocities increase with
confining pressure and decrease with temperature (Fig. 4a). Figure 5a shows the P- and S-waveforms
recorded on the dry sample.

Under fluid-saturated conditions:
Only nitrogen- and CO
2
-saturated measurements are provided in this sub-section. The measurements
conditions were set up so that any variation in the recorded waveforms was due to change in the pore
fluid physical properties. Thus, for each sample, a constant differential pressure of 15 MPa was
maintained by varying the confining pressure accordingly to the pore pressure. Assuming that the equal
amount of pore pressure increase cancelled the equal amount of confining pressure increase, any
waveform variation observed was solely caused by pore fluid effects.
Nitrogen-saturated measurements: Prior to CO
2
-saturated measurements, each sample was subjected to
constant differential pressure measurement of 15 MPa under inert nitrogen gas saturation. In general
within a fluid, the bulk modulus and density are the physical properties that will influence wave
propagation; bulk modulus has a direct proportional effect while density has an inverse proportional effect
on wave velocities. As change in the nitrogen properties (density, bulk modulus) only slightly affects
sound waves travel times under the range of pressure and temperature conditions used in this study
(Lemmon et al., 2011), no substantial change in the P- and S- wave velocities meant that the applied
differential pressure was the main factor controlling the variation of the velocity. This behaviour of the
wave velocities did lend confidence to selecting the set of high-quality measurements that were suitable
for the purpose of our study. Figure 4b shows that P- and S-wave velocities remained constant during the
nitrogen-saturated measurements.
CO
2
-saturated measurements: Results presented here are from measurements made at constant
temperature (23 C and 50 C) with the pore pressure varying from 1 MPa to 25 MPa in each case. As
pore pressure increases, for the lower temperature runs (23C), CO
2
changed from a gas phase to a liquid
phase in the pore spaces while for the higher temperature runs (50C), CO
2
changed from a gas phase to
Comment [DRS1]: You need to say a
bit more about the effective pressure being
constant here not sure this really said too
much.
the supercritical fluid phase (See arrows (i) and (ii) in Fig. 1). The wave velocities of the two constant
temperature runs are plotted in figure 6. On these plots, velocities of the dry sample at a differential
pressure of 15 MPa are also shown. P- and S-waveforms of the CO
2
-saturated sample are shown in
figures 5b and 5c.
Once CO
2
gas was injected into the sample, irrespective of the temperatures, both P-and S-wave
velocities decreased with the pore pressure. The drop of the P- and S-wave velocities was more
immediate for the low temperature run; the large drop occurred at a pore pressure of about 5 to 7 MPa and
10 to 15 MPa for the low temperature (T=23 C) and high temperature (T =50 C) runs, respectively. At
23 C, the immediate nature of the wave responses was likely due to the sharp contrast in the density of
the CO
2
when the gas-liquid boundary was crossed. Overall, the decrease of the P- and S- waves travel
times right before the gas-liquid and gas-supercritical fluid transitions were up to 3%. Over the phase
transition, P wave velocity recovered while S-wave speed kept decreasing. The recovery of the P-wave
velocity was gradual and was up to 1 % at a pore pressure of 25 MPa. The total drop of the S-wave travel
times was up to 4 % over the entire 1 to 25 MPa pore pressure interval.

4. CONCLUSIONS
An experimental procedure has been set up to inspect the effects of CO
2
on the seismic properties of
rocks. The series of measurements allows for the assessment of the effects of the CO
2
phase changes (gas
to liquid and gas to supercritical fluid) on the overall rock seismic behaviour. CO
2
phase changes (gas to
liquid and gas to supercritical fluid) are marked by a drop in velocities of up to 4% over the entire 1 to 25
MPa pore pressure interval used in this study. The abruptness in the velocity change differs significantly
between the gas-liquid and gas-supercritical fluid transitions; and the gradual variations seen for the latter
suggest that this will be difficult to detect using seismic reflection methods. Real situations sometimes
differ from the experimental conditions set up in this study as CO
2
may coexist with other in-situ fluids in
the pore space. However, this study provides an end member understanding of the possibilities with CO
2

in the pore space, which is critical for the setup of an accurate monitoring of CO
2
plume in the
subsurface. In fact, any change in the conditions of pressure and temperature of a CO
2
storage area or any
upward migration of the CO
2
towards the surface could lead to the transformation of CO
2
into either a
gas or a liquid or a supercritical fluid.

ACKNOWLEDGMENTS

The protocols were developed here under the support of the IEA Weyburn-Midale CO
2
Monitoring and
Storage Project for Njiekak and Kofman with data analysis funded by research grants from the Carbon
Management Canada project to Schmitt. The development of the CO
2
measurement system was
supported by NSERC and the Canada Research Chair program. Yam was supported in part by an NSERC
CGS. The technical assistance of Len Tober and Lucas Duerksen was invaluable.

REFERENCES

Seismic Behavior of CO
2
Saturated Fontainebleau
Sandstone Under In Situ Conditions

Md. Mizanul Huq Chowdhury*
University of Alberta, Edmonton, Alberta, Canada
mhchowdh@ualberta.ca
and
Douglas R. Schmitt
Department of Physics, Edmonton, Alberta, Canada

Summary:

Understanding the seismic response of a rock in the CO
2
sequestration is important for
the societal acceptance of this process. We have done an ultrasonic pulse transmission
experiments on a fully CO
2
saturated porous Fontainebleau sandstone within a range of
pressures and temperatures which crosses the boundaries of gas, liquid and
supercritical phases of the CO
2
. This situation is most likely in the uppermost kilometer
of most sedimentary basins. P- and S-wave velocities were determined from laboratory
data and were plotted against the different pressures and temperatures to check the
behavior of the sandstones in those situations.

Introduction:

In a geological CO
2
sequestration project, CO
2
leakage from the reservoir is a vital
concern for which monitoring and verifying the subsurface movement and phase
behavior of the injected CO
2
is important to ensure the storage integrity. Seismic
methods are perhaps the best way to monitor the changes in subsurface in a CO
2
sequestration project as seismic velocities are equally sensitive to a rocks mineralogical
composition, porosity and pore fluid contents. This work gives simultaneous
measurements of ultrasonic compressional and shears wave velocities on Fontainebleau
sandstones. Fontainebleau sandstones (collected from Paris region, France) are clay
free, very pure (99.8% quartz) with a significant variation in their porosity, permeability
and state of cracking. All these properties make them good candidates for the test. The
main motivation of this work is to get a good understanding on the rock physics involved
with CO
2
as pore fluid.

Physical and elastic properties of CO2:

2
are shown in the Fig. 1 as
function of pressure and temperature on the basis of Span and Wagners (1996)
thermodynamic model. The critical point of CO
2
0
C and at
7.4 MPa. Depending on the subsurface situations CO
2
can be either gas or liquid under
this temperature and pressure. A sudden change in physical properties of CO
2
clearly
indicates the gas-liquid boundary in the diagram, which gradually vanishes as the red
point (critical point) reached. The supercritical fluid phase state starts just after that point.
The supercritical fluid phase has distinct property-it shows the physical behavior of gas
and liquid in the same time. This results a smooth transition of liquid-supercritical or gas-
supercritical phases.

Fig1: CO
2
phase diagrams as function of pressure and temperature according to the
thermodynamic model of Span and Wagner (1996). The left panel gives the bulk
modulus and the right panel is for density phase diagram of CO
2
.The critical point of
CO
2
is clearly indicated by the red dot in both phase diagrams. The gas-liquid boundary
is easily noticeable because of the sudden change in physical properties. The white
dotted line gives the boundaries for the supercritical fluid phase. (Figure after Yam,
2011).

Experiment:

Experiments are carried out on a Fontainebleau cylindrical shaped sample of 4.2 cm
long and 3.7 cm diameter, cut from a block of the sandstone collected from the Ile de
France region near Paris. This sample has a 10% porosity (measured by He
porosimeter). It shows a significant variation in the pore geometry that makes it a
suitable candidate to observe the changes of pores in rocks under seismic conditions.

Fig. 2: Fontainebleau sandstone sample (porosity 10%).

Ultrasonic pulse transmission method was used to find out the elastic behavior of the
sample. The experimental set-up is shown is the Fig.3.

Fig. 3: A simplified schematic diagram of the experimental setup.

The pulse transmission method involves the travel time measurement through the
sample. The experimental set up consists of a number of functional units. There is a
pulse generator, source/receiving transducers, a digital oscilloscope (NI USB-5122
100MS/s), a pressure vessel that can apply confining pressure up to 200 MPa, a fluid
reservoir and a thermocouple as shown in Fig.3. The sample was placed between the
transducers made of P-and S-piezoelectric ceramics with center frequencies of 1MHz
mounted on an aluminum buffer cap. A Tygon (a flexible clear plastic) tube is used to
jacket the sample for sealing it from hydraulic oil contamination.

We carried out several ultrasonic measurements on the sample under dry and CO
2

saturated conditions. During the dry runs the pore space was under vacuum (P
p
=0
MPa) while the confining pressure varies from 2 MPa to 55 MPa. We did several dry
runs to check the repeatability of the tests. Unfortunately, repeated cycling of the
confining pressure up and down displays a good deal of hysteresis of the observed
velocities with pressure that may be due to closing of pores and micro cracks in the
sample. In the case of the CO
2
saturated measurements, 2 different temperature runs
were done while pore pressure varied from 2 MPa to 25 MPa. The 2 constant
temperatures are ~23
0
C (room temperature) and 50
0
C. A 15 MPa differential pressure
is maintained in the each measurements of constant temperature by changing the
confining pressure according with the pore pressure. Another 2 measurements are
carried out at constant pore pressure (P
p
~7 MPa) while temperature is changed from
50
0
C to room temperature (~23
0
C). From the phase diagram (Fig. 1) we can see for
the constant room temperature (~23
0
C) runs where pore pressure increases illustrates a
transition from a gas phase to liquid phase while the constant pore pressure and
changing the temperature from higher (50
0
C) to room temperature gives the transition
from Gas to liquid too. In the case for constant 50
0
C while changing the pore pressure
shows the changes from gas phase to supercritical fluid phase.

Examples:

Figure 4 shows the P and S-wave velocities as a function of confining pressure in dry
runs that on the sample from 2 MPa to 55 MPa pressure at room temperature. We can
see some variation in the velocities between each dry run, which occurred due to the
closing of micro cracks in the sample.

1800
2400
3000
3600
4200
4800
5400
6000
0 10 20 30 40 50 60
V
e
l
o
c
i
t
y

(
m
/
s
)

P-Wave - FB Dry Summery
Up-Cycle 1
Up-Cycle 2
Down-Cycle 1
Down-Cycle 2

Figure.4: Ultrasonic P-wave and S-wave (up and down respectively) velocities measured
when sample is not saturated with CO
2
.

Fig. 5: Ultrasonic P-and S-wave velocities measured on the CO
2
saturated sample with
15MPa differential pressure during the constant low temperature (~room temperature).

In Fig 5. We can see the transition of gas to liquid phase with constant room temperature
and changing pore pressure. We can see the changes happening in the P
p
~7.4 MPa and
the velocity become significantly low after the transition happened.

1500
1900
2300
2700
3100
3500
0 10 20 30 40 50 60
V
e
l
o
c
i
t
y

(
m
/
s
)

S-Wave - FB Dry Summery
Up-Cycle 1
Up-Cycle 2
Down-Cycle
1
Down-Cycle
2
5120
5160
5200
5240
5280
5320
0 5 10 15 20 25 30
V
e
l
o
c
i
t
y
,

m
/
s

Pore Pressure, MPa
P-Wave
3250
3270
3290
3310
3330
3350
3370
0 10 20 30
V
e
l
o
c
i
t
y

(
m
/
s
)

Pore Pressure (MPa)
S-Wave

Fig.6: Ultrasonic P-and S-wave velocities measured on the CO
2
saturated sample with
constant differential pressure 25MPa and changing the temperature to get the liquid to
supercritical phase transition.

Fig.6 shows the transition of liquid to supercritical phase transition. The P-wave velocity-
temperature graph is not giving much informatio but in S-wave profile we can see a clear
transition happening in ~31
0
C.

Conclusions:

We have measured P-and S-wave velocities for various pressures and temperatures
conditions in a porous Fontainebleau sample. Some variation in elastic wave velocities
during phase transitions of CO
2
is noted. From these results we can see a sensitiveness
of elastic waves with the change of pore pressure and CO
2
phase transitions. However
in the real situation the rock may not only saturated with CO
2
, there may be other pore
fluid present there too.

Acknowledgements:
Mizan Chowdhury would like to thank professor Schmitts Lab technician Randy Kofman
and postdoctoral fellow Dr. Gautier Njiekak for their technical support in making this
experiment.

References:

Yam, Helen, CO
2
rock physics: a laboratory study, 2011,MSc thesis, Department of Physics, University of
Alberta.

Span, R., and Wagner, W., 1996, A new equation of state for carbon dioxide covering the fluid region from
triple-point temperature to 1100K at pressure up to 800 MPa, J . Phys. Chem. Ref. Data, 25, 1509-1596.

Benson, S. M., 2006, Monitoring Carbon dioxide sequestration in deep geological formation for inventory
verification and carbon credits, 2006 SPE Annual Technical Conference and Exhibition in San Antonio, USA,
24-27 September 2006.
5180
5190
5200
5210
5220
5230
20 25 30 35 40 45 50 55
V
e
l
o
c
i
t
y

(
m
/
s
)

Temperature, 'C
P-Wave
3000
3050
3100
3150
3200
20 30 40 50 60
V
e
l
o
c
i
t
y

(
m
/
s
)

Temperature, ' C
S-Wave
Improved understanding of CO
2
Rock physics: a laboratory study

Regular Paper

G. Njiekak
1
, H. Yam
1
, R. S. Kofman
1
, M. Chowdhury
1
, D. R. Schmitt
1
1. Department of Physics, Institute for Geophysical Research, Univ. of Alberta, Edmonton, AB,
Canada, Email: njiekak@ualberta.ca

To get quantitative estimates of CO
2
saturation for seismic monitoring purposes requires a rock
physics model that can properly handle effects of CO
2
and other in-situ fluids (e.g. oil and brine)
on seismic waves. However, today, the small number of laboratory core measurement involving
CO
2
saturation (e.g., Shi et al., 2007; Lei and Xue, 2009; Purcell et al., 2010) limits the building
of a robust theoretical framework to support seismic monitoring surveys. Quantitative
assessment of injected CO
2
remains one of the main challenges for seismic monitoring studies.
In this work, we focus on the effect of lone CO
2
on elastic properties of rocks, investigating the
changes of waves velocities and densities in response to CO
2
's varying phase state (gas-liquid-
supercritical fluid). Real situations may sometimes differ from the experimental conditions set up
in this study as CO
2
may coexist with other in-situ fluids in the pore space. However, this work
provides an end member understanding of the possibilities with CO
2
in the pore space.
Experiments were conducted under a variety of temperatures and (confining and pore fluid)
pressures representative of conditions expected in volcanic edifices and geological sequestration
projects.

References:
Lei, X., and Xue, Z., 2009, Ultrasonic velocity and attenuation during CO2 injection into water-saturated
porous sandstones: Measurements using difference seismic tomography: Physics of the Earth and
Planetary Interiors, 176, 224-234

Purcell, C., Mur, A., Soong, Y., Mclendon, R., Haljasmaa, I. V., and Harbert, W., 2010, Integrating
velocity measurements in a reservoir rock sample from the SACROC unit with an AVO proxy for
subsurface supercritical CO2: The Leading Edge, 29, 192-195.

Shi, J ., Xue, Z., and Durucan, S., 2007, Seismic monitoring and modelling of supercritical CO2 injection
into a water-saturated sandstone: Interpretation of P-wave velocity data: International J ournal of
Greenhouse Gas Control, 1, 473-480.

G. Njiekak: PhD (Alberta); Postdoctoral Researcher, University of Alberta
H. Yam: M.Sc. (Alberta); University of Alberta
R. S. Kofman: M.Sc. (Alberta); Research Assistant, University of Alberta
M. Chowdhury: M.Sc. (Alberta); Graduate Student, University of Alberta
D.R. Schmitt: PhD (Alberta); Professor, University of Alberta

Oral presentation

E1: Geophysical Applications in CO
2
Storage
Regular Paper
Experimental Measurements Of Seismic Wave Speeds And Attenuation In CO2 Saturated Porous Rocks
G. Njiekak;
1
; H. Yam;
1
; R. S. Kofman;
1
; M. Chowdhury;
1
; D. R. Schmitt;
1
;
1. Department of Physics, Institute for Geophysical Research, Univ. of Alberta, Edmonton, AB, Canada.

Body: Due to the sensitivity of seismic waves to pore fluid contents, time lapse seismology is regarded as a promising
monitoring method for geological CO2 sequestration projects and is employed in all industrial scale projects (Sleipner,
Weyburn, In Salah). Therefore, understanding the effect of CO2 as a pore fluid on the overall rock seismic response is
critical, and it is particularly interesting as CO2 can be in gas, liquid, or supercritical phases even at the relatively
modest pore pressures and temperatures in the uppermost kilometer of the earth's crust. To address this issue,
ultrasonic P- and S-wave pulse transmission experiments were carried out on fully CO2 saturated samples of a
synthetic porous ceramic, Berea and Fontainebleau sandstones, and carbonates under a variety of temperatures and
pressures representative of conditions expected in volcanic edifices and geological sequestration projects. The
synthetic sample was chosen because of its lack of microcracks, which can complicate the acoustic behavior of real
rocks. Although this sample is extremely porous (58%) and is not reflective of real reservoir rocks, its large porosity
allows the overall rock behavior to be more susceptible to the changes in the physical properties of the pore fluid; this
could provide an extreme end member understanding on the rock physics involved with CO2 as the pore fluid.
Laboratory results show waveform variations (velocity, amplitude, attenuation) in response to CO2's varying phase
state. For the ceramic rod, CO2 phase changes (gas to liquid and gas to supercritical fluid) are marked by a drop in
velocities of 4-5% likely due to the increased density of the liquid or the supercritical fluid relative to the gas. Wave
attenuation increases with pore pressure and with frequency. The measured elastic wave velocities showed good
agreement with Biot's model in this highly porous sample. The real sandstones, in contrast, display more complicated
behaviour at the point of the phase transition. The abruptness in the velocity change differs significantly between the
gas-liquid and gas-supercritical fluid transitions; and the gradual variations seen for the latter suggest that this will be
difficult to detect using seismic reflection methods.
Simultaneous ultrasonic measurements of wave velocities under conditions of
CO
2
saturation

Helen Yam, Gautier jiekak
1
, Randy Kofman, Douglas R. Schmitt
Institute for Geophysical Research, Department of Physics, University of Alberta
1
Email: njiekak@ualberta.ca

Introduction

Simultaneous ultrasonic measurements of wave velocities are being carried out on a number of
materials including synthetic samples, shales, sandstones, and carbonates. These studies aim at
improving our knowledge on the variation of the velocity as a function of confining and pore
pressures, temperature, and different saturation conditions. The present paper dealt with studies of the
effect of CO2 as a pore fluid on the overall rock seismic response. Laboratory work using a fully
saturated synthetic sample has been conducted for a variety of temperatures and pressures that are
representative of subsurface conditions in which CO2 can be in gas, liquid, supercritical fluid phases
(figure 1). This is particularly important in seismic monitoring as CO2 transformations can occur at
quite modest temperatures and pore pressures; all three phases are readily obtained within the
uppermost kilometer of a sedimentary basin. Although full CO2 saturation is not expected as in real
cases in the earth the pore space would also contain brine or hydrocarbons, this work provides an end
member understanding in the evolution of seismic responses resulting from changes in the phase state
of CO2.

Experiment

A porous ceramic rod, derived from a high fired alumina body (Al2O3) was used as the host sample in
this study. This sample has a porosity of 58%, with an air permeability of about 100mD. This synthetic
sample was chosen for this study because of its lack of microcracks, which can complicate the acoustic
behavior of real rocks. Although this sample is extremely porous and is not reflective of real reservoir
rocks, its large porosity will allow the overall rock behavior to be more susceptible to the changes in
the physical properties of the pore fluid. This reinforces the purpose of our work on establishing an
extreme end member understanding on the rock physics involved with CO2 as the pore fluid.
Ultrasonic pulse transmission method was applied to determine the elastic behavior of the sample. A
cylindrical sample 50 mm in length and 25.4 mm in diameter was placed in between a transmitting and
a receiving transducer made of P- and S- piezoelectric ceramics with center frequencies of 1MHz
Figure 1: Bulk modulus (left) and
density (right) phase diagrams of CO
2
as
a function of pressure and temperature
based on the thermodynamic model of
Span and Wagner (1996). The red dot
on both phase diagrams marks the
critical point of CO
2
(T = 31 C and P =
7.4 MPa). The vapor-liquid boundary is
clearly discernible from the sudden
change in physical properties. The
boundaries of the supercritical fluid
phase state are marked by the white
dotted line. The white arrows shows the
temperature and pressure conditions
applied to the CO
2
during ultrasonic
pulse transmission measurements.
mounted on an aluminum buffer cap. The sample-transducer assembly was set inside a pressure vessel
that was filled with hydraulic oil as the hydrostatic pressure medium. The sample was jacketed in a
Tygon tube, a flexible clear plastic to seal it from hydraulic oil contamination. CO2 is introduced into
the initially dry and vacuumed sample from pressure bottle of hospital grade CO2 located outside the
vessel. An electrical resistance tape is wrapped around the outside of the vessel and controls the
temperature of the experiment and is measured by K-type thermocouple placed inside the pressure
vessel.
The transmitted signal is generated by exciting the transmitting transducer with a fast-rising 200 V
square wave from a pulse generator. The propagated signal was recorded by a digital oscilloscope
system at a sampling rate of 10 nanoseconds. The final waveform is a stack of over 500 traces to
reduce random noise effects. From the final waveforms the transit time of the signals can be deduced,
and with the length of the sample the elastic wave velocities can be calculated.
Ultrasonic measurements were made on the sample under dry and CO2 saturated conditions. The dry
measurements were done with the pore space under vacuum (Pp =0 MPa) while the confining pressure
varied from 2 MPa to 40 MPa. The resulting elastic wave velocities showed no differential pressure
dependence due to the lack of microcracks. For the CO2 saturated measurements, 4 different constant
temperature runs were conducted while pore pressure varied from 2 MPa to 25 MPa. The constant
temperature runs were done at 23 C, 28 C, 40 C, and 45 C (See white arrows in figure 1). For each
measurement run, a constant differential pressure of 15 MPa was maintained by varying the confining
pressure accordingly to the pore pressure. In figure 1, superimposed on the phase diagrams of CO2 are
the different temperatures and pore pressures conditions explored during the series of CO2 saturated
measurements. As pore pressure increases, for the lower temperature runs (23C and 28C), CO2
changes from a gas phase to a liquid phase while for the higher temperature runs (40C and 45C), CO2
changes from a gas phase to the supercritical fluid phase.

Results

Figure 2 shows the P- and S-wave waveforms recorded for one of the lower temperature and one of the
high temperature runs, T=28 C and T=40 C, respectively. The variations of the waveforms with pore
pressure for the T = 23 C run is similar to T = 28 C run and so is T = 45 C run with T = 40 C run,
and therefore they are not shown to avoid redundancy. The phase transition out of the gaseous phase
state as pore pressure increases are all marked by a slower arriving and weakened P- and S-wave
signal. The change is more drastic for the lower temperature runs and is related to the sharp contrast in
gas to liquid physical properties see in the bulk modulus and the density phase diagrams of CO2. The
transition for the higher temperature runs is less abrupt and occurs at a higher pore pressure as expected
since the critical pressure marks the limit of the vapor-liquid boundary.

p

Figure 2: P- and S-wave (left and right, respectively) waveforms collected for T=28 C and T=40 C (top and
bottom, respectively) constant temperature runs.

Apart from velocity analyses, attenuation analyses have also been done on the waveforms acquired.
Figure 3 shows the differential attenuation coefficients for the constant temperature runs of T=28 C
and T=40 C as a function of pore pressure and frequency. Differential attenuation coefficient is the
attenuation coefficient of each signal minus the attenuation coefficient of the signal when Pp=2MPa of
that measurement run. The signal of Pp=2MPa was used because it is the least attenuated signal of the
set of signals. In all plots of figure 3, two general trends of the differential attenuation coefficient can
be noted. First, the differential attenuation coefficients is the lowest at low pore pressures (gaseous
phase state) and increases at high pore pressures (liquid or supercritical), which is anticipated based on
the variations of the waveforms observed in figure 2. Second, higher frequencies are more attenuated
than lower frequencies especially at higher pore pressures; Q is not constant over this range of
frequencies.

Figure 3: Differential attenuation coefficient of P- and S-wave (left and right, respectively) for T=28 C and
T=40 C (top and bottom, respectively) constant temperature runs from laboratory data.
Biots theory (see Bouzidi and Schmitt, 2009) was used to model the signal response under the same
conditions applied in the laboratory. In figure 4, an excellent agreement between laboratory results and
velocities modeled from Biot's theory is observed.

Figure 4: P- and S-wave velocities measured in the laboratory (dots) and modeled from Biots theory (solid
lines) for T=28 C and T=40 C constant temperature runs.

Conclusions

P- and S-wave laboratory data were acquired for various pressure and temperature conditions in a
porous ceramic rod. Results show clear variations of elastic wave velocities and differential attenuation
coefficient with the phase transitions of CO2. Biots model was used to replicate the elastic wave
velocities and differential attenuation coefficient. There is good agreement for the P-and S-wave
velocities. These results show that elastic waves are sensitive to changes in the pore space and CO2
phase change is evident from signal variations. However, these results are for an extreme end member,
and they do not imply the same results can be attained with a real rock that is less porous, and may
contain more than just CO2 in the pore fluid.

Acknowledgements
The authors would like to thank the NSERC-CGS M.Sc. award for support, the NSERC Discovery
Program for support of the laboratory work and to UofA Physics Dept. technicians, Len Tober and
Lucas Duerksen for their technical support in making these experiments possible.

References
Bouzidi, Y., and Schmitt, D. R., 2009, Measurement of the speed and attenuation of the Biot slow wave using a large ultrasonic
transmitter, J. Geophys. Res., 114, B08201.
Span, R., and Wagner, W.,1996, A new equation of state for carbon dioxide covering the fluid region from the triple-point
temperature to 1100 K at pressures up to 800 MPa, J. Phys. Chem. Ref. Data, 25, 1509-1596.

Recovery 2011 CSPG CSEG CWLS Convention 1
CO
2
Rock Physics: A Laboratory Study
Helen Yam*
hyam@ualberta.ca
and
Douglas R. Schmitt

Summary
It is expected that 4D seismic monitoring will play a critical role in monitoring geologically sequestered
CO
2
but understanding the seismic rock physics of sequestered CO
2
remains a key knowledge gap.
Ultrasonic pulse transmission experiments were conducted on a fully CO
2
saturated porous ceramic rod
subjected to a range of pressures and temperatures that crosses the CO
2
phase boundaries between gas,
liquid, and supercritical, such pore fluid conditions are found within the uppermost kilometer of most
sedimentary basin. P- and S-wave velocities and differential attenuation coefficients were determined from
the laboratory data. The measured elastic wave velocities showed good agreement with Biots model where
as the measured differential attenuation coefficient showed similar trends with changing pore pressure and
frequencies, but is 6-7 times greater in absolute value.
Introduction
Main concerns in geological CO
2
sequestration are whether the injected CO
2
will stay in place over time,
and the ability to observe and verify the injected CO
2
from the surface (Benson, 2006; Chadwick et al,
2009). Therefore, monitoring the subsurface movement and phase behavior of the injected CO
2
is a crucial
aspect of any carbon capture and storage project. Due to the sensitivity of seismic waves to pore fluid
contents, time lapse seismology is regarded as a promising monitoring method for geological CO
2

sequestration projects (Benson and Surles, 2006) and is employed in all industrial scale projects (Sleipner,
Weyburn, In Salah). To utilize seismic methods effectively as a monitoring tool, it is therefore essential to
understand the effect of CO
2
as a pore fluid on the overall rock seismic response. Laboratory work using a
fully saturated synthetic sample has been conducted for a variety of temperatures and pressures that are
representative of subsurface conditions in which CO
2
can be in gas, liquid, supercritical fluid phases. This
is particularly important in seismic monitoring as CO
2
transformations can occur at quite modest
temperatures and pore pressures; all three phases are readily obtained within the uppermost kilometer of a
sedimentary basin. Elastic wave velocity variations and wave strength dampening are observed with phase
transitions, signifying that the physical property changes of the pore fluid due to changing in situ conditions
can be detected using seismic. Although full CO
2
saturation is not expected as in real cases in the earth the
pore space would also contain brine or hydrocarbons, this work provides an end member understanding in
the evolution of seismic responses resulting from changes in the phase state of CO
2
.

CO
2
elastic properties and behaviours
Figure 1 displays the bulk modulus and density phase diagrams of CO
2
as a function of pressure and
temperature based on the thermodynamic model of Span and Wagner (1996). The accepted critical point of
CO
2
is at 31 C and at 7.4 MPa. Below this pressure and temperature, CO
2
can be either a gas or liquid in
respect of the possible subsurface conditions. The vapor-liquid boundary is clearly marked by an abrupt
change in the physical properties of CO
2
where this sudden change disappears as the critical point is reached
and denotes the beginning of the supercritical fluid phase state. A main characteristic of supercritical fluids
is that it has physical behaviors of both a gas and a liquid. As a consequence of this characteristic, any gas
or liquid phase transitions into the supercritical phase results in a smooth variation, unlike the gas to liquid
transition.

Figure 1: Bulk modulus (left) and density (right) phase diagrams of CO2. The red dot on both phase diagrams marks the
critical point of CO
2
. The vapor-liquid boundary is clearly discernible from the sudden change in physical properties.
The boundaries of the supercritical fluid phase state are marked by the white dotted line. The white arrows shows the
temperature and pressure conditions applied to the CO
2
during ultrasonic pulse transmission measurements.
Experiment
A porous ceramic rod, derived from a high fired alumina body (Al
2
O
3
) was used as the host sample in this
study. This sample has a porosity of 58%, with an air permeability of about 100mD. This synthetic sample
was chosen for this study because of its lack of microcracks, which can complicate the acoustic behavior of
real rocks. Although this sample is extremely porous and is not reflective of real reservoir rocks, its large
porosity will allow the overall rock behavior to be more susceptible to the changes in the physical properties
of the pore fluid. This reinforces the purpose of our work on establishing an extreme end member
understanding on the rock physics involved with CO
2
as the pore fluid.

Ultrasonic pulse transmission method was applied to determine the elastic behavior of the sample. A
cylindrical sample 50 mm in length and 25.4 mm in diameter was placed in between a transmitting and a
receiving transducer made of P- and S- piezoelectric ceramics with center frequencies of 1MHz mounted on
an aluminum buffer cap. The sample-transducer assembly was set inside a pressure vessel that was filled
with hydraulic oil as the hydrostatic pressure medium. The sample was jacketed in a Tygon tube, a
flexible clear plastic to seal it from hydraulic oil contamination. CO
2
is introduced into the initially dry and
vacuumed sample from pressure bottle of hospital grade CO
2
located outside the vessel. An electrical
resistance tape is wrapped around the outside of the vessel and controls the temperature of the experiment
and is measured by K-type thermocouple placed inside the pressure vessel.

The transmitted signal is generated by exciting the transmitting transducer with a fast-rising 200 V square
wave from a pulse generator. The propagated signal was recorded by a digital oscilloscope system at a
sampling rate of 10 nanoseconds. The final waveform is a stack of over 500 traces to reduce random noise
effects. From the final waveforms the transit time of the signals can be deduced, and with the length of the
sample the elastic wave velocities can be calculated. The general scheme of the experiment is shown in
figure 2.

Ultrasonic measurements were made on the sample under dry and CO
2
saturated conditions. The dry
measurements were done with the pore space under vacuum (Pp =0 MPa) while the confining pressure
varied from 2 MPa to 40 MPa. The resulting elastic wave velocities showed no differential pressure
dependence due to the lack of microcracks. For the CO
2
saturated measurements, 4 different constant
temperature runs were conducted while pore pressure varied from 2 MPa to 25 MPa. The constant
temperature runs were done at 23 C, 28 C, 40 C, and 45 C. For each measurement run, a constant
differential pressure of 15 MPa was maintained by varying the confining pressure accordingly to the pore
pressure. In figure 1, superimposed on the phase diagrams of CO
2
are the different temperatures and pore
pressures conditions explored during the series of CO
2
saturated measurements. As pore pressure increases,
for the lower temperature runs (23C and 28C), CO
2
changes from a gas phase to a liquid phase while for
the higher temperature runs (40C and 45C), CO
2
changes from a gas phase to the supercritical fluid phase.

Figure 2: Experimental setup for ultrasonic pulse transmission for CO2 saturated measurements.
Examples
Figure 3 shows the P- and S-wave waveforms recorded for one of the lower temperature and one of the high
temperature runs, T=28 C and T=40 C, respectively. The variations of the waveforms with pore pressure
for the T = 23 C run is similar to T = 28 C run and so is T = 45 C run with T = 40 C run, and therefore
they are not shown to avoid redundancy. The phase transition out of the gaseous phase state as pore
pressure increases are all marked by a slower arriving and weakened P- and S-wave signal. The change is
more drastic for the lower temperature runs and is related to the sharp contrast in gas to liquid physical
properties see in the bulk modulus and the density phase diagrams of CO
2
. The transition for the higher
temperature runs is less abrupt and occurs at a higher pore pressure as expected since the critical pressure
marks the limit of the vapor-liquid boundary.


Figure 3: P- and S-wave (left and right, respectively) waveforms collected for T=28 C and T=40 C (top and bottom,
respectively) constant temperature runs.

From analyzing the above waveforms, the P-and S-wave velocities are determined and displayed in figure 4.
On these plots, the velocity of the dry sample at a differential pressure of 15MPa is shown by the black dot.
Once CO
2
is introduced into the pore space, irrespective of the phase state of the CO
2
, the elastic waves
traveling through the sample are slower than when the pore space is at vacuum. In each constant
temperature run the CO
2
phase change is marked by a drop in both P- and S-wave velocity of 4-5%, with
the drop being more gradual for the higher temperature runs. Within a given phase state (gas, liquid, or
supercritical fluid), as pore pressure increases the elastic wave velocities lessens and this is due to the
increased density experienced.

Apart from velocity analyses, attenuation analyses have also been done on the waveforms acquired. Figure
5 shows the differential attenuation coefficients for the constant temperature runs of T=28 C and T=40 C
as a function of pore pressure and frequency. Differential attenuation coefficient is the attenuation
coefficient of each signal minus the attenuation coefficient of the signal when Pp=2MPa of that
measurement run. The signal of Pp=2MPa was used because it is the least attenuated signal of the set of
signals. The advantages of dealing with differential attenuation coefficients over the conventional method
of determining absolute attenuation coefficients is that the former requires less time and uses only one
sample, while the purpose of demonstrating signal strength variation is still preserved.


Figure 4: Ultrasonic P- and S-wave (left and right, respectively) velocities measured of the sample saturated with CO2
under various temperatures and pore pressures and when the sample is not saturated (black dot at Pp=0MPa). All
measurements on the two plots were completed with a differential pressure of 15MPa. The error of the P-and S-wave
velocities is about the size of the marker.
In all plots of figure 5, two general trends of the differential attenuation coefficient can be noted. First, the
differential attenuation coefficients is the lowest at low pore pressures (gaseous phase state) and increases at
high pore pressures (liquid or supercritical), which is anticipated based on the variations of the waveforms
observed in figure 3. Second, higher frequencies are more attenuated than lower frequencies especially at
higher pore pressures; Q is not constant over this range of frequencies.

Figure 5: Differential attenuation coefficient of P- and S-wave (left and right, respectively) for T=28 C and T=40 C (top
and bottom, respectively) constant temperature runs from laboratory data.

For the S-wave plot of T=28 C, there is large variability of the differential attenuation coefficient at the top
right corner corresponding to high pore pressure and high frequencies and this is caused by noise effects.
The more attenuated a signal is, the smaller the signal will be and the more susceptible it is to the effects of
noise.

It is important to compare the laboratory results to rock physics models to see whether the responses can be
predicted from existing theories; this is particularly true should we wish to make predictions of behaviour at
typical seismic frequencies (~5 Hz to 300 Hz). If the responses can be predicted, then other signal
responses can be determined for various conditions without replicating it in the laboratory. Biots theory
(see Bouzidi and Schmitt, 2009) was used to model the signal response under the same conditions applied in
the laboratory by using the fluid properties from Span and Wagners CO
2
thermodynamic model and the
sample dry bulk and shear moduli characterized from the porous ceramic rod. The modeled P- and S-wave
velocities using Biots equations are shown in figure 6. There is excellent agreement of the modeled and
laboratory measured Biot fast P and the shear wave velocities. Figure 7 shows the modeled differential
attenuation coefficient over the same frequency interval, temperature and pore pressure conditions as
analyzed with the laboratory data. The two general trends observed in figure 5 with respect to pore
pressure and frequency is also seen in figure 7. However the trends from Biots model appear more linear
than what is observed with our data. In addition, when comparing the actual values of the differential
attenuation coefficients, the modeled values are about 6-7 times less than the measured values from
laboratory data.

Figure 6: P- and S-wave velocities measured in the laboratory (dots) and modeled from Biots theory (solid lines) for
T=28 C and T=40 C constant temperature runs.
Conclusions
P- and S-wave laboratory data were acquired for various pressure and temperature conditions in a porous
ceramic rod. Results show clear variations of elastic wave velocities and differential attenuation coefficient
with the phase transitions of CO
2
. Biots model was used to replicate the elastic wave velocities and
differential attenuation coefficient. There is good agreement for the P-and S-wave velocities but not for the
differential attenuation coefficient. However, the general trend with pore pressure and frequency is in
agreement. These results show that elastic waves are sensitive to changes in the pore space and CO
2
phase
change is evident from signal variations. However these results are for an extreme end member, and they
do not imply the same results can be attained with a real rock that is less porous, and may contain more than
just CO
2
in the pore fluid.


Figure 7: Differential attenuation coefficient of P- and S-wave (left and right, respectively) for T=28 C and T=40 C (top
and bottom, respectively) constant temperature runs from Biots theory.

Acknowledgements
HY would like to thank the NSERC-CGS M.Sc. award for support, and the NSERC Discovery Program for
support of the laboratory work and purchase of some of the equipment that allowed us to appropriately
control fluid pressures in the samples. Also,a big thanks to UofA Physics Dept. technicians, Len Tober and
Lucas Duerksen for their technical support in making these experiments possible.

References
Benson, S. M., 2006, Monitoring Carbon dioxide sequestration in deep geological formations for inventory verification and carbon credits, 2006
SPE Annual Technical Conference and Exhibition in San Antonio, USA, 24-27 September 2006.
Benson, S. M., and Surles, T., 2006, Carbon dioxide capture and storage: An overview with emphasis on capture and storage in deep geological
formations, Proceedings of the IEEE, 94, 1795-1805.
Bouzidi, Y., and Schmitt, D. R., 2009, Measurement of the speed and attenuation of the Biot slow wave using a large ultrasonic transmitter, J.
Geophys. Res., 114, B08201.
Chadwick, R. A., Arts, R., Bentham, M., Eiken, O., Holloway, S., Kirby, G. A., Pearce, J.M., Williamson, J. P., Zweigel, P., 2009, Review of
monitoring issues and technologies associated with the long-term underground storage of carbon dioxide (in Underground gas storage;
worldwide experiences and future development in the UK and Europe): Geological Society of London Special Publications, 313, 257-275.
Span, R., and Wagner, W.,1996, A new equation of state for carbon dioxide covering the fluid region from the triple-point temperature to 1100
K at pressures up to 800 MPa, J. Phys. Chem. Ref. Data, 25, 1509-1596.

Rock physics model for different phase states of CO
2
: Combining laboratory P and S wave
measurements with a model of CO
2
state behaviour
Helen Yam*, Douglas R. Schmitt, University of Alberta

Summary

Given the thermo-physical properties of CO
2
and the elastic
properties of a dry rock, the seismic elastic properties of a
fully CO
2
saturated rock may be modeled by standard
Gassmann fluid substitution. A model is developed for the
elastic properties of CO
2
over a range of pressures,
temperatures, and phase states anticipated in the upper
sections of a sedimentary basin. The fluid properties
obtained are then employed in a simple Gassmann
substitution analysis that uses measurements in our
laboratory of effective pressure dependant P and S wave
velocities on a Berea sandstone to predict the values under
in situ saturated conditions. Although a more realistic
situation would need to include mixtures of CO
2
with other
liquids and gases, this exercise illustrates some of the
variations in velocity containing CO
2
in the pore space.

Introduction

Sequestering CO
2
geologically may be a viable method to
reduce the amount of CO
2
released in our atmosphere.
However, a major concern with geologically sequestered
CO
2
is whether the injected CO
2
will stay in place over
time, otherwise rendering our efforts useless and possibly
giving rise to additional problems. Thus, a crucial aspect of
CO
2
sequestration projects is to carefully monitor the
movement and phase behavior of the subsurface CO
2
over
time particularly to ensure that any leakage or unpredicted
migrations can be detected. Time lapse seismology is the
most promising monitoring approach in obtaining
information about changes in the subsurface without the
direct use of well bores (Arts et al., 2002). To properly
monitor using seismic methods, it is necessary to
understand not only how seismic elastic properties (density
and velocity) of a rock change in response to CO
2

saturation, but also how important changes in the effective
pressure might be. Understanding the changes under in situ
conditions is complicated by the fact CO
2
can exist in gas,
liquid and supercritical fluid states.

Fluid properties: CO
2
Phase and Physical Property
Behaviour

In the range of pore pressures and temperatures
encountered in the upper crust, CO
2
can be in gas, liquid, or
supercritical state phases (Fig. 1a). Once within the
reservoir, CO
2
can exist as free gas or as a dissolved
constituent in the already existing pore fluid such as brine
or oil (e.g., Carcione et al, 2006), but here for purposes of
illustration CO
2
is the only saturant. The phase state (and
consequent variations in density and fluid compressibility)
can have a large influence on the overall effective seismic
properties of the saturated porous rock.
Since the 1960s, the thermodynamic properties of
CO
2
have been extensively studied resulting in many
proposed empirical equations of state to accurately describe
the behavior of CO
2
. It is accepted that the critical pressure
and temperature of CO
2
is 7.4MPa and 31C, respectively.
Using data derived from the thermodynamic models
developed by Span and Wagner (1996) within the
implementation by Lemmon et al., (2007) plots of elastic
property variations of CO
2
as a function of pressure and
temperature were made. Fig. 1b shows the sound speed of
pure CO
2
over a pressure range of 1-10 MPa and a
temperature range of 1-50C. The vapor-liquid boundary
for CO
2
can be easily discerned by an abrupt discontinuity
in sound speed that disappears above the critical point. At
high pressure and low temperature, CO
2
is in liquid phase
resulting in higher sound speeds with iso-velocity contours
running steeply and flattens as the temperature increases.
At the gas phase, the velocity values show a significant
decrease with iso-velocity contours running closer to the
horizontal. The steepness of iso-velocity lines increases as
higher pressure and temperature is reached. Approaching
the top right corner, the iso-velocity contours of gas and the
liquid phase begin to merge and the distinction between
these two phases disappears as the critical point is reached,
signifying the start of the supercritical phase of CO
2
. In
this state, both the properties of gas and liquid can coexist;
the supercritical fluid has a compressibility like that of a
gas and a density comparable to a liquid.
Figure 1c shows the density of CO
2
as a function of
pressure and temperature. Many of the same trends seen in
the sound speed plot appear here. Additionally, however,
near the critical point small changes in pressure or
temperature result in rapid changes in density, this is an
expected behavior of supercritical fluids.
With this information on the thermo-physical
properties of CO
2
available, the fluid bulk moduli, of CO
2

as a function of pressure and temperature can be
determined by,

K
f
=
CO2
Vp
CO2
(1)

Host rock properties: Laboratory measurements of dry
elastic wave velocities

Berea sandstone is predominantly a quartz held together by
silica and clay. Due to its uniform properties, Berea
sandstone has been extensively used as a standard in
petrophysical testing (e.g., Green and Wang, 1994; OHara,
Rock physics model for free CO
2
Phase Diagram of CO2
Liquid
Supercritical
Gas

Figure 1: a) Phase diagramof CO2 with critical point noted by red
circle. b) P-wave velocity of CO2 in m/s as function of pressure
and temperature. c) P-wave density of CO2 in m/s as function of
pressure and temperature.
1985 among many others). The sample used had a porosity
of 18.6% and a dry bulk density of 2160 kg/m
3
. To
describe the elastic properties of this rock, simultaneous
ultrasonic P and S wave, pulse-transmission velocities were
made on the dry sample with the confining pressure
varying from 0 to 100 MPa, while the sample was
unsaturated with the pore space maintained under vacuum
during the measurements. Transit times were picked from
the waveforms obtained and the dry elastic wave velocities
were simply found fromthe dividing the sample length
over the transit times. Fromthese calculated dry elastic
wave velocities (Fig 2), the dry frame bulk and shear
moduli were calculated as a function of differential
pressure. As the differential pressure increased, both the
bulk and shear moduli increased correspondingly with the
well known nonlinear variations in elastic moduli with
pressure.

Figure 2: P and S wave velocities obtained in laboratory
measurements of a dry Berea sandstone. Pore space was
maintained under vacuumas confining pressure was varied.

Modeling: Gassmanns fluid substitution

Since the properties of the pore fluid is dependent on
temperature and pressure, which is in turn dependent on
depth, reasonable temperature and pore pressure to be
expected at a given depth , z, can be found by equation (2)
and (3) respectively,
Gz T T
o
+ = , (2)

gz P
w p
= . (3)

c)
b)
a)
2
Equation 2 and 3 are descriptive for a simplistic case,
where the subsurface temperature is only affected by the
surface temperature T
o
and by a constant geothermal
gradient,G =20C/km. For pore pressure, it is assumed that
there are no permeability barriers, allowing fluid to freely
flow fromthe surface to depth z such that the pore pressure
is then equal to the hydrostatic pressure, with
w
being the
density of water and g, the acceleration of gravity.
Typically, reservoirs for storing CO
2
are at depths of
at least a 1000 m. At a depth of 1000m, according to
equation 2 and 3, the temperature and pore pressure will be
35C and 9.81MPa respectively; CO
2
is in the supercritical
fluid phase state. If the injected CO
2
finds an escape and
moves up towards the surface, at a depth of 790m the
original supercritical CO
2
will transform into the liquid
state. If this upward escape continues, at depths shallower
than 720m, the gaseous phase state will eventually be
reached.
With the pressure dependent elastic property of the
host rock characterized fromlaboratory measurements and
with the fluid properties of CO
2
available as a function of
temperature and pressure, the bulk modulus of a fully
saturated Berea sandstone with CO
2
can be modeled by
fluid substitution using Gassmanns equation, (Gassmann,
1951).

s
f s f
f
s
f
s sat
K
K
K K K
K
K
K
K K
K K
+
+ +
=

) 1 (
) 1 (
(4)

In equation (4) K, K
s
, K
f
is the dry frame modulus of the
rock, grain modulus, and fluid modulus respectively. The
composition of the Berea sandstone is nearly pure quartz,
therefore a grain modulus of 36.6GPa was used.
Since fluids do not have a shear modulus, the shear
modulus of the wet rock is simply the modulus of the dry
rock. Consequently by combining the calculated saturated
bulk modulus with the dry shear modulus of the Berea
sandstone obtained fromlaboratory measurements, the P-
wave and S-wave velocities of the saturated rock can be
found by

dry sat
p
K
V
3
4
+
= and
dry
s V = (5)

where is the bulk density and is given by

f s + = ) 1 ( . (6)

Results:

Fig 3 displays the measured dry elastic velocities of the
host rock and the modeled elastic velocities of a CO
2

saturated Berea sandstone of different fluid phase states,
determined based on the encountered depth described
above and also for contrast, the modeled elastic wave
velocities of a water saturated rock. Under these 3 different
depths that results in 3 different phases of CO
2
, water exists
only as a liquid with little physical property change.
For S-wave velocities (fig. 3a), the order of density
related to the pore fluid is clearly illustrated based on the
order of velocity at any given differential pressure. Since
the shear modulus does not change for a given rock, the
only effect on the varying velocity is the inverse relation it
has with bulk density due to saturation. Of all the types of
pore fluid, water has the lowest velocity due to its highest
density, followed by the supercritical phase state of CO
2
,
the liquid state of CO
2
, and then lastly the gaseous CO
2
.
For P-wave velocities (fig 3b), the order seen in the
S-wave is not completely reflected in the same manner. P-
wave velocities just like S-wave velocities are affected by
the bulk density of the rock, but the resulting elastic
velocities are also complicated by an overall rock bulk
modulus that is affected by pore fluids. J ust like the S-
waves, the velocity due to the saturation of the gaseous
CO
2
is the highest of the 3 phase states, followed by the
liquid state and lastly, the supercritical phase state.
Although the presence of fluid in the pore space should
stiffen the rock, the similar trend to S-waves implies that
for CO
2
saturations, the effect of bulk density on velocity
dominates over the effect of bulk modulus on the overall
rock. On the other hand, with water as the pore fluid, it is
observed that the higher bulk modulus of water stiffens the
rock much more than the case of CO
2.
The effect of the
overall bulk modulus dominates over the bulk density
allowing it to have the highest velocity, even higher than
that of the dry rock.
In figure 3, the velocities of the CO
2
saturated rock
in the liquid state and that of the supercritical state lie very
close together, unlike the larger step up for the gaseous
CO
2
state. The supercritical CO
2
experiencing a pressure
of 9.81MPa and a temperature of 35C lies very close to
the transition boundary of liquid to supercritical phase.
Supercritical fluids have behaviors of both a gas and a
liquid, where the dominating behavior ruling the
supercritical fluid depends on where it lies in the phase
diagram. Thus the close proximity of the supercritical and
liquid CO
2
saturated velocities is reflective of a liquid
dominated supercritical fluid behavior.

2

Figure 3: a) S-wave b.) P-wave velocities of a Berea sandstone dry
and of various pore fluids and phase states. The CO2 properties,
and consequently phase states were determined fromequation 2
and 3. The behavior of the supercritical fluid is liquid dominated.

So far, the modeling of the velocity behaviors as
function of CO
2
phase state has been determined based on
the oversimplified description of the subsurface conditions
using equations 2 and 3. Realistically, the factors
controlling the in situ temperature and pore pressure is far
more complex; there may be a set of conditions which
causes the behavior of the supercritical CO
2
to be gas
dominated. In such a case, if the supercritical CO
2

experiences a pore pressure of 8MPa and a temperature of
50C, placing it very close to the gas to supercritical
transition boundary, the supercritical CO
2
saturated rock
velocity changes drastically, figure 4. The velocity of this
gas-dominated supercritical CO
2
saturated rock is now the
highest of all the different CO
2
phase states.
Fromthe two extreme behaviors of a supercritical
fluid, gas versus liquid dominated physical properties, the
modeling illustrates how significantly the velocity of a
supercritical fluid can vary. Frombeing the slowest in the
liquid-dominated supercritical phase, to the being the
fastest in the gas-dominated supercritical phase state, the
actual phase state of the CO
2
can not be determined by its
velocity or fromits relative velocity to other known phase
states values.

Figure 4: a) S-wave b.) P-wave velocities of a Berea sandstone dry
and of various pore fluids and phase states. The supercritical fluid
has a gas dominated behavior.

Summary

Understanding the elastic property of CO
2
saturated rocks
is vital to monitoring CO
2
sequestration projects using time
lapse seismology. With thermo-physical properties of the
pore fluid and elastic properties of a dry rock frame, rock
properties of the saturated rock can be modeled using
Gassmanns formula. Modeled results show effective
pressure and fluid phase dependency for wave velocities of
a fully CO
2
saturated Berea sandstone. As well, the
dominating behavioral effects of a gas or liquid on the
supercritical CO
2
phase state can vary the saturated rock
velocity substantially. Thus, velocity alone or its relative
value with respect to other known phase states cannot
determine the actual phase state of the pore fluid.
Laboratory experiments to confirm this modelling are
currently underway.
a)
a)
b)
b)
2

Abstract

In any geological sequestration projects, monitoring and verification are
essential components in ensuring storage integrity. Seismic methods are regarded
as a feasible way to monitor the subsurface CO
2
because of their sensitivity to a
rocks pore space content. Therefore understanding the effects of CO
2
and its
variability on seismic response is important.
Ultrasonic pulse transmission measurements were conducted on a porous
ceramic sample and on a Berea sandstone sample. P-and S-waveforms were
collected under various pressures, temperatures, and fluid-type saturation. The
wave velocity and attenuation under full CO
2
saturation and under a constant
differential pressure were analyzed. The presence of differing phase states and
some phase transitions were notable from wave velocity and wave attenuation
changes. Only the observed wave velocities of the porous ceramic sample were in
good agreement with Biots modelled results. Generally, CO
2
s density plays a
more dominant role than its bulk modulus on controlling the P-wave velocity.

Table of Contents
Chapter 1: Introduction ........................................................................................ 1
1.1 Climate Change ...................................................................................... 1
1.2 Aim of Research and Chapter Description ............................................ 4
Chapter 2: Background on Carbon Storage ....................................................... 8
2.1 Types of CO
2
sequestration.................................................................... 8
2.2 Examples of Geological Sequestration ................................................ 14
2.3 Safety and Monitoring ......................................................................... 15
2.4 Direct monitoring measurements ......................................................... 17
2.5 Indirect monitoring measurements (Geophysical) ............................... 17
2.5.1 Geodetic ................................................................................ 18
2.5.2 Electromagnetic (EM) ........................................................... 19
2.5.3 Gravitational ......................................................................... 20
2.6 Seismic ................................................................................................. 21
2.6.1 Surface Seismic ..................................................................... 22
2.6.2 Vertical Seismic Profiling ..................................................... 24
2.6.3 Cross-well Imaging ............................................................... 26
2.7 Review of existing literature on CO
2
laboratory measurements .......... 27
2.8 Summary .............................................................................................. 32
Chapter 3: Theoretical Background .................................................................. 34
3.1 Carbon dioxide ..................................................................................... 35
3.1.1 Introduction to Thermodynamics .......................................... 35
3.1.2 CO
2
Phase States ................................................................... 36
3.1.3 Geological Sequestration Approximate Physical Conditions ...

........................................................................................................ 38
3.2 Seismic Theory .................................................................................... 40
3.2.1 Theory of Elasticity ............................................................... 41
3.2.2 Wave propagation in an isotropic linear elastic medium ...... 45
3.2.3 Viscoelasticity ....................................................................... 47
3.3 Rock Physics ........................................................................................ 50
3.3.1 Gassmanns Equation ............................................................ 51
3.3.2 Biots Formulation ................................................................ 57
3.4 Introduction to Elastic Wave Velocity and Attenuation Determination
in the laboratory ......................................................................................... 64
3.4.1 Laboratory Measurement Method ......................................... 65
3.4.2 Attenuation ............................................................................ 67
3.4.3 The Attenuation of Ultrasonic Waves through Acrylic ........ 72
3.5 Summary .............................................................................................. 77
Chapter 4: Experimental Setup and Procedure ................................................ 79
4.1 Piezoelectric Transducers .................................................................... 79
4.2 Sample Preparation .............................................................................. 82
4.3 Experimental Apparatus ....................................................................... 84
4.4 Experimental Procedure ....................................................................... 87
4.5 Velocity and Attenuation Analysis Methodology ................................ 91
4.5.1 Velocity and Error Analysis .................................................. 92
4.5.2 Attenuation and Error Analysis............................................. 96
4.6 Chapter Summary .............................................................................. 100
Chapter 5: Sample Characterization ............................................................... 102
5.1 Samples .............................................................................................. 102
5.2 Density and Porosity .......................................................................... 103
5.2.1 Grain Volume and Density ................................................. 104
5.2.2 Bulk Density and Porosity .................................................. 105
5.2.3 Pore Size Distribution ......................................................... 106
5.3 Permeability ....................................................................................... 110
5.4 Tortuosity ........................................................................................... 112
5.5 Scanning Electron Microscope Imaging (SEM) ................................ 114
5.6 Summary ............................................................................................ 118
Chapter 6: Results and Discussion: Synthetic sample .................................... 119
6.1 Dry ..................................................................................................... 119
6.1.1 Dry Waveforms ................................................................... 120
6.1.2 Dry Wave Velocities ........................................................... 121
6.2 CO
2
..................................................................................................... 122
6.2.1 CO
2
Saturated Waveforms .................................................. 123
6.2.2 Observed CO
2
Saturated Wave Velocity ............................ 125
6.2.3 Modelled CO
2
6.2.4 Observed CO
2
Saturated Wave Attenuation ....................... 134
6.2.5 Modelled CO
2
6.3 Water .................................................................................................. 143
6.3.1 Water Saturated Waveforms ............................................... 143
6.3.2 Observed Water Saturated Wave Velocities ....................... 144
6.3.3 Modelled Water Saturated Wave Velocities ....................... 144
6.4 Results and Discussion ...................................................................... 146
6.4.1 CO
2
...................................................................................... 146
6.4.2 Water ................................................................................... 148
6.5 Summary ............................................................................................ 149
Chapter 7: Results and Discussion: Non-synthetic sample ............................ 152
7.1 Dry ..................................................................................................... 152
7.1.1 Dry Waveforms ................................................................... 153
7.1.2 Dry Wave Velocities ........................................................... 153
7.2 CO
2
..................................................................................................... 154
7.2.1 CO
2
Saturated Waveforms .................................................. 155
7.2.2 Observed CO
2
7.2.3 Modelled CO
2
7.2.4 Observed CO
2
7.2.5 Modelled CO
2
7.3 Water .................................................................................................. 175
7.3.1 Water Saturated Waveforms ............................................... 175
7.3.2 Observed Water Saturated Wave Velocities ....................... 176
7.3.3 Modelled Water Saturated Wave Velocities ....................... 176
7.4 Results and Discussion ...................................................................... 179
7.5 Summary ............................................................................................ 184
Chapter 8: Conclusion ....................................................................................... 187
8.1 Summary of Work .............................................................................. 187
8.2 Contributions of this Work ................................................................ 190
8.3 Future Work ....................................................................................... 191
References ...................................................................................................... 194
Appendix A: Observed and modelled Results ................................................. 208
A.1 Porous Ceramic Rod ......................................................................... 208
A.1.1 Dry Condition Results ........................................................ 209
A.1.2 CO
2
Saturated Results ........................................................ 213
A.1.3 Water Saturated Results ..................................................... 233
A.2 Berea Sandstone ................................................................................ 235
A.2.1 Dry Condition Results ........................................................ 235
A.2.2 CO
2
Saturated Results ........................................................ 239
A.2.3 Water Saturated Results ..................................................... 263

List of Tables

3.1 Relationships among the elastic constants for a homogeneous isotropic
material ..................................................................................................... 45
4.1 The measurements carried out for a sample during ultrasonic pulse
transmission experiments. .......................................................................... 91
5.1 The measured envelope volume of the samples ...................................... 106
5.2 Petrophysical properties characterized of the samples ............................ 115
A.1 Petrophysical properties determined of the porous ceramic rod .............. 208
A.2 Dry P-and S-wave velocities of the porous ceramic rod at T = 23 C under
various confining pressures ...................................................................... 212
A.3 Dry P-and S-wave velocities of the porous ceramic rod at Pc = 10 MPa
while under various temperatures ............................................................ 213
A.4 CO
2
saturated P-and S-wave velocities of the porous ceramic rod for the
T= 23 C constant temperature run .......................................................... 216
A.5 CO
2
A.6 CO
2
A.7 CO
2
A.8 CO
2
Pp= 7 MPa constant pore pressure run .................................................... 224
A.9 CO
2
Pp= 10 MPa constant pore pressure run .................................................. 225
A.10 CO
2
Pp= 25 MPa constant pore pressure run .................................................. 226
A.11 Water saturated P-and S-wave velocities of the porous ceramic rod at
T=23C for varying pore pressures .......................................................... 235
A.12 Petrophysical properties determined of the Berea sandstone................... 235
A.13 Dry P-and S-wave velocities of the Berea sandstone at T = 23C under
various confining pressures of the first measurement run ....................... 238
A.14 Dry P-and S-wave velocities of the Berea sandstone at T = 23C under
various confining pressures of the second measurement run ................... 239
A.15 CO
2
saturated P-and S-wave velocities of the Berea sandstone for the T=
23 C constant temperature run ................................................................ 243
A.16 CO
2
A.17 CO
2
A.18 CO
2
A.19 CO
2
A.20 CO
2
saturated P-and S-wave velocities of the Berea sandstone for the Pp=
7 MPa constant pore pressure run ............................................................ 255
A.21 CO
2
10 MPa constant pore pressure run .......................................................... 256
A.22 CO
2
25 MPa constant pore pressure run .......................................................... 257
A.23 Water saturated P-and S-wave velocities of the Berea sandstone at T=23C
for varying pore pressures ........................................................................ 265

List of Figures

1.1 Atmospheric concentrations of CO
2
............................................................ 3
2.1 CO
2
versus sea water density .................................................................... 10
2.2 Time lapse seismic images of pre- and post-CO
2
injection ....................... 23
2.3 Amplitude difference maps of the Midale-Marly horizon ......................... 23
2.4 Amplitude differenced image between pre-and post injection at SACROC
........................................................................................................ 25
2.5 P-wave tomogram from cross-well survey in Frio, Texas ......................... 27
2.6 Depiction of Xue and Oshumis transducer array ...................................... 30
3.1 A generic phase diagram ............................................................................ 38
3.2 CO
2
phase diagrams ................................................................................... 40
3.3 Creep and recovery response of a viscoelastic medium ............................ 48
3.4 Relaxation and recovery response of a viscoelastic medium ..................... 49
3.5 Stress and strain varying sinusoidally with time ........................................ 51
3.6 The basic principle behind ultrasonic pulse transmission method ............. 66
3.7 Illustration of phase velocity determination............................................... 71
3.8 The time series signals recorded through varying lengths of acrylic ......... 73
3.9 The amplitude spectra of the times series signals through acrylic ............. 74
3.10 P-wave attenuation coefficient and quality factor...................................... 76
3.11 S-wave attenuation coefficient and quality factor...................................... 76
4.1 P-and S-wave piezoelectric ceramics used ................................................ 81
4.2 Cartoon and photograph of a transducer .................................................... 83
4.3 A fully prepared and assembled sample ready for measurement............... 84
4.4 A simplified schematic and photograph of the laboratory equipment ....... 86
4.5 Waveforms from buffer measurements over a confining pressure range of
5 MPa to 50 MPa ....................................................................................... 93
4.6 P-wave signals through only the buffer and through the buffer with sample
under a confining pressure of 15 MPa ....................................................... 93
4.7 P-wave time series signals of the porous ceramic rod acquired at dry
conditions under 5 MPa and 40 MPa confining pressure .......................... 99
4.8 The P-wave amplitude spectra and the resulting differential attenuation
coefficient ............................................................................................... 100
5.1 Photographs of the samples used ............................................................. 103
5.2 Cumulative and incremental mercury porosimetry intrusion curves for the
porous ceramic rod ................................................................................... 108
5.3 Cumulative and incremental mercury porosimetry intrusion curves for the
Berea sandstone ....................................................................................... 109
5.4 Photograph of Tiny Perm II, portable air permeameter system ............... 111
5.5 Schematic diagram of the four electrode circuit ...................................... 114
5.6 SEM images of the porous ceramic rod at 5 kX and 20 kX..................... 116
5.7 SEM images of the Berea sandstone at 500 X and 10 kX ....................... 117
6.1 Normalized waveforms under dry conditions at room temperature for the
porous ceramic rod displayed as wiggle traces and color amplitudes ..... 120
6.2 Dry P-and S-wave wave velocities with errors as a function of confining
pressure for the porous ceramic rod ......................................................... 122
6.3 CO
2
s bulk modulus and density phase diagram with the applied
measurement conditions superimposed for the porous ceramic rod ........ 123
6.4 Normalized CO
2
saturated P-and S-wave waveforms of T = 28 C and T =
40 C constant temperature runs .............................................................. 124
6.5 Normalized CO
2
saturated P-and S-wave waveforms of P
P
= 7, P
P
=10, and
P
P
=25 MPa constant pore pressure runs .................................................. 126
6.6 Dry and CO
2
saturated P-and S-wave velocities with error for all constant
temperature runs ....................................................................................... 127
6.7 Dry and CO
2
pore pressure runs .................................................................................... 128
6.8 Observed and Gassmanns modelled P-and S-wave velocities for T = 28
C and T = 40 C constant temperature runs ........................................... 130
6.9 Observed and Gassmanns modelled P-and S-wave velocities for P
P
= 7,
P
P
=10, and P
P
=25 MPa constant pore pressure runs .............................. 132
6.10 Observed and Biots modelled P-and S-wave velocities for T = 28 C and
T = 40 C constant temperature runs ....................................................... 133
6.11 Observed and Biots modelled P-and S-wave velocities for P
P
= 7, P
P
=10,
and P
P
=25 MPa constant pore pressure runs ........................................... 134
6.12 Observed P-and S-wave differential attenuation coefficient for T = 28 C
and T = 40 C constant temperature runs................................................. 135
6.13 Observed P-and S-wave differential attenuation coefficient for P
P
= 7, P
P
=10, and P
P
=25 MPa constant pore pressure runs ................................... 137
6.14 Biots modelled P-and S-wave differential attenuation coefficient for T =
28 C and T = 40 C constant temperature runs ...................................... 139
6.15 Biots modelled P-and S-wave differential attenuation coefficient for P
P
=
7, P
P
=10, and P
P
=25 MPa constant pore pressure runs .......................... 140
6.16 Biots critical frequency computed for the CO
2
saturated porous ceramic
rod ...................................................................................................... 142
6.17 Normalized water saturated P-and S-wave waveforms as a function of
pore pressure. ........................................................................................... 143
6.18 Dry and water saturated P-and S-wave velocities with error ................... 144
6.19 Observed and Gassmanns modelled water saturated P-and S-wave
velocities .................................................................................................. 145
6.20 Observed and Biots modelled water saturated P-and S-wave velocities 146
7.1 Normalized waveforms under dry conditions at room temperature for the
Berea sandstone displayed as color amplitudes ....................................... 153
7.2 Dry P-and S-wave velocities with error as a function of confining pressure
for the Berea sandstone ............................................................................ 154
7.3 CO
2
s bulk modulus and density phase diagram with the applied
measurement conditions superimposed for the Berea sandstone ............. 155
7.4 Normalized P-and S-wave waveforms for T = 23 C and T = 55 C
constant temperature runs ........................................................................ 156
7.5 Normalized P-and S-wave waveforms for the P
P
= 7, P
P
=10, and P
P
=25
MPa constant pore pressure runs ............................................................. 158
7.6 Dry and CO
2
7.7 Dry P-and S-wave velocities with error as a function of confining pressure
of two dry measurement runs of the Berea sandstone ............................. 160
7.8 Corrected dry and CO
2
saturated P-and S-wave velocities with error for all
constant temperature runs. ....................................................................... 161
7.9 Dry and corrected CO
2
constant pore pressure runs ...................................................................... 163
7.10 Observed and Gassmanns modelled P-and S-wave velocities for T = 23
C and T = 55 C constant temperature runs ........................................... 164
7.11 Observed and Gassmanns modelled P-and S-wave velocities for P
P
= 7,
P
P
=10, and P
P
=25 MPa constant pore pressure runs .............................. 165
7.12 Observed and Biots modelled P-and S-wave velocities for T = 23 C and
T = 55 C constant temperature runs ....................................................... 167
7.13 Observed and Biots modelled P-and S-wave velocities for P
P
= 7, P
P
=10,
and P
P
=25 MPa constant pore pressure runs ........................................... 167
7.14 Observed P-and S-wave differential attenuation coefficients for T = 23 C
and T = 55 C constant temperature runs................................................. 169
7.15 Observed P-and S-wave differential attenuation coefficients for P
P
= 7, P
P
=10, and P
P
=25 MPa constant pore pressure runs ................................... 170
7.16 Biots modelled P-and S-wave differential attenuation coefficient for T =
23 C and T = 55 C constant temperature runs ...................................... 172
7.17 Biots modelled P-and S-wave differential attenuation coefficient for P
P
=
7, P
P
=10, and P
P
=25 MPa constant pore pressure runs .......................... 173
7.18 Biots critical frequency computed for the CO
2
saturated Berea sandstone175
7.19 Normalized water saturated P-and S-wave waveforms as a function of
pore pressure ............................................................................................ 176
7.20 Dry and water saturated P-and S-wave velocities with error ................... 177
7.21 Observed and Gassmanns modelled water saturated P-and S-wave
velocities .................................................................................................. 177
7.22 Observed and Biots modelled water saturated P-and S-wave velocities 178
A.1 Normalized P-and S-wave waveforms under dry conditions at T = 23C
for the porous ceramic rod ....................................................................... 209
A.2 Normalized P-and S-wave waveforms under heated dry conditions at a
constant confining pressure of 10 MPa for the porous ceramic rod ........ 209
A.3 Dry P-and S-wave wave velocities with errors as a function of confining
pressure for the porous ceramic rod ......................................................... 210
A.4 Heated dry P-and S-wave velocities with errors as a function of
temperature for the porous ceramic rod under a Pc=10 MPa .................. 211
A.5 Normalized P-and S-wave waveforms of the CO
2
rod for all measured constant temperature runs ....................................... 214
A.6 Dry and CO
2
temperature runs for the porous ceramic rod ........................................... 215
A.7 Observed and Gassmanns modelled P-and S-wave velocities for all
constant temperature runs for the porous ceramic rod ............................. 218
A.8 Observed and Biots modelled P-and S-wave velocities for all constant
temperature runs for the porous ceramic rod ........................................... 219
A.9 P-and S-waves amplitude spectra for all constant temperature runs for the
A.10 Observed P-and S-wave differential attenuation coefficient for all constant
A.11 Biots modelled P-and S-wave differential attenuation coefficient for all
constant temperature runs ........................................................................ 222
2
rod for all measured constant pore pressure runs ..................................... 223
A.13 Dry and CO
2
pore pressure runs for the porous ceramic rod ......................................... 227
A. 14 Observed and Gassmanns modelled P-and S-wave velocities for all
constant pore pressure runs for the porous ceramic rod........................... 228
A. 15 Observed and Biots modelled P-and S-wave velocities for all constant
pore pressure runs for the porous ceramic rod ......................................... 229
A.16 P-and S-waves amplitude spectra for all constant pore pressure runs for the
A. 17 Observed P-and S-wave differential attenuation coefficient for all constant
pore pressure runs of the porous ceramic rod .......................................... 231
constant pore pressure runs of the porous ceramic rod ............................ 232
A.19 Normalized water saturated P-and S-wave waveforms as a function of
pore pressure for the porous ceramic rod. ................................................ 233
A.20 Dry and water saturated P-and S-wave velocities with error of the porous
ceramic rod ............................................................................................... 234
for the Berea sandstone of the first run .................................................... 236
for the Berea sandstone of the second run ............................................... 236
A.23 Dry P-and S-wave wave velocities with errors as a function of confining
pressure for the Berea sandstone of the first and second run ................... 237
2
saturated Berea
sandstone for all measured constant temperature runs............................. 240
A.25 Dry and corrected CO
2
constant temperature runs of the Berea sandstone ................................... 242
A.26 Observed and Gassmanns modelled P-and S-wave velocities for all
A.27 Observed and Biots modelled P-and S-wave velocities for all constant
temperature runs of the Berea sandstone ................................................. 247
A.28 P-and S-waves amplitude spectra for all constant temperature runs of the
Berea sandstone ....................................................................................... 248
A.29 Observed P-and S-wave differential attenuation coefficient for all constant
temperature runs of the Berea sandstone ................................................. 249
2
saturated Berea
sandstone for all measured constant pore pressure runs .......................... 253
A.32 Dry and corrected CO
2
constant pore pressure runs of the Berea sandstone ................................. 254
A. 33 Observed and Gassmanns modelled P-and S-wave velocities for all
A. 34 Observed and Biots modelled P-and S-wave velocities for all constant
pore pressure runs of the Berea sandstone ............................................... 259
A.35 P-and S-waves amplitude spectra for all constant pore pressure runs for the
Berea sandstone ....................................................................................... 260
A. 36 Observed P-and S-wave differential attenuation coefficient for all constant
pore pressure runs of the Berea sandstone ............................................... 261
A.38 Normalized water saturated P-and S-wave waveforms as a function of
pore pressure for the Berea sandstone...................................................... 263
A.39 Dry and water saturated P-and S-wave velocities with error of the Berea
sandstone ................................................................................................. 264

List of Symbols

Symbol Description
A
0
Initial wave amplitude
A Wave amplitude, Surface area
A() Amplitude spectrum of a signal
A, Q, R Biot parameters
ber Real part of Kelvin function
bei Imaginary part of Kelvin function
C Elastic stiffness constant
C Storage modulus
C Loss modulus
D
pore
Pore throat diameter
D() Amplitude spectrum of receiver
d Grain diameter
E Youngs modulus
E
stored
Stored energy
E Energy dissipated in a cycle
f Frequency
f
C
Critical frequency
F() Viscosity correction factor
G Geothermal gradient
G() Geometric effects
g Gravity, pore size parameter
J
n
Bessel functions of order n
K Bulk modulus
K
dry
Bulk modulus of dry sample
K
f
Bulk modulus of pore fluid
K
s
Bulk modulus of mineral grains
K
sat
Bulk modulus of saturated sample
k Real wavenumber
k
*
Complex wavenumber
L, L
s
Sample length
L
C
Actual path through network of pores
L
0
Straight path through sample
L Amount of sample shortening
l Path traveled by electrical current
M P-wave modulus
m Mass of sample
P Pressure
P
c
Confining Pressure
P
d
Differential Pressure
P
p
Pore Pressure
P Change in pressure
Q Quality factor
q Darcy velocity
R
Electrical Resistance, Reflection
coefficient
R
saline
Resistance of saline
R
sample
Resistance of saturated sample
S() Amplitude spectrum of outgoing pulse
T Temperature, period, response function
T Change in temperature
t time
t
b
Travel time through buffer
t
bs
Travel time through buffer and sample
t
s
Travel time through sample
t
phase
Phase lag in time between signals
u Displacement
V volume
V
e
Envelope volume
V
g
Grain volume
V
group
Group velocity
V
HG
Volume of mercury
Vp P-wave velocity
V
phase
Phase velocity
V
S
S-wave velocity, Volume of sample cell
V
R
Volume of reference cell
x Traversed distance
z Depth

Attenuation coefficient, tortuosity,
linear expansion coefficient

Biot coefficient

Coefficient of fluid content, surface
tension of mercury
Delayed response, error
strain
Fluid viscosity

Volumetric strain, contact angle of
mercury
Permeability
Lam first parameter
Lam second parameter Shear modulus
dry
Shear modulus of dry saturated
sat
Shear modulus of saturated sample
Poissons ratio
Porosity
c
Critical Porosity
Phase difference of two waveforms
Density, Electrical Resistivity
b
Bulk density
g
Grain density
12
Coupling mass density
Stress
Tortuosity
Angular frequency

1

Chapter 1
Introduction

1.1 Climate Change
Scientists have long announced the observed increase of the overall
Earths surface temperature in the last century. According to the Fourth
Assessment Report (AR4) released in 2007, the Intergovernmental Panel on
Climate Change (IPCC) concluded that the average global surface temperatures
have increased by 0.74 0.18C in the last century, with accelerated warming
during the last two decades. Increases in sea levels are consistent with the
warming; global sea levels have risen at an average rate of 1.8 mm/yr from 1961
to 2003, with the average rate accelerating to 3.1 mm/yr from 1993 to 2003. It
has been estimated that thermal expansion accounts for 57% of the sea level
increase and the remaining 43% is from the melting of glaciers, ice caps, and
polar ice sheets. In addition to the average temperature rise, evaporation increase
has caused winds and ocean currents to shift impacting global weather systems.
Precipitation has significantly increased in eastern parts of North and South
America, northern Europe, and northern and central Asia while precipitation has
decreased in the Sahel, the Mediterranean, southern Africa and southern Asia over
the last century. Severe weather events such as storms, floods, hurricanes,
tornadoes, and blizzards have been more frequent and more intensive in the last
few decades. Changes in ecosystems have also been documented with the
warming. More evidence are showing an earlier arrival of spring events (bird
migration, leaf unfolding), a poleward shifts of plants and animals, changes in the
abundance of algal, plankton, and fish in high latitude oceans, and an earlier
migration of fish in rivers.
2

The Earths climate depends on a balance of energy from the amount
received from the sun and the amount radiated back out to space. The most
favoured explanation for this climate change concerns greenhouse gases (GHG)
such as carbon dioxide (CO
2
), methane (CH
4
), nitrous oxide (N
2
O), and
halocarbons (gases containing fluorine, chlorine, or bromine). Greenhouse gases
tend to trap heat in the atmosphere, and therefore increased concentrations of
GHG in the atmosphere would consequently yield a warmer surface on the Earth.
Some of these GHG are released into the atmosphere naturally but human
activities have significantly increased the emissions. CO
2
is the most important
anthropogenic GHG by contributing 77% of the total anthropogenic GHG
emissions in 2004. CO
2
emissions have dramatically increased since the start of
the industrial revolution in 1750. Preindustrial levels of atmospheric CO
2
were
270 parts per million (ppm) compared to current date concentrations of 390 ppm,
and levels are currently increasing at 2 ppm per year (figure 1.1). This increase is
blamed on anthropogenic releases through the combination of fossil fuel usage
and land practices, with the former holding a more dominant role. The annual
emission in 2004 was 38 gigatonnes of CO
2,
mainly coming from electricity
production, transportation, heating buildings and industrial processes. Energy
demand is expected to rise as global population grows and developing nations
expand their economies. Energy consumption is predicted in some scenarios to
double between 2010 and 2050. Fossil fuels (oil, natural gas, and coal) are
currently the primary source for power generation and account for 85% the
worlds energy needs. With such substantial energy usage predicted for the future,
the current dependence on using fossil fuels to drive the world economies, and the
growing evidence linking atmospheric CO
2
concentrations to climate change,
methods to mitigate CO
2
emission are direly needed to avoid catastrophic
environmental effects.
The different CO
2
mitigation methods can be classified as either reducing
CO
2
sources or increasing CO
2
sinks. Increased energy efficiency, energy
conservation, low- or zero- carbon alternatives for energy generation, and carbon
capture and storage (CCS) are all ways to mitigate CO
2
emissions. As research
3

Figure 1.1: (Figure reproduced from IPCC (2007))* Atmospheric concentrations of CO
2

inferred from ice core measurements (different colored symbols for different core studies)
and from atmospheric measurements (red line).

and development continues on increasing the supply of alternative energy to a
level that can support the worlds growing population and economies, CCS
methods are deemed as the next most effective way, albeit stopgap, to combat the
issue at hand. A variety of CCS techniques exist; each utilizing a different
component of the Earths natural system and each with their own problems to
overcome to be effective. The different carbon sequestration methods available
will be briefly described in the next chapter. Geologic sequestration is the most
practiced storage technique to date and involves the injection of concentrated CO
2

into a subsurface rock formation for containment. Given the range of temperature
and pressure conditions possible in the upper sedimentary basin, CO
2
can exist as
a gas, liquid, or a supercritical fluid when in the subsurface. The largest factor
controlling the acceptance of geological storage is the safe and secured
containment of the injected CO
2
. Therefore the ability to monitor, to verify, and
*Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II, and III to the
Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Figure 2.3. IPCC,
Geneva, Switzerland.

4

to account for the injected CO
2
, is very important. Geophysics will play a key role
in site selection through reservoir characterization and, in the monitoring and
verification of the injected CO
2
in deep geologic formations over time.

1.2 Aim of Research and Chapter Description
Time lapse seismic surveys are an integral component of any monitoring
and verification program in a geological sequestration projects. Seismic methods
are used because the seismic behavior of a rock formation is not only sensitive to
its mineralogical composition and porosity, but also its pore fluid content. Using
seismic methods for subsurface monitoring has two aims in geological
sequestration: 1.) qualitatively locate and track the movement of the CO
2
plume
,
and 2.) quantitatively determine the amount of CO
2
in place. CO
2
s seismic
properties, bulk modulus and bulk density, are considerably different than those of
other in-situ pore fluids such as oil and brine. Therefore introduction of CO
2
into
the pore space would yield a change in the overall seismic signature allowing the
movement of the subsurface CO
2
to be qualitatively tracked from seismic surveys.
Quantitatively tracking by determining the amount of CO
2
from the changes in a
seismic signal however is a much harder issue to deal with; seismic velocity
change has been demonstrated to be very slight for CO
2
saturations past ~20%.
Furthermore, unlike typical pore fluids encountered in-situ, CO
2
can exist as three
different phase state which adds complexity to the accurate interpretation of
seismic. Using other seismic attributes to accurately determine the saturation
levels of CO
2
and thus quantifying CO
2
currently are subjects of ongoing
research.
Understanding the effects of fluid saturation and its variability on seismic
responses is crucial to the improvement of monitoring the movement and to the
quantification of CO
2
. CO
2
rock physics will undoubtedly play an essential part
in accurate seismic interpretation over geological sequestration fields. Rock
physics will enable us to relate the geological properties (e.g. porosity, lithology,
saturation) of a rock at certain physical conditions (e.g. pressure, temperature)
with its the corresponding elastic and seismic properties (e.g. elastic modulus,
5

velocity, impedance). The earliest systematic measurements of variations in the
acoustic properties of rocks appeared in the 1950/1960 period and in the last 30
years, many studies have been made to understand and relate the physical
properties of rocks to geophysical observations. Relationships between seismic
wave velocities (P- and S- wave) and porosity, saturation, and effective pressure
have been largely investigated, especially on sandstones. Some of these
relationships involve the effects of pore pressure, temperature, lithology, pore
structure, and fluid type. Studies on the subject of intrinsic attenuation have also
been investigated, but to a lesser extent than wave velocities. The fluids used in
rock physics studies typically involve water/brine or oil, and sometimes glycerol.
The first known rock physics study using CO
2
as a pore fluid was inspired by its
role in enhanced oil recovery processes and was not conducted until 1989 by
Wang and Nur. Only recently have more CO
2
rock physics studies appeared but
fundamental studies on the effects of CO
2
fluids under in-situ pressure and
temperature conditions are still lacking. All related studies to date involve the
injection of CO
2
into a porous media pre-saturated with another in-situ fluid and
the acoustic variations observed are usually from a combination of pore pressure
and fluid substitution effect. Problems with these studies are the partial saturation
of CO
2
is usually unknown and the observed changes solely due to CO
2
cannot be
quantitatively determined.
To differ from previous studies, my work involves full CO
2
saturation and
pore fluid effects will be separated from pore pressure effects. Here, a systematic
study is conducted on P- and S-wave velocities and attenuations in synthetic and
real rock samples under in-situ conditions when fully saturated with CO
2
in the
gas, liquid, or supercritical fluid phase states. Realistically, full saturation is
unlikely in geological sequestration projects, but it is important to have an end
member study in order to constraint the more complex and more probable cases of
partial saturation.
Chapter 2 provides background information on carbon storage. A brief
overview is provided on the types of carbon storage available with special
emphasis given to geological CO
2
sequestration. The concerns of safety and
6

monitoring associated with geological sequestration are discussed, along with the
monitoring methods that are currently available. Examples of the various seismic
methods that have been employed in geological sequestration projects around the
world are given. This chapter concludes with a literature review on previous
laboratory rock physics work that have been conducted with CO
2
.
Chapter 3 discusses the theoretical background that is relevant to the work
conducted in this thesis. To begin with, CO
2
s thermodynamic behavior is
introduced and the plausible phase states that can occur in a sequestration project
are discussed. Following this brief introduction on CO
2
, the basic concepts
behind the theory of elasticity, wave propagation, and the effect of viscoelasticity
on a materials behavior are provided. The theoretical fluid substitution
formulations of Gassmann and Biot, their assumptions, and their applicability are
reviewed and contrasted. Finally, an introduction to the foundation of the
ultrasonic pulse transmission laboratory technique, an introduction to the analysis
of the resulting wave velocity and wave attenuation, and an example to
demonstrate attenuation analyzes conclude the chapter.
Chapter 4 focuses on the experimental setup and the laboratory procedures
followed to obtain the laboratory results for this thesis. The construction of
piezoelectric transducers, the required sample preparation, and the experimental
configuration for the measurements are presented. A detailed description of the
experimental procedure and of the data analysis followed in this study is also
provided. The errors associated wave velocity and differential attenuation
coefficient are assessed in this chapter.
Chapter 5 examines the petrophysical properties of the samples used for
measurements. The bulk density, grain density, porosity, pore size distribution,
air permeability, and tortuosity were addressed using a variety of laboratory
techniques. In addition, scanning electron microscope images were used to reveal
the microstructure of the samples.
Chapter 6 and chapter 7 presents the laboratory results and discuss the
findings of the synthetic and non-synthetic sample, respectively. The wave
velocitys dependencies on pressure, temperature and fluid-type saturation are
7

highlighted. In addition, the laboratory results are compared to the theoretical
predictions of Gassmann and Biot. The similarities and differences between the
observed results of the two samples are given in chapter 7.
Chapter 8 is the concluding chapter of this thesis where an overview of the
work is provided, the contributions of the findings are discussed, and directions
for future work are suggested.

8

Chapter 2
Background on Carbon Storage

In the previous chapter, a brief introduction on climate change was made.
Increasing energy efficiency, energy conservation, alternative energy
development and usage, and carbon capture and storage are ways to help reduce
greenhouse gas emissions. Carbon capture and storage involves capturing
anthropogenic CO
2
(usually from a large nearby point source) and storing such
that it does not enter the atmosphere. In this chapter issues related to CO
2
storage,
monitoring options and studies pertaining to CO
2
laboratory velocity
measurements in porous media will be briefly reviewed. The material presented
here in this chapter related to geological sequestration is the underlying
motivation for my work.

2.1 Types of CO
2
sequestration
The scientific community has suggested various methods to store the
captured CO
2
in response to mitigating global warming associated with increased
anthropogenically induced CO
2
. The feasibility of each proposed method depends
on its scale of implementation, costs, additional research and development, and
permanence. The more promising methods are oceanic, terrestrial, mineral
carbonation, and geological sequestration.

1. Oceanic
Atmospheric CO
2
is regularly taken up in the ocean waters naturally through
dissolution and it is estimated that 500 gigatons (Gt) of the 1300 Gt of emitted
anthropogenic CO
2
have been naturally taken up by the worlds ocean in the last
9

two centuries. It is also estimated that the worlds ocean has a capacity to hold
39,000 Gt-C (gigatons of carbon) and is therefore considered a prime candidate
because of this large storage potential (Rackley, 2010). Storage options in the
ocean consist of injecting CO
2
into the ocean, from a fixed pipeline or from a
moving ship in waters greater than 1000 m depth, and storage is attained by direct
dissolution in the seawater.
One alternative to direct dissolution, storage in the ocean can be attained
by injecting CO
2
at depths greater than 3000 m where the CO
2
can pool and form
a lake in the deep waters of the ocean. Under the pressure and temperature
conditions of very deep waters, CO
2
is denser than the surrounding saline
seawater and therefore by the effects of gravity the injected CO
2
is presumed to
stay on the ocean floor (Figure 2.1). Over time the CO
2
will dissolve and disperse
in the surrounding waters.
Both methods of oceanic disposal of CO
2
are not permanent because the
ocean plays a major part in the global carbon cycle. The surface of the ocean is in
continual exchange with the atmosphere for CO
2
and therefore the dissolved CO
2

in the ocean will eventually equilibrate with the atmospheric CO
2
. Furthermore,
critics of this method argue that the injection of large amounts of CO
2
into the
ocean will cause major negative ecological impact such as the increased
acidification of the oceans from the dissociation of CO
2
in water and the extent of
biological impact is unknown ( Seibel and Walsh, 2001, Fabry et al, 2008).
Oceanic disposal, therefore may create additional environmental problems while
only temporary solving the original issue.

2. Terrestrial ecosystem:
CO
2
is naturally taken up from the atmosphere by photosynthesis. Some of this
captured CO
2
will be released to the atmosphere through respiration while some
may reside longer in the soils and biomass. A rough estimate of the net retention
of CO
2
due to terrestrial processes is 2.8 Gt-C per year (Rackley, 2010).
Terrestrial sequestration of CO
2
can be achieved by increasing the transfer of CO
2
from the atmosphere into soils and vegetation, and by limiting CO
2
emissions
10

Figure 2.1: (Figure reproduced from Rackley (2010))* CO
2
versus sea water density over a
range of ocean depth.

from terrestrial ecosystems into the atmosphere.
To be effective, CO
2
needs to be held in long-lived terrestrial carbon pools
such as above- and below ground biomass (large capacity vegetations such as
trees), long-lived products from biomass (from wood), and stabilized organic
carbon fractions in soils. Practices in minimizing soil disturbance (no-till farming
and soil erosion control), increase return of crop residues to soils, afforestation on
degraded and marginal agricultural soils, growing plants with a large capacity for
carbon storage, and refined logging techniques to reduce forest disturbances are
ways to enhance carbon sequestration.
Current problems associated with this type of sequestration are
determining the storage permanence and quantifying the global impact from the
ecosystem on climate change with confidence. The interactions and feedbacks
between ecosystems and the atmosphere is complex, and the current
understanding of the factors that may interrupt and affect intermediate processes
is limited.
*This figure was published in Carbon Capture and Storage, Stephen A. Rackley, Copyright
Elsevier (2010).
11

3. Mineral carbonation:
In mineral carbonation, CO
2
is locked away by reacting it with naturally occurring
Mg and Ca minerals from highly abundant silicates and oxides in the Earths crust
to form stable carbonates. The conversion process is an exothermic reaction and
therefore is thermodynamically favorable because the formed carbonates are in a
lower energy state than free CO
2.
Geologically, this is a naturally occurring
process in respect of weathering and formation of rocks and occurs on a scale of a
hundred thousand years. For the purpose of CO
2
sequestration, this process can be
accelerated from geological time scales by exposing the materials to an optimal
reaction temperature and pressure conditions. Also, pre-treatment (heat treatment,
grinding, etc) and additives or catalysts (aqueous solutions with Na
2
CO
3
, molten
MgCl
2
salts, hydrochloric acid) can be applied to the raw materials and have been
shown to facilitate the reaction time in producing carbonates. The stable
carbonates formed can be stored in the mines where the original silicates or oxides
were extracted. The new products are volumetrically larger than the original
minerals, therefore terrain profile change maybe expected. Mineral carbonation is
the only storage method where leakage is not a concern because the produced
carbonates are in an inert, stable, solid form.

4. Geological Sequestration:
Geological sequestration is any means of injecting CO
2
into the subsurface and
containing it in a rock formation. Suitable rock formations for storage are ones
with high pore volume, located at depths greater than 1000 m, and are capped by a
layer of non-porous rock above. A typical working scheme for geological storage
involves capturing CO
2
at a point source (e.g. coal plant, drilling rig platform),
processing for near pure CO
2
, shipping to injection site via pipeline, pressurizing
into liquid or supercritical form and injecting into suitable subsurface formations.
Depleting oil and gas reservoirs, coal beds, and saline aquifers are all suitable
sites for sequestering CO
2
, each with a suite of trapping mechanisms.
Globally, depleted oil and gas reservoirs have an estimated storage
capacity between 675 to 900 Gt-C (IPCC, 2005). The estimated storage in
12

western Canada (British Columbia, Alberta, Saskatchewan, and Manitoba) is 1.04
Gt-CO
2,
with Alberta contributing 75% to this total (Bachu and Shaw, 2004).
Depleted oil and gas reservoirs are favorable sites since they have been highly
characterized by the hydrocarbon industry, and the sealing cap has prove to be
effective from housing hydrocarbons for millions of years before oil or gas
production.
Also, some oil reservoirs considered as depleted in the past, usually still
contain a significant amount of residual oil for production. This residual oil can
be recovered by injecting fluids such as CO
2
into the reservoir to build up in situ
pressure and to reduce the viscosity of the residual oil for easier mobilization.
Some of this injected CO
2
will be produced along with oil, while some will be
permanently stored from this process. Enhanced oil recovery using CO
2
has been
in use since the 1970s and its impact as a carbon storage option is currently the
main focus of the IEA GHG Weyburn-Midale EOR project in Saskatchewan,
Canada. It is argued that the usual high cost of geological CO
2
sequestration can
be offset from the profits of the recovered oil. The trapping mechanism for
depleted oil and gas reservoirs initially consists of structural trapping via
impermeable cap rock as the injected CO
2
buoyantly rises in the geologic
formation, followed by solubility trapping from the dissolution of the trapped CO
2

with the in-situ pore fluids. Given time, mineral trapping will be the ultimate
carbon locking mechanism for carbon storage where new and stable minerals will
form from the chemical reactions between the dissolved CO
2
and surrounding
minerals.
Saline aquifers hold un-potable waters unsuitable for consumption and
irrigation, but these aquifers could also become important for carbon storage with
an estimated capacity to sequester between 1000 and 10,000 Gt-C. The trapping
mechanisms of CO
2
in saline aquifers are essentially the same as in depleted oil
and gas reservoirs except saline aquifers have an additional trapping mechanism
called residual-phase trapping. Residual-phase trapping is also known as capillary
trapping and this process involves the imbibition of water into the CO
2
plume.
13

This mechanism will immobilize the trailing edges of CO
2
and will slow up-dip
migration.
Since they did not contain hydrocarbons, saline aquifers are not as well
characterized as to hydrocarbon reservoirs due to the lack of obvious financial
benefits. However saline aquifers have the advantages of vastly larger reservoir
capacity, lower risk of leak paths from the fewer penetrations wells, and are more
widespread in occurrences. The probability that a suitable saline aquifer
sequestration site is close to a large CO
2
producer is also higher leading to
reduced transportation cost. For example, at Statoils Sleipner field sequestration
project, the saline aquifer used for CO
2
injection is in the same geographic
location as the natural gas reservoir that produces the CO
2,
but located above the
reservoir. The CO
2
is produced, removed from the natural gas, and injected into
the saline aquifer from the same rig, on adjacent platforms (Kongsjorden et al,
1998).
Deep un-mineable coal beds or coal beds used for methane production are
also suitable geologic options for sequestration. In coal beds, the retention of CO
2

is accomplished by adsorption trapping from the preferential adsorption of CO
2

onto coal than that of the natural occurring methane on coal. It has been shown
that CO
2
s affinity to coal is at least twice that of methane due to its larger heat of
adsorption (Liang, 2003; Rackley, 2010). Natural occurring methane on coal will
then be desorbed and displaced by the inject CO
2
thereby allowing for coal bed
methane recovery to occur while sequestrating CO
2
. Similar to EOR, enhanced
coal bed methane (ECBMR) has financial benefits to help offset the costs for
sequestration. A notable problem with ECBMR using CO
2
is that the
permeability of the coal beds may decline due to the swelling of the coal matrix
from the absorption of CO
2
. This will restrict the flow of CO
2
for sequestering
and impede the recovery of the displaced methane. The carbon sequestering
capacity of un-mineable coal beds is up to 200 Gt-C (IPCC, 2005).

14

2.2 Examples of Geological Sequestration
Currently, geological sequestration appears to be the most economically
and technically feasible long term storage for achieving significant reductions in
atmospheric CO
2
emissions. It is the only carbon storage method applied on a
commercial scale to date. Worldwide there are numerous geologic storage
projects varying on small and large scales, mostly involving oil fields for EOR or
saline aquifers. Currently there are 4 commercial scale storage projects operating
around the world aimed at greenhouse gas mitigation.

Sleipner:
The Sleipner project is the worlds first industrial-scale CO
2
injection project,
operated by StatOil and partners offshore in the North Sea. In 1996 CO
2

injections began in the Slepiner field into a major saline aquifer, the Utsira
formation. The Utsira formation is a 200 m thick sandstone formation located
800 m below the seafloor and is capped by a thick shale layer while containing
multiple thin intra-reservoir shale layers. The CO
2
used for injection is produced
from a natural gas reservoir in the Heimdal formation that is located in the same
field but 1200 m deeper. The CO
2
is stripped from the natural gas produced and
processed on the same platform for injection. Today, about 12 Mt-CO
2
have been
injected at a maximum rate of 1 Mt per year. The injections are expected to
proceed for 20 years.

Weyburn:
The IEA GHG Weyburn-Midale CO
2
Monitoring and Storage Project is an
international collaboration with Cenovus Energys and Apache Corporations
enhanced oil recovery operations in Saskatchewan, Canada. CO
2
injections
started in late 2000 in the Weyburn and the Midale oil fields. Injections at both
fields are in carbonates at depths of about 1500 m with general thickness less than
30 m. The reservoirs have caprocks composed of anhydrites from the Midale
Evaporite and of anhydritized dolostones, which in turn are overlain by aquitards.
Currently over 15 million tons have been stored between the two fields at a rate of
15

more than 2 million tons per year. The CO
2
for the injection comes from a
synfuel plant located in north Dakota that supplies 2.7 million cubic meters of
CO
2
per day to the Weyburn-Midale fields via a 325 km pipeline. It is estimated
that approximately 35 Mt of anthropogenic CO
2
will have been stored in the
reservoir by the end of the operations.

In Salah:
The In Salah CO
2
storage project began operations in 2004 and is a joint venture
between BP, Sonatrach, and StatOil located in Algeria. CO
2
is injected into a 20
m thick saline formation located 1900 m below the ground, and is positioned
down dip from the producing gas horizon. The storage formation, the Krechba
reservoir, is composed of Carboniferous sandstone and is overlain by 950 m of
impermeable mudstone, the same caprock that keeps the up dip natural gas in
place. The CO
2
used for injection is produced from surrounding gas fields,
processed at nearby gas plants and transported to the injection site. By 2008, 2.5
Mt-CO
2
have been stored with a future rate anticipated of 1.2 Mt per annum.

Snohvit:
Operated by StatOil, the first CO
2
injection at Snohvit in the Barent Sea was in
April 2008. The CO
2
is stored in a sandstone layer can the Tubaen formation that
is located 2500 m beneath the seabed and is underneath the gas containing
formations where the CO
2
is produced from. The Tubaen formation is between
45 m to 75 m thick and is sealed by an overlying shale formation. It is estimated
that 0.7 Mt-CO
2
will be stored annually.

2.3 Safety and Monitoring
The definition of storage permanence or successful sequestration is one
that retains 99% of the injected CO
2
over 1000 years (IPCC, 2005).
Understanding the geologic, hydrologic, geomechanical and geochemical
processes in the subsurface are the keys to a successful and secure sequestration,
as these are important factors that will influence the fate of the sequestered CO
2
.
16

But nonetheless, monitoring will be required and will always be an essential
component to any CO
2
sequestration project. The greatest resistance for a
widespread implementation of geological sequestration is the public acceptance of
the associated risks. Concerns with geological sequestration involve the safety of
the practice for both the public and nearby natural resources, and its effectiveness
in mitigating CO
2
levels in the atmosphere.
Carbon dioxides effect on human health at low concentration is benign,
composing 0.039% of the atmosphere, which is approximately 390 ppm and is
currently increasing at a rate of 2ppm/yr. At an increase to 1000 ppm, health
symptoms including headaches, nausea, and drowsiness can appear in some
people. The Occupational Safety and Health Administration maximum
acceptable level for long exposures is 5000 ppm. Exposures at levels surpassing
5000 ppm can lead to oxygen deprivation resulting in brain damage, coma or
death.
Another concern with the leakage of geologically sequestered CO
2
is the
impact on local environments. CO
2
seepage to the surface, will affect all living
organisms apart from just humans. Local ecologies, habitats such as plant health
and plant species distributions can drastically change with an influx of CO
2
. Tree
kills and tree plant stress caused by elevated CO
2
soil concentration levels, caused
by subsurface magma interactions in Mammoth Mountain, CA, demonstrate the
possible environmental catastrophic effect if CO
2
were to escape to the surface
through hidden faults. In the subsurface, migration of CO
2
to a fresh water
aquifer can contaminate the potable groundwater resource by creating an acidic
solution when CO
2
dissociates in water.
If in an environment where both human and natural resources are not in
harm, monitoring the subsurface is still essential. If a storage project is failing,
this needs to be known so the injection process could be stopped immediately to
avoid wasting resources and effort. For governments, policy makers, and
businesses, knowing how much CO
2
have been injected and retained in the
subsurface will allow for inventory verification for national accounting, carbon
fines and taxes assessments, and carbon credit trading.
17

Monitoring CO
2
sequestration projects will then serve a few purposes in
this regard. Successful monitoring will detect and allow for advance action in
response to a leakage that may put human health or natural resources in jeopardy,
and in otherwise successful containment, serve the purpose of inventory
verification and the assurance of the storage integrity.
There are various measurement technologies that can be applied for
monitoring in the case of leakage. Essentially they can be split between direct and
indirect measurement methods.

2.4 Direct monitoring measurements
For direct measurements, CO
2
can be sampled in the air, water, or soil. In
the air at the storage sites, sensors similar to ones found in occupational settings
such as infrared gas analyzers are used that can allow for continuous and real time
monitoring of the level of CO
2
. Gas sampling bags, gas chromatography, and
mass spectrometry are also used in checking air quality (Benson and Myer, 2002).
In water, monitoring can occur from geochemical sampling from the surface or at
observational wells. In observation wells, fluid samples can be collected and
analyzed for changes in brine composition and for the presence of natural and
injected tracers. By using pH levels, gas compositions, and tracers, CO
2

breakthrough at the well location can be determined as well as the possible
chemical reactions that have taken place in the subsurface from the CO
2
-water-
rock interaction (Gunter et al., 2001, Blencoe et al., 2001). For soil monitoring
gas surveying, probes or accumulation chambers are placed in or on the soil, and
samples are analyzed periodically to determine gas composition and fluxes.
Seasonal and local variations of CO
2
flux need to be accounted for.

2.5 Indirect monitoring measurements (Geophysical)
The methods listed above are only effective if there is a CO
2
leak or there
is access to an observational well. Therefore direct methods limit advance
preparation in the case of a leakage and their application is inadequate for
providing information on the subsurface movement of the CO
2
away from the
18

wells. To remedy these shortcomings, indirect monitoring techniques, namely
geophysical methods, can be applied in conjunction with direct measurements.
The various geophysical methods that can be applied to detect underground CO
2

from the surface are land surface deformation, electromagnetic, gravitational, and
seismic.
Geophysical methods are the most favored and applied monitoring
techniques in geologic sequestration because of the widespread use and
experience from their application in the hydrocarbon industry. The feasibility of
any geophysical technique depends on the inherent resolution of the technique and
the magnitude of change in the measured geophysical properties produced by
increasing the CO
2
concentration.

2.5.1 Geodetic
Subsurface pore pressure changes causing land surface deformation can
occur when injecting CO
2
underground. This displacement of the Earths surface
can be used to infer the migration of CO
2
and can be monitored using tiltmeters,
global positioning system (GPS) and satellite/airborne interferometric synthetic
aperture radar (InSAR).
Tiltmeters can be placed on the ground surface or in boreholes and have
the potential to measure very small tilts, as small as one nanoradian (Sweatman
and McColpin, 2009). Modern day tiltmeters use a highly sensitive electrolytic
bubble level where an arrangement of electrodes senses the precise position of the
bubble in the electrolytic solution. In 2010, tiltmeters were employed for the first
time in a commercial scale injection project where 71 tiltmeters were deployed in
the In Salah project over one of the CO
2
injectors.
GPS, a satellite-based positioning system, is well-known for providing
reliable position coordinates when 4 or more satellites are accessible. To achieve
higher precision than the conventional GPS, differential GPS (DGPS) is being
utilized in CO
2
sequestration projects. In DGPS more than one GPS receiver is
used. One of the receivers is placed in an area with minimal displacement while
the other receiver(s) is/are placed where displacement is expected. DGPS allows
19

for compensation of atmospheric variations and errors, permitting millimeter
horizontal and vertical readings to be achieved.
InSAR can produce high spatial resolution surface deformation maps (a
typical coverage is 10, 000 km
2
) and is capable of detecting up to millimeter-scale
changes on Earth. A beam of microwave energy is projected to the surface of the
earth and when this energy is reflected back, images of the surface can be
generated. The tilt of a surface caused by volumetric change can be measured by
the phase shift between reflected pulses recorded at two different times. InSAR is
currently used for monitoring at the In Salah CO
2
injection project, and have
detected uplifts centered on the injection wells up to 21 mm after 3 years of
injection (Falorni et al, 2010).

2.5.2 Electromagnetic (EM)
Electromagnetic techniques are responsive to the electrical resistivity of
earth materials where the electrical resistivity of a rock formation is affected by its
porosity, fluid saturation, and pore fluid resistivity. Therefore monitoring CO
2
in
saline aquifers based on electrical methods is feasible, because changes in brine
saturation would be notable due to the highly-resistive electrical nature of CO
2

with respect to the surrounding brine. In petroleum reservoirs however, both
hydrocarbons and CO
2
are electrically resistive. As a result of this small electrical
contrast between the original and the new pore fluid, the detection of CO
2
would
be less sensitive in a hydrocarbon-bearing reservoir than in a saline aquifer.
Using EM as a monitoring method is the most optimal in saline aquifers.
Electromagnetic methods involve measuring an induced or secondary
electric and magnetic field from the subsurface generated by a time varying
source or primary electric field. The measured fields hold information about the
electrical distribution of the subsurface. A commonly exercised electromagnetic
method in geologic sequestration is the cross well survey which consist of two
adjacent wells, one containing transmitters and the other containing receivers.
The tomography of the conductive structure between the transmitting and
receiving wells can be mapped, and the ongoing replacement of original pore fluid
20

with CO
2
can then be monitored. Cross well EM imaging have been successful in
measuring subsurface distribution of CO
2
in Frio brine formation in south Texas
(Hovorka et al., 2005) and the migration of CO
2
during an EOR flood in Lost
Hills oil field in southern California (Hoversten et al, 2003). In west Texas
Vacuum Oil field CO
2
flood, Wilt et al (2003) used EM to map CO
2
floods and
they have shown good correlation between porosity distribution and resistivity
variation.

2.5.3 Gravitational
Gravity measurements detect variations in the subsurface rock and pore
fluid density from gravitational acceleration changes. CO
2
for most of the depth
interval of interest for sequestration is less dense than that of oil or water. So an
introduction of CO
2
into a brine or oil filled reservoir will decrease the overall
bulk density which in turn will produce a change in the gravitational attraction
from the reservoir. For optimal detection and the success of surface gravity, large
material contrasts such as high porosity, shallow reservoir and high temperatures
causing low densities would be ideal. From the size of the gravity variations,
subsurface mass changes can be inferred allowing quantification of the amount of
CO
2
in place. From spatial gravity variations, the lateral movement and
distribution of CO
2
can be reflected.
Gravity stations are easily deployable at the surface and are relatively
cheap for large spatial coverage. However, because gravity anomalies decay with
the inverse square of the distance from the source, measured responses at the
ground surface from the changes in gravitational attraction of the reservoir would
yield weak response and could quite possibly be in the background noise level
(Gasperikova et al., 2008, Hare et al., 1999). Alternatively, gravity stations can
be placed inside boreholes in existing wells, thereby reducing the distance to the
reservoir allowing for larger magnitude of change to be detected. However this is
feasible only at the cost of substantially reduced data coverage due to the limited
distribution of existing borehole.
21

Modelled time lapse results using Schrader Bluff oil field in Alaska by
Gasperikova and Hoversten (2006) over a time interval of 20 years show direct
correlation of spatial pattern change in the vertical component of gravity with net
change in reservoir density. In the offshore sequestration project the Sleipner
field, gravimetric measurements were conducted on the seafloor in 2002 and
2005. Notable gravimetric change is observed from the time lapse measurements
and is being used to constrain the in situ CO
2
density estimates (Nooner et al,
2007).

2.6 Seismic
Of these geophysical methods, seismic imaging is by far the most highly
developed and extensively applied. Seismic resolution is generally recognized to
be higher than gravity and electromagnetic methods and is deemed as a feasible
time-lapse method for monitoring (Arts et al. 2002, Newmark et al. 2002,
Hoversten et al 2003, Gritto et al, 2004, Saito et al 2006, Shi et al 2007). Seismic
waves depend on both the density and the elastic stiffness of the medium of
propagation where the pore space content can have a large effect on these
parameters. Varying pore space content can lead to sharp contrasts in acoustic
impedance, consequently generating distinct high amplitude reflections. CO
2
is
generally more compressible and, as stated earlier, is less dense than oil or brine.
Seismic methods offer the potential to identify migration or leakage from the
storage reservoir and the potential to quantify the amount of CO
2
in place based
on reflected amplitudes and velocity pushdown effects. This quantitative
capability of seismic methods is currently under ongoing research.
The variations of seismic methods that can be used for monitoring are
surface seismic, vertical seismic profiling (VSP), and cross well seismic. For
surface seismic, the source and the receivers are placed along the surface and 2D
or 3D data can be acquired. In terms of lateral coverage, surface seismic can
provide the most amount of spatial and geometric information for the CO
2
plume,
the reservoir and the overburden. However, surface seismic when compared to
the other two wellbore seismic methods, is the lowest in resolution. In vertical
22

seismic profiling, the source is located on the surface while the receivers are
placed along the wellbore and higher resolution of the near borehole environment
can be obtained. Even higher resolution can be obtained by cross well seismic.
Cross well seismic, similar to cross well EM, requires a pair of boreholes close to
or in the reservoir with a string of sources placed in one and a string of receivers
are in the other.

2.6.1 Surface seismic
Time lapse surface seismic is the most commonly used monitoring
technique in CO
2
sequestration or CO
2
enhanced oil recovery projects. All four
commercial scale CO
2
sequestration projects, Sleipner, Weyburn, In Salah, and
Snohvit have a time lapse surface seismic component in their monitoring
program. In Sleipner, surface seismic was acquired in 1994 for pre-injection and
then repeated in 1999, 2001, 2002, 2004, 2006, and 2008 for post-injection (figure
2.2). Significant differences can be noted between the pre and post injection
surveys by the presence of bright reflections and their growth with time. The
horizontal reflectors are interpreted to be thin layers of CO
2
accumulating
underneath the many thin mudstone layers inside the reservoir and the reservoir
caprock. Underneath the interpreted CO
2
plume, velocity push-down effect is
apparent due to the slower wave travel through the CO
2
. The plume is interpreted
to be roughly 200 m thick, and in plan view the plume is elongated with a major
axis of 3000 m from the 2008 seismic survey.
In Weyburn, pre-injection surveys were acquired in 1999 and subsequent
post-injection surveys were acquired in 2001, 2002, 2004, and 2007 to monitor
the movement of the CO
2
flood as brine and oil are produced (figure 2.3). In plan
view, the negative amplitude differences in the sequential 3D surveys are clearly
visible. The largest reflection amplitude changes are around the horizontal
injecting wells. Generally, there is a good agreement between injection volumes
and the areal extent and intensity of the amplitude and travel time anomalies.
Rostron et al (2009) suggest the observed seismic response is mainly from CO
2

saturation effects rather than from pressure induced effects. They argue that, first,
23

Figure 2.2: (Figure reproduced from Chadwick et al (2010))* Time lapse seismic images of
pre- and post-CO
2
injection: (top) North-south inline through plume; (bottom) Plan view of
total reflection amplitude in plume.

there is a weak correlation between the P-wave amplitude anomalies and
monitored or predicted pressure changes, second, time-travel delays can be
modelled from saturation effects by using the reservoir thickness and velocity
variation, and last, there is no correlation between the P-wave and S-wave time
lapse anomalies suggesting the P-wave anomalies are primarily saturation related
because S-waves are relatively insensitive to fluids.

Figure 2.3: (Figure reproduced from White (2009))* Amplitude difference maps of the
Midale-Marly horizon. Only negative amplitude differences are shown.
*Reprinted with permission from the Society of Exploration Geophysicists.
24

The In Salah and Snohvit storage project in relation to Sleipner or
Weyburn is a relatively much younger and the monitoring to date is still in its
early stages. For In Salah, the first time lapse 3D seismic survey to monitor pore
saturation and pressure changes was conducted in 2009. So far there has been
little published on the results.

2.6.2 Vertical Seismic Profiling:
VSP generally provides higher frequency (higher resolution of 10-30 m)
data than surface seismic methods. In comparison, VSP waves travel shorter
distances and the sensors are placed below the near surface, therefore leading to a
less attenuated signal.
The repeatability of VSP is also higher than surface seismic because the
receivers exact location in the borehole is known and can be replicated accurately
for each survey. From these advantages of VSP, smaller scale plumes could be
detected which would otherwise not be seen with surface seismic. Moreover, the
VSP method can be useful for interpreting surface seismic by aiding the tie of
surface seismic to borehole measurements. Since VSP allows measurements of
both direct and reflected waves, the reflected waves are from the same interfaces
that also produces the reflected waves for surface seismic. Application of VSP in
imaging CO
2
has succeeded at various sites:
In 2001, a CO
2
EOR pilot project was launched by Anadarko Petroleum at
the Patrick Draw Field in Wyoming, U.S in the Monell sandstone unit. A baseline
VSP survey was acquired in 2002 and another VSP survey was recorded at the
end of the project in 2003 with 430 million cubic feet of CO
2
injected in between
the two surveys. From the differenced volume of the time-lapse surveys, a clear
amplitude change associated with the CO
2
flood in the vicinity of the well is
apparent. Apart from the change in reflection strength in the second survey,
velocity push-down is also present underneath the CO
2
flood. From analyzing the
amount of velocity push-down present, it has been determined that the P-wave
velocity decreased between 14-19%. Also from the time-lapse surveys, it was
determined that the injected CO
2
has migrated 215-275 m (700-900 ft) radially
25

away from the injection well, with a preferential up-dip movement. This
observation is in agreement with production data.
In Frio, Texas, about 1600 tons of CO
2
were injected into an onshore
saline aquifer for a small scale pilot test of CO
2
injection at a depth of 1500 m. A
70% increase in reflection amplitude and an increase of wave travel time were
observed between pre- and post- CO
2
injection for the Frio horizon (Daley et al,
2008). Surveys were conducted pre-injection in July 2004 and post-injections
were surveyed in November 2004.
In another project in Texas, two sets of VSP surveys were acquired in the
SACROC oil field in west Texas; a baseline in 2008, and the repeated survey in
2009. CO
2
was injected at a depth of 2000 m into a reef structure and the amount
injected between the two surveys was not mentioned. Figure 2.4 shows the
differenced image between 2008 and 2009 surveys. A preliminary observation of
the image shows an identifiable change in reflectivity at the injection zone.
Further analysis is currently in progress.

Figure 2.4: (Figure reproduced from Cheng (2010))* Amplitude differenced image between
pre-and post-injection at SACROC oil field. Green circle highlights the injection zone.

*Reprinted with permission from the Society of Exploration Geophysicists.
26

2.6.3 Cross-well Imaging:
Of all three seismic methods, cross-well seismic survey has the highest
resolution of 1 to 5 m, compared to 10-30 m for VSP and 30-100 m for surface
seismic. The source along with the receivers is placed in the subsurface; therefore
the high frequency components of a signal are preserved even more from the
attenuating near surface materials. Likewise to VSP, survey repeatability is
enhanced by permanently embedding the receivers and the sources (in this case)
in the borehole. The purpose of using crosswell seismic surveys is to provide a
spatial, high-resolution map of the subsurface CO
2
and from this attempt to
estimate the CO
2
saturation between the two wells by using P- and S-wave
velocity tomographic imaging. Velocity tomography between the wells is
produced from the analysis of the one-way wave travel times recorded for each
source-receiver pair and ray theory. A major limitation of using crosswell
imaging is that only the region between the source and receiver wells is imaged.
Examples of using crosswell seismic surveys in CO
2
sequestration projects are:
In Frio, Texas crosswell seismic surveying was conducted between source
and receiver wells located 30 m apart. High resolution tomographic images of P-
wave velocity change illustrated the CO
2
distribution between the two wells.
Seismic P-wave velocity decreased up to 500 m/s and the S-wave velocity showed
minimal change (Daley et al, 2008). The velocity change follows the dip of the
local stratigraphy and is caused by buoyant forces acting on the injected CO
2

leading to an up-dip migration (figure 2.5). Velocity variation is not
homogeneous in the injected layer with less variation near the top, which agrees
with the lowered permeability observed near the top of the sands from well logs.
In Nagaoka, Japan between 2003 and 2006 multiple cross well seismic
surveys were conducted during the injection of 12700 tonnes (14000 tons) of CO
2

at an onshore CO
2
sequestration pilot test in a porous sandstone reservoir at 1100
m depth. The source and receiver wells have slight deviation from the vertical
where the horizontal separation between the top source and receiver wells is 92 m,
while the horizontal separation of the bottom source and receiver is 200 m.
Crosswell seismic data were measured before injection started and after 2902,
27

Figure 2.5: (Figure reproduced from Daley (2008))* P-wave tomogram from cross-well
survey in Frio, Texas.

5623, 8072, and 9434 tonnes (3200, 6200, 8900, and 10400 tons, respectively) of
CO
2
were injected. The greatest time-lapse changes occurred between pre-
injection and 3200 and 6200 tons, with minor differences between the 6200 and
10400 ton injections (Spetzler et al, 2008). From the time lapse tomographic
velocity images, Spetzler et al (2008) determined that there was an -18% velocity
anomaly.

2.7 Review of existing literature on CO
2
laboratory measurements

As reviewed above, seismic techniques are the most promising methods in
monitoring the subsurface related to CO
2
storage. Although CO
2
injections into
the subsurface have been conducted for some time now, there exist a very limited
number of laboratory studies on examining the elastic effects of CO
2
on pore fluid
*With kind permission from Springer Science+Business Media: Environmental Geology, Time-
crosswell seismic and VSP monitoring of injected CO2 in a brine aquifer,54, 2008, 1662, Daley,
T. M., Myer, L. R., Peterson, J. E., Majer, E.L., and Hoversten, G. M., figure 6.
28

seismic responses. Initial rock physics studies involving CO
2
had a focus directed
to assisting EOR processes and only recently has the underlying focus of the
existing studies been shifted to monitoring, verification, and accounting for
carbon storage purposes. All laboratory work to date has exercised the ultrasonic
pulse transmission technique and demonstrated that the presence of CO
2
in the
pore space will lower seismic velocities due to its higher compressibility
compared to other possible in-situ fluids. A review of existing rock physics
literature on seismic velocity measurements with CO
2
will follow.
One of the first published laboratory studies involving CO
2
as a pore fluid
saturant was done by Wang and Nur (1989). Their investigation was motivated
by the lack of understanding of CO
2
flooding effects on the seismic properties of
rocks saturated by hydrocarbons and using seismic methods in the field to assist
EOR processes. P- and S-wave velocities (V
p
and V
s
, respectively) in seven
sandstones of various porosities (6 to 29%) initially saturated with n-hexadecane
C
16
H
34
were measured before and after CO
2
flooding. For each sample CO
2

measurements were completed at two temperatures, 21C and 60C as pore
pressure varied from 2 to 16 MPa while confining pressure was maintained at 20
MPa. They determined that P-wave velocities decreased greatly due to CO
2

flooding, and the extent of decrease was dependent upon porosity, temperature,
and effective pressure whereas S-wave velocities were less affected. The authors
concluded that the amount of decrease in P-wave velocities caused by the CO
2

flooding may be seismically resolvable so that the CO
2
zones can be mapped and
tracked such that the flooding processes can be controlled to optimize EOR
process.
In 1998, another rock physics study of CO
2
flooding in the laboratory was
published by Wang et al. The authors were trying to understand the velocity
changes observed from the crosswell seismic imaging surveys conducted at the
CO
2
injection pilot project at the McElroy field in west Texas. In their
experiment, 8 dolostone core samples were obtained from the reservoir,
29

representing the range of porosities and permeability available. The samples were
initially saturated by an oil/water mixture, and then flooded with CO
2
. CO
2

velocity measurements on each sample were done at a constant confining pressure
of 20 MPa with the pore pressure varying from 8.3 MPa to 17.9 MPa while
subjected to the reservoir temperature of 31C. From the combined effect of CO
2

fluid substitution and pore pressure buildup, V
p
reduction varied from 3 to 10.9%,
while reductions in V
s
varied from 3.3 to 9.5%. These magnitudes of velocity
change are to the same extent as the observed velocity changes from field results
using time-lapse crosswell tomographic imaging. For CO
2
accounting purposes,
the authors suggested that the changes in pore fluid property effects can be semi-
quantitatively separated from pore pressure buildup effects in field seismic by
comparing changes in V
p
with V
s
. V
p
is sensitive to both CO
2
saturation and pore
pressure change while V
s
is particularly sensitive to the latter. Consequently if a
significant change in V
p
is observed and is not mirrored by V
s
observations, this
would imply that pore pressure build up is not the main cause of the velocity
change. As well, during the experiments the largest V
p
and V
s
changes are
observed with high porosity and high permeability rocks and the authors
suggested that it may be possible to distinguish high porosity and high
permeability zones seismically if the size of the zones are within seismic
resolution based on velocity changes.
From the Research Institute of Innovative Technology for the Earth
(RITE) in Japan, in the last 10 years a suite of literature has been published as a
part of the research effort for the Research and Development of Underground
Storage Technology for Carbon dioxide initiative. One of the first studies was
done by Xue and Oshumi (2002) where they monitored the movement of a CO
2

front during the flooding of gaseous CO
2
in a water saturated porous sandstone
carried out by measurements of P-wave velocities under hydrostatic pressure.
Their experiment consisted of a dense array of sixteen piezoelectric transducers
arranged in a 44 configuration at two orthogonal directions along the axial
direction of the core (figure 2.6). Two pore pressure tubes were connected to the
sample, one on each end of the sample and served as either the inlet or the outlet
30

for the pore fluid. The outlet valve was kept opened during the CO
2
flooding to
minimize velocity changes due to pore pressure build up. They demonstrated
significant reduction in P-wave velocities (typically on the order of 10%) due to
the displacement of water by the injected gaseous CO
2
.

Figure 2.6: (Figure reproduced from Xue et al (2005))* Depiction of Xue and Oshumis
(2002) transducer array on their core in their experiment.

Xue and Oshumi (2004) further modified and developed their work by
rearranging the configuration of their piezoelectric arrays so that the sixteen
piezoelectric transducers now formed two (instead of four) parallel lines along the
axial direction of the core and they conducted their tests with the injection of
different phase states of CO
2
. It was observed that the largest velocity reductions
were recorded when using supercritical CO
2
.
In another study, Xue and Lei (2006) used differential arrival time to
display the velocity changes during the injection tests by P-wave velocity
tomograms. Time lapse velocity images for each injection showed variations in
P-wave velocity across the sample as the CO
2
swept through the core. The
highest velocity decrease was observed during the supercritical CO
2
injection and
the lowest velocity reduction was during gaseous CO
2
injection. They concluded
that the heterogeneity of the pore distribution in the rocks and the different
injected CO
2
phases contributed to observed differences in P-wave velocity
images. Expanding on these preliminary results related to P-wave velocity, Lei
*Reprinted from Energy, 30, Xue, Z., Oshumi, T., and Koide, H., An experimental study on
seismic monitoring of a CO
2
flooding in two sandstones, 2352-2359, 2005, with permission from
Elsevier.
31

and Xue (2009) developed tomographic images of relative attenuation coefficient
to refine their work in mapping CO
2
migration and water displacement. They
reported that on average, P-wave velocity fell by 7.5, 12, 14.5% and the
attenuation coefficient increased by factors of 3.3, 2.7, and 3.7 during the
injection of gaseous, liquid, and supercritical CO
2
respectively.
Also from the same data set obtained, Shi et al (2007) focused on the
dynamic behavior of the P-wave velocity during the supercritical CO
2
injection
tests. The authors attributed the considerable variations in velocity reductions
across the sample to the final state of saturation (uniform, patchy, or in between)
of the two-phase fluids. Based on the amount of velocity reductions observed
and from comparing them to Gassmanns modelled velocities corresponding to
patchy and uniform saturation, they suggest regions of the core closest to the
injection point reached uniform saturation while large patchy saturations occurred
farthest from the injection point. The authors tried to integrate the laboratory data
with a numerical simulation of a simplified 1D model to provide an estimation of
the phase saturation changes underlying the measured P-wave velocity change
observed. However, when they compared their empirically established
relationship between measured P-wave velocities (corrected for velocity
dispersion) and estimated CO
2
saturation to Gassmanns prediction for uniform
and patchy saturation, there is a clear deviation from both velocity curves.
From the Korea Institute of Geoscience and Mineral Resources, Park et al
(2009) have also carried out ultrasonic laboratory measurements on a sandstone
sample to characterize the effects of CO
2
injection on seismic velocity and
amplitude. They noted a reduction of 6% in velocity or more until full CO
2

saturation has been reached and the wave amplitude decreased more than the
velocity decrease. In their work, they suggests that the Vs/Vp ratio is more
sensitive to CO
2
saturation than the current approaches and this attribute can
potentially allow quantitative assessment of CO
2
.
Purcell et al (2010) performed lab-scale velocity measurements under in-
situ conditions on a reservoir rock sample from the SACROC unit and on other
32

representative rock samples (shale and Berea sandstone). The work is intended to
assist interpretation of repeated 3D reflection seismic surveys over the SACROC
oil field to effectively monitor the movements of CO
2
after injection. Laboratory
measurements were conducted at room temperature and reservoir temperatures
(~50C) varying pore pressure (from 0 to 30 MPa) while maintaining constant
confining pressure (of 30, 40, and 50 MPa). For all 3 constant confining pressure
runs at room temperature, it was shown that as pore pressure increased the
velocities decreased. This observation with pore pressure change however, is a
combined effect from variations in pore fluid property and the changes in
differential pressure on the rock. Furthermore, their velocity results showed large
fluctuations and the effects of fluid phase change on their velocities cannot be
inferred from their data.
Despite the growing number of geological CO
2
sequestration projects and
in view of the above laboratory studies involving CO
2
, there lacks an extensive,
systematic study to inspect the seismic effects from:
1.) CO
2
as a lone pore fluid,
2.) the different phase states of CO
2
,
3.) pore fluid property changes that are definitively separated from pore
pressure build-up effects.
The first effect on seismic is important for providing an end member study to
constrain the possibilities with CO
2
in the pore space. The second and third
effects have implications on quantifying CO
2
using seismic. These three seismic
effects will be investigated in this thesis.

2.8 Summary
As concerns over climate change loom, governments, policy makers and
businesses around the world strive for ways to combat increasing atmospheric
CO
2
. This chapter provided a brief overview on the types of carbon storage that
can be exercised to mitigate CO
2
emissions and issues that can arise from the
different storage options. Geological sequestration is the most practiced type of
carbon storage with numerous projects on various scales around the world. A
33

major component of all geological sequestration projects is monitoring the
subsurface movement and the behavior of the injected CO
2
for
environmental/societal safety concerns and for carbon credit accounting purposes.
The different monitoring techniques available were presented and the role of
geophysics was discussed. Seismic techniques are deemed as an effective
monitoring method and are widely exercised in geological sequestration projects.
Published laboratory studies on seismic with porous media containing CO
2
were
also reviewed. It was shown that upon CO
2
introduction in the pore space,
ultrasonic P- and S- wave velocity reductions were always observed, regardless of
the original saturation condition. However, with the minimal amount of rock
physic studies to date that have been conducted with CO
2
, there are some seismic
effects that have yet to be addressed. This background introduction to CO
2

sequestration and the role of geophysics will lead to the next chapter in which the
theories relevant to CO
2
rock physics will be discussed.

34

Chapter 3
Theoretical Background

Whether the motivation behind interpreting seismic data is driven by
hydrocarbon/mineral exploration or is from the need to monitor subsurface
containments related to safety and reservoir integrity, the behaviour of seismic
waves must be well understood in order to be accurately informed about the
subsurface. Seismic waves are known to be influenced by both the physical
properties of the propagating medium and the ambient conditions such as pressure
and temperature. Furthermore wave propagation is highly sensitive to fluids in
the pore space and to their variability with changing conditions, leading to the
valued use of seismic methods in CO
2
sequestration. This chapter will provide
relevant background on the influences of seismic waves in geologic CO
2

sequestration and to the work conducted in this thesis. To begin, a brief
introduction is given on the thermodynamics of CO
2
, particularly on the elastic
properties pertinent to seismic wave propagation. To follow, the basic concepts
of elasticity and its relation to wave velocity is provided, along with the concept
of viscoelasticity and its implication on wave travel. Next, the rock physics
concept of fluid substitution is discussed by the widely used Gassmanns
formulation and also the limitations of his equation. This discussion is expanded
to include Biots formulation. Finally, the foundation to the laboratory technique,
ultrasonic pulse transmission, and the analysis methodologies applied in this
thesis are provided.

35

3.1 Carbon Dioxide
Carbon dioxide (CO
2
) that was first discovered in the 1750s by a Scottish
chemist and physician, Joseph Black, is an odourless and colorless gas at standard
temperature and pressure. In a CO
2
molecule, the carbon atom is bonded to the
two oxygen atoms in a linearly symmetric form and results in a non-polar
molecule. It is a commonly encountered substance that plays a major role in the
process of photosynthesis for producing carbohydrates, is emitted from volcanoes,
hot springs, and geysers, is an important ingredient in carbonate rocks, and as
introduced in chapter 1 is a major by-product of combustion. Human uses of CO
2

include the carbonation in beverages, fire extinguishers, the decaffeination of
coffee by supercritical fluid extraction, as refrigeration via dry ice, and in
pneumatic systems as a compressed gas. Naturally, CO
2
is exchanged between
the atmosphere, hydrosphere, biosphere, pedosphere, and lithosphere by a set of
complex biogeochemical interactions, known as the carbon cycle.
As discussed in the last chapter, in order to combat the fast- rising
atmospheric concentration of CO
2
, geological sequestration is conducted by
injecting CO
2
into the subsurface. Concerns about the safety and the integrity of
the underground containment have enlisted various methodologies for monitoring
the behaviour of the injected CO
2
, where seismic methods are highly favourable.
The presences of a pore fluid have been shown to have a strong impact on the
elastic properties of a porous medium (Timur, 1968; De Vilbiss, 1980;
Christensen and Wang, 1985; Batzle and Wang, 1992). Given the possible
pressure and temperature conditions of the upper sedimentary basin, CO
2
can be
in a gas, liquid, or supercritical fluid phase state, where each will yield a different
effect on seismic waves from the consequent variations in density and fluid
compressibility. Therefore the thorough understanding of the elastic properties of
CO
2
is crucial to the accurate interpretation of a seismic response.

3.1.1 Introduction to Thermodynamics
Thermodynamics is an area of study that deals with energy
transformations and of the relationships between the physical properties of
36

substances which are affected by these transformations. For a substance that is
under a set of physical conditions, the resulting physical behaviour or its state of
matter can be described by an equation of state (EOS) through the state variables
of pressure, temperature, and specific volume. The simplest and most known
EOS is the ideal-gas equation. However this EOS is only adequate at low
pressures because the intermolecular forces that become significant at other
physical conditions (particularly at higher pressures and lower temperatures) are
unaccounted for. Other more accurate and well known general EOS for fluids are
the Van der Waals equation, the Virial, the Redlich-Kwong, the Peng-Robinson,
and the (Benedict-Webb-Rubin) BWR (Wark and Richards, 1999). At present,
there is no single equation of state that accurately predicts the properties of all
substances under all conditions.
A phase diagram of a generic substance is shown in figure 3.1. The triple
point of a substance marks the condition of where the gaseous, liquid, and solid
phase states can coexist in thermodynamic equilibrium. In contrast to the triple
point, the critical point of a substance denotes the conditions above which the
phase boundary between the gas and liquid phase state ceases to exist. Beyond
this point, the properties of the gas and liquid phases are indistinguishable,
resulting in only one phase known as the supercritical fluid phase state. A
supercritical fluid has physical behaviours of both a gas and a liquid such that its
density is comparable to that of liquid and it has a compressibility similar to that
of a gas. Above the critical pressure for any temperature, a distinct gas cannot
form and above the critical temperature for any pressure, a distinct liquid cannot
form. However, for a sufficiently high enough pressure and above the critical
temperature, a transition into a solid phase can occur from the supercritical fluid.
Furthermore small changes in pressure or temperature near the critical point can
yield large changes in density.

3.1.2 CO
2
Phase States
In the last few decades, the thermodynamic properties of CO
2
have been
studied extensively and new empirical equation of state (EOS) have been
37

continuously proposed as more experimental data are acquired and more
sophisticated correlation methods are developed (Kesselman et al, 1965; Stein,
1972; Huang et al, 1985; Ely et al, 1989; Pitzer and Sterner, 1994; Zhang and
Duan, 2005; Kim, 2007). The developed correlation equations varied in the base
structure of their equation, in the data used for the correlation, and on the
temperature/pressure regions of interest. Span and Wagner (1996) presented an
empirical EOS for CO
2
that extended from the triple point temperature to 1100 K
with pressures up to 800 MPa in the form of the fundamental equation Helmholtz
free energy. Span and Wagners empirical EOS overcame many limitations faced
by previous efforts such as being descriptive over a limited physical condition
range, data were not represented within their experimental uncertainty,
unreasonable results near the critical point, and unstable extrapolations. The
estimated uncertainty in their equation for pressures up to 30 MPa and
temperatures up to 250C ranged from 0.03 to 0.05% in density and 0.03% to 1%
in the speed of sound. These errors were estimated by comparisons to
experimental data and to other equations of state in regions where there were no
experimental data. Their empirical EOS for CO
2
is widely used and is currently
implemented by the National Institute of Standards and Technology (NIST)
chemistry webBook (Lemmon et al, 2011). From their review of existing data,
the triple point for CO
2
is -56.558 0.003C and 0.51795 0.00010 MPa, where
the critical point is 30.9782 0.015C and 7.3773 0.0030 MPa. Based on the
typical temperatures and pressures encountered in the subsurface, the solid phase
CO
2
is not a concern in geological sequestration. At standard temperature and
pressure, CO
2
is about 1.5 times denser than air and the speed of sound is
approximately 0.25 times slower than in air.
Based on the thermodynamic model of Span and Wagner, phase diagrams
of CO
2
s bulk modulus, density, and fluid viscosity (figure 3.2) were created over
a temperature range of 0-60C and a pressure up to 30 MPa. The bulk modulus
and the density of a fluid are the main physical properties that influence elastic
wave propagation. The viscosity property of a fluid, however, is only relevant to
38

high frequency wave propagation in a saturated porous medium. The vapour-
liquid boundary is clearly discernible by an abrupt change in all the physical

Figure 3.1: (Figure reproduced from Wikipedia)* A generic phase diagram. The dotted
green line is the anomalous behaviour of water.

properties. As the critical point is approached, the properties of the gas and liquid
phases converge and no distinction is apparent thereafter, yielding the
supercritical fluid phase. Consequently, a phase change from a gas or a liquid
state into a supercritical fluid state is a smooth transition in physical properties.
Furthermore, in the supercritical region close to the critical point, large density
variations are clearly seen over a small range of pressure and temperature.

3.1.3 Geological Sequestration Approximate Physical Conditions
Subsurface temperatures and pressures conditions can be influenced by a
number of factors (radiogenic rocks, plate tectonic activity, or aquitards,
hydrocarbon caps, and sealing faults, respectively) but are generally dependent on
the depth from the surface. To provide a simple reference, the conditions that can
be anticipated in a geological sequestration project could be modelled by an
oversimplified situation where they are only dependent on depth. The
temperature variation with depth for a constant geothermal gradient, G, is
Gz T T
o
+ = , (3.1)
*attribution: Matthieumarechal: http://en.wikipedia.org/wiki/File:Phase-diag2.svg)
39

where T
o
is the surface temperature, z is the depth from the surface, and the
geothermal gradient is typically 15-30C/km for sedimentary basins and areas
away from tectonically active zones. The pore pressure with depth can be
assumed to be from the hydrostatic pressure if a fluid can freely flow from the
surface to at depth without any permeability barriers (such as aquitards),
gz P
w p
= , (3.2)
where
w
is the density of water and g is the acceleration of gravity. As
mentioned in the previous chapter, suitable sequestration formations are typically
at depths greater than 1000 m. Using the Earths average surface temperature of
15C, and the average typical geothermal gradient of 22.5C/km, a water density
1000 kg/m
3
, and a gravity of 9.81 m/s
2
, the minimum temperature and pore
pressure to be encountered in a geological sequestration project under the average
condition is 37.5C and 10 MPa, respectively, and is shown by the red dot in
figure 3.2d. Based on this simplistic depth model relating subsurface temperature
and pressure under the average condition (profile shown by the red-dashed line in
figure), free CO
2
in a geologic containment is therefore nearly always expected to
be in a supercritical fluid phase. If the CO
2
migrates out of its containment into a
shallower formation, it can be in a liquid or a gas phase in the respect of
temperature and pressure conditions. As can be seen in the figure, this simplistic
profile barely traverses into the liquid regime, indicating that liquid CO
2
would
only be expected within a short depth interval. The liquid phase CO
2
could be
completely bypassed if a higher geothermal gradient is used, or have a greater
presence if a lower geothermal gradient is applied, where the geothermal gradient
changes the slope of the profile in a phase diagram plot. In figure 3.2d, the dotted
lines are the resulting profiles from using the lower and upper limit of the typical
geothermal gradient range, 15C/km and 30C/km, respectively. Furthermore, the
surface temperature will also influence the position of the profile. A lower or
higher surface temperature will shift the profile such that the liquid phase will
have a significant presence or will be completely bypassed, respectively. For the
purpose of providing a general idea of the type of pressures and temperatures

40

Figure 3.2: CO
2
phase diagrams of (a) bulk modulus, (b) density, (c) and viscosity. The phase
boundaries of the supercritical fluid are shown by the white dotted lines. The critical point is
marked by the red dot. (d) A basic CO
2
phase diagram with the possible CO
2
sequestration
conditions shown by the red dashed-line, based on the simple model described.

that can occur, and consequently the type of phase states, such a simplistic view is
sufficient as this is not a case study.

3.2 Seismic Theory
An elastic wave is a mechanical wave that travels in an elastic medium by
displacing the particles of the medium. When an elastic wave travels through
Earth materials, they are referred to as seismic waves. In this section, the theory
of wave travel will be discussed by a brief introduction to the theory of elasticity
by the pertinent stress-strain relationship, the resulting wave equation, and the
41

elastic stiffness constants for an isotropic medium. In addition, viscoelasticity and
its implication on wave travel will be introduced. These sections will be briefly
reviewed as the in depth discussion have been extensively presented in literature
such as Aki and Richards, 1980, Bourbi et al (1987), Telford et al (1990), Lay
and Wallace (1995), Lowrie (1997), Lakes (2009) and in any Rock Physics or
introductory Geophysics textbooks.

3.2.1 Theory of Elasticity
When a force acts upon an elastic body, physical deformation of the elastic
body will occur. This applied force can be further characterized as the applied
stress, where stress is defined as the ratio of the applied force to the involved area.
Fundamentally, there are two types of stresses and they are defined based on the
direction of the applied force in relation to the involved area. If the direction of
the force is perpendicular to the area, the stress is called a normal stress and if the
direction of the force is tangential to the area, the stress is called a shear stress.
The resulting physical deformation such as changes in the dimensions of the
elastic body or a change in shape is described as strain; where normal stress will
result in normal strain and shear stress will give shear strain. For a perfectly
elastic material, an applied stress will produce in an immediate, reversible strain.
The energy used to deform the material is stored so when the stress is removed,
the stored energy is used to rebound the material back to its original shape. All
displaced material points will resume their original positions and no energy is lost
from storing to rebounding.
The mechanical behaviour for a volume of an elastic material can be
described by Hookes law, succinctly in Einsteins summation over repeated
indices, as
kl ijkl ij
C c o = i, j, k, l=1, 2, 3 (3.3)
which relates stress,
ij
, linearly to strain,
kl
, by the elastic stiffness constant, C
ijkl
.
The stress and strain are second-order tensors composed of nine components that
represents the three-dimensionality of a volume, where the indices, i, j, k, l,
represents the three orthogonal axes. For a volume that is in static equilibrium,
42

the total moment is zero, thereby the stress and strain tensors are symmetrical
about the main diagonal allowing
ji ij
o o = , (3.4)
lk kl
c c = . (3.5)
This symmetry reduces the nine component matrices to six independent
components. The elastic stiffness constant, C
ijkl
, is a fourth-order tensor,
composed of eighty-one components. Owing to the symmetries of the stress and
strain tensors and strain energy considerations, the elastic stiffness tensor has the
following equivalency,
klij jilk ijlk jikl ijkl
C C C C C = = = = . (3.6)
Consequently, the number of independent components is reduced to twenty-one.
Furthermore, the representation of equation 3.3 can be simplified by employing
the Voigts notation via reducing the rank of a symmetric tensor. Using Voigts
notation, the double indices of 3.3 are transformed to only one index by:
i,j or l,k I or J
11 = 1
22 = 2
33 = 3
23, 32 = 4
13, 31 = 5
12, 21 = 6.
As a result, equation 3.3 is simplified to
J IJ I
C c o = . (3.7)
The once fourth-order elastic stiffness tensor is rewritten as a 66 symmetric,
second-order tensor, C
IJ
, and the once second-order stress and strain tensor are
now 61 stress and strain vectors,
I
and
J
, respectively. The Hookes law shown
explicitly with these new index adaptations is,

43

(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
=
(
(
(
(
(
(
(
(
6
5
4
3
2
1
66 65 64 63 62 61
56 55 54 53 52 51
46 45 44 43 42 41
36 35 34 33 32 31
26 25 24 23 22 21
16 15 14 13 12 11
6
5
4
3
2
1
c
c
c
c
c
c
o
o
o
o
o
o
C C C C C C
C C C C C C
C C C C C C
C C C C C C
C C C C C C
C C C C C C
(3.8)
For the most general elastic material, twenty-one independent constants is
the maximum number of constants required for its behavioural description.
Materials that require the full twenty-one elastic constants for description are the
least symmetric material and are called triclinic solids. Materials of higher
symmetry will contain fewer independent elastic constants. The most simplistic
case is an isotropic material, where their physical properties are independent of
direction such that they hold infinite symmetry. For an isotropic material only
two independent elastic constants, C
11
and C
44
are needed and the structure of the
Voigt elastic stiffness matrix has the following form:
(
(
(
(
(
(
(
(

=
44
44
44
11 44 11 44 11
44 11 11 44 11
44 11 44 11 11
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
0 0 0 2 2
0 0 0 2 2
0 0 0 2 2
C
C
C
C C C C C
C C C C C
C C C C C
C
IJ
(3.9)
The elastic constants C
11
and C
44
can be related to the materials bulk modulus, K,
and its shear modulus, , or its Lam parameters, and by:
2
3
4
11
+ = + = K C , (3.10)
=
44
C . (3.11)
In rock physics and seismology, wave propagations are usually given in
terms of the bulk modulus and the shear modulus. The bulk modulus is
essentially a measure of the mediums resistance to uniform compression
(stiffness), and is given as the ratio of the hydrostatic stress,
0
, to the volumetric
strain, :
44

u
o
0
= K ,
3
3 2 1
0
o o o
o
+ +
= , and
3 2 1
c c c u + + = . (3.12)
The reciprocal of the bulk modulus will give the compressibility of a material.
The shear modulus, modulus of rigidity, or the Lams second parameter is a
measure of the mediums resistance to shear strain, and it is the ratio of the shear
stress to the shear strain:
i
i
c
o
2
= , i=4, 5, 6. (3.13)
The Lams first parameter, however does not have any physical interpretation,
but is useful for the simplification of the elastic stiffness matrix. If under a
uniaxial stress (all stresses are zero, except in one direction), other elastic moduli
can be also used to describe an isotropic material and they are Youngs modulus
and Poissons ratio. The Youngs modulus, E, is the ratio of the extensional stress
to the extensional strain:
,
i
i
E
c
o
= i=1, 2, 3. (3.14)
The Poissons ratio, , is the negative ratio of the lateral strain to the axial strain:
,
j
i
c
c
v = i, j=1, 2, 3 and ij. (3.15)
If in a uniaxial strain state (all strains are zero, except in one direction), the P-
wave modulus can be used instead and it is also defined as the ratio of the axial
stress to the axial strain:
,
i
i
M
o
c
= i=1, 2, 3. (3.16)
For an isotropic, linear elastic medium any one of the isotropic constants (K, , ,
E, , M) can be derived in terms of the others. The relations between the given
elastic moduli are listed in table 3.1.

45

K E M
(, )
3
2
+

+
+ ) 2 3 (

- -
) ( 2
+

2 +
(E, )
) 3 ( 3 E
E

-
E
E

3
) 2 (

-
1
2

E

E
E

3
) 4 (

(K, )
-
K
K K
3
) ( 9

-
2
) ( 3 K

K 3

2 3 K
(K, )
-
+ K
K
3
9

3
2
K
-
) 3 ( 2
2 3
K
K

3
4
+ K
(, )
v
v
3
) 1 ( +
v
v v ) 2 1 )( 1 ( +

-
v
v
2
) 2 1 (

-
v
v ) 1 (

(, )
) 2 1 ( 3
) 1 ( 2
v
v
+

) 1 ( 2 v +
v
v
2 1
2

- -
) 2 1 (
) 1 ( 2
v
v

(E, )
) 2 1 ( 3 v
E

-
) 2 1 )( 1 ( v v
v
+
E

) 1 ( 2 v +
E

-
) 2 1 )( 1 (
) 1 (
v v
v
+
E

(K, )
-
) 2 1 ( 3 v K
v
v
+ 1
3K

) 1 ( 2
) 2 1 ( 3
v
v
+
K

-
) 1 (
) 1 ( 3
v
v
+
K

(K, E)
- -
E K
E K K
9
) 3 ( 3

E K
KE
9
3

K
E K
6
3

E K
E K K
+
9
) 3 ( 3

(M, )
3
4
M
M
M ) 4 3 (
2 M
-
) ( 2
2
M
M

-
Table 3.1: (After Birch (1961))* Relationships among the elastic constants for a
homogeneous isotropic material

3.2.2 Wave propagation in an isotropic linear elastic medium
Detailed derivation of wave propagation in a linear elastic medium will
not be presented here since this topic has been reviewed in numerous textbooks.
In this section, only the basic concepts required for later use and discussion are
provided. For a more in depth discussion and derivation, the reader is referred to
the literature listed in the introduction of section 3.2.
From the previous section, the material dependent relationship between
the applied stress and the immediate resulting strain has been established by
*Reproduced with permission from the American Geophysical Union
46

Hookes law. However, Hookes law on its own does not give the variation in the
displacement of material points with time. Instead, the propagation of a wave
with space and time can be described if the volume of material considered is not
in static equilibrium, so that there is a non-zero net force (unbalanced) on the
volume. Newtons second law of motion states that a non-zero net force on a
body is equal to the product of the mass and the acceleration of the body. By
substituting Hookes law into the equation of motion and expressing the strains in
terms of displacement, the scalar one-dimensional wave equation in an elastic
medium for a displacement, u, is
u C
t
u
2
2
2
V =
c
c
. (3.17)
where u is given in terms of position and time, u(x,t), is the bulk density of the
elastic medium, and C is the elastic stiffness constant or elastic modulus
pertaining to the type of wave in consideration. The wave velocity for the most
general case in equation 3.17 is,
C
V = . (3.18)
Essentially, the wave equation relates a time derivative of a displacement to
spatial derivatives through the proportionality constant of V
2
.
In a homogeneous isotropic medium, there are two types of waves that can
propagate, a P-wave (also known as primary, dilatational, longitudinal,
irrotational, or compressional wave) and a S-wave (also known as secondary,
tranverse, rotational, or shear wave). Because P- and S-waves travel in the
interior of a medium, they are also referred to as body waves. For the velocity of
P-and S-waves, equation 3.18 would be

M
K
C
V
p
=
+
=
+
= =
3
4
2
11
, (3.19)
= =
44
C
V
s
, (3.20)
respectively. Since fluids cannot sustain shear forces such that the shear moduli
of fluids are zero, only P-waves can travel in fluids where the wave velocity is
47

K
V
p
= . (3.21)

3.2.3 Viscoelasticity
So far discussed, the material under stress is perfectly elastic meaning
when the applied stress is removed the strain is completely recovered, and no
energy is lost. In contrast, there are some materials when under stress will deform
by flow and the deformation is permanent. Unlike an elastic material, the energy
used to strain the material is completely dissipated to the surroundings without
any storage. With no energy to recover the original shape, the material stays
permanently deformed when the stress is removed. This type of material has a
viscous behaviour and is exhibited by fluids under shear stress. For a viscous
fluid under shear stress the resulting behaviour is
t
ij
ij
o
oc
q o = , ij (3.22)
where the applied stress is linearly related by the fluid viscosity, , to the strain
rate, t
ij
o oc , and there is now a time dependency on the stress-strain relationship.
Between the two end member cases of elasticity, in reality materials usually
exhibit both an elastic and a viscous response and they are called viscoelastic
materials. Some phenomena in viscoelastic materials are (Lakes, 2009):

1.) If the stress is held constant, the strain increases with time and is called a
creep response: Figure 3.3 displays the creep and recovery of a material
experiencing a step stress. While under a constant stress, the deformation of a
viscoelastic material will continually progress with time. On the strain plot, from
examining the behaviours of all material responses, it is obvious that the
viscoelastic response is the combined response between a perfectly elastic and a
purely viscous material. The non-zero intercept of the strain curve is a result of
the elastic behaviour of the material, where for a given stress there will be an
instantaneous strain. Thereafter, the effect of a viscous deformation is evident by
the increasing strain with time. When the stress is removed the strain
48

immediately drops a substantial amount due to the elastic rebound of the material
and then tapers off without ever fully recovering the full strain, as governed by the
viscous response of the material.

Figure 3.3: (Figure reproduced from Lakes (2009))* Creep and recovery response of a
viscoelastic medium. Top plot is a step stress over time, bottom plot is the resulting strain
with time. Superimposed on the bottom plot in dotted lines are the elastic and viscous
responses to a step stress.

2.) If the strain is held constant, the stress decreases with time and is referred to
as relaxation: Figure 3.4 displays the relaxation and recovery of a material
experiencing a step strain. Since time is increasing, stress must decrease to keep
strain constant. When the step strain ends at zero, implying this material has
recovered its full deformation, the applied stress must need to reverse directions to
attain this. Otherwise due to the viscous response, there would be residual strain.

3.) If cyclic loading is applied, hysteresis occurs, leading to dissipation of
mechanical energy: In one dimension, for an applied stress that varies
sinusoidally with time, t, and angular frequency, , given by,
t i
e t
e
o o
0
) ( = , (3.23)

*Reprinted with the permission of Cambridge University Press.
49

Figure 3.4: (Figure reproduced from Lakes (2009))* Relaxation and recovery response of a
viscoelastic medium. Top plot is step strain over time, bottom plot is the resulting stress with
time. The dotted line is elastic response to the step strain.

the responding strain in a viscoelastic material will also vary sinusoidally at the
same frequency but not immediately,
) (
0
) (
o e
c c

=
t i
e t . (3.24)
The delayed strain response to the applied stress gives rise to the phase lag, .
Figure 3.5 displays the relation between the sinusoid varying stress and strain of a
viscoelastic material. With both the stress and strain functions varying at the
same period T, the phase lag is a result of the time delay, t, of the strain
response in respect of the stress,
T
t A
=
t
o
2
. (3.25)
, is also referred to as the loss angle and is a measure of the internal friction and
the mechanical damping of a material. It can also be given as the loss tangent
where the tangent is taken of the loss angle, tan. If there is no phase lag between
the stress and strain of a given material, =0, this material is an elastic solid.
Furthermore, the stiffness modulus, C, that relates stress and strain seen in section
*Reprinted with the permission of Cambridge University Press.
50

3.2.1 for a perfectly elastic material is a real number. For a viscoelastic material,
as a consequence of the intrinsic phase lag between stress and strain, the stiffness
modulus is now a complex number,
" ' *
) (
) (
iC C C e
t
t
i
o
o
+ = = =
o
c
o
c
o
, (3.26)
with C and C being the real and imaginary parts, respectively, of the complex
modulus C*. C is commonly called the storage modulus and it is responsible for
the elastic response of the material or the preservation of energy. C is the loss
modulus and it is responsible for the viscous response of the material or the
dissipation of energy. For a propagating wave, the storage modulus supports the
physical travel of a wave where the loss modulus provides the attenuation or the
disappearance of the wave. Moreover, the loss tangent can be related to the
storage and loss modulus by.
'
"
tan
C
C
= o , (3.27)
The physical meaning of the loss tangent is associated with the ratio of energy
dissipated to the energy stored.

3.3 Rock Physics
From the last section, the background on wave propagation was given for
a homogeneous medium. In geophysics, the medium of concern is usually Earth
materials which are not homogeneous at a micro-scale or even at macro-scale.
Earth materials or rocks are of a porous medium, consisting of a solid matrix
(mineral grains) and a void volume (pores) that may be filled with fluid. The
elastic properties pertaining to wave propagation in a porous medium are
therefore influenced by both the solid matrix and the pore fluid constituents.
In the discipline of rock physics, the aim is to establish relations between
rock properties and the observed seismic response so that predictive theories can
be developed to assist seismic interpretation. Therefore the physical properties of
interest are ones that will affect how seismic waves physically travel through the
51

Figure 3.5: (Figure reproduced from Lakes (2009))* Stress and strain varying sinusoidally
with time. The time delay of the strain response in respect of the stress is given by t.
Subpanel is a plot of complex modulus E*, its real (E) and imaginary parts (E) in relation
to the loss angle, .

rocks, namely the compressibility, the rigidity, the porosity, the mass density, and
the pore fluid. To establish such relation requires knowledge about the elastic
properties of the rock matrix and pore fluid, and rock-fluid interaction models.
Fluid substitution refers to the prediction or modeling of wave velocities in a
porous medium that is saturated with a fluid from the information obtained of the
dry porous medium or from the saturated porous medium with a different fluid.
Two theoretical rock-fluid interaction models will be discussed in this section.

3.3.1 Gassmanns equation
In rock physics, Gassmanns equation is the most widely used fluid
substitution relation due to its simplicity and presumed applicability in the seismic
frequency range (~100 Hz). His theoretical equation was derived by
considerations of elementary elasticity of the pore fluid and the mineral grains of
a saturated rock sample. In his formulation, Gassmann (1951) assumed:
*Reprinted with the permission of Cambridge University Press
52

1) The rock is macroscopically homogeneous and isotropic;
2) The rocks minerals all have the same bulk and shear moduli;
3) The fluids can freely move about in the pore space where pores are
interconnected or communicating and the fluid viscosity is zero;
4) The pore space is fully saturated at all times;
5) The pore fluid does not interact with the rock minerals thereby changing
the rock frames stiffness.
6) Quasi-static conditions where the frequencies are sufficiently low such
that the induced pore pressures can be equilibrated through the pore space.
Gassmanns equation will be presented here without derivation; detailed
discussion on the origin of Gassmanns equation can be found in Bourbi et al
(1987) and Berryman (1999).
By Gassmanns formulation, the bulk modulus of a saturated rock, K
sat
, is
related to the rock frames bulk modulus (frame modulus), K
dry
, the mineral
grains bulk modulus (solid constituent of the rock), K
s
, the pore fluids bulk
modulus (fluid constituent of the rock), K
f
, and the porosity, , of the rock
through:
) (
f s
f
dry s
dry
sat s
sat
K K
K
K K
K
K K
K
=
|
, (3.28)
where rearranging gives the predicted saturated bulk modulus of the rock as,
2
2
1
) 1 (
s
dry
s f
s
dry
dry sat
K
K
K K
K
K
K K
+
+
/
+ =
| |
. (3.29)
If the rock frames bulk modulus is unavailable (usually the case for in-situ
measurements), the bulk modulus of a rock saturated with a fluid, K
sat
, can instead
be related to the bulk modulus of the rock saturated with a different fluid, K
sat2
:
) ( ) (
2
2
2
2
1
1
f s
f
sat s
sat
f s
f
sat s
sat
K K
K
K K
K
K K
K
K K
K
| |
. (3.30)
For the shear modulus of a saturated rock,
sat
, the computation is even simpler.
In the Gassmann formulation, the shear modulus is regarded to be independent of
53

the pore fluid because fluids are incapable of sustaining shear forces. Therefore,
the shear modulus of the saturated rock is the same as the shear modulus of the
dry rock or, of the saturated rock with another fluid:
2 sat dry sat
= = . (3.31)
From the determined saturated bulk or shear moduli, the corresponding saturated
P-or S-wave velocity can be found from using equation 3.19 or 3.20 in section
3.2.2, where the bulk density of the saturated rock is given by:
f s sat
| | + = ) 1 ( (3.32)
In order to predict the bulk modulus of a saturated rock, 1) the mineral
grains bulk modulus, 2) the pore fluids bulk modulus, 3) the rock frames bulk
modulus, and 4) the rocks porosity needs to be known. The determination of a
samples porosity will be given in Chapter 5 on Sample Characterization and the
determination of the different bulk moduli will be given below.
1.)Estimating Mineral Grain Elastic Moduli
The mineral grain refers to the rock-forming minerals constituting the rock
matrix or simply the solid component of a rock. To determine the mineral grains
bulk modulus, the minerals making up the rock need to be identified. Once
identified, the bulk modulus of the constituting minerals can be determined by
referring to literature on mineral and material properties such as Bass (1995). For
samples whose composition is of one or is largely dominated by one mineralogy
(>95%), determining K
s
is straightforward and suffice. Otherwise for samples
consisting of more than one mineral, the volume fraction of the various mineral
constituents of the matrix needs to be identified. The mineral type and volume
fractions can be determined from microscopic imaging such as thin-sections or X-
ray diffraction/fluorescence (XRD/XRF) methods. With the mineral
compositions known, the bulk modulus of the samples matrix that is composed of
a mixed mineral composition can be estimated by using Hills average (1952),
( )
R V S
K K K + =
2
1
(3.33)
54

where the K
V
is Voigts average (1928)and K
R
is Reuss average (1929). These
two averages respectively give the maximum and minimum values possible for a
mixture:
=
=
n
i
i i V
K f K
1
(3.34)
=
=
n
i i
i
R
K
f
K
1
1
. (3.35)
where f
i
is the fraction amount of the i
th
mineral, and K
i
is the bulk modulus of
that corresponding mineral.
Alternatively, the mineral grains bulk modulus can be determined without
knowing the exact mineral composition of the rock matrix. Biot and Willis
(1957) introduced a method of determining the mineral grains bulk modulus in
the laboratory and it is commonly referred to as the unjacketed test. This method
is only appropriate if there are no occluded pores. Briefly, a sample is immersed
in a fluid without being sealed such that the surrounding fluid can enter the pore
space. With raised fluid pressure, the samples pore space will be fully saturated
and the pressure inside the sample is equal to the pressure on the outside. Any
volumetric change (strain) of the sample from increased fluid pressure is due to
the compression of the grains. Such unjacketed quasi-static strain tests were
conducted by Bakhorji (2010) on carbonate rocks and Schmitt and Zoback (1989)
on low porosity metamorphic gneisses. However, recently Hart and Wang (2010)
have questioned this assumption.
2.)Estimating Pore Fluid Elastic Moduli
The pore fluid refers to any fluid that resides in the void space of the rock.
It can be a gas, a liquid, a supercritical fluid, of hydrocarbons, of water, of CO
2
, of
a mixture of fluids, or even of a vacuum. The bulk modulus of the pore fluid, K
f
,
can be theoretically calculated from equation 3.21 if auxiliary thermodynamic
properties (from EOS, or empirical relations such as Batzle and Wang, 1992) are
available to provide the fluid density,
f
, and the speed of sound through the fluid,
V
Pf
:
55

2
pf f f
V K = . (3.36)
In the case of fluid mixtures, the overall pore fluid bulk modulus can be
calculated from averaging the contribution of the individual fluids using Voigts
or Reuss average equation. If the bulk modulus from Reuss average is used this
is also known as Woods relation. In addition, the overall fluid bulk density from
the mixture is
fi
n
i
i f
f

=
=
1
. (3.37)
3.)Estimating Dry Frame Elastic Moduli
The term frame refers to the rock sample with empty pores or
alternatively, the dry rock. The bulk and shear modulus of the dry rock, K
dry
and
dry
, can be determined from wireline log data, from a theoretical or an empirical
relationship or from laboratory measurements.
In using wireline data, K
dry
can be determined if K
sat
is measured for a
different fluid. If the parameters , K
f
, K
s
are known before hand, K
dry
can be
determined by rewriting the Gassmann equation for K
dry
,
|
|
| |

+
|
|
.
|
\
|
+
=
1
1
s
sat
f
s
s
f
s
sat
dry
K
K
K
K
K
K
K
K
K . (3.38)
This back calculation for K
dry
is an intermediate step that was skipped when
applying fluid substitution of two fluids in equation 3.30.
To determine K
dry
from using theoretical and empirical relations, there are
a variety of equations that can be employed. From a theoretical standpoint, by
considering the stress and strain under dry conditions, the dry rock modulus is
(Bourbi et al, 1987; Mavko et al, 2009):
) 1 ( | =
s dry
K K . (3.39)
is formally defined as the ratio of the pore volume change to the bulk rock
volume change while under constant pore pressure and is sometimes called the
Biot coefficient.
56

In terms of using an empirical relation, Nur et al (1995) suggested that
between low porosity and critical porosity (the limiting cases for rocks), there is a
linear trend for the dry moduli as a function of porosity. The critical porosity
separates whether the rock is in a grain- load bearing (<
c
) or a fluid-load
bearing (>
c
) domain. For most reservoir rocks, the grains are load bearing and
the critical porosity serves as a limiting case. The bulk and shear moduli for dry
rocks are:
|
|
.
|
\
|
=
c
s dry
K K
|
|
1 , (3.40)
|
|
.
|
\
|
=
c
s dry
|
|
1 . (3.41)
An alternative empirical relation to estimate the bulk modulus in dry rocks is
( ) | 50 1
1 1
+ =
s dry
K K
, (3.42)
developed by Geertsma (1961) for porosities 0 < < 0.3.
Lastly the dry bulk and shear moduli can be determined in the laboratory.
In the laboratory, ultrasonic P- and S-wave velocities through the dry rock and the
dry bulk density are measured. From rearranging equation 3.19 and 3.20, the dry
bulk and shear modulus can be found:
)
3
4
(
2 2
s P
V V K = (3.43)
2
s
V = . (3.44)
For a more in depth discussion on laboratory determination of ultrasonic P-and S-
wave velocities and dry bulk density, the reader is referred to Chapter 4 and
Chapter 5, respectively.
Gassmanns equation as mentioned is widely used due to its ease and also
because it works reasonably well when compared to certain data. However it
should be noted that for rocks with low-aspect ratio pores (cracks or fractures), or
is saturated by high viscosity pore fluids, Gassmanns prediction will be lower
than what is actually observed (Dewar, 2001). The overall rock sample will
appear to be stiffer than assumed in the formulation because the pore shape and
57

viscosity of the fluid make it harder for the pore fluid to equalibrate within the
appropriate time frame. Also as noted before, Gassmanns relation is only valid
for low frequencies such that the motion of the pore fluid and the rock matrix are
perfectly coupled when responding to a passing wave. Comparison to high
frequency data such as sonic log data (~KHz) or laboratory data (~MHz) where
differential movement can arise between the fluid and the solid constituents of a
rock during the passing of a wave would yield a discrepancy in values. If higher
frequency analysis is of interest, then Biots equation should be employed.

3.3.2 Biots Formulation
In contrast to Gassmann, Biot predicted the frequency-dependence of
wave velocities for saturated rocks. In 1956, Biot presented the theory of
propagation of elastic waves in a fluid saturated porous solid in the low frequency
range and the higher frequency range in two publications (Biot, 1956a; Biot
1956b). For his low frequency limit, Biots formulation reduces to Gassmanns
relation and the equation is sometimes referred to as the Gassmann-Biots
equation. For the higher frequency range, his formulation is valid up to the limit
where grain scattering becomes important and the rocks can no longer be viewed
as homogeneous. In his formulation, Biot assumed:
1) The rock is macroscopically homogeneous and isotropic;
2) The rocks minerals all have the same bulk and shear moduli;
3) The fluids can freely move about in the pore space where pores are
interconnected or communicating;
4) The pore space is fully saturated at all times;
5) The pore fluid is Newtonian;
6) The pore fluid does not react with the rock minerals which otherwise
could change the stiffness of the rock frame.
In his derivation Biot accounted for the frequency dependence of wave
behaviours by incorporating viscous and inertial interactions between the pore
fluid and the solid matrix of the rock. At low frequencies, the slow rate of the
oscillating stress from a passing seismic wave allows for a sufficient amount of
58

time for pore fluids to react to the change in stress and reach equilibrium with the
solid matrix. Substantial coupling exists between the pore fluid and the pore wall
and no differential motion will occur between them. The controlling factor on the
pore fluid motion then is the internal resistance of the fluid to flow, namely its
viscosity. Therefore, fluid motion in the low frequency regime of Biots theory is
dominated by viscous effects.
On the other hand at higher frequencies, pore fluids will not have enough
time to react and relax to the passing stress or, rather the pore fluids are resisting
the sudden change to its current state of motion. Therefore, fluid motion in the
higher frequency regime of Biots theory is dominated by inertial effects. The
pore fluid and the pore wall will be decoupled from one another, therefore giving
rise to relative motion between them. As a consequence of the resistance of fluid
flow in the pore space, the overall saturated rock will appear to be stiffer at higher
frequencies compared to at lower frequencies and this phenomenon is called
velocity dispersion. Furthermore, the resulting relative motion between the pore
fluid and pore wall will give rise to wave attenuation.
The low and high frequency range of a saturated porous medium is
determined by a reference frequency, or more commonly known as the Biots
critical frequency, f
c
, given by,
k t
|q
f
c
f
2
= . (3.45)
is the viscosity of the pore fluid and is the absolute permeability of the rock.
For a wave that propagates with a frequency, f, it is in the low frequency regime if
f<<f
c
and it is in the high frequency regime if f>>f
c
. A physical interpretation of
the critical frequency defines it as the frequency where the viscous force equals
the inertial forces acting on the pore fluid (Mavko et al, 2009).
Moreover, as a result of relative motion of the pore fluid and the pore wall
Biot predicted that there are three different body waves generated inside a porous
medium; one shear wave and two compressional waves where the two
compressional waves are called the fast P-wave and the slow P-wave. The fast
and slow P-waves correspond to the overall in-phase or out of phase motion,
59

respectively, between the rock frame and the pore fluid. The fast P-wave is the
most easily observed in the laboratory and in the field. The slow P-wave,
sometimes referred to as Biots slow wave, is highly attenuating and has only
been observed in the laboratory (e.g., Johnson and Plona, 1982; Bouzidi and
Schmitt, 2009). In our study, the slow P-wave is not of importance because only
the fast compressional wave and shear wave will be considered.
Here, a brief overview of the Biots formulation is given; a more extensive
derivation can be found in Bouzidi (2003). The equations of motion governing
wave propagation in porous materials saturated with a single fluid were derived
by considering the total force due to stress acting on the solid constituent and that
acting on the fluid constituent, and incorporating:
1.) the stress-strain relations from considering potential energy effects;
2.) the kinetic energy of the solid, of the fluid, and of the induced mass from
the solid-fluid coupling;
3.) and the dissipation function.
The resulting general wave equations of motion in terms of the displacement
vector for the solid and fluid constituents,
s
u and
f
u respectively, are (Bouzidi,
2003):
) ( ) ( ) (
) ( ) ( ) ( ) 1 (
2
2
2
2
12
2
2
f s dry s dry
f s f s s s
u Q u A u
u u
t
F u u
t
u
t
V V + V V + + V
=
c
c
+
c
c
c
c

k
q|
e |
(3.46)
) ( ) (
) ( ) ( ) (
2
2
2
12
2
2
f s
f s f s f f
u R u Q
u u
t
F u u
t
u
t
V V + V V
=
c
c

c
c
+
c
c
k
q|
e |
(3.47)
where is the viscosity of fluid, is the permeability of the sample,
12
is the
coupling mass density, F() is a viscosity correction factor that is dependent on
frequency, , and A, Q, and R are the Biot parameters. The coupling mass
density, the viscosity correction factor and the Biot parameters will be further
defined later. The solid and fluid displacement vectors can also be expressed
through Helmholtz decomposition in terms of the scalar potential, o , and the
60

vector potential, , to simplify the equations of motion (Bouzidi, 2003). The
displacement vector for the solid and fluid constituents respectively are:
s s s
u o V + V = (3.48)
f f f
u o V + V = . (3.49)
For only compressional waves to occur the divergence operator is applied
to equation 3.46 and 3.47. In terms of the potential fields defined in equation 3.48
and 3.49, the longitudinal wave equation of motion is then

| |
f s dry
f s f s s s
Q A
t
F
t t
o o
o o
k
q|
e o o o |
+ + V
=
c
c
+
c
c
c
c
) 2 (
) ( ) ( ) ( ) 1 (
2
2
2
2
12
2
2
(3.50)

| |
f s
f s f s f f
R Q
t
F
t t
o o
o o
k
q|
e o o o |
+ V
=
c
c

c
c
+
c
c
2
2
2
2
12
2
2
) ( ) ( ) (
. (3.51)
If a regular plane wave solution is applied to both partial differential equations,
the following results

0 ) (
) ( ) 1 ( ) 2 (
2
12
2 2
2
12
2 2 2
=
(

+
(
+ + +
f
s s dry
F i Q k
F i A k
o
k
q|
e e e
o
k
q|
e e e | e
, (3.52)
0 ) (
) (
2
2
12
2 2
2
12
2 2
=
(
+ +
+
(

f f
s
F i R k
F i Q k
o
k
q|
e e | e e
o
k
q|
e e e
. (3.53)
In the equations k is the wavenumber and is the angular wave frequency, where
both are obtained from the plane wave solution. If a ratio of the solid to fluid
scalar potential is defined,
f
s
l
o
o
, = , (3.54)
61

then by combining equation 3.52 and 3.53, and eliminating k, a quadratic equation
can be then obtained in terms of
l

0
2
= + + c b a
l l
, , (3.55)
where the coefficients a, b, and c are:

s dry
Q iF Q A a |
ke
q|
e ) 1 ( ) ( ) 2 (
2
12
+
|
|
.
|
\
|
+ + + = (3.56)
s f d
dry
R A
iF R A b
| |
ke
q|
e
) 1 ( ) 2 (
) ( ) 2 (
2
12
+ +
|
|
.
|
\
|
+ + =
, (3.57)
f
Q iF R Q c |
ke
q|
e
|
|
.
|
\
|
+ + =
2
12
) ( ) ( . (3.58)
Equation 3.55 is a polynomial of the second degree and therefore yields two
solutions by
a
ac b b
l
2
4
2

= , . l=1, 2 (3.59)

Furthermore, if equation 3.52 and 3.53 were combined without eliminating the
wavenumber, then the wavenumber can then be defined as
2
1
) 2 (
) 1 (
(
(
+ + +
+
=
Q R A
k
dry l
f s l
l
,
| | ,
e l=1, 2. (3.60)
The wavenumber is complex and has two possibilities as a result of the two
l

solutions of equation 3.59. The two possible complex wavenumbers corresponds
to the fast and slow compressional waves, as introduced earlier. The real part
of k
1,2
allows the determination of the phase velocity via
2 , 1
2 , 1
k
V
P
e
= , (3.61)
while the imaginary part directly provides the attenuation coefficient of which
will be discussed in the next section.
62

For rotational waves to occur, the curl operator is applied to the general
equation of 3.46 and 3.47. In terms of the potential fields defined in equation 3.48
and 3.49, the shear wave equation of motion is then,

s dry f s
f s s s
t
F
t t

k
q|
e
|
2
2
2
2
12
2
2
) ( ) (
) ( ) 1 (
V =
c
c
+
c
c
c
c
, (3.62)
0 ) ( ) ( ) (
2
2
2
12
2
2
=
c
c

c
c
+
c
c
f s f s f f
t
F
t t

k
q|
e | (3.63)
Applying a plane wave solution to both partial differential equations, the
following results:

0 ) (
) ( ) 1 (
2
12
2
2
12
2 2 2
=
(
+ +
f
s s dry
F i
F i k
k
q|
e e e
k
q|
e e e | e
(3.64)
0 ) ( ) (
2
12
2
12
=
(
+
(
+
f f s
iF iF
ke
q|
e |
ke
q|
e (3.65)
Using equation 3.65 and the defining ratio of the solid to fluid vector potential as,
f
s
, =
3
, (3.66)
the following can be obtained

ke
q|
e
ke
q|
e |
,
2
12
2
12
3
) (
) (
iF
iF
f
+

= . (3.67)
Similar to compressional waves, if equation 3.64 and 3.65 are combined, the
wavenumber can be defined as
2
1
3
3
3
) 1 (
(
(
+
=
dry
f s
k
,
| | ,
e . (3.68)
The wavenumber is also complex and has only one possibility as a result of the
singular solution of
l
for rotational waves (where subscript l=3). Similar to
63

compressional waves, the real part of the complex wavenumber, k
3
, allows the
phase velocity of the shear wave to be determined, while the imaginary part
directly gives the attenuation coefficient.
From the general wave equation of motion, equation 3.46 and 3.47, the
coupling mass density,
12
, the viscosity correction factor, F(), and the Biot
parameters, A, Q, and R were introduced but were not defined. The coupling
mass density describes the induced mass resulting from inertial drag caused by the
relative motion between the solid frame and the pore fluid and is given by
f
| o ) 1 (
12
= (3.69)
where is tortuosity and is a geometrical factor that describes the structure of the
pore space.
The viscosity correction factor or also known as the viscodynamic
operator accounts for the fluid flow that deviates from Poiseuille type flow that
occurs at really high frequencies. At these high frequencies however, the
wavelength is still larger than the pore size. The viscosity correction factor is
|
|
|
.
|
\
|
+
=

e
) (
2 1
) (
4
1
) (
T
i
T
F , (3.70)
where is given by
2
1
2
|
|
.
|
\
|
=
q
e
f
g
(3.71)
and T() is
) (
) (
) ( ) (
) ( ' ) ( '
) (
4
0
4
1
4
3
t
t t
i
i i
e J
e J e
ibei ber
ibei ber
T

=
+
+
= . (3.72)
The parameter g is a pore-size parameter that is a characteristic of the dimensions
and the geometry of the pore space. For spherical pores with a grain diameter
given by d, Hovem and Ingram (1979) obtained
) 1 ( 3 |
|
=
d
g . (3.73)
64

The ber and bei are the real and imaginary parts of the Kelvin function,
respectively, and denotes the first derivative with respect to . J
0
( ) and J
1
( ) are
the Bessel functions of order 0 and 1, respectively.
Finally, to link the stresses in the solid and fluid constituents together Biot
parameters were introduced and they are:
dry dry
dry s s
dry s s
K
K K K
K K K
A
|
3
2
) (
2
2
+
+

= , (3.74)
dry s s
dry s s s
K K K
K K K K
Q
+

=
2
) (
| |
, (3.75)
dry s s
dry
K K K
K
R
+
=
2
2 2
|
, (3.76)
where is called the coefficient of fluid content and is given by
|
|
.
|
\
|
=
s f
K K
1 1
| . (3.77)
The parameters in the above equations, K
s
, K
f
, K
dry
,
dry
, and
f
, are of the
same nomenclature as in Gassmanns equation. Therefore they can be determined
through the methods outlined earlier. In addition, Biots formulation requires the
permeability and the tortuosity of the porous sample, and the viscosity of the fluid
to be known. Similar to porosity, the determination of permeability and of
tortuosity will be given in Chapter 5 on Sample Characterization. For fluid
viscosity, just like fluid bulk modulus and fluid density, it can be found from
auxiliary thermodynamic information.

3.4 Introduction to Elastic Wave Velocity and Attenuation
Determination in the Laboratory
The laboratory method employed in this thesis to study elastic wave
properties and a brief background on wave attenuation will be given in this
section. The ultrasonic pulse transmission is the technique used for obtaining the
waveforms through our samples, from which wave velocity and wave attenuation
can be estimated. The wave velocity is calculated from the arrival time of the
65

propagated signal and wave attenuation is determined by using the log spectral
ratio method. Concluding this section is an experimental study on the wave
attenuation of acrylic that demonstrates the use of the log spectral ratio method.

3.4.1 Laboratory Measurement Method
There are a variety of laboratory methods that can be used to determine the
elastic properties in a rock sample such as stress-strain curves (Gordon and Davis,
1968; Mckavanagh and Stacey, 1974; Bakhorji, 2010), forced
oscillations/resonant bar (Winkler, 1979; Murphy, 1982, OHara, 1985; Yin et al,
1992; Zadler et al, 2004), pulse echo (Winkler and Plona, 1982; McCann and
Sothcott, 1992; Mashinskii, 2005), and pulse transmission. The bases of these
investigative methods vary from cyclic loading and unloading to standing and
traveling waves. The pulse transmission method is one of the most widely used
ultrasonic methods in rock physics (Wyllie et al, 1958; Mobarek, 1971; Nur,
1971; Timur, 1977; Toksz et al, 1979; De Vilbiss, 1980; Tosaya and Nur, 1982;
Blair, 1990; Wang et al, 1991;Cadoret et al, 1995;Thurman et al, 2002;Adam et
al, 2006; Verwer et al, 2008; Fujimoto et al, 2010; Wolf, 2010) and is the only
known method applied to date for CO
2
laboratory studies on elastic waves. The
large usage of the pulse transmission technique in laboratory experiments is
associated with its relatively easy and straightforward application in comparison
to the other techniques. Variables such as pressure, temperature, and saturation
can be easily and are commonly manipulated to examine their effects on seismic
responses. Thereby, this technique is employed in this study.
The basic premise of a pulse transmission measurement requires the
sample to be placed in between a source and a receiver (figure 3.6). The source
and receiver are usually piezoelectric ceramic transducers. When the source is
excited, a wave is produced and propagates through the sample. The arrival of the
wave at the receiver is detected and recorded where the required transit time of the
elastic wave to travel between the source and the receiver through the sample is
determined. With knowledge of the length of the sample, that can be measured,
the wave velocity through the sample can be calculated.
66

Figure 3.6: The basic principle behind ultrasonic pulse transmission method. The white-
dashed arrow denotes the wave path through the sample, traveling from the source to the
receiver.

The wave velocities associated with Section 3.2 and 3.3 are the phase
velocities and are fundamentally different from group velocities. The phase
velocity of a wave is the velocity at which a given phase in a wave propagates,
k
V
phase
e
= (3.78)
where is the angular frequency and k is the wave number . The group velocity
is the velocity of the envelope of a group of interfering waves having different
frequencies and phase velocities.
dk
d
V
group
e
= . (3.79)
In a non-attenuating medium where the waveform shape does not evolve with
time, the two velocities would be the same. However, realistically this is usually
not the case. The type of wave velocity determined from pulse transmission is
debatable and dependent on the method used to determine the first arrival of a
signal. Molyneux and Schmitt (2000) experimentally determined the phase and
group velocities in highly attenuating glycerol-saturated, unconsolidated packs of
glass beads (of two different sizes) and quartz sand. They compared these
fundamental velocities to the velocities estimated from the more common and
simpler methods of determining the first arrival from picking the first amplitude
extremum, the first peak of the Hilbert Transform amplitude envelope, cross-
correlation lags, and the first onset of energy. The authors concluded that the
67

velocities determined from picking the signals first amplitude extremum, the
peak of the amplitude envelope, and from cross-correlation, appear to provide
reasonable estimates of the group velocities. However they also showed that the
velocity determined from using a signals first amplitude extremum compared to
the phase velocity agreed to better than 1% for the two glass beads packs, but
differed by 15% for the quartz sand. If the waves velocity determined from using
a signals first amplitude extremum is actually the waves group velocity, it
should be noted that the propagating medium in their study is highly attenuating;
for a less attenuating medium, the separation between group and phase velocity
would be minimal. The estimate quality factors of the different packs ranged
from 2 to 6 over a frequency range of 400 kHz, with the quartz sand being more
attenuating than the two glass bead packs at any given frequency. Therefore the
large difference between the quartz sand packs phase velocity and the determined
group velocity from using the signals first extremum may be caused by the
higher attenuation of the quartz sand pack. Typical Earth materials would exhibit
less attenuation than that observed in this study. Therefore the wave velocities
determined in the laboratory from pulse transmission by using the signals first
extremum can be a reasonable measure of the waves phase velocity.

3.4.2 Attenuation
When a seismic signal propagates through Earth materials, the amplitude of
the seismic signature decreases as it temporally broadens. This phenomenon is
referred to as physical attenuation. As already introduced, an attenuating medium
will cause a distinction between phase and group velocities. Quantifying physical
attenuation is of great interest because of its implication on seismic analysis and
imaging. Quantitative amplitude studies such as amplitude versus offset (AVO)
requires the effects from attenuation on the amplitude and the phase of signal to
be mitigated or modelled for proper amplitude analysis. In vertical seismic
profiling (VSP), quantifying attenuation can help improve the interpretation of
seismograms by serving as another seismic parameter available for petrological
interpretation. Apart from signal analysis, knowledge of seismic attenuation is
68

also valuable in seismic acquisition. Survey planning can be optimized by
understanding how much signal can be preserved in the designed geometry. In
CO
2
sequestration projects, in addition to the already listed benefits, knowledge of
the attenuation will assist with the identification and quantification of the
subsurface CO
2
plume.
The causes of attenuation can generally be classified as either extrinsic or
intrinsic. Extrinsic attenuation is caused by geometrical effects and material
heterogeneities. The resulting wave dampening is not due to the loss of
mechanical energy, but is instead caused by the defocusing of wavefront energy.
Examples of extrinsic attenuation are wave scattering, internal multiples (multi-
pathing), and geometric divergence. Intrinsic attenuation on the other hand is
from the anelasticity of the propagating medium and results in a loss of wave
energy. It is related to the interaction between a passing wave, the traversed
porous medium, and its saturating fluids. Intrinsic attenuation is thus a function
of the propagating medium and can be used as an additional seismic property in
characterizing Earth materials.
There are a number of factors that can give rise to intrinsic attenuation, and a
few proposed mechanisms for saturated rocks will be briefly mentioned. Walsh
(1966) proposed inter-granular friction or frictional dissipation as a mechanism
and is related to the relative grain-grain boundary slip motion. This mechanism is
amplified in the presence of pore fluid from the lubrication effect between the
grains. Another favoured explanation for intrinsic attenuation is global flow, the
mechanism associated with Biots theory (1956a, b). It accounts for the
differential movements between the fluid and solid constituents in a saturated
porous medium as a wave passes through, therefore causing energy dissipation.
Local flow or squirt flow is another possible mechanism of intrinsic attenuation
and it involves the local motion of the fluid in the pores (OConnell and
Budiansky, 1977; Mavko and Nur, 1979; Palmer and Traviolia, 1981). This is a
result from the compression of grains and consequently the capillary tubes from a
passing wave, thereby inducing pore pressure gradients and causing fluid inside
the heterogeneous pore cavities to squirt.
69

For a harmonic plane wave propagating through an attenuating medium, the
decay of the wave amplitude A(x,t) with space and time can be written as
) (
0
) , (
kx t i x
e e A t x A

=
e o
. (3.80)
A
0
is the initial wave amplitude before traversing a distance, x, and is the
frequency dependent attenuation coefficient (given in units of m
-1
or nepers-m
-1
),
where the frequency of the wave is given by, , and k is the wavenumber. By this
representation, the elastic travel of the wave with space and time is given by the
cyclic nature of the second exponential power where the effective damping of the
wave amplitude is governed by the first exponential containing the negative
product of the attenuation coefficient and the traversed distance. Equation (3.80)
can be, instead, rewritten under a single exponential by using a complex
wavenumber, k
*
,
o i k k =
*
, (3.81)
so that in a simpler form, the wave amplitude can represented by
) (
0
*
) , (
x k t i
e A t x A

=
e
. (3.82)
From equation 3.81, the attenuation coefficient is essentially the imaginary
wavenumber.
Alternatively, instead of describing the attenuation of a waveform by the
attenuation coefficient, the quality factor is used more commonly. The quality
factor, Q, characterizes the dissipation of a medium by a dimensionless quantity
that generally compares the elastic to the inelastic behaviour of a seismic wave. A
large quality factor value indicates low attenuation or loss, where as a small
quality factor indicates the material is very dissipative. The quality factor can be
defined in a number of ways (Lakes, 2009).
Rheologically, for a linear viscoelastic medium, the modulus relating
stress and strain is mathematically complex due to a phase lag between them
(refer to equation 3.26). The real and the imaginary parts of the complex modulus
are the storage modulus, C, and the loss modulus, C, respectively, and are
related to the quality factor as a ratio,
70

"
'
C
C
Q =
. (3.83)
In terms of energy, the quality factor can be expressed as the amount of energy
dissipated in a cycle, E, for a given amount of stored energy in the system,
E
stored
,
E
E
Q
stored
A
=
t 2
. (3.84)

Harmonically, in terms of wave propagation, the quality factor is
|
|
.
|
\
|
=
2
2 2
) (
1
) ( 2 e
e o
e o
e V
V
Q
(3.85)
which can be simplified to
) ( ) ( 2 e e o
e
V
Q =
, (3.86)
for Q>>1, where V() is the frequency dependent phase velocity of the wave.
The log spectral ratio method is a common technique employed to determine
wave attenuation in both field and laboratory seismic by analyzing the times
series of two signals in the frequency domain (Gladwin and Stacey, 1974; Toksz
et al, 1979; Tonn, 1991; Wepfer and Christensen, 1991;Tutuncu et al, 1994;
Sarma and Ravikumar, 2000; Diallo et al, 2003; Rickett, 2006; ; Santos et al,
2009; Jaya et al, 2010). For a signal acquired in the laboratory via the ultrasonic
pulse transmission technique, the Fourier amplitude spectrum of a pulse through a
sample length x
i
can be expressed as (after Yin, 1993):
) , ( ) 1 ( ) ( ) ( ) , (
2 ) (
i
x
i
x G R e D S x A
i
e e e e
e o
=

. (3.87)
S() is the amplitude spectrum of the outgoing pulse, D() is the amplitude
spectrum of the receiver, G(,x
i
) is geometric effects, and R is the reflection
coefficient from the aluminum end cap of the transducer to the sample interface.
The spectra of two signals corresponding to different traveling distances of the
same viscoelastic medium can be related by
) , (
) , (
) , ( ) , (
1
2 ) )( (
1 2
1 2
x G
x G
e x A x A
x x
e
e
e e
e o
= (3.88)
71

if the same signal transmitting and receiving equipment is used. Therefore if the
amplitude spectra can be determined, the attenuation coefficient of a viscoelastic
medium for a range of frequencies can be calculated by
) (
) , ( ) , (
) , ( ) , (
ln
) (
1 2
2 1
1 2
x x
x G x A
x G x A
|
|
.
|
\
|
=
e e
e e
e o , (3.89)
as hinted by the name of the method. Additionally, the quality factor can be
found from the attenuation coefficient if the phase velocity, V
p
(), of the
propagated signal is known,
) ( ) ( 2 e e o
e
p
V
Q = . (3.90)
The phase velocity at a given frequency is
phase
p
t
x x
V
A

=
) (
) (
1 2
e , (3.91)
where t
phase
is the phase lag in time between the two signals found by
differencing their corresponding phase spectra over frequency (figure 3.7)
e
|
e
A
= A ) (
phase
t . (3.92)

Figure 3.7: Illustration of phase velocity determination (Figure reproduced from Qi (2008)
after Molyneux and Schmitt (2000))*. a) Elastic wave pulses through two different sample
lengths. b) Unwrapped phase of the Fourier transforms of the times series pulses. c)
Determination of the frequency-dependent time phase.
* Reprinted with permission from the Society of Exploration Geophysicists.
72

In attenuation measurements, large errors are quite common (Bourbie et al,
1987, White, 1992; Molyneux and Schmitt, 2000) due to uncertainties in the
spectral decomposition of a signal. From the conventional log spectral ratio
method, the associated errors can be estimated statistically from redundant
attenuation measurements if multiple (more than 2) lengths of the same samples
are used (Molyneux and Schmitt, 2000; Qi, 2008). An example for demonstrating
the analysis of waveform attenuation with errors using the log spectral ratio
method will be shown in the next section.

3.4.3 The Attenuation of Ultrasonic Waves through Acrylic
A series of ultrasonic pulse transmission measurements were performed on
four different lengths of acrylic. Acrylic is a soft, viscoelastic, synthetic polymer
of methyl methacrylate that is commonly known as Plexiglass. For each sample
length, the acrylic sample was placed in between a pair of transmitting and
receiving transducer composed of 1 MHz P- and S-wave piezoelectric ceramics.
The time series of the transmitted signal through a sample is received by an
oscilloscope at a sampling interval of 10 nanoseconds. During data acquisition,
the sample with the pair of transducers was subjected to a confining pressure to
improve the transducer-sample coupling, thereby enhancing a received signals
quality. Further details on the transducers, the experimental set up, and data
acquisition system used in these measurements can be found in Chapter 4.
P- and S-wave time series signals were acquired under a confining
pressure of 20 MPa for acrylic lengths of 2.00, 3.00, 4.00, and 5.00 cm. The
recorded P- and S-wave traces are shown in figure 3.8 where the dampening of
wave strength with increasing sample length is apparent. However, temporal
broadening of the waveform is less obvious. The times series signals were
windowed and the Fast Fourier transform (FFT) was applied to obtain their
equivalent in the frequency domain. The corresponding amplitude spectra of the
signals are shown in figure 3.9. As expected, the strength of the Fourier amplitude
spectrum is reflective of their signal strength and decreases with longer acrylic
sample lengths. The peak of all P-wave spectra is centered at 0.70 MHz, while
73

Figure 3.8: The time series signals recorded through varying lengths of acrylic: P-waves
(top) and S-waves (bottom).

for the S-wave spectra the peak frequency generally shifts to a lower frequency
with increasing sample length. The peak frequency is 0.59, 0.56, 0.52, and 0.54
MHz for the 2.00, 3.00, 4.00, and 5.00 cm sample, respectively. A detailed
examination of the S-wave signal corresponding to the 4.00 cm long acrylic
shows a slightly different morphology than the rest of the acquired S-wave
signals. Thereby a skewed amplitude spectrum from this signal is evident through
comparison to the rest of the S-wave signals. This variation can be attributed to
misalignment of the pair of transducers (the S-wave transducer is orientation-
sensitive) or to the poor connection of electrical wires in the experiment. For both
the P- and S-wave amplitude spectra the majority of the energy is spread between
74

0.4 to 1.4 MHz. To avoid large errors, only the data in this useful frequency
range is analyzed.

Figure 3.9: The amplitude spectra determined of the time series signals shown in figure 3.8:
P-waves (left) and S-waves (right).

For all the possible combinations of a pair of samples with different
lengths, the natural logarithm was taken of their signals amplitude spectra ratio at
each frequency, and plotted against their respective length difference,
independently for P-and S-waves. The P- or S-wave attenuation coefficient of
acrylic was determined from the plot based on the slope of the best fit line, as a
function of frequency. In this study, the diffraction loss from the geometry of the
samples was ignored. The corresponding attenuation coefficient error was
75

estimated by finding the maximum absolute difference between the slope of the
best fit line and the slopes of all the possible lines that can fit through any two
data points. The computed P-wave attenuation coefficient as a function of the
frequency is shown in figure 3.10 (top plot). From 0.4 MHz to 1.4 MHz, the P-
wave attenuation coefficient varies from 14.4 m
-1
to 24.9 m
-1
, and at the peak
frequency of 0.70 MHz it has a value of 15.3 m
-1
. The largest P-wave attenuation
coefficient error estimated is 12.1 m
-1
. The S-wave attenuation coefficient varies
from 16.6 m
-1
to 38.3 m
-1
, and at the averaged peak frequency of 0.55 MHz it has
a value of 21.5 m
-1
. The largest S-wave attenuation coefficient error estimated is
18.2 m
-1
(figure 3.11, top plot). With increasing frequency, both the P-wave and
S-wave attenuation coefficient generally increases.
The phase velocity was found using the procedure outlined in section 3.4.2
for all the possible length combination pairs, and the averaged value was used to
find the quality factor. The phase velocity for both P-and S-wave varied
minimally (~ 10 m/s) over the range of frequencies analyzed. The P-wave quality
factor, Q
p
, with errors propagated from the attenuation coefficient and phase
velocity is displayed in figure 3.10 (bottom plot). Over the useful frequency
range, the quality factor varies from 21.4 to 65.4, and at 0.70 MHz the quality
factor is 45.2. The largest error of the P-wave quality factor estimated is 26.6.
The computed S-wave quality factor, Q
s
, is shown on bottom plot of figure 3.11.
The S-wave quality factor varies from 49.7 to 131.4, and at the averaged peak
frequency of 0.55 MHz, it is 57.3. The largest S-wave quality factor error is
118.7.

76

Figure 3.10: P-wave: (top) Attenuation coefficient and (bottom) quality factor, as a function
of frequency.

Figure 3.11: S-wave: (top) Attenuation coefficient and (bottom) quality factor, as a function
of frequency.
77

3.5 Summary
In order to understand seismic wave behaviour pertaining to geologic CO
2

sequestration, factors that will influence subsurface wave propagation were
reviewed. From reviewing the thermodynamics of CO
2
and from using a
simplified temperature and pore pressure model of the upper section of the
sedimentary basin, CO
2
can exist as a gas, a liquid, or a supercritical fluid for a
geological sequestration project. The physical properties of these phase states
relevant to wave travel were obtained from Span and Wagners (1996) empirical
equation of state and phase diagrams of these properties were generated. The
behavioural variations of the bulk modulus, density, and viscosity of CO
2
with the
corresponding phase state were consistent; a large change in properties for a gas-
liquid transition and a subtle change for a gas- or liquid-supercritical fluid
transition.
To describe the basic behaviour of elastic wave travel, the theory of
elasticity was first reviewed. When a volume of perfectly elastic material is
subjected to a force or is stressed, deformation will result and it is fully
recoverable when the force is removed. Stress and strain relations were reviewed
through Hookes law where the elastic stiffness constant was established for a
linearly elastic isotropic medium. Starting from 81 elastic constants, the tensor
was reduced to 2 independent constant for an isotropic medium. From
incorporating Hookes law into Newtons second law of motion, the wave
equation for an isotropic medium was shown. P- and S-wave velocities in terms
of the elastic constants and the density of the material were defined. Furthermore,
in contrast to a perfectly elastic material the basic concept of viscoelasticity along
with its implication on wave travel was discussed. The viscous behaviour
component of a viscoelastic material will give rise to the attenuation of a
propagating wave.
Rock physics, the branch of geophysics that relates geophysical
observation to their physical cause, was reviewed by the fluid substitution
formulations of Gassmann (1951) and Biot (1956a, b). Gassmanns equation is
easy to apply in relation to Biots equation; however Gassmanns formulation has
78

limitations that are particularly concerning for our application, namely the
restriction to low frequencies. In contrast Biots formulation is valid for a wide
frequency range and differentiates the pore fluid behaviour with respect to the
mineral grain, i.e. decoupling motion. Physical velocity dispersion, wave
attenuation, and two compressional waves corresponding to the in and out of
phase motion between the rock frame and pore fluid, are the distinct results
produced from Biots formulation.
Lastly, an introduction to the foundation of ultrasonic pulse transmission
was given. Phase and group velocity of transmitted signal was discussed. Based
on Molyneux and Schmitts (2000) work, it was determined that from picking the
first arrival of a signal, the resulting velocity is a reasonable approximation for
phase velocity. In addition to analyzing wave velocities from ultrasonic pulse
transmission, the attenuation of an ultrasonic wave was discussed. The complex
wave number, the attenuation coefficient and the quality factor used for
characterizing wave dampening were formally defined. The actual application of
the ultrasonic pulse transmission method and the analysis of wave attenuation
experimentally were carried out by using varying lengths of acrylic. The
amplitude dampening of waves through longer sample lengths is clearly evident in
both the time and frequency domain of the signals. The log spectral ratio method
was applied, and the attenuation coefficient and the quality factor of acrylic were
determined. The typically large uncertainties associated with attenuation
measurements are observed here, where the largest the errors observed with our
determined P-and S-wave attenuation coefficients are 12.1 m
-1
and 18.2 m
-1
,
respectively.

79

Chapter 4
Experimental Setup and Procedure

As previously discussed in Chapter 3, ultrasonic pulse transmission is the
underlying methodology in our study applied to determine the wave velocity and
attenuation in CO
2
saturated samples. The application of this method allows the
recording of a P- or S-wave after propagation through a sample. Ultrasonic pulse
transmission is a commonly applied technique to determine the dynamic elastic
properties in rocks and to the best of our knowledge, has been the only laboratory
method applied to date concerning CO
2
rock physics. In this chapter, the working
application of this method will be introduced through a description of the
experimental setup and the procedures for the laboratory measurements. First, the
construction and the workings of a piezoelectric transducer will be discussed.
Then the sample preparation, assemblage and the experimental apparatus will
follow. Next, the experimental procedure and the acquisition of data will be
given. Finally, examples of waveforms will be shown and a discussion on the
determination of wave velocities, wave attenuation, and associated errors will
ensue.

4.1 Piezoelectric Transducers
Transducers are the main component of pulse transmission measurements.
In order to transmit and receive an elastic wave in the laboratory, ultrasonic
transducers to accommodate cylindrical core samples with 2.54 cm diameter were
made. In this section, the materials and construction of the transducers used in
this study are reviewed. The main components of our ultrasonic transducer are
80

the aluminum buffer caps, P- and S-wave piezoelectric ceramics, copper foil
electrodes, and the damping material.
To generate P- and S-waves in ultrasonic pulse transmission experiments,
piezoelectric ceramics are typically employed. Piezoelectric materials are
materials that are capable of producing an electrical potential when experiencing
an applied stress such as a mechanical vibration, and this effect is linear and
reversible such that when an electrical potential is applied the material will exhibit
mechanical strain or a mechanical vibration. Piezoelectricity is based on the
internal structures of the material and is caused by the linear interaction between
the electrical and mechanical behaviour of the medium. This phenomenon can
only be exhibited in materials whose crystal structure has no center of symmetry
(Zhu and Meng, 2002). The type of vibration generated is determined by the
polarization of the piezoelectric crystal. Longitudinal waves (P-waves) are
generated if the material is axially polarized such that axial compression and
expansion can occur. Transverse waves (S-waves) are generated if the material is
laterally polarized allowing for shear motion to occur. The piezoelectric material
used in this study (shown in figure 4.1) is a ceramic made from lead zirconate
titanate with a resonant frequency of 1 MHz, manufactured by Omega Piezo
Technologies Inc.
From the direct and reversible characteristic of piezoelectricity mentioned
above, both the transmitting and receiving ultrasonic transducers were built in the
same way by using the same materials and arrangements. On an aluminum buffer
end cap, the different vibration mode ceramics were mounted in a stacked
configuration to allow for the simultaneous measurement of P-and S-waves.
Before construction, the surface of the aluminum buffers was polished for a
smooth and flat surface to prevent scattering of the generated waves. The S-wave
piezoelectric ceramic was placed directly on the surface of an aluminum buffer
cap and then the P-wave piezoelectric ceramic was placed on top of the S-wave
ceramic. Separating the P- and S-wave ceramics is a piece of copper foil which
acts as a common electrode between the two piezoelectric

81

Figure 4.1: P-(circle) and S-wave (square) piezoelectric ceramics used in our transducers.
The larger increments on the topside of the ruler are in centimetres.

ceramics. It serves as the positive terminal for the S-wave ceramic and acts as the
negative terminal for the P-wave ceramic. The negative terminal for the S-wave
ceramic will come from the contact with the buffer cap, while the positive
terminal for the P-wave ceramic comes from another piece of copper fixed onto
the top of the ceramic. Each piece mentioned above in this assembly was attached
to one another using CircuitWorks CW2400 silver conductive epoxy. To
prevent any short-circuiting of the positive and negative terminals, excess
conductive epoxy was removed from the edges of the ceramics and tested for
conductivity between each mounting. The copper foils attached only extended a
short amount past the stacked ceramics and insulated electrical wires were
soldered onto the exposed copper foils for extension. In addition, after the silver
epoxy from the last attachment has dried, a thin film of non-conductive, general
purpose epoxy was applied to the surface of the P- and S-wave ceramics and the
exposed copper.
After this applied epoxy has dried, damping material made from a mixture
of urethane rubber (Flexane 80 Liquid) and tungsten filings were placed on top
of the transducer arrangement in a cylindrical mould. To prevent the quick
deterioration of the transducers from hydraulic oil intrusion during the lab
measurements, the hardened damping material, the exposed ceramics, and parts of
82

the aluminum buffer were set in liquid urethane rubber for sealing. Hydraulic oil
is the confining medium in the pressure vessel.
The construction of the transducers were completed at least a week prior
to use because the urethane rubber requires at least 7 days to cure to its full
strength. A cartoon depicting the different components and a photograph of a
completed source and receiver transducer pair are shown in figure 4.2. One of the
transducers in the pair contains a stainless steel tube that extends from the top of
the aluminum buffer cap. The end of the tube that is not shown in the figures is
connected to a hole that traverses the full, vertical length of the buffer and allows
pore pressure control and fluid saturation of the sample.

4.2 Sample Preparation
Ultrasonic pulse transmission measurements were conducted on a
synthetic and a non-synthetic sample, a porous ceramic rod and a Berea
sandstone, respectively. Cylindrically shaped samples of 2.54 cm in diameter and
of lengths greater than 4 cm were obtained for the measurements. The non-
synthetic sample was cored under wet conditions using a 2.54 cm diameter
diamond encrusted bit. Water was continually flushed through the drill bit and
the sample during the coring process for cooling and lubrication. Afterwards, the
end faces of the core sample were made parallel with first using a wet saw to
achieve approximate parallelism, and then the end faces were fined tuned and
smoothed with the use of a wet grinder. The parallelism for the cored sample was
measured using a dial gauge and was only deemed acceptable if it was within 2
thousandths of an inch (within 0.025 mm). Smooth and parallel end faces are
important for enhanced signal transmission and to minimize velocity
measurement inaccuracy. The synthetic sample was purchased and came in a
cylindrical dimension of 2.54 cm in diameter and 5 cm in length. The end faces
were flat and parallel, and no additional work was required for the synthetic
sample. Prior to any lab measurements, both samples were dried in an oven at
70C under vacuum for at least 48 hours and kept in a desiccator jar afterwards
until in use.
83

Figure 4.2: (top) Cartoon depiction of the various components in a transducer. (bottom) A
photography of a completed pair of transducers (pulser and receiver). The transducer on
the right has the pore pressure inlet.

The final preparation for ultrasonic pulse transmission measurements
involves assembling the sample with the transducers and pieces that will provide
sealing of the sample from hydraulic oil inside the pressure vessel. The sample
was placed inside a flexible, 2.54 cm wide Tygon tubing and a piezoelectric
transducer was placed on each end of the sample, with a part of the transducer
underneath the Tygon tubing. Since S-waves are polarized, the transducers on the
ends of the sample were aligned with care to achieve proper polarization of the S-
wave ceramics. Improper alignment of the shear ceramics would lead to a weak
84

S-wave signal and even an absent signal if the two shear ceramics were positioned
90 to each other. To prevent hydraulic oil from contaminating the sample by
seeping in from between the transducer and the overlapping Tygon tubing, rubber
O-rings were slipped around the aluminum buffers and underneath the tubing. In
addition, metal hose clamps were placed around the outside of the Tygon tubing,
positioned directly above the underlying rubber O-rings, and were tightened to
maximum. A fully assembled sample ready to be placed inside the pressure
vessel for ultrasonic measurements are shown in figure 4.3. The workings of the
rest of the experimental apparatus will be discussed in the next section.

Figure 4.3: A fully prepared and assembled sample ready for measurement. The sample is
placed in between a pair of transducers and is jacketed in a clear Tygon tubing. Iron hose
clamps are tightened against the tube and the rubber o-rings that surrounds the aluminum
buffer.

4.3 Experimental Apparatus
Aside from the sample and transducer assemblage, the main components
of the experimental set up consists of the pressure vessel, the pore fluid source,
pressure gauges, a thermocouple, an electrical resistance tape, a pulse generator,
and a digital oscilloscope. Figure 4.4 shows the schematic and the photograph of
the pressure vessel and the rest of the experimental setup. The sample and
transducer assemblage is placed inside the pressure vessel, in a 10 cm wide and 40
85

cm deep cylindrical cavity filled with hydraulic oil. As mentioned earlier, the
hydraulic oil serves as the pressurizing medium for providing hydrostatic
confining pressure on the sample.
The desired pore fluid is introduced into the sample via stainless steel
tubing that connects the pore space of the sample to the pore fluid reservoir
located outside of the vessel through the vessel lid. The confining and pore
systems are independent of each other such that different pressure conditions can
be applied in irrespective of each other by using different pumps located outside
of the pressure vessel. A Quizix
TM
Q5000 pump system was used to control both
pressure systems. Both the confining pressure (pressure vessel) and the pore
pressure system are capable of reaching a maximum pressure of 70 MPa from this
pump system. Higher confining pressures, up to 200 MPa, could also be achieved
by using a separate air pump connected to the pressure vessel.
For CO
2
saturated measurements, a CO
2
cylinder of 99.9% purity was
used. The pressure inside the tank is essentially buffered at the vapour-liquid
transition pressure at the given room temperature; normally the CO
2
leaving the
tank is gaseous. To generate pore pressures higher than the pressure in the tank,
theoretically the pump can be used to pressurize gaseous CO
2
into a liquid phase
state at room temperature. However practically this requires a lot of pumping and
time. Since at room temperature the CO
2
in the tanks are partly in the gaseous
phase and partly in the liquid phase state, the denser liquid state can be obtained
directly by simply inverting the tank thereby allowing liquid CO
2
to be the first to
leave the nozzle of the tank.
The ambient temperature of the pressure vessel can be raised from room
temperature from the heating of an electrical resistance tape that is wrapped
around the outside of the pressure vessel. Desired temperature of the pressure
vessel is controlled by the turning on and off of the power supply to the electrical
resistance tape. The actual temperature inside of the pressure vessel is determined
from a K-type thermocouple and is located immediately next to the sample.

86

Figure 4.4: a) A simplified schematic of the experimental set up. b) Photograph of the
laboratory equipment used. The acquisition system behind the pressure vessel and is not
captured in the photograph.

87

The experimental set up pertaining to generating and recording the elastic
waves consists of a pulse generator, a digital oscilloscope, a switchbox and
electrical wire feed-throughs. The electrical wire feed-throughs are soldered onto
the electrical wires of the transmitting and receiving transducers and are fed
through the lid of the pressure vessel to allow for the external excitation and
recording of the pulsed signal. The transmitted signal is generated by triggering
the transmitting transducer with a fast rising 200 V square wave using a JSR-
PR35 pulse generator. The propagated signal is recorded by a digital oscilloscope
made by National Instrument, at a sampling rate of 10 nanoseconds. To reduce
random noise effects, the final waveform recorded is a stack of over 500 traces.
The switchbox was used to facilitate the switch between exciting and recording of
the P- and S-waves at a given condition.

4.4 Experimental procedure
The samples were subjected to a series of measurements including
measurements made under dry conditions, various CO
2
saturated conditions, and a
water saturated condition. Before describing the different temperature, pressure,
and fluid-type saturation conditions undertaken during this study, it is important
to define the different types of pressures that can be applied to a sample during a
measurement. As mentioned in the previous section, the pressurizing system
controls two types, confining pressure and pore pressure. Confining pressure is
the external pressure applied onto the surface of the sample. Pore pressure is the
pressure of the fluids inside the pore space of a sample. These two types of
pressures are exerted in opposite directions in relation to each other, where the
confining pressure pushes the grains of sample together and the pore pressure
pushes the grains out. The net effect of these two pressures applied on a sample is
the differential pressure, P
d
, and is defined as
p c d
P P P = , (4.1)
where P
c
is the confining pressure and is usually larger, in-situ, than the pore
pressure, P
p
. Indeed, P
p
in excess of P
c
will lead to hydraulic fracturing of the
rock mass.
88

During the measurements for a given sample or between the two samples,
the conditions applied differed slightly, but fundamentally the measurements were
the same and the intended conclusions could be still drawn even with the slight
differences. These differences in applied procedure are due to unexpected
circumstances that have arisen during the measurements and to new learning
along the way that have resulted in procedural tuning between the different
measurement runs to increase more relevant data and reduce less useful data. Re-
measurement of the samples for perfect procedural consistency was not done due
the time intensity required for a complete set of measurements for a sample and
the high demand for the use of the pressure vessel and system. To follow are the
set of measurements and procedures attempted for a given sample. Both the
synthetic and non-synthetic samples have slight measurement deviations from this
list and their deviations are mentioned.
The first set of ultrasonic measurements conducted on the samples was the
dry measurements. For dry conditions, the pore space is ideally empty and the
pore pressure is consequently zero. Air inside the pore space of the sample and
inside the pore tubings of the pressure vessel system are pumped out over a
duration of 12 hours by attaching a vacuum to the pore pressure system. When
applying the vacuum to the pore pressure system, the drop in pressure follows an
exponential curve with a rapid drop in the first hour followed by a slower and
steady drop in the second hour and remains unchanged for the next 10 hours. The
final pressure reading on the pressure gauge is usually about 0.0024 KPa.
Compared to the initial pressure of about 101.3 KPa, the final pressure is only
0.002% of the original atmospheric pressure. Even though after 12 hours of
vacuuming the pore pressure is not zero but, the pore space is essentially empty
for the purposes of these experiments, as the mass and the gaseous compressibility
are negligible. These dry measurements were first conducted at room
temperature as confining pressure was varied between 5 MPa and 40 MPa at 2.5
MPa intervals during both pressurization and depressurization cycles, while the
pore pressure was under vacuum. Afterwards, heated dry measurements were
conducted at a constant confining pressure of 10 MPa while the temperature of the
89

pressure vessel was raised. During the heated attempt for the non-synthetic
sample, the connector of the thermocouple to the pressure vessel malfunctioned
and this was not realized until the vessel reached 55C. Therefore the changes in
the waveform as the sample warmed were not mapped, instead only the
waveforms at two temperatures, 23C and 55C, were recorded. Nonetheless, the
temperature dependency of the sample is still exhibited by these two
measurements.
Following the dry measurements, a large set of CO
2
saturated
measurements were made under various pressure and temperature conditions.
The samples were saturated with CO
2
by connecting the pore pressure system
under vacuum to the CO
2
tank, and left overnight at a pore pressure of 10 MPa for
the sample to be fully saturated. For both samples, 5 different constant
temperature runs were attempted while pore pressure varied from 2 MPa to 25
MPa. The constant temperature runs were done at 23 C, 28 C, 40 C, 45 C,
and 55 C. As pore pressure increased for the 2 lower temperature runs (23C and
28C), CO
2
changed from a gas phase to a liquid phase while for the higher
temperature runs (40C ,45C, and 55 C), CO
2
changed from a gas phase to the
supercritical fluid phase. For the synthetic samples 55C run, it was only
partially completed because there was a lack of CO
2
in the tank to build higher
pressures during the measurements. Unfortunately, the sample was later
contaminated by hydraulic oil leakage and the measurements could not be
repeated.
To reach the higher temperatures from room temperature (~23C), the
heating typically required 2 or more hours. To use time effectively, velocity
measurements were also conducted as the temperature changed during heating
while holding both confining and pore pressure constant for the synthetic sample.
Constant pressure runs were conducted at 7 MPa, 10 MPa, and 25 MPa as the
pressure vessel was heated from room temperature (~23C) to the desired end
temperature for the constant temperature run. For the non-synthetic sample,
constant pressure runs were conducted separately from the heating up of the
vessel for constant high temperature runs so that a final temperature of 50C could
90

be reached for consistency. At 7 MPa, the CO
2
changed phase states from liquid
to gas, while for the 10 MPa and 25 MPa the CO
2
underwent a liquid to
supercritical phase transition. For both constant temperature and constant pore
pressure runs, a constant differential pressure of 15 MPa was maintained by
varying the confining pressure accordingly to the pore pressure.
After the series of CO
2
saturated measurement runs were completed and
prior to water saturated measurements, the pore system along with the pore space
of the samples was vacuumed for 12 hours and dry measurements at room
temperature were repeated under the same conditions as before. The purpose of
this second dry run was to verify if the nature of the sample has changed. This
second dry run for the synthetic sample was done after the partial completion of
the CO
2
measurements at T = 55C, as we waited for the arrival of a new CO
2

tank.
The final run for the samples were water saturated measurements.
Distilled water was introduced into the vacuumed system from a water reservoir.
The sample was left overnight at a pore pressure of 10 MPa to reach full
saturation. Water saturated measurements were conducted the next day at room
temperature with the pore pressure varying from 2 MPa to 25 MPa while
maintaining constant differential pressure of 15 MPa. As mentioned above due to
oil contamination of the synthetic sample, water saturated measurements were not
conducted. However, water saturated measurements on the sample were
conducted in an earlier set of measurements. The conditions applied were
different than the ones listed above. They were done at room temperature with a
constant differential pressure of 5 MPa while pore pressure varied from 5 to 50
MPa.
All measurements listed above involved the acquisition of both P- and S-
waveforms. For the measurements that involved changing pressures, 15 minutes
were allowed before recording the waveform to allow conditions inside the
pressure vessel to equilibrate. To summarize, the measurements conducted for a
sample in terms of the type of saturation, temperature, and pressure conditions
applied are given in table 4.1.
91

Type of
Measurement
(saturation)
Temperature

(C)
Pressure
Confining
(MPa)
Pore
(MPa)
Differential
(MPa)
Dry
23
a
5-40 0 5-40
23-50
b
10 0 10
CO
2

23 17-40 2-25 15
28 17-40 2-25 15
40 17-40 2-25 15
45 17-40 2-25 15
55
c
17-40 2-25 15
23-50 23 7 15
23-50
d
25 10 15
23-50
e
40 25 15
Water
23
f
17-40 2-25 15
23
g
10-50 5-45 5
Table 4.1: The measurements carried out for a sample during ultrasonic pulse transmission
experiments.
a
measurement run conducted twice, prior and after CO
2
saturated
measurements.
b
Porous ceramic rods end temperature was 40C and only the beginning
and end temperature waveforms were recorded for the Berea sandstone.
c
For the porous
ceramic rod, the maximum pore pressure reached was 7 MPa for this measurement run.
d
Porous ceramic rods end temperature is 40C.
e
Porous ceramic rods end temperature is
45C.
f
measurement run for the Berea sandstone.
g
measurement run for the Porous
ceramic rod.

4.5 Velocity and Attenuation Analysis Methodology
As per the experimental procedure reviewed above, a large suite of P- and
S-waveforms was acquired under various saturation, pressure, and temperature
conditions for each sample. From these collected waveforms the P-and S-wave
velocities and attenuations can be analyzed as a function of the applied conditions.
The method behind wave velocity and wave attenuation analysis from the
acquired signals will be described in this section.

92

4.5.1 Velocity and Error Analysis
Before determining the P- and S-wave velocities from the acquired
waveforms, the time axis of the waveforms first need to be calibrated. The arrival
time of a P- or S-wave determined from the recorded waveforms is a combination
of the time required for the signal to travel through both the sample and the
aluminum buffer caps. To determine the actual velocity through just the sample,
the time effects from the aluminum buffer caps therefore need to be eliminated
and are referred to as the delay time. To determine this delay time for a
transducer set, calibration measurements were conducted by performing ultrasonic
pulse transmission through just the pair of buffers prior to conducting the suite of
measurements on the sample. The delay time is affected by pressure, therefore
the calibration measurements were done over the range of pressures that would be
encountered during the measurements using a sample. Figure 4.5 is a plot of the
normalized traces collected during the buffer measurements under confining
pressures of 5 MPa to 50 MPa at room temperature. Each waveform is
normalized with respect to the largest amplitude of the trace. At first glance, the
arrival times of the signals show little pressure dependency over this pressure
range. A closer and more detailed look reveals a slight linear dependence on
pressure with the signals at higher pressures arriving earlier than at lower
pressures. If this arrival time pressure dependency is ignored, this can translate to
an incorrect velocity determination of ~50 m/s for the samples. Therefore care
must be taken in applying the correct calibration measurements to the
measurements made under the corresponding pressure on the samples. Figure
4.6a shows a signal through just the buffers under a confining pressure of 15 MPa
and figure 4.6b shows the signal at the same confining pressure traveling through
the buffer and the dry sample. The delay time of the signal in plot b is determined
from the arrival time of the signal in plot a.
The arrival times of the signals were picked using the first extremum and
are indicated by the red circle in figure 4.6. By determining the difference in
arrival times of the signal through the buffers with sample (t
bs
) and the signal
through just the pair of buffer (t
b
), the travel time of the signal through the sample
93

Figure 4.5: Buffer measurements over a confining pressure range of 5 MPa to 50 MPa: (a)
normalized P-waveforms, (b) normalized S-waveforms. The actual S-wave arrivals are
marked by the red box.

Figure 4.6: P-waves under a confining pressure of 15 MPa (a) through only the aluminum
buffer cap (b) through both the aluminum buffer and synthetic sample. The red circle
marks the extremum picked for the arrival time of the signal.

94

can be determined (t
s
). Consequently, the signals velocity, v, through a sample is
determined from this travel time in conjunction with the samples length, L
s
, via
the simple relation:
b bs
s
s
s
t t
L
t
L
v
= = (4.2)
In any type of work involving measurements, error will always be present.
This error is associated with imprecision and uncertainties of the measurement. In
order to calculate the velocity in equation 4.2, three variables had to be measured
and therefore three sources for error exist in contributing to the error in velocity.
The obvious uncertainties in the variables of equation 4.2 are determining the
sample length and picking a signals arrival time for both the buffer calibration
measurements and the sample measurements. For the former source of error, the
end faces of the samples were ground until a parallelism was within a thousandth
of an inch, therefore the error in sample length is 210
-5
m. For the latter source
of error, the error is variable depending on the quality of signal. There is a
minimum error of 10 ns (1010
-9
s) for high quality signals (the sampling rate is
1010
-9
s in our data acquisition system) and a maximum error of 3010
-9
s for
low quality signals (the time window where the largest peak or trough value is
certainly within for the poorest quality signals).
Apart from these 3 obvious sources of errors, other less apparent factors
and errors also need to be addressed that may be significant and will contribute to
the total velocity error. Additional error to length determination could be sample
shortening under pressure. Hemsing (2007) suggested that sample shortening is
not a great concern as he estimated a 0.1-0.2% velocity change over the range of
pressures he subject his rock samples to in his measurements. The amount of
sample shortening, L, under a certain change in pressure, P, can be estimated
from the samples Youngs modulus, E, and the sample length measured at room
conditions, L, by
E
P
L
L A
=
A
. (4.3)
95

Using equation 4.3, a sample shortening of 1.310
-4
m was estimated for the least
stiff sample under a pressure of 50 MPa for our study. Furthermore, during
calibration measurements the temperature dependency of the buffer delay time
was not considered. All materials will experience thermal expansion when
heated, with our aluminum buffer caps expanding more readily than our porous
medium samples which are consisted of either alumina or quartz grains. For
measurements conducted at 50C, using the calibration measurements conducted
at room temperature (~23C) is therefore inaccurate. The linear expansion of a
material can be calculated using:
T
L
L
A =
A
o (4.4)
where L is expanded length, L is the original length of the sample, is the linear
expansion coefficient, and T is the temperature change. The linear expansion
coefficient for aluminum is 2.3410
-5
/C. A temperature change from 23C to
50C then translates to a 4.310
-5
m lengthening of the aluminum buffer caps. If
it is assumed that the elastic wave velocities through aluminum stays relatively
constant over this 27C temperature change, then the lengthening of the aluminum
buffers at 50C would increase the delay time by 710
-9
s for P-waves and 110
-8
s for S-waves, compared to the delay times measured at 23C. The thermal
expansivity of alumina and quartz is 410
-6
/C and 0.5510
-6
/C, respectively,
therefore the lengthening of our samples in this study is considered negligible
from the effects of heat.
After considering these different possibilities for error, the error in
velocity can be calculated by the propagation of error using equation 4.2:
2 2
|
|
.
|
\
|
+
|
|
.
|
\
|
=
s
s
s
s
t
t
L
L
V
V o o o
(4.5)
where is the error of the particular parameter. The largest velocity error possible
for a measurement involves the shortest sample length and the shortest travel
time. The non-synthetic sample, Berea, has the shortest length of 43.45 mm and it
has the shortest travel time under dry conditions and the highest confining
pressure. The error for length is 210
-5
m from parallelism and 1.310
-4
m from
96

sample shortening, totalling an error in length of 1.510
-4
m. The maximum error
in picking arrival times for buffer calibration and sample measurements due to
poor signal quality is 310
-8
s for each, and the error in the arrival time for buffer
calibration due to thermal expansion of the buffer caps is 7.010
-9
s for P-waves
and 1.010
-8
s for S-waves. This totals an error in time of 6.710
-8
s and 710
-8
s
for P-and S-waves, respectively. Using equations 4.5, the largest percentage error
in P-wave velocity is 0.7% and in S-wave velocity is 0.5% for this study. These
percentage errors were used in determining all P-and S-wave velocity
uncertainties in chapter 6 and 7.

4.5.2 Attenuation and Error Analysis
In our study, measurements were conducted only for one length of each
sample. In order to study waveform attenuation, a modified approach to the
conventional log spectral ratio method was developed. The conventional log
spectral ratio described in Chapter 3 is based on comparing two waveforms
obtained from two different sample lengths and all other factors are held constant
including the viscoelastic behaviour of the samples. Therefore the attenuation
between waveforms is investigated through the absolute attenuation coefficient
via differential sample lengths. However, absolute measurements of any kind are
always problematic and here a differential measure technique was developed. In
our study, when using only one sample length but subjecting the sample to
different conditions (saturation, pressure and temperature), various viscoelastic
behaviours will arise and hence different waveforms will result. The varying
factor is the viscoelasticity of the overall propagating medium while the sample
length is held constant. By using equation 3.87 to describe the amplitude
spectrum of a signal due to attenuation, the attenuation of two waveforms
corresponding to traveling in two different viscoelastic media but of the same
distance, x, can be related by their differential attenuation coefficient, [
2
()-
1
()],
97

| |
x
A
A
|
|
.
|
\
|
=
) (
) (
ln
) ( ) (
1
2
1 2
e
e
e o e o . (4.6)
Therefore the attenuation between waveforms in our study is investigated via
differential attenuation coefficient, and not the absolute attenuation coefficient.
In using only one sample length and analyzing attenuation from
determining the differential attenuation coefficient, multiple advantages arise over
the conventional log spectral ratio method. First, the geometric diffraction loss
does not need to be determined in equation 3.87 because the sample length is
invariant between the two signals and geometric effects are therefore eliminated
in equation 4.6. Second, carrying out the complete set of measurements listed in
section 4.4 is quite time intensive, therefore running the full set of measurements
just once, in our approach, will save at least half the time. Third, the integrity of
the sample over the course of the measurements will be altered due to saturation
and hysteresis effects, which will be discussed in chapter 7. In the conventional
log spectral ratio method, the same viscoelastic behaviour may not be repeatable
for the varying lengths and therefore can lead to inaccurate attenuation analysis.
Fourth, the signal strength through a sample is also affected by how the sample is
assembled in the experimental setup prior to measurements. Minor
inconsistencies during set up such as slight S-wave misalignments and the
connection in the soldered joints of the electrical wires will cause signal strength
variations among the different sample lengths that is not caused by viscoelastic
attenuation. Therefore by using only one sample length, these inconsistencies are
not a concern. Finally, any systematic errors that are inherent to determining the
absolute attenuation coefficient will be cancelled out when the difference is taken
in differential attenuation coefficient.
To analyze the attenuation of waveforms, all of the waveforms acquired were
first windowed and then the Fast Fourier Transform (FFT) was applied to get the
amplitude spectra of the windowed signals. Figure 4.7 shows an example of two
windowed P-wave time series signal recorded under dry conditions at room
temperature and experiencing a confining pressure of 5 MPa and 40 MPa. The
98

amplitude spectra of the windowed signals from FFT are shown in figure 4.8a.
Applying the natural logarithm to their spectral ratio and dividing by the length of
the sample yields the differential attenuation coefficient as a function of frequency
for the pair of signals (figure 4.8b). From the amplitude spectra of the signals, the
peak frequency of the signals have shifted from the nominal 1 MHz resonant
frequency of the piezoelectric ceramics and only a limited frequency band carries
most of the energy (0.4-1.4 MHz). This shift in peak frequency is likely attributed
to the construction of the transducers from the compilation and mounting of
multiple pieces with the piezoelectric ceramics, therefore shifting the energy from
higher frequencies. Outside of this high energy band of frequencies, the spectrum
is weak and is considered unreliable due to noise contamination. Consequently,
the differential attenuation coefficient outside of this range behaves erratically and
this information should be disregarded, or if it must be analyzed it should proceed
with great caution.
Since analyzing waveform attenuation was conducted from using only one
sample length, statistical error analysis was not possible for our set of
measurements. In addition, applying the propagation of errors for equation 4.6
was met with difficulty from the inability to accurately assess the error in
determining the amplitude spectrum of a signal. Therefore the exact errors of the
differential attenuation coefficients could not be estimated. However, as
mentioned in chapter 3, the errors in attenuation measurements are usually quite
large so interpreting just the differential attenuation coefficient values without any
guidance on the possible amount of error is inadequate. Since the experimental
method and setup applied in the acrylic attenuation measurements introduced in
chapter 3 is the same as the measurements described in this chapter, the errors
estimated statistically for the acrylic measurements would be an appropriate
estimate for our differential attenuation coefficients errors. The attenuation of
waveforms in the acrylic measurements was analyzed using the conventional log
spectral ratio method and the errors found were for the absolute attenuation
coefficients. The maximum errors found of the attenuation coefficient were 12 m
-
1
for P-waves and 18 m
-1
for S-waves. For the errors in the differential
99

attenuation coefficients, they can be taken directly from the errors determined for
the absolute attenuation coefficient and do not need to be propagated for the
subtraction in its definition. This is because the computations for these two
different quantities are actually exactly the same. To recap, the absolute
attenuation coefficient is determined from two different sample lengths of the
same viscoelastic medium by

x
x A
x A
A
|
|
.
|
\
|
=
) , (
) , (
ln
) (
1
2
e
e
e o ,
which has the same format to equation 4.6 for the differential attenuation
coefficient. Furthermore from the reasons stated earlier, the error associated with
absolute attenuation coefficient is greater than the errors in differential attenuation
coefficient. Therefore these approximated errors should be more than suffice to
account for the actual error in determining the differential attenuation coefficient.

Figure 4.7: The P-wave time series signals of the porous ceramic rod at dry conditions under
two confining pressures: a) at 5 MPa b) at 40 MPa. The red box is shows the windowed
signal used for the FFT.
100

Figure 4.8: (a) The amplitude spectra of the windowed-part of the two signals shown in
figure 4.7. (b) The resulting differential attenuation coefficient.

4.6 Chapter Summary
This chapter gave a detailed description of the experimental setup, the
procedures that were used to obtain waveforms and the methods that were applied
to analyze the data for the results reported in chapter 6. Ultrasonic pulse
transmission is the technique applied to acquire P-and S-waveforms for velocity
and attenuation analysis. This chapter provided the details behind building a set
of transducers for ultrasonic pulse transmission, the procedure used to prepare and
assemble the samples prior to measurement, a description of the experimental
apparatus, and a description of the different measurements conducted. Under a
variety of pressure and temperature conditions, measurements were carried out on
all samples for three saturation conditions: dry, CO
2
saturated, and water
saturated. Examples of the waveforms acquired during the measurements were
introduced, and the methods for analyzing them for signal velocity and the
attenuation were discussed. Signal velocity was determined from picking the time
of the first extremum and applying a delay time correction. The error in the signal
velocity was calculated from the propagation of errors and a maximum error of
0.7% and 0.5% were estimated for P-waves and S-waves, respectively. In order
101

to study waveform attenuation, a modified approach to the conventional log
spectral ratio method was applied, where the differential attenuation coefficient
between waveforms is analyzed instead of the absolute attenuation coefficient.
Analyzing waveform attenuation with this new approach provides many
advantages over the conventional log spectral ratio method, from reducing
laboratory time to minimizing measurement errors. The errors in differential
attenuation coefficients were not determined directly but were adopted from the
statistically estimated errors in the acrylic attenuation measurements from chapter
3. The adopted errors for the P- and S-wave differential attenuation coefficient
are 12 m
-1
and 18 m
-1
, respectively.

102

Chapter 5
Sample Characterization

In this chapter the porous media used in this thesis will be introduced and
characterized, mainly through petrophysical parameters. The petrophysical
parameters measured are density (grain and bulk), porosity, pore throat size, air
permeability, and tortuosity. Furthermore, images of the micro-structure of the
samples were examined. All sample characterizations were conducted at the
University of Alberta and their characterization methodologies will be reviewed.
The samples were dried under vacuum in an oven at 70C for more than 48 hours
before the various characterizations were performed. The main petrophysical
properties determined will be reported in table 5.2 at the end of the chapter.

5.1 Samples
In this thesis laboratory measurements were focused on two samples, a
synthetic sample and a natural rock. The synthetic sample is a porous ceramic rod
purchased from Hoskin Scientific Ltd. and was manufactured by Soilmoisture
Equipment Corporation. The porous ceramic rod is developed from a high fired,
alumina (aluminum oxide-Al
2
O
3
) body. It is consisted of open pore structure
thereby permitting fluid to move from one end of the sample to the other through
interconnected network of channels. The porous ceramic rod is white coloured,
appears to be very homogeneous, and composed of very fine grains. The pores or
the grains are not readily visible to the naked eye.
The natural sample is a porous sandstone from the Berea formation in
Ohio. This sample was obtained by material provided for a round-robin physical
testing campaign associated with the San Andreas Fault Observatory at Depth
103

(SAFOD) project. The Berea sandstone is predominantly composed of quartz
with minor traces of clay minerals for holding the quartz grains together. The
porosity and permeability of Berea sandstone is known to be relatively high, and
consequently it has been extensively used in petrophysical testing (Wyllie et al,
1958; Timur, 1968; Mobarek, 1971; De Vilbiss, 1980; Johnston and Toksoz,
1980; Nur and Murphy, 1981; Winkler and Nur, 1982; Winkler, 1985; Jones and
Nur, 1983, Shankland and Johnson, 1993; Hart and Wang, 1995; Zhan et al,
2010). The Berea sandstone appears as a light grey, homogeneous coloring with
no layering or noticeable features in macro-view, and is composed mainly of
sand-sized grains. Photographs of the two samples are shown in figure 5.1.

Figure 5.1: Photographs of a) Porous ceramic rod and b) Berea sandstone.

5.2 Density and Porosity
The grain volume, grain density, bulk density, porosity, and pore size of
the samples will be examined in this section. From helium porosimetry the grain
volume of the samples is determined, and consequently the grain density and the
porosity can be inferred. Bulk density is simply determined from the weighed
mass of the dry sample and measured dimensions of the sample. Mercury
porosimetry will also be discussed and it is used to describe the pore size
distribution.

104

5.2.1 Grain Volume and Density
Helium porosimetry is used to determine the grain volume of the samples.
The grain volume is the volume of the solid constituents of the porous sample (i.e.
rock matrix). The governing principal behind determining grain volume from
helium porosimetry is Boyless gas law. This ideal gas law states that under
conditions of fixed gas quantity and constant temperature, the product of the
pressure and volume stays constant and is expressed as

2 2 1 1
V P V P = (5.1)

where P
1
and V
1
are the original gas pressure and volume, respectively, and P
2

and V
2
are the new gas pressure and volume, respectively.
A helium pycnometer, Quantachrome Instruments MVP- D160-E, was
used to perform the porosimetry measurements. The machine essentially consists
of a sample cell compartment and a reference cell compartment, which are
connected by a valve. To commence the measurement, a sample is placed in the
sample cell compartment and the entire pycnometer system is vacuumed, flushed
through with helium gas, and vented to the atmosphere. Then the reference
compartment is closed off from the sample compartment and only the reference
cell is filled with helium. The resulting pressure from this compartment is logged
(P
1
). Next, the valve to the sample cell from the reference cell is switched open,
allowing the helium that was originally in the reference cell to expand to the new
available volume and the resulting new pressure is then logged (P
2
). A helium
gas of 99.995% purity was used for these porosimetry measurements.
Based on the above Boyles gas law, a more descriptive equation of the
different volumes involved in our measurement can be written and equation 5.1
becomes,
) (
2 1 g s R R
V V V P V P + = . (5.2)
V
R
is the volume of the reference cell, V
s
is the volume of the sample cell, and V
g

is the grain volume. The volumes of the reference and sample cells are
105

predetermined; the sample cell compartment used in our measurement has a
volume of 147.903 cm
3
, while the reference cell compartment has a volume of
88.52 cm
3
. From this known information and with the original and new pressures
recorded, the grain volumes of our samples were calculated from equation 5.2.
From the determined grain volume and knowing the mass of the dry sample, m,
the grain density can be consequently calculated,
g
g
V
m
= . (5.3)
The dry mass of a sample was found from weighing the sample prior to helium
porosimetry measurements and is reported in table 5.2.

5.2.2 Bulk Density and Porosity
The bulk density of a sample can be determined relatively easily and does
not require the use of the helium pycnometer. The bulk density of the sample is
simply the ratio of the samples dry mass to the samples envelope volume, V
e,

e
b
V
m
= . (5.4)
The envelope volume of the sample is found from measuring the length and
diameter of the sample using a Vernier caliper and is reported in table 5.1.
However, this seemingly simple measurement is often not that accurate due to
even relatively small variations in sample shape.
To determine the porosity of a sample, the information on grain volume
found from helium porosimetry is used. Porosity is usually the most important
characteristic used to describe a porous medium. The porosity of a sample is
defined as the ratio of the total void volume of a porous medium to the envelope
volume of the sample, or the ratio left over if the ratio of the grain volume to
envelope volume is subtracted from unity,
e
g
V
V
=1 | . (5.5)

106

Sample Diameter
(cm)
Length
(cm)
Envelope Volume
(cm
3
)
Porous Ceramic Rod 2.54 5.061 25.64
Berea Sandstone 2.54 4.345 22.02
Table 5.1: The measured envelope volume of the samples.

5.2.3 Pore Size Distribution
To determine the distribution of pore size in our samples, mercury injection
porosimetry is used. This method operates by injecting liquid mercury into the
pore space of the porous medium. Liquid mercury is a non-wetting fluid, with a
contact angle, , of about 140. Unlike wetting fluids, without external pressure
liquid mercury would not enter the pore space of a porous medium. The smaller
the pore size, the larger the pressure is required to force the liquid mercury in.
Therefore by monitoring the amount of mercury that is injected into the sample as
the applied pressure increases, the pore sizes of the sample can be inferred. The
relationship relating pressure and pore size from the penetration of mercury is
given by the Washburn equation,
P
D
pore
u cos 4
= (5.6)
where D
pore
is the pore throat diameter, P is the exerted pressure, is the surface
tension of mercury which is 0.485 Nm
-1
at 25C, and is the contact angle of
mercury. The maximum applied pressure on the mercury in our measurements is
413 MPa, therefore the smallest pore that can be investigated is about 3.5 nm. It
should be noted that mercury porosimetry measurements measures the largest
entrance into a pore, and not the actual inner size of a pore. So the determination
of the pore size distribution is more accurately the pore throat size distribution, if
the access to the larger pore is through a smaller pore. Also, closed pores cannot
be analyzed by this method because mercury has no way of entering the pore
space.
To conduct these measurements, the mercury porosimeter, Autopore IV
9500 from Micromeritics, was used. A small piece of sample is placed into the
107

sample cell of the penetrometer which has a long hollow stem connected to the
sample cell. The long hollow stem is coated with a metal sheet. Initially, the
penetrometer with the sample is placed into the low pressure cycle part of the
porosimeter. The penetrometer is evacuated to remove air and then it is filled
with mercury under low pressure. In this low pressure cycle, mercury will
penetrate the largest pores and fill the spaces between the pieces of the sample, if
more than one piece is used. At the completion of the low pressure cycle, the
penetrometer is removed and placed into the high pressure cycle part of the
porosimeter where the actual intrusion measurements begin. In this cycle as the
exerted pressure increases, the total amount of pore space intruded with mercury
also increases while the size of the pore spaces being progressively invaded
decreases. The amount of mercury used up from intrusion into the pore space of
the sample is monitored throughout pressurization by the capacitances changes
between the mercury in the stem and the metal coating on the stem of the
penetrometer. Therefore the cumulative volume of mercury injected into the
sample is recorded as a function of pressure. Incremental intrusion as a function
of pressure can also be determined from the recorded cumulative volume data by
taking the derivative,
P
dP
dV
V
Hg
Hg
A = c . (5.7)
It basically represents the change in volume of mercury that has intruded into the
pore space from the last pressure to the new pressure. This quantity reveals the
relative distribution of the samples porosity among the pore sizes, and at which
pressure the contribution of mercury is the greatest. Consequently, a peak in the
incremental intrusion curve would coincide with the steepest increase in the
cumulative intrusion curve. Figure 5.2a and 5.3a shows the cumulative and
incremental intrusion with pressure for the porous ceramic rod and Berea
sandstone respectively. Converting pressure to pore size via Washburns
equation, Figure 5.2b and 5.3b shows the incremental intrusion with pore size,
instead. Any large peaks in incremental intrusion indicate a dominant pore size.
For both samples, the incremental curve exhibits a modal behaviour indicating
108

that majority of the pores is of one size. The modal pore size of porous ceramic
rod is 2 m while for the Berea is 11 m. Contrasting the morphology of the two
Figure 5.2 Cumulative and incremental intrusion curves for the porous ceramic rod a) with
pressure b) with pore size
109

Figure 5.3 Cumulative and incremental intrusion curves for the Berea sandstone a) with
pressure b) with pore size

110

incremental intrusion curves, the porous ceramic rod exhibits a smoother and
cleaner curve compared to the Berea sandstone, especially at pores smaller than
the dominant pore size. This signifies that the Berea sandstone pore size
distribution has more variability than the porous ceramic rod. In particular, the
Bereas pore size distribution extends to quite small dimensions; and this is likely
indicative of the small aspect ratio (i.e. microcrack) porosity that leads to highly
nonlinear elastic behaviour.

5.3 Permeability
Another important characteristic used often to assess the nature of a
porous medium is permeability. This petrophysical property describes how easily
fluid can flow through the voids of the medium, where open and interconnected
pores make a medium more permeable. Permeability, k, is physically defined by
relating the fluid volume flow per second and per area (Darcy velocity), q, of fluid
flow to pressure gradient, dP/dx, experienced by the fluid and the viscosity of the
fluid, ,

dx
dP k
q
q
= . (5.8)
To measure the permeability of our samples a portable air permeameter,
Tiny Perm II made by New England Research Inc. and courtesy of Professor
David Potter at the University of Alberta, was used. This measuring system is
consisted of a plunger connected to a vacuum cylinder with an attached pressure
transducer, and this transducer is connected to a microprocessor and control unit
by an electrical cable. Figure 5.4 shows a photograph of the Tiny Perm II used.
Prior to measurement with the plunger withdrawn and no vacuum is indicated on
the control unit, the rubber nozzle on the tip of the vacuum cylinder is pressed up
against the sample. The measurement begins by depressing the plunger and the
control unit shows a reading on the current vacuum in the cylinder and the
measurement status. When the plunger is completely depressed and the reading
indicates that there is no vacuum, a response function result will appear. The
manufacturer indicates that the micro-processor unit computes the response
111

function from monitoring the syringe volume and transient vacuum pulse created
as the plunger is depressed. The permeability of the sample in millidarcies is
related to the response function, T, by the following calibration relationship,
8737 . 12 ) ( log 8206 . 0
10
+ = k T . (5.9)

Figure 5.4: Tiny Perm II, the portable air permeameter system used in measuring the air
permeability of our samples. The two arrows are pointing out the connection between the
pressure transducer in the black box to the yellow microprocessor and control unit.
Photograph from NER Tiny Perm II users manual.

Several readings were taken for each sample, and the average permeability value
was taken and is shown in table 5.2. The permeability values fluctuated within
5% of the average permeability value for both samples.
Strictly speaking the permeability value determined this way of a sample
is its air permeability, and it is generally known to be higher than its liquid
permeability (Klinkenberg, 1941; Kilmer et al, 1987; Springer, 1998; Wu, 1998),
especially in low-permeable rocks. This difference in value of gas permeability to
liquid permeability for the same medium is caused by gas slippage at the walls of
the pores, known as the Klinkenberg effect. Liquid flow and gas flow at high
pressure is laminar, following Darcys Law, with a zero flow velocity at the pore
112

walls. Gas flow at low pressures on the other hand moves in a non-Darcy fashion
where the flow at the pore walls is not zero. Klinkenberg (1941) developed an
equation to correct for this slippage effect so that gas permeability can be
converted to its equivalent liquid permeability if the gas pressures are known
during the permeability measurements. No correction for the Klinkenberg effect
was carried out on our air permeability values because the permeameter unit used
in our study, strangely, does not provide this required information. However, in
the manual of the equipment, the permeability values determined from the
calibration curves are referred to as absolute permeability. The term absolute
permeability is a bit ambiguous as it can be referred to the permeability measured
when there is only one type of fluid present in the pore space, or it can be
synonymous to liquid permeability. Nonetheless, if the Klinkenberg effect
correction is not applied by the air permeameter, this effect on our samples would
only cause a slight difference between the two permeability values. As mentioned
earlier this effect is only significant for a low-permeable sample and as a result, a
large majority of its correction in scientific studies are performed on tight samples
(<0.0001 mD). Klinkenberg (1941) states for samples of moderately high
permeability, the discrepancies gas and liquid permeability are not of first
importance for practical purposes. He showed the differences in permeabilities
are about ~10% for samples with liquid permeabilities greater than 100 mD.

5.4 Tortuosity
Tortuosity in a porous medium is a measure that describes how twisted,
crooked, and convoluted the shortest passage is between two points taken through
only the voids in the medium. It is defined as the ratio of distances of the actual
path taken, L
C
, through the network of channels to that of the straight path, L
0
,
between the same starting and ending points,
1
0
> =
L
L
C
t . (5.10)
This parameter of a porous medium impacts how easily a fluid can move within
the void space; the less tortuous the path, the less resistant in the movement of the
113

fluid. The tortuosity of a sample in our study was measured using an electrical
method with an experimental set up similar to that described by Garrouch et al
(2001). Essentially, by comparing the resistance of a sample saturated with a
saline solution to the resistance of only the saline solution over the same distance
and knowing what the porosity of our sample is, the tortuosity of the sample can
be inferred. This method can only be applied if the matrix of the sample is
electrically non-conductive. The principal behind this method comes from the
relationship between electrical resistance, R, and resistivity, :
A
l
R

= (5.11)
where l is the distance of the path traveled by a current in a conductive medium,
and A is the surface area of the conductive medium subjected to an applied
current.
For the measurements, the sample was vacuumed and then saturated with
a saline solution of 15,000 ppm. The four electrode technique was adopted to
avoid contact resistance from developing between the sample and the electrodes.
The four electrodes consist of two current electrodes and two voltage probes. The
sample is placed between the two current electrodes that covers the parallel end
surfaces exactly and the two needle voltage probes are placed on the surface of the
sample. A 60 Hz AC source was connected to the current electrodes and an
ammeter was used to determine the current following through this circuit. From
measuring the voltage drop across the two probes and knowing the current in the
circuit, the resistance of the saturated sample, R
sample
, is determined by Ohms
law. The same electrical measurement was repeated with only saline solution and
the resulting resistance, R
saline
, was found. Figure 5.5 depicts the electric circuit
and electrical current path taken in each measurement. Care was taken to ensure
the distances between the four electrodes were the same for both measurements.
From these measured resistances and using equation 5.11 to describe each
measurement scenario, the tortuosity of a sample is simply
| t
saline
sample
R
R
= . (5.12)
114

Due to the need of this parameter for the analysis of the porous ceramic rods
ultrasonic pulse transmission measurement results, tortuosity was only measured
of the porous ceramic rod.

Figure 5.5: Schematic diagram of the four electrode circuit. The red line denotes the path
taken by the electrical current. a) Electrical measurement with the sample saturated with
saline solution, b) electrical measurement with only saline solution.

5.5 Scanning Electron Microscope Imaging (SEM)
A scanning electron microscope (SEM) was used to obtain images of our
samples to examine their grain structures and micro-features qualitatively. SEM
is a microscope that can produce highly magnified and resolved microscopic
images of the surface of an object through scanning the object with a beam of
electrons. The Zeiss EVO MA 15 SEM used in imaging our samples had a
magnification up to 100 000 X. During an SEM imaging, a beam of electrons is
emitted from an electron gun fitted with a filament and is accelerated by attractive
forces towards an anode. This electron beam is focused on a small area of the
sample by a magnetic field. When this beam interacts with the surface of the
sample, secondary electrons are created from ionization and scatters. These
secondary electrons are recognized and accounted by a detector, and an image is
produced from the number of electrons scattered from each spot on the samples
surface. The electron beam is moved across the sample by a set of scanning coils.
Prior to imaging, the sample specimen is prepared by coating the surface with
gold powder. This gold powder is to provide conductivity to the surface of the
115

sample to prevent electrons from being trapped, thereby creating an overly bright
image. SEM imaging was carried out at the Earth and Atmospheric Science
department at the University of Alberta.
Figure 5.6 and figure 5.7 are the SEM images for the porous ceramic rod
and Berea sandstone, respectively, for two different magnifications. The 5 kX
magnification image of the porous ceramic rod shows that the overall sample is
fairly uniform. At a magnification of 20 kX, the image reveals the aluminum
oxide grains and some secondary material attached to them. This secondary
material is believed to be a binder material used in the making of the sample
during firing. For the Berea sandstone SEM image at 500 X magnification, the
image shows a generally uniform rock sample. The large grains are quartz and
they are surrounded by another mineral. At 3.5 kX magnification, the surface of
the quartz grains reveals evidence of dissolution and sheet-like minerals. These
sheet-like minerals have been identified as kaolinite, a type of clay.
From these SEM images of the two samples, the grains and pore throats of
the porous ceramic rod are evidently smaller than that of the Berea sandstone.
Also despite the irregular shape of the alumina grains, the pores of the porous
ceramic rod appear more open, in contrast to the large amounts of crack-like pores
of the Berea sandstone. These images are in qualitative agreement with the
mercury porosimetry curves of Figs. 5.3 above.

Sample Porous ceramic Rod Berea Sandstone
Mass (g) 39.6 43.24
Bulk Volume (cm
3
) 25.33 22.02
Bulk Density (g/cm
3
) 1.55 2.15
Grain Volume (cm
3
) 10.63 17.79
Grain Density (g/cm
3
) 3.70 2.64
Porosity (%) 58.4 19.0
Modal Pore Size (m) 2.06 11.33
Air Permeability (mD) 96.94 237.65
Tortuosity (dimensionless) 1.73 3.9
a

Table 5.2: Petrophysical properties characterized of the samples.
a
tortuosity value taken
from Garrouch et al (2001)
116

Alumina
Binder
material
b)
a)
Figure 5.6: SEM images of the porous ceramic rod at two magnifications a) 5 kX b) 20 kX.
117

Kaolinite
Quartz
b)
a)
Figure 5.7: SEM images of the Berea sandstone at two magnifications a) 500 X b) 10 kX.

118

5.6 Summary
The synthetic sample, porous ceramic rod, and the non-synthetic sample,
Berea sandstone, used in this thesis were petrophysically characterized using a
variety of methods. Helium porosimetry was used to determine the bulk grain
volume, grain density and porosity of the two samples. Both samples have high
porosity with the porous ceramic rod being extremely porous. Mercury
porosimetry was used to reveal the pore size distribution in the samples and both
samples exhibits a modal pore distribution with a dominant pore diameter on the
order of microns for both samples. From the variability of the mercury intrusion
curves, the pore structure of the porous ceramic rod displays more uniformity than
the Berea sandstone. Air permeability of the samples was determined using a
portable air permeameter. Just like the dominant pore size, the permeability of the
Berea sandstone is higher than the porous ceramic rod. The tortuosity was only
measured of the porous ceramic rod and was done by using an electrical method.
Finally, SEM images were examined for both samples at various magnifications.
The Berea sandstone showed a lot of compliant pores where as the porous ceramic
rod displayed only non-compliant pores.

119

Chapter 6
Results and Discussion: Sintered
Alumina

In this chapter the laboratory results, the modeling results, and a
discussion on the wave behaviours of the porous ceramic rod with the
petrophysical characteristics described in table 5.2 and that was subjected to the
pressure, temperature, saturation conditions outlined in table 4.1 are given. The
P-and S- waveforms and their determined wave velocities will be provided for all
three saturation states of dry, CO
2
saturated, and water saturated. Their
comparison to modelled velocities will be conducted for only the CO
2
saturated
and water saturated states. The laboratory results and the modelled predictions of
the differential attenuation coefficient will be given for only the CO
2
saturated
measurements.
All waveforms in this study were acquired using the ultrasonic pulse
transmission method introduced in chapter 3. The methods in determining the
wave velocity, the wave attenuation, and their associated errors from the acquired
waveforms were discussed in chapter 4. Gassmanns and Biots fluid substitution
models, for predicting wave velocities of saturated porous medium, were
introduced in chapter 3.

6.1 Dry
P-and S-waveforms were obtained of the porous ceramic rod under dry
(i.e. vacuum in the pore space) conditions at room temperature (T = 23C) while
120

confining pressure varied between 5 to 40 MPa, undergoing through both
pressurization and depressurization cycles.

6.1.1 Dry Waveforms
The full set of normalized P-and S-waveforms are shown in figure 6.1.
The signals are normalized with respect to the largest amplitude within the given
set of waveforms. In the figure, the waveforms are displayed conventionally by
wiggle traces and also alternatively by coloured amplitudes to highlight patterns
and subtle features. All waveform plots hereafter will be only be displayed as
normalized coloured amplitudes. In the waveform plot, the arrival for both the P-

a) b)
c) d)
Figure 6.1: Normalized waveforms obtained under dry conditions at room temperature
plotted as a function of confining pressure for the porous ceramic rod. a) and b) shows the P-
and S-waves as wiggle traces. c) and d) shows the P-and S-waves as coloured amplitudes.
121

and S- waveforms for the dry porous ceramic rod do not vary much over this
pressure interval. However, minor signal strength changes can be seen.

6.1.2 Dry Wave Velocities
The P-and S-wave velocities with their uncertainties estimated from the
waveforms (method discussed in section 4.5.1) shown in figure 6.1 are plotted in
figure 6.2. As applied confining pressure increases on the sample, the wave
velocities increase. However over the entire 40 MPa pressure range investigated,
the amount of change in wave velocities is quite insignificant, less than 1% for
both P-and S-waves. Furthermore, at a given confining pressure the wave
velocities during pressurization cycle (upward-pointing triangle) are always lower
than the wave velocities during the depressurization cycle (downward-pointing
triangle). This behavioural hysteresis can be attributed to the closing of compliant
pores during pressurization that are not reopened at the same pressure during
depressurization due to frictional forces. However, the differences between the
wave velocities of the pressurization and depressurization cycles are minor where
the velocities are well within error.
Overall the velocity changes and hysteresis observed here with pressure
are insignificant and can be considered as negligible, in comparison to that of real
rocks (Gardner et al, 1965). The vanishingly small changes observed are due to
the lack of low-aspect ratio pores, or alternatively micro-cracks, in the sample.
Low-aspect ratio pores are compliant meaning they are inclined to collapse under
pressure. As pores close up in a porous sample, the overall sample will become
stiffer. If the presence of micro-cracks is large, this will translate to large
increases in wave velocity with pressure due to the increased sample stiffness (as
will be seen with the Berea sandstone in chapter 7). Therefore from the wave
velocity behaviour observed with pressure, the porous ceramic rod studied here
contains mainly of high-aspect ratio, non-compliant pores. This inference is in
agreement with the non-cracked nature of the sample that was first suggested
from reviewing SEM images in chapter 5 and from the mercury injection
porosimetry.
122

a) b)
Figure 6.2: Dry P-wave (a) and S-wave (b) velocities with their respective errors as a
function of confining pressure for the porous ceramic rod.

6.2 CO
2
The measurements made under a CO
2
saturated state can essentially be
classified as either a constant temperature or a constant pressure run. To recapture
the measurement conditions applied, figure 6.3 displays the different temperatures
and pore pressures that were explored, superimposed on the bulk modulus and
density phase diagram of CO
2
. In the figure the vertical arrows are of constant
temperature measurement runs, and the horizontal arrows are of the constant
pressure (confining and pore pressure) measurement runs. Each measurement run
conducted is expected to cross a phase boundary. During all CO
2
saturated
measurements, a constant differential pressure of 15 MPa was maintained.
Because a constant differential pressure is maintained, any waveform variations
observed should solely caused by pore fluid effects. For all the constant
temperature runs, the pore pressure varied from 2 MPa to 25 MPa. CO
2
is a gas at
low pore pressures and is a liquid or a supercritical fluid at higher pore pressures
depending on the temperature. For all the constant pressure (pore, confining, or
differential) runs, the measurements began at a temperature of 23C where CO
2
is
a liquid state. As temperature increased CO
2
eventually transforms into a gaseous
or a supercritical fluid state, depending on the pore pressure shown in figure 6.3.

123

a) b)
Figure 6.3: Phase diagrams of CO
2
s: (a) bulk modulus (b) and density. The temperature
and pressure conditions applied to the CO
2
saturated porous ceramic rod during ultrasonic
pulse transmission measurements are shown by the white arrows.

6.2.1 CO
2
Saturated Waveforms
Figure 6.4 shows the P-and S- waveforms as a function of pore pressure
for the constant temperature measurements at T = 28C and at T = 40C. The
major trends and features in waveform variations and wave attenuation with pore
pressure for the T= 23C and T = 28C runs are quite similar, as are the T = 45C
and T = 40C runs; therefore not all are shown to avoid redundancy. However,
the laboratory results of the T = 23C and T = 45C measurements are available in
the appendix. In all the P-and S-wave waveform plots, within a given CO
2
phase
state the observed variations of the waveform such as amplitude and arrival time
are minor. Between CO
2
phase states the observed differences in waveforms are
significant with the phase transitioning of the CO
2
being quite obvious. Both P-
and S-waves arrive earlier and more strongly in the gaseous phase (lower pore
pressures) than in the liquid or supercritical phase (higher pore pressures). The
large changes to the waveforms are more immediate for the T = 28C run where
as the changes to the waveforms for the T = 40C run are less abrupt. Also the
phase transition for the T = 40C run appears to take place at a higher pore
pressure, past 7 MPa, than the T = 28 C run, under 7 MPa.
124

a) b)
c) d)
Figure 6.4: Normalized P- and S-wave waveforms as a function of pore pressures collected
during the constant temperature runs of a) and b) T = 28C and, c) and d) T = 40C. A
constant differential pressure of 15 MPa was maintained throughout each measurement run.

Figure 6.5 shows the P-and S-waveforms as a function of temperature at
a constant pore pressure (P
P
) of 7 MPa, 10 MPa, and 25 MPa while the
differential pressure was held constant at 15 MPa. The P
P
= 7 MPa P-and S-
waveform plot shows large changes with the signals arriving earlier and with
greater amplitude in the gaseous phase (higher temperature) than in the liquid
phase (lower temperature). In contrast, for the P
P
= 10 MPa and P
P
= 25 MPa
waveform plots, there are no substantial changes to the waveforms even over the
anticipated liquid to supercritical fluid phase transition. However for the P
P
= 10
MPa run, the arrivals of the P-and S-waveforms do appear to arrive slightly earlier
and slightly higher amplitude at the higher temperatures. For the P
P
= 25 MPa
run, the P-waves display little noticeable change over the range of temperature
125

change while the S-waves show slightly earlier arrivals at higher temperatures.
The S-waves of the P
P
=25 MPa, figure 6.5f, also appear to be highly attenuated as
is demonstrated by the noise overprint.

6.2.2 Observed CO
2
Saturated Wave Velocity
The P- and S-wave velocities with their estimated errors for the four
constant temperature runs are shown in figure 6.6. The wave velocities of the two
runs whose waveforms were not shown in the previous section are also shown
here on the same velocity plot for the purpose of reinforcing the observed trends.
Also, the dry wave velocities obtained under a confining pressure of 15 MPa are
plotted at P
P
= 0 MPa to provide a reference for the changes seen. With CO
2

saturating the pore space, irrespective of the phase state, both P-and S-wave
velocities are lower than the dry wave velocities. The CO
2
phase transition out of
the gaseous state is marked by a significant drop in wave velocity, with the drop
being smaller and less immediate for the two higher temperature runs. Comparing
velocities right before and right after the large drop, namely where the velocities
change becomes minor, the change is about 4.5% and 5.2% for the P-and S-waves
of T = 28C respectively, and the change is about 4.2% and 4.8% for the P-and S-
waves of T = 40C respectively. Over the entire 2 to 25 MPa pore pressure
interval, the overall T = 28C velocity change is 6.7% and 8.1%, for P-and S-
wave, respectively. The overall T = 40C velocity change is 6.1% and 7.1% for
P-and S-wave, respectively. Furthermore, the large drop in wave velocities for
the two higher temperature runs occur at pore pressures that are higher than the
pore pressures of the two lower temperature runs. The large drop in wave
velocities for the two higher temperature runs occur at pressures close to the
CO
2
s critical pressure of 7.4 MPa. While the large drop in wave velocities for
the two lower temperature runs occur at pore pressures that approximate CO
2
s
temperature dependent-vapour pressure. For both cases, the exact agreement to
the vapour pressure or to the critical pressure cannot be determined due to the
coarse pore pressure increments taken of the laboratory measurements.
Within a given phase state (gas, liquid, or supercritical) as pore pressure
126

a) b)
c) d)
e) f)

Figure 6.5: Normalized P- and S-wave waveforms as a function of temperature collected
during the constant pore pressure runs of a) and b) P
P
=7 MPa, c) and d) P
P
=10 MPa and e)
and f) P
P
= 25 MPa. A constant differential pressure of 15 MPa was maintained throughout
each measurement run.

increases the elastic wave velocities decrease, even though the differential
pressure is constant for all measurements. When all four constant temperature
runs are compared at a high pore pressure, the velocities of the higher temperature
127

runs, which are in the supercritical fluid phase, are generally greater than the
lower temperature runs, which are in the liquid phase. However the actual
velocity differences between the liquid and supercritical fluid phase are small and
are within error, in contrast to large velocity change associated with the gas to
liquid or gas to supercritical phase transition.

a) b)
Figure 6.6: Ultrasonic P-wave (a) and S-wave (b) velocities with their respective errors
measured of the porous ceramic rod saturated with CO
2
for all constant temperature runs
and when the sample is not saturated (black dot at Pp = 0 MPa).

The P-and S-wave velocities with their errors for the three constant pore
pressure runs are shown along with the dry wave velocities obtained under a
confining pressure of 15 MPa plotted at T = 23C, in figure 6.7. Similarly to
above, with CO
2
in the pore space the wave velocities decrease with respect to the
dry wave velocities. The CO
2
phase transition from liquid to gas is marked by a
significant increase in wave velocity of about 3.8% and 4.3% for both P-and S-
waves, respectively. The overall increase in P-and S-wave velocities over the
entire temperature interval is 4.1% and 4.8%, respectively. The phase transition
from liquid to supercritical fluid does not exhibit much variation on wave velocity
but does show a slight overall increase in velocity with higher temperatures. The
P
P
= 10 MPa run exhibits a 0.4% and 0.9% overall increase in P-wave and S-wave
velocity, respectively. The P
P
= 25 MPa run exhibits a 0.1% and 0.4% overall
increase in P-wave and S-wave velocity, respectively. When all three constant
128

pore pressure runs are compared at a given temperature, the velocities of a low
pore pressure run are generally higher than a high pore pressure run.

a) b)
Figure 6.7 Ultrasonic P- wave (a) and S-wave (b) velocities with their respective errors
measured of the sample saturated with CO
2
for all constant pore pressure runs and when the
sample is not saturated (black dot at Pp = 0 MPa).

The wave velocity change observed over the vapour-liquid boundary for
the constant pore pressure run is less immediate than that observed for the
constant temperature runs. Also, this liquid to gas phase transition occurs at a
delayed temperature in respect to the vapour temperature expected from CO
2
s
phase diagram. The large P-and S-wave velocity variations occurred after 33C in
our data, while CO
2
s vapour temperature is 29C at a fluid pressure of 7 MPa.
This temperature delay maybe caused by the lack of time provide at each new
temperature for the pore fluid to equilibrate during data acquisition.
Alternatively, it may be related to the influence of the finite pore size on the phase
transition, occurring in a porous medium due to energy cost associated with
surface energy effects. It should be noted that during constant pore pressure runs,
the measurements were taken as the vessel was continually heated. The thermal
inertia of the system makes changing the temperature a slow and difficult to
control process.

129

6.2.3 Modelled CO
2
In chapter 3, two fluid substitution models used for predicting wave
velocities of a saturated porous medium were introduced. Gassmanns and Biots
theoretical predictions are computed here over the same pressure and temperature
conditions as applied in the laboratory. In order to compute both models a series
of sample and fluid properties need to be known. The bulk modulus, the density,
the fluid viscosity of the pore fluid (CO
2
) were determined from the
thermodynamic properties obtained from NISTs online chemistry webBook, the
density and the bulk modulus of the mineral grains were determined from Bass
(1995), the bulk modulus of the frame was determined from ultrasonic pulse
transmission measurements on the dry sample, and the bulk density, the porosity,
and the tortuosity of the sample were determined in chapter 5. The bulk modulus
of the frame was determined as a function of temperature to account for the
softening of the sample frame with heat. Incorporating the temperature
dependency of the frame will allow for the modelled results to be more accurate.
The modelled results from their predictions are compared to the laboratory
observed results.

Gassmanns Model
Here, the modelled P-and S-wave velocities of the CO
2
saturated porous
ceramic rod are presented by applying the widely employed Gassmanns fluid
substitution formulation given by equation 3.29 and 3.31. Figure 6.8 shows the
modelled P-and S-wave velocities along with the observed velocities under CO
2

saturation for two of the four constant temperature runs, T = 28C and T = 40C.
The results and comparisons for all four constant temperature runs can be found in
the appendix. For both P-and S-waves, Gassmanns modelled velocities show
similar behaviours with varying pore pressure, specifically: the large drop in wave
velocities; where the wave velocities begin to change significantly; and the
immediate or gradual nature of the velocity change over the phase transition. In
terms of the absolute velocity values, the modelled velocities are almost in
agreement with the observed velocities over low pore pressures. However, as
130

pore pressure increases the discrepancy between the modelled and observed
velocities enlarges. The largest discrepancy is 6.6% and 6.0% for P-and S-waves,
respectively. This discrepancy is larger than the drop in wave velocities over the
phase transition of the observed laboratory data. Essentially, Gassmanns model
predicts a larger change in wave velocities than that is observed. Over the phase
transition interval for P-and S-waves, respectively, Gassmann predicts a 7.5% and
7.6% change for T = 28C, and a 7.4% and 7.5% change for T = 40C. For the T
= 28C run the overall change in wave velocity over the entire interval is 12.0%
and 13.2% for P-and S-wave velocity, respectively. For the T = 40C run the
overall change in wave velocity over the entire interval is 11.7% and 12.7% for
P-and S-wave velocity, respectively. Generally for both P-and S-waves, the
modelled velocities are always lower than the observed velocities.

a) b)
Figure 6.8: Observed and Gassmanns modelled P- wave (a) and S-wave (b) CO
2
saturated
velocities for T = 28C and T = 40C constant temperature runs. To avoid clustering, only
the error at a pore pressure of 25 MPa are shown. The errors at all other data points are
approximately the same size.

Figure 6.9 shows the modelled P-and S-wave velocities with the observed
velocities under CO
2
saturation for the three constant pore pressure runs. Starting
with the P
P
= 7 MPa results, Gassmanns modelled velocities do show a large
change over the vapour-liquid boundary; a 6.1% and a 6.2% increase in P-and S-
wave velocities, respectively. However for the modelled velocities, the large
change in wave velocities in response to the phase transition appears to be more
131

immediate and it occurs at an earlier temperature than the observed velocities.
Also there is a large discrepancy between the absolute velocity values where the
discrepancy is largest with low temperatures and the discrepancy decreases with
higher temperature. Unlike in the constant temperature runs, the modelled and
observed velocities here are nowhere near agreement for any data point. Over the
entire interval, Gassmann predicts an overall P-wave increase of 8.8% and an
overall S-wave increase of 9.0%.
For the P
P
= 10 MPa and P
P
= 25 MPa runs in terms of behavioural trends,
Gassmanns modelled results also show that there is no significant wave velocity
change over the liquid to supercritical fluid phase transition. For the P
P
=10 MPa
run, the modelled velocities show a smooth increase with temperature where the
increase is more than the observed velocities are showing with temperature. A
2.0% and 2.4% increase in wave velocity is modelled for the P-and S-wave,
respectively. For the P
P
= 25 MPa run, the modelled S-waves velocities also show
a smooth increase with temperature but the increase is slighter; this is consistent
with the observed S-wave velocity increase. The modelled P-wave velocities
however appear to be invariant with temperature where this behaviour is
somewhat consistent with the observed P-wave velocities. A 0.0% and 0.5%
increase in wave velocity is modelled for the P-and S-wave, respectively. In
terms of absolute velocity values, there is a large discrepancy between the
modelled and observed for all the temperatures explored. The discrepancy for the
P
P
= 10 MPa run does appear to decrease with higher temperatures, where as the
discrepancy for the P
P
= 25 MPa run does not show any perceivable change with
temperature. For all constant pore pressure runs of both P-and S-waves the
largest discrepancy in velocity values between modelled and observed is 6.3%,
and the modelled velocities are always less than the observed velocities.
Overall Gassmanns predictions do not adequately describe the wave
behaviour observed in the laboratory. Gassmanns formulation always under-
predicts the wave velocities and it does not account for the wave attenuation
observed. This is a direct consequence of applying Gassmanns formulation to
the inappropriate frequency regime; Gassmanns equation is only valid for
132

describing wave behaviour pertaining to low frequencies and strictly only at zero-
frequency (static deformation). High wave frequency effects that can give rise to
differential movements between the pore fluid and pore wall leading to frequency-
dependent wave attenuation that Gassmanns zero-frequency limit was never
intended to account for.

a) b)
Figure 6.9: Observed and Gassmanns modelled P-wave (a) and S-wave (b) CO
2
saturated
velocities for P
P
=7 MPa, P
P
=10MPa, and P
P
=25 MPa constant pore pressure runs. To avoid
clustering, only the error at the highest temperature in each run is shown. The errors at all
other data points are approximately the same size.

Biot Model
The P-and S-wave velocities of the CO
2
saturated porous ceramic rod are
modelled by using Biots formulation given in section 3.3.2. As a reminder, there
are two advantages in using Biots formulation over Gassmanns formulation:
1.) Biots model is valid for a wide range of frequencies; for low frequencies,
where Gassmann is valid, and for the high frequencies, which are
encountered in our laboratory data.
2.) Considers wave attenuation.
Figure 6.10 shows the modelled P-and S-wave velocities along with the
observed velocities under CO
2
saturation for the same two of the four constant
temperature runs shown before for Gassmanns model. The results and
comparisons for all four constant temperature runs again can be found in the
appendix. For both the P-and S-waves the modelled velocities are generally
133

within the error of the observed velocities and for where there is a disagreement
between results, the discrepancy is usually small (< 1%). Of all the velocity
comparisons the S-wave velocities of T = 28C show a disagreement between
results the most frequent with them all occurring in the higher pore pressures.
Furthermore as a result of the predominantly good agreement between Biots
modelled and observed results, Biots modelled velocities show the same
magnitude in the drop of wave velocities with the phase transition, the same pore
pressures of where the wave velocities begin to change significantly, and the same
immediate or gradual nature of the velocity change over the phase transition.

a) b)
Figure 6.10: Observed and Biots modelled P- wave (a) and S-wave (b) CO
2
saturated
velocities for T = 28C and T = 40C constant temperature runs. No error bars are shown,
however the size of errors can be referred to in figure 6.6 and 6.8.

Figure 6.11 shows the modelled P-and S-wave velocities with the
observed velocities under CO
2
saturation for the three constant pore pressure runs.
Starting with the P-wave velocities, for both the higher pore pressure runs the
agreement between the modelled and observed is good for all temperatures
explored. For the P
P
= 7 MPa run, the results are only in agreement at
temperatures well away from the phase transition. Near the phase transition
temperatures, the discrepancy is quite large (4.2%) where the modelled response
to the phase transition is again more immediate than observed response. For the
S-wave velocities, the general agreement is less than that observed with P-wave
velocities for all three runs. Where there is a disagreement in velocity values, for
134

the two higher pore pressure runs the discrepancies are not substantial (< 1%).
For the P
P
= 7 MPa run the manner of the S-wave discrepancies are similar to that
seen with P-waves; insignificant away from the phase transition but is quite large
over the phase transition (4.8%).

a) b)
Figure 6.11: Observed and Biots modelled P-wave (a) and S-wave (b) CO
2
saturated
velocities for P
P
= 7 MPa, P
P
= 10MPa, and P
P
= 25 MPa constant pore pressure runs. No
error bars are shown, however the size of errors can be referred to in figure 6.7 and 6.9.

6.2.4 Observed CO
2
Saturated Wave Attenuation-
From the waveform plots pertaining to the CO
2
saturated state, displays of
large signal strength variations were observed. Here the earlier qualitative
observations made on wave attenuation will be examined quantitatively. The
attenuation of waveforms was studied by determining the differential attenuation
coefficient as described in section 4.5.2. For all CO
2
saturated measurements, the
differential attenuation coefficient of each waveform was determined with respect
to the waveform acquired at 2 MPa of CO
2
and 23C. The waveform acquired at
2 MPa and 23C was used as the reference signal because it has the greatest signal
strength of all the waveforms collected.
Figure 6.12 displays the differential attenuation coefficient for P-and S-
waves for the two constant temperature runs as a function of pore pressure and of
frequency. Here, only the differential attenuation coefficients for the T = 28 C
and T = 40 C constant temperature runs are presented. The computed amplitude
135

spectrum of the signals used for determining the differential attenuation
coefficient for all the measurement runs is given in the appendix, along with the
differential attenuation coefficient results of T = 23C and T = 45C measurement
runs. The frequency range used for analysis is limited by the band of frequencies
that carries most of the energy in their amplitude spectra. The useable frequency
range considered not to be contaminated by too much noise is from 0.4 to 1.4
MHz for P-waves and 0.5 to 1.7 MHz for S-waves. The peak in the spectra is at
about 0.67 MHz and 1.2 MHz for P-and S-waves, respectively.

a) b)
c) d)
Liquid Liquid
Supercritical
Fluid
Supercritical
Fluid
Gas Gas
Gas Gas
Figure 6.12: Observed P- and S-wave differential attenuation coefficient determined for a)
and b) T = 28C, and c) and d) T = 40C constant temperature runs as a function of
frequency and pore pressure. The expected CO
2
phase state is indicated at the appropriate
physical conditions.

The CO
2
phase transition out of the gaseous state and into the liquid or
supercritical states is marked by a significant increase in wave attenuation. When
136

contrasting the T = 28C to the T = 40C plots for both P-and S-waves, especially
at high pore pressure and high frequencies the attenuation is slightly greater for
the T = 28C than for the T = 40C. Furthermore, similar to that seen with wave
velocities, the large changes in wave attenuation occurs at a lower pore pressure
for the T = 28C results than the T = 40C results.
When comparing the P-wave differential attenuation coefficient to the S-
wave differential attenuation coefficient under the same condition, the S-wave
generally appears to be more attenuated than the P-wave. Also, the differential
attenuation coefficients of S-waves appear to be more variable than the P-waves
especially at high pore pressure and frequencies. This is because the S-waves are
more susceptible to noise effects due to their lower amplitude strength.
From examining all the differential attenuation coefficient plots, two
general attenuation trends can be established. First, the attenuation increases with
pore pressure where it is the lowest in the gaseous phase state and becomes higher
in the liquid or supercritical fluid phase states. Second, as would be expected the
attenuation is greater at higher frequencies than at lower frequencies and is
particularly more obvious at high pore pressures.
waves for the constant pressure runs with pore pressure at 7 MPa, 10 MPa, and 25
MPa as a function of temperature and frequency. When the P-and S-wave plots of
P
P
=7 MPa are compared to the P- and S-wave plots of P
p
=10 MPa and of P
p
= 25
MPa the overall character of the plots are different. The P
P
= 7 MPa displays a
sharp decrease in differential attenuation coefficients. The P
P
= 10 MPa plots and
P
P
= 25 MPa plots, just like their wave velocity behaviours, do not exhibit any
significant attenuation variations that signifies a phase transition. When all three
constant pressure runs are compared at a given temperature, the attenuation of a
low pore pressure run is generally lower than the attenuation for a high pore
pressure run. Again, S-wave generally appears to be more attenuated than P-
waves when compared at the same condition.
In general, two attenuation trends can be established from these constant
pressure runs. First, however slight, the attenuation decreases with increasing
137

a) b)
c) d)
Supercritical
Fluid
Supercritical
Fluid
Gas Gas
e) f)
Liquid Liquid
Liquid
Supercritical
Fluid
Supercritical
Fluid
Liquid Liquid
Liquid

Figure 6.13: Measured P- and S-wave differential attenuation coefficient determined for a)
and b) P
P
= 7 MPa, c) and d) P
P
= 10 MPa, and e) and f) P
P
= 25 MPa constant temperature
runs as a function of frequency and pore pressure. The expected CO
2
phase state is
indicated at the appropriate physical conditions. The black dashed-line indicates where the
phase transitions are suppose to occur.

temperature where it is the highest in the liquid phase and becomes lower in the
gas phase or supercritical fluid phase. Second, the attenuation is frequency
138

dependent where attenuation is greater at high frequencies than at lower
frequencies and is especially obvious at lower temperatures.

6.2.5 Modelled CO
2
Saturated Wave Attenuation
As mentioned earlier, Biots formulation also considers wave attenuation.
Figure 6.14 shows the computed differential attenuation coefficient from Biots
model for two constant temperature runs. A large change to both P-and S-wave
attenuation in response to the phase transition is modelled. Comparing these two
temperature runs at the same pore pressure and frequency, Biots model shows
that the attenuation is always greater for the lower temperature run than the higher
temperature run. Between the P-and S-wave differential attenuation coefficients
for any constant temperature runs, Biots model shows that the S-wave results are
larger than the P-waves results when compared under the same conditions. Both
of these comparisons between modelled results (low against high temperature; P-
against S-waves) are consistent with the comparisons between observed results.
Furthermore, the two general attenuation trends of attenuation increasing with
pore pressure and with frequency seen with the observed results are also
supported by Biots modelled results. However when the absolute differential
attenuation coefficient values are compared between the modelled and observed
results there are some discrepancies where Biots modelled attenuation are
generally much lower than the observed attenuation. From the discussion in
section 4.5.2, the errors of the observed P-and S-wave differential attenuation
coefficient for this study are 12 m
-1
and 18 m
-1
respectively. With these errors
considered, the differential attenuation coefficients are in agreement at only low
pore pressures before the phase transition. At high pore pressures after the phase
transition, disagreement between the results occurs for all frequencies although
the discrepancy is less at lower frequencies. The largest discrepancy in
differential attenuation coefficient values is 68% for P-waves and 63% for S-
waves.
Figure 6.15 shows the computed differential attenuation coefficient from
Biots model for the three constant pore pressure runs. The modelled differential
139

attenuation coefficients do show large changes in response to the gas to liquid
phase transition whereas the modelled attenuation response to the liquid to
supercritical phase transition shows only a smooth variation. When the three
constant pore pressure runs are contrasted with each other at a given condition

a) b)
c) d)
Gas Gas
Liquid
Gas Gas
Supercritical
fluid
Supercritical
fluid
Liquid
Figure 6.14: Biots modelled P- and S-wave differential attenuation coefficient determined
for a) and b) T = 28C, and c) and d) T = 40C constant temperature runs as a function of
frequency and pore pressure. The prevailing CO
2

irrespective of the wave type, the attenuation is always higher for the higher pore
pressure. Again when the differential attenuation coefficient of the S-waves is
compared to the P-waves at the same condition of any pore pressure run, the
attenuation of S-waves is always greater than P-waves. Overall Biots modelled
results show that attenuation decreases with temperature and increases with
frequency, and they are in agreement with our observed attenuation trends.
140

All these trends deduced from Biots modelled differential attenuation
coefficients are in qualitative agreement with the observed differential attenuation
a) b)
c) d)
Supercritical
Fluid
Supercritical
Fluid
Gas Gas
e) f)
Liquid Liquid
Supercritical
Fluid
Supercritical
Fluid
Liquid Liquid
Liquid Liquid
Figure 6.15 Biots modelled P- and S-wave differential attenuation coefficient determined for
a) and b) P
P
P
P
2
phase state is

coefficients. However in terms of quantitative agreement, the modelled and the
observed results disagree. Similar to the constant temperature runs there is a
141

discrepancy of absolute differential attenuation coefficient values. Apart from
gaseous region where the results are in agreement, the discrepancy between
results increases with decreasing temperature and increasing frequency. The
largest discrepancy in differential attenuation coefficient values is 72% for P-
waves and 69% for S-waves. Also, the decrease in attenuation with increasing
temperature for observed results is significantly less pronounced than that
displayed by Biots results, particularly for the two high pore pressure runs. For
example from 23C to 40C at a frequency of 1 MHz for the P
P
= 10 MPa run,
Biots results show a 40% change while the observed results yield only a 4%
change.
Overall, Biots modelled results show much improvement over
Gassmanns modelled results. For the majority of wave velocities, Biots results
are in agreement with the observed laboratory results, where as Gassmanns
results were never in agreement. When Biots modelled results do not agree with
the observed results, the disagreement is usually substantially less than the
disagreement between the observed and Gassmanns modelled values. In
addition, Biots formulation accounts for wave attenuation, which is apparent in
our data and was not considered by Gassmanns formulation. The significant
improvements of Biots model over Gassmanns model are because Biots model
considers a wide range of wave frequencies.
Recall, Biots formulation describes wave behaviour based on the
prevailing frequency regime. A wave is in the low frequency regime if its
frequency is less than the critical frequency and a wave is in the high frequency
regime if its frequency is greater than the critical frequency, defined by equation
3.45. The critical frequency given here again is
k t
|q
f
c
f
2
=
where is porosity, is the viscosity of the pore fluid,
f
is pore fluid density, and
is the absolute permeability of the porous sample. The computed critical
frequency for the porous ceramic rod saturated with CO
2
over a range of pressures
and temperatures are shown in figure 6.16. Over the actual CO
2
conditions
142

explored for the porous ceramic rod (shown by the white arrows in the diagram)
the critical frequency ranges from 0.07 MHz to 0.43 MHz where a large majority
of the critical frequencies are below 0.10 MHz. The higher critical frequencies
are actually comparable to the wave frequencies of our data and all the higher
critical frequencies occur at conditions where CO
2
is a gas. This implies that our
data in the gaseous CO
2
state should also display behaviours relevant to low
frequencies. From our laboratory data results, there are two observations that
support low frequency behaviours. First, in the gaseous CO
2
state wave
attenuation is low. Second, in the gaseous CO
2
state, when the observed wave
velocities are compared with Gassmanns modelled velocities, the wave velocities
are almost within error.

Figure 6.16: Biots critical frequency for the CO
2
saturated porous ceramic rod. The
superimposed white arrows represent the conditions investigated in the study.

In regards to wave attenuation, Biots predictions are qualitatively in line
with observed wave attenuation behaviour. Quantitatively however Biots wave
attenuation predictions are substantially less than the observed wave behaviour.
The wave attenuation mechanism considered in Biots formulation may not be the
only wave attenuation mechanism occurring in the laboratory. Alternatively,
143

measurements of attenuation are notoriously prone to error and this may be the
reason for the discrepancy (e.g., White (1992)).

6.3 Water
Water saturated measurements were conducted on the porous ceramic rod
under full water saturation at T = 23C with pore pressure varying from 5 to 50
MPa while differential pressure was maintained at 5 MPa. It should be noted that
this water saturated measurement run was not obtained during the same series
conducted for the dry and CO
2
saturated measurements (see chapter 4), but was
acquired earlier on a different sample of the same material.

6.3.1 Water Saturated Waveforms
The P- and S-waveforms obtained under full water saturated conditions of
the porous ceramic rod are shown in figure 6.17. In this figure, both the P-and S-
waveforms do not vary significantly over this pressure interval. The signal
strength appears relatively constant throughout, with very slight weakening at the
lowest pore pressure. As pore pressure increases the P-waves arrive slightly
earlier, but the S-wave arrival times do not appear to vary.
a) b)
Figure 6.17: Normalized P- (a) and S-wave (b) water saturated waveforms as a function of
pore pressure. These water saturated waveforms were collected from another porous
ceramic rod sample instead of the one used for the dry and CO
2
saturated measurements. A
constant differential pressure of 5 MPa was maintained throughout the measurement run.

144

6.3.2 Observed Water Saturated Wave Velocities
Figure 6.18 displays the P-and S-wave velocities with error as a function
of pore pressure using the waveforms shown in figure 6.17. The dry P-and S-
wave velocity obtained under a confining pressure of 5 MPa is shown at P
P
= 0
MPa. Similar to that observed with CO
2
, with water in the pore space the wave
velocities decreases with respect to the wave velocities under dry conditions. The
wave velocity decrease from dry to saturated is much more than that for CO
2
.
This could be because water is denser than CO
2
in general. However, as the pore
pressure of water increases the P-wave velocities also increases. Over this 5 to 50
MPa pore pressure interval, the P-wave velocities increased linearly of about
1.2%. In contrast over the same pressure range the S-wave velocities decreased
linearly, but of only 0.3%.

a) b)
Figure 6.18: Ultrasonic P- (a) and S-wave (b) velocities with their respective errors measured
of the sample saturated with water under various pore pressures at T = 23C and when the

6.3.3 Modelled Water Saturated Wave Velocities
As for the CO
2
saturated results, Gassmanns and Biots theoretical
predictions are computed here over the same pressure and temperature conditions
as applied in the laboratory under water saturation. The water properties essential
to the computations such as bulk modulus, density, and fluid viscosity, were also
obtained from NISTs online chemistry webBook.
145

Gassmann Model
Figure 6.19 shows the modelled and observed P-and S-wave velocities of
the water saturated state. The observed linear increase of P-wave velocities with
increasing pore pressure and the observed opposite behaviour of a linear decrease
of S-wave velocities with increasing pore pressure are both modelled by
Gassmanns fluid substitution. The modelled increase in P-wave velocity is 0.6%
and the modelled decrease in S-wave velocities is 0.3%. However neither the P-
nor S-wave modelled velocities are in agreement with the observed velocities; the
modelled velocities are consistently lower than the observed velocities. The
discrepancy is 1.6% for P-waves and 3.6% for S-waves.

a) b)
Figure 6.19: Observed and Gassmanns modelled P-waves (a) and S-wave (b) water
saturated velocities.

Biot Model
Figure 6.20 displays the water saturated modelled and observed P-and S-
wave velocities. The linearly increasing and decreasing trend with pore pressure
for the P-and S-wave velocities respectively, are also predicted by Biots model.
The increase for P-wave velocities is 0.5% while the decrease for S-wave
velocities is 0.1%. However, overall there is a disagreement between the velocity
results. The modelled velocities are always higher than the observed velocities
and the disagreements are greater for the S-waves than the P-waves; the largest
discrepancy for P-waves is 1.3% while for S-waves the largest discrepancy is
146

1.4%. These are still reasonably close. Again, Biots modelled results show a
closer fit with the observed data than Gassmanns modelled results.

a) b)
Figure 6.20: Observed and Biots modelled P- wave (a) and S-wave (b) water saturated
velocities.

6.4 Results and Discussion
In this section, the pore fluid effects of CO
2
and water on the wave
behaviours observed and modelled are reviewed, generalized and discussed.

6.4.1 CO
2

In general from all the wave velocity behaviours seen in both laboratory
and modelled results and from considering the simple equations of 3.19 and 3.20,
it appears the controlling factor is the bulk density of the sample. For S-waves,
this observation is obvious because theoretically the changes to S-wave velocities
can only be caused by density changes, if the rigidity (shear modulus) of the
sample remains unaffected by the fluid. As the bulk density of the sample (which
is dependent on CO
2
s density) increases the S-wave velocities decreases due to
the inversely proportional effect of bulk density to wave velocity. For P-waves,
the changes to wave velocity are theoretically affected by the changes in both bulk
modulus and bulk density of the sample. The effects of bulk modulus and bulk
density on P-wave velocities oppose each other; a bulk modulus change will yield
a directly proportional change to the wave velocity whereas a density change will
147

cause an inversely proportional change to wave velocity. From the observed and
modelled results, the P-wave velocity decreases whenever the bulk modulus and
bulk density of CO
2
are supposed to increase (due to increase in pore pressure or
decrease in temperature, or going from a dry to CO
2
saturated state). Therefore
the changes of CO
2
s density dominate those of the bulk modulus in contributing
to the P-wave velocity. Furthermore, because S-waves velocity variations are
only affected by the bulk density changes, the fact that all P-wave velocity
variations mirror the S-wave variations under the same conditions also suggests
the behaviour of P-wave velocities is largely controlled by the variation in the
bulk density for the sample.
For the constant temperature runs, the wave responses (velocity and
attenuation) to the gas to liquid phase transition are more immediate than the
wave responses to the gas to supercritical fluid phase transition. Also as the
temperature of the run increases, the levelling of the wave response from the large
changes caused by the phase transition do not occur until a higher pore pressure.
These responses are seen with laboratory observed results, Gassmanns modelled
results, and Biots modelled results. The immediate or gradual nature of the wave
responses are related to the sharp or subdue contrast in the bulk modulus and
density of the CO
2
when the phase transition boundaries are traversed. Because
supercritical fluids have behaviours of both a gas and liquid, any transition from a
gas or liquid will yield a smooth physical property change and consequently a
smooth wave response. Furthermore, the different pore pressures where the large
changes in wave responses begin to taper off are controlled by where the large
changes to the density of CO
2
begins to level off. In figure 6.3 large changes to
CO
2
s density are seen to occur at a higher pore pressure for a higher temperature
run.
For both observed and modelled results, the P-and S-wave changes (of
velocity and attenuation) are always greater for the gas-liquid phase transition
than for the gas-supercritical or liquid-supercritical fluid transition, when
comparing within constant temperature or constant pressure runs, respectively.
The larger wave variation for the gas-liquid transition is also due to the same
148

reasoning above; supercritical fluids have behaviours of both a gas and a liquid.
The absolute change in physical properties for the gas-supercritical or liquid-
supercritical transition will always be less than the gas-liquid transition, therefore
causing a smaller wave variation.
Furthermore, over the temperature and pressure conditions explored the
phase transition between a gas and liquid phase or a gas and supercritical fluid
phase is detectable from both P-and S-wave velocities and attenuation changes.
Based on observed data, a sudden drop in velocity and large increase in
attenuation is seen over the conditions close to when gas transitions into a liquid
or supercritical fluid. However the phase transition between a liquid and a
supercritical fluid phase state is not abrupt; there are no sudden wave changes.
The undetectable liquid supercritical phase change in wave velocities is a
consequence of the more gradual CO
2
density change across this transition.

6.4.2 Water
Both the observed and modelled behaviour of the water saturated sample
show a linear increase in P-wave velocities and shows a slight linear decrease in
S-wave velocities. The smooth nature of the velocity variation seen is expected
since water will not undergo any phase transition based on the applied conditions.
From a pore pressure of 5 MPa to 50 MPa, according to NISTs online chemistry
webBook the density of water increases from 999.8 kg/m
3
to 1019.1 kg/m
3
, while
the bulk modulus of water increases from 2.25 GPa to 2.52 GPa. Both the density
and bulk modulus of water are greater than those for CO
2
over the pressure and
temperature conditions investigated. The slight linear decrease of the S-wave
velocities is due to the minor increase in waters density over the investigated
pressure interval. For P-waves the increases in velocities with pressure suggests
that the changes in the bulk modulus of water have a greater effect on influencing
the overall wave velocity than the changes in fluid density. This is a behaviour
that was not seen when the sample was saturated with CO
2
. However similar to
the behaviour under CO
2
saturation, both the observed and modelled results show
that once water is introduced into the pore space, P-and S-waves velocity is
149

lowered significantly in comparison to the dry wave velocities in this highly
porous material. The effect of the changes to the bulk density is greater than the
effect of the changes to bulk modulus only as the sample goes from dry to water
saturated. The reader should not take this as a general trend for all porous media
as this will reverse in lower porosity materials.
The physical properties of the porous ceramic rod are not typical of real
reservoir rocks. It should be remembered that the porous ceramic rod is highly
porous. With a porosity greater than 58%, its high porosity have allowed the
overall wave response to be more sensitive to the changes in the physical
properties of CO
2
than a less porous sample would. Furthermore this synthetic
sample contains a negligible amount of micro-cracks, which in a larger presence
can complicate the acoustic behaviour of a sample as will shortly be seen in the
next chapter. By using a sample with such physical properties, in addition to
applying full CO
2
saturation, we have provided the possible wave behaviours for
an extreme case of having CO
2
in the pore space. In the next chapter, a less
extreme and a more realistic study on the CO
2
wave response is given.

6.5 Summary
Ultrasonic pulse transmission experiments were performed on the porous
ceramic rod under dry, CO
2
saturated, and water saturated conditions. The P-and
S-waveforms and the corresponding wave velocities of all three saturated states
and the differential attenuation coefficient of just the CO
2
saturated state were
presented for this sample. As a function of applied pressure the dry wave
velocities of the porous ceramic rod showed little change. This behaviour with
pressure provided further evidence, in conjunction to SEM images and mercury
injection porosimetry results, indicating this material contains a lack of micro-
cracks.
CO
2
saturated measurements were conducted for a variety of temperatures
and pressures where the gas, liquid, and supercritical phase state of CO
2
were
explored. Both the P-and S-wave velocities of the CO
2
saturated sample are
always lower than the dry wave velocities. From all the observed wave velocity
150

changes, changes to the bulk density of the CO
2
saturated sample appear to be the
controlling factor on the response of both P-and S- wave velocity. For the gas to
liquid and gas to supercritical phase transition, significant drops in wave velocity
and wave amplitude were observed. The porous ceramic rod displayed a wave
velocity drop of 4-5% for both P-and S-waves. The differential attenuation
coefficients showed a large increase over the phase transition interval and
remained relatively constant thereafter. The wave changes related to the gas-
liquid transition were always greater and more immediate than the gas-
supercritical transition. The nature of the wave responses was reflective of the
changes in CO
2
s physical properties. For the liquid to supercritical fluid phase
transition, the waveforms showed minor and smooth wave velocity and wave
amplitude increase. No signs of CO
2
phase transition could be detected from
observing the waveforms due to the lack of distinct waveform changes. The
observed smooth wave velocity change leading to an undetected liquid to
supercritical phase transition is a consequence of the lack of significant CO
2

density change over the conditions investigated.
Fluid substitution using Gassmanns and Biots equations were applied to
the same CO
2
saturated conditions as conducted in the laboratory for this sample.
Gassmanns modelled results show poor agreement with the observed wave
velocities except at the low pore pressures. The agreement or disagreement
between Gassmanns and the observed results could be understood through the
critical frequency of the porous ceramic rod. Moreover the Biots modelled wave
velocities agreed well with the observed wave velocities and Biots modelled
differential attenuation coefficients agreed only qualitatively with the observed.
Water saturated measurements on the porous ceramic rod were only
conducted as a function of pore pressure while under a constant differential
pressure. Minor linear changes in wave velocities with pore pressure were
observed. Similar to the CO
2
saturated results, the Biots modelled wave
velocities agree better than Gassmanns modelled velocities, however there is still
a disagreement between modelled and observed wave velocities.
151

The good agreement between the observations and the Biot modeling
suggests that the Biot theory reasonably predicts the wave behavior.
Consequently, and perhaps paradoxically, this further suggest that the Gassmanns
equations, that are the same as Biots formulations at zero frequency, should
describe the low frequency behavior of the sintered alumina.

152

Chapter 7
Results and Discussion: Natural Berea
Sandstone

In the previous chapter, the results and discussion pertaining to a synthetic
sample were given. Instead in this chapter the laboratory results, the modeling,
and a discussion on the Berea sandstone are given. Similarly, the P-and S-
waveforms, the determined wave velocities, and their comparison to modelled
velocities will be given for all three states: dry, CO
2
saturated, and water
saturated. The laboratory results and the modelled predictions of the differential
attenuation coefficient will be given for only the CO
2
The methodology behind signal acquisition, wave velocity and attenuation
analysis, and fluid substitution modeling applied to the Berea sandstone are the
same as applied to the porous ceramic rod. The natural Berea sandstone is
petrophysically quite different from the porous ceramic rod. The Berea sandstone
is two-thirds less porous and contains micro-cracks (refer to chapter 5s
observations). In comparison to the porous ceramic rod, this sample provides a
more realistic study for geological sequestration.

7.1 Dry
P-and S-waveforms were obtained of the Berea sandstone under dry (i.e.
vacuum in the pore space) conditions at room temperature (T = 23C) while the
confining pressure varied between 5 to 40 MPa. Both pressurization and
depressurization were applied to the sample over this pressure interval.

153

7.1.1 Dry Waveforms
The full set of normalized P-and S-waveforms displayed as coloured
amplitudes are shown in figure 7.1. The signals are normalized with respect to
the largest amplitude within the given set of waveforms. In contrast to the
waveform plots of the porous ceramic rod subject to the same conditions (figure
6.1), the Berea sandstone displays remarkable nonlinear changes in travel times
with pressure as has been seen by many authors in the past (e.g. Christensen and
Wang, 1985). In addition, a slight asymmetry of the wave arrivals is apparent
especially in the S-wave plot; the waves arrive earlier during the depressurization
cycle than the pressurization cycle for the same pressure.

a) b)
Figure 7.1: Normalized P- wave (a) and S-wave (b) waveforms obtained under dry
conditions at room temperature plotted as a function of confining pressure as coloured
amplitudes for the Berea sandstone.

7.1.2 Dry Wave Velocities
The P-and S-wave velocities with error of the waveforms obtained under
dry conditions are shown in figure 7.2. Both P-and S- wave velocities exhibit a
nonlinear increase with pressure. The wave velocities increase rapidly at low
pressures that taper off with higher pressures. This behavioural increase is caused
by the progressive closure of compliant pores where most of the compliant pores
will close at low pressures, therefore causing a rapid velocity increase. The
velocity increase diminishes at higher pressure because as the applied pressure
increases, the amount of compliant pores available for collapse reduces.
154

Furthermore from the velocity plots the wave velocities obtained during
depressurization are always higher than the velocities taken during pressurization,
thereby demonstrating wave velocity hysteresis.
Due to the cracked nature of the Berea sandstone, its pressure dependent
behaviour differs from that of the porous ceramic rod. The Berea sandstone
exhibits large velocity variations with pressure and large velocity hysteresis in
comparison to the synthetic sample where the presence of micro-cracks is
minimal.

a) b)
Figure 7.2: Dry P-and S-wave velocities with their respective errors as a function of
confining pressure for the Berea sandstone.

7.2 CO
2
For the Berea sandstone, five constant temperature runs and three constant
pore pressure runs were conducted. The measurement conditions applied to the
CO
2
saturated sample are shown in figure 7.3. Each measurement run conducted
is expected to cross a phase boundary. Constant temperature runs were conducted
at 23C, 28C, 40C, 45C, and 55C while pore pressure varied from 2 to 25
MPa. CO
2
is a gas at low pore pressures and is a liquid or a supercritical fluid at
higher pore pressures depending on the temperature. Constant pressure runs were
conducted at a pore pressure of 7 MPa, 10 MPa, and 25 MPa while temperature
varied from 23C to around 50C. For each of the three constant pressure (pore,
confining, or differential) runs, CO
2
is a liquid state at low temperatures. As
155

temperature increases CO
2
eventually transforms into a gaseous or a supercritical
fluid state, depending on the pore pressure as shown in figure 7.3. Similar to the
porous ceramic rod, during all CO
2
saturated measurements a constant differential
pressure of 15 MPa was maintained for each measurement run so that any
waveform variation observed is solely caused by pore fluid effects. Maintaining a
constant differential pressure during CO
2
measurements for the Berea sandstone is
especially important since the wave behaviours of this sample display strong
pressure dependence.
To avoid redundancy, only the most representative measurement runs for
this sample are shown here. Therefore for the constant temperature runs, only the
waveforms and differential attenuation coefficient of T = 23C and T = 55C run
are shown. The laboratory results for T = 28C, T = 40C, and T = 45C
measurement runs are available in the appendix.

Figure 7.3: Phase diagrams of CO
2
s: (a) bulk modulus (b) and density. The temperature
and pressure conditions applied to the CO
2
saturated Berea sandstone during ultrasonic
pulse transmission measurements are shown by the white arrows.

7.2.1 CO
2
Saturated Waveforms
Figure 7.4 shows the P-and S-waveforms as a function of pore pressure for
the lowest and highest temperature run, T = 23C and T = 55C respectively. In
general, as pore pressure increases the signals arrive later and within a given plot,
two arrival trends can be seen. Furthermore, the waveforms signal strength
156

shows a slight decrease near the anticipated phase transition. For the lower
temperature run (T = 23C), the elastic wave response to CO
2
s phase transition is
sharp and significant. In contrast, the elastic wave responses of the higher
temperature run (T = 55C) to CO
2
s phase transition are gentle and lack any
abrupt change.

a) b)
c) d)
Figure 7.4: Normalized P- and S-wave waveforms as a function of pore pressures collected
during the constant temperature runs of a) and b) T = 23C and, c) and d) T = 55C. A

Figure 7.5 displays the P-and S-waveforms as a function of temperature
from 23C to 50C for all three measurement runs. The P-and S-waveform plot
of P
P
= 7 MPa exhibits a sharp change with the signals arriving earlier and
stronger at higher temperatures than lower temperatures. Away from the
transition, both P-and S-waveforms do not noticeably change. For both the P
P
=
157

10 MPa and P
P
= 25 MPa P-and S-waveform plots, no sharp or large changes to
the waveforms occur. For the P
P
= 10 MPa waveform plots, both travel times and
signal strengths seem constant over the entire temperature range. For the P
P
= 25
MPa waveform plots no obvious signal strength changes are seen, however the
signals do arrive later with increasing temperature.

7.2.2 Observed CO
2
The P- and S-wave velocities with their estimated errors for all five
constant temperature runs, and the dry wave velocities obtained at a confining
pressure of 15 MPa plotted at P
P
= 0 MPa are shown in figure 7.6. Unlike that
seen before for the porous ceramic rod, the distribution of the five measurement
runs with respect of each other is somewhat scattered with no reasonable
relations.
Both the P-and S-wave velocities obtained under dry conditions are slower
than those obtained when the sample is saturated with gaseous CO
2
. For the P-
wave, this response is hard to validate quickly since both bulk modulus and bulk
density can affect the wave velocity and in opposite manners. However for the S-
wave, the response is not expected because the change in S-wave velocity should
only be affected by changes in bulk density (equation 3.20), which increases from
a dry to a saturated state and has an inverse effect on velocity, if the integrity of
the sample is assumed to be unchanged. Instead seen here, the dry S-wave
velocity is lower than all the saturated S-wave velocities at under low pore
pressures. This indicates that the material property of the sample has changed in
between the measurement runs.
Further evidence for a substantial change in the properties of the Berea
sandstone is found once the velocities under dry conditions are repeated. The dry
wave velocity shown in figure 7.6 was obtained prior to all the CO
2
saturated
measurements. A second set of dry measurements was conducted after all the
CO
2
saturated measurements and the resulting wave velocities are significantly
higher at all confining pressures than the initial set, see figure 7.7. The greatest
158

a) b)
c) d)
e) f)
Figure 7.5: Normalized P- and S-wave waveforms as a function of temperature collected
P
P
= 10 MPa and e)
and f) P
P

159

a) b)
Figure 7.6: Ultrasonic P-wave (a) and S-wave (b) velocities with their respective errors
measured of the Berea sandstone saturated with CO
2
for all constant temperature runs and
when the sample is not saturated (black dot at Pp = 0 MPa).

difference in wave velocities occurs at low confining pressures and the velocity
difference lessens with increasing confining pressure. The second set of dry wave
velocities show a smaller nonlinear increase of wave velocities with pressure and
displays smaller velocity hysteresis between the pressurization and
depressurization cycle. These observed differences between the two dry
measurement runs suggest sample consolidation, the permanent change in
physical properties of the material under pressure due to the irreversible closure of
micro-cracks. Sample consolidation would cause the sample undergoing the
second dry measurement run to contain less collapsible pores with respect to the
first dry measurement run.
In order to obtain a reasonable relation between the different measurement
runs, the absolute velocities were corrected with respect of each other by
considering the change to the nature of the sample. CO
2
saturated measurements
for each run were also conducted at a pore pressure of 1 MPa. It is assumed that
at this low pore pressure the amount of influence from the pore fluid on the
overall sample behaviour is minor so if the integrity of the sample has not
changed, all wave velocities at 1 MPa would be about the same. Using the wave
velocity of the T = 23C as the reference velocity, the difference in wave
velocities from all constant measurement runs were determined at 1 MPa and then
160

a) b)
Figure 7.7: P-wave (a) and S-wave (b) velocities with their respective errors as a function of
confining pressure for the two dry measurement runs of Berea sandstone at room
temperature, conducted before and after the series of CO
2

applied to the rest of the wave velocities of their respective measurement runs.
Figure 7.8 shows the wave velocities after applying the velocity corrections. The
dry wave velocity shown at P
p
= 0 MPa is the velocity determined from the
second set of dry measurements. The reasonability of the five measurement runs
distribution in relation to each other and the dry wave velocity has improved.
Similar to that seen with porous ceramic rod, with CO
2
in the pore space
irrespective of the phase state, both the new P-and S-wave velocities are lower
than the dry wave velocities. In addition, the new distribution of the five
measurement runs displays a positive trend with temperature, if their errors are
considered. At a given pore pressure, as the temperature of the run increases the
wave velocity of the run also increases. This observation generally holds except
for the S-waves of the T = 28C run at high pore pressures, where some of the
velocities are below the velocities obtained at T = 23C. For P-waves, a similar
discrepancy is also seen, however the discrepancy is still within error. This
disagreement of wave velocities could be attributed to the deterioration of one of
the transducers. The day after conducting the T = 28C measurement run, it was
determined that one of the transducers has failed where S-waves could not be
transmitted. Therefore the data for to the T = 28C measurement run should be
interpreted with caution.
161

a) b)
Figure 7.8: The corrected ultrasonic P-wave (a) and S-wave (b) velocities of the Berea
sandstone saturated with CO
2
for all constant temperature runs.

Furthermore, as pore pressure increases, the wave velocities generally
decrease with the exception of P-waves at high pore pressures of the two lower
temperature runs, which display velocity increase with increasing pore
pressure. The phase transition out of the gaseous state is clear for the two lower
temperature runs where an immediate and large drop in wave velocities occurs
near their respective vapour pressure (the pressure that marks the transition
between a gas and a liquid for a given temperature). The drop in wave velocity
right before and after the phase transition for T = 23C run for both P-wave and S-
wave is 1.8%. Over the entire investigated pore pressure interval the drop in
overall wave velocity is 3.0% and 4.2% for P-and S-wave respectively. In
contrast, the phase transition for the three higher temperature runs is less obvious
with the wave velocities displaying a very gradual response. The exact
occurrence of the phase transition cannot be identified from the velocity plot.
Therefore the amount of velocity change can only be reported over the entire
investigated pore pressure interval. From a pore pressure of 2 to 25 MPa, the P-
wave velocities dropped 3.3% for the T = 55C run and the S-wave velocities
dropped 3.8%. The overall P-wave velocity change for the T = 55C is greater
than the T = 23C because the P-wave velocities of the T = 23C actually recovers
after the large drop.
162

The three constant pore pressure measurement runs, are also affected by
the varying nature of the sample due to consolidation. Unlike the constant
temperature measurements, for the conditions applied to the three constant pore
pressure runs there are no similar pore fluid physical properties which would have
allowed similar corrections of the constant temperature runs to be employed.
However, from reviewing figure 7.3 of the various conditions applied to the Berea
sandstone, all three constant pore pressures do have points of overlap with four
constant temperature measurements. Therefore all the wave velocities of the
constant pore pressure measurements were corrected so that the wave velocities
obtained under the same conditions as the constant temperature run of T = 23C
are the same.
Figure 7.9 shows the corrected P-and S-wave velocities with their errors
for the three constant pore pressure runs along with the dry wave velocity
obtained under a confining pressure of 15 MPa from the second set of dry
measurements plotted at T = 23C. Similarly to above, with CO
2
in the pore
space the wave velocities are lower with respect of the dry wave velocity. The
CO
2
phase transition of the P
P
= 7 MPa run from liquid to gas is evident from an
obvious increase in wave velocity of 1.1% for both P-and S-wave velocities.
Over the entire investigated temperature interval an overall increase of 1.0% and
0.8% for P-and S-wave, respectively is seen. For the P
P
= 10 MPa and P
P
= 25
MPa the phase transition is not apparent at all; the wave velocities variations are
smooth and there are no changes to the velocity trend. The P
P
= 10 MPa run
exhibits a 0.2% overall increase with temperature for both P-wave and S-wave
velocity. In contrast, the P
P
= 25 MPa run exhibits a decrease in both P-and S-
wave velocities with increasing temperatures; P-waves show an overall decrease
of 1.7% in velocity and S-waves show an overall decrease of 1.3% in velocity.

7.2.3 Modelled CO
2
Here, the observed results for the CO
2
saturated Berea sandstone are
163

a) b)
Figure 7.9: The corrected ultrasonic P- wave (a) and S-wave (b) velocities of the Berea
2
for all constant pore pressure runs.

compared to modelled results computed from Gassmann and Biots formulations
over the same pressure and temperature conditions. To compute the models, the
series of parameters required as inputs were obtained from the same sources or by
the same methods as mentioned for the porous ceramic rod. However unlike the
porous ceramic rod, for the Berea sandstone the temperature dependency of the
frame is not included in the modeling results. This correction was not needed as
all the observed velocities shown in figure 7.8 were already corrected with respect
to the T = 23C data, thereby eliminating the temperature dependence of the
frame in the observed velocities.

Gassmann Model
Using equation 3.29 and 3.31 the Gassmanns modelled P-and S-wave
velocities of the CO
2
saturated Berea sandstone was calculated. Figure 7.10
shows the modelled and the observed P-and S-wave velocities under CO
2

saturation for the constant temperature runs of T = 23C and the T = 55C. The
rest of the modelled constant temperature run results can be found in the
appendix. Unexpectedly slight agreements can be seen between some observed
and modelled results, such as at low pore pressures of the T = 55C for both P-and
S-waves. Also, the amount of change in wave velocity with changing pore
pressure is somewhat comparable between the two results for both P-and S-waves
164

and for both the low and high temperature runs. For T = 23C, over the phase
transition Gassmann predicts a 2.5% and 2.7% drop in P-and S-wave velocity,
respectively and an overall drop of 2.9% and 3.8% for P-and S-wave velocity,
respectively. For T = 55C, Gassmann predicts an overall drop of 2.8% and 3.2%
drop in P-and S-wave velocity. Strangely, when there is a disagreement between
results, Gassmanns modelled wave velocities are always greater than the
observed wave velocities. The largest discrepancy between modelled and
observed results is 2.0% and 1.8% for P-and S-wave respectively. Furthermore
for the T = 23C results the observed P-waves velocity curve overall tends to
have a greater slope than the modelled velocity curve. For S-waves this only
occurs at lower pressures before the phase transition. For the T = 55C results for
both P-and S-waves the shape of the observed and modelled velocity curves
follow each other quite well.

a) b)
Figure 7.10: Observed and Gassmanns modelled P- wave (a) and S-wave (b) CO
2
saturated
velocities for T = 23C and T = 55C constant temperature runs. To avoid clustering, only
the error at a pore pressure of 25 MPa are shown. The errors at all other data points are
approximately the same size.

Figure 7.11 displays the modelled and observed P-and S-wave velocities
under CO
2
saturation for all three constant pore pressure runs. None of the
observed and modelled results are in agreement with each other, where
Gassmanns modelled results are always higher than the observed. As the
temperature increases, the discrepancy between observed and modelled results
165

increases. Starting with P
p
= 7 MPa, similar to that seen with the porous ceramic
rod, the observed results have a lagged temperature response to the occurrence of
the liquid to gas phase transition. The phase transition theoretically should occur
around 29 C while the observed results do not show a transition until after 33 C.
Possible reasons for this discrepancy were provided in section 6.2.2. Similar to the
constant temperature runs, the amount of change in wave velocity in response to
the phase transition is roughly comparable between the observed and modelled
results for both P- and S-waves; Gassmanns predicts a 1.5% and 1.6% P-and S-
wave velocity increase. Over the entire temperature interval, a 2.4% and a 2.6%
overall increase is predicted. Aside from the region close around the phase
transition, the largest discrepancy between the results is 2.8% for both P-and S-
waves. Also there is a discrepancy on the slope of the velocity curves; away from
the phase transition at both low and high temperatures, the observed velocity
curve has a negative slope while the modelled shows a positive slope.

a) b)
Figure 7.11: Observed and Gassmanns modelled P-wave (a) and S-wave (b) CO
2
saturated
velocities for P
P
= 7 MPa, P
P
= 10MPa, and P
P
= 25 MPa constant pore pressure runs. To
avoid clustering, only the error at the highest temperature in each run is shown. The errors
at all other data points are approximately the same size.

For both the P
p
= 10 MPa and P
p
= 25 MPa runs, the modelled results show
no sudden changes in wave velocities in response to the anticipated liquid to
supercritical phase transition. For the P
p
= 10 MPa results, the largest velocity
discrepancy is 2.6% for both P-and S-waves. For this constant pore pressure run,
166

both the modelled and observed velocity curves have positive slopes, except the
observed results slope is much slighter. Over the entire temperature interval,
Gassmanns model predicts a 1.5% P-wave velocity change and a 2.1% S-wave
velocity change while observed results show only a 0.2% velocity change for both
P-wave and S-waves. For the P
p
= 25 MPa results, the largest discrepancy is 2.9%
for both P-and S-waves. Similar to P
p
= 7 MPa results the velocity curves of the
observed and modelled results slope in opposite directions; observed results
slopes negatively while modelled results slopes positively but gently, with
temperature. Over the entire temperature interval the modelled results show less
than 0.1% increase in P-wave velocity and a 0.5% increase in S-wave velocity
while the observed results show a 1.7% and 1.3% decrease in P-and S-wave
velocity, respectively.

Biot Model
Even though Gassmann formulation deals with wave frequencies that are
much lower than the wave frequencies dealt with in the laboratory, surprisingly
Gassmanns prediction sometimes agrees with the observed results (the constant
temperature runs). Also, where there is a poor agreement in this study for the
Berea sandstone, Gassmanns modelled results are usually higher than the
observed results. Therefore if Biots formulation is applied even higher wave
velocities would result from the modelling. However Gassmanns theory, unlike
Biots theory, does not provide a description for wave attenuation which is present
in our data. Therefore, for the completeness of applying Biots model, here Biots
modelled wave velocity is still computed and compared for the Berea sandstone.
Biots modelled P-and S-wave velocities of the CO
2
saturated Berea
sandstone of the constant temperature and constant pore pressure runs are
displayed in figure 7.12 and 7.13, respectively. All of Biots modelled velocity
curves show the same behaviour trends and variation with changing pore pressure
or temperatures as that shown by Gassmanns modelled results. As expected, the
modelled velocities are all higher than the observed velocities. Apart from the
results at low pore pressures of the T = 55C constant temperature run, all other
167

modelled and observed results are not in agreement. The majority of the
discrepancies are greater than 2.0%.

a) b)
Figure 7.12: Biots modelled and observed P- wave (a) and S-wave (b) CO
2
saturated
velocities for T = 23C and T = 55C constant temperature runs. No error bars are shown,
however the size of errors can be referred to in figure 7.8 and 7.10.

a) b)
Figure 7.13: Biots modelled and observed P-wave (a) and S-wave (b) CO
2
saturated
velocities for P
P
= 7 MPa, P
P
= 10 MPa, and P
P
= 25 MPa constant pore pressure runs. No
error bars are shown, however the size of errors can be referred to in figure 7.9 and 7.11.

7.2.4 Observed CO
2
As with the porous ceramic rod, the attenuation of the acquired waveforms
of the Berea sandstone was studied by determining the differential attenuation
coefficient as described in section 4.5.2. For all CO
2
saturated measurements, the
differential attenuation coefficient of each waveform was determined with respect
168

to the waveform acquired at 2 MPa and 23C. Unlike that of the porous ceramic
rod, the waveform acquired at 2 MPa and 23C does not have the greatest signal
strength of all the measurement runs, especially for the S-wave. However it was
still used as the reference signal for the purpose of maintaining procedural
consistency with the analysis conducted for the porous ceramic rod.
Figure 7.14 display the differential attenuation coefficient for P-and S-
waves for the two constant temperature runs, T = 23C and T = 55C, as a
function of pore pressure and of frequency. For the lower temperature run, the
phase transition from gas to liquid is met by a sudden increase in attenuation and
is especially obvious at the higher frequencies. For the higher temperature run,
the phase change from gas to supercritical fluid is also marked by an increase in
attenuation, and is again more pronounced at higher frequencies. In contrast to
the lower temperature run, the large change in differential attenuation coefficients
is less drastic. Also, large changes to attenuation do not begin until well after the
critical pressure.
From examining these differential attenuation plots some general
behaviour can be established for both P-and S-waves. First, wave attenuation is
usually lower in the gaseous phase (low pore pressures) than the liquid phase or
supercritical fluid phase (higher pore pressures). Second, wave attenuation is the
greatest immediately after the phase transition into the liquid or supercritical
phase state but then lessens with increasing pore pressure. Third, wave
attenuation increases as the frequency increases. Four, S-waves attenuate more
than P-waves. However when comparing the low and the high temperature run to
each other at the same conditions, for both P- and S-waves the attenuation of the
T = 23C is usually higher than the T = 55C.
On the color bar index, the scale sometimes indicates negative differential
attenuation coefficient values. This is because the reference signal is not the
strongest signal of all measurement runs. In addition, for the T = 23C run even
though the reference signal is the strongest signal of the entire run, however the
amplitude spectrum of the reference signal is not consistently the greatest for all
frequencies.
169

wave for the three constant pressure runs, with pore pressure at 7 MPa, 10 MPa,
and 25 MPa as a function of temperature and frequency. The character of both
the P-and S-waves of the P
p
= 7 MPa plot is much different than the other two

a) b)
c) d)
Supercritical
Fluid
Supercritical
Fluid
Supercritical
Fluid
Supercritical
Fluid
Liquid Liquid
Gas Gas
Gas Gas
and b) T = 23 C, and c) and d) T = 55 C constant temperature runs as a function of
frequency and pore pressure. The expected CO
2

constant pressure plots by displaying a sharp change in wave attenuation in
response to the liquid to gas phase transition. The wave attenuation peaks near
the phase transition and then decreases with increasing temperature thereafter.
For the other two constant pressure plots there are no significant changes in
attenuation with respect to the phase transition. The P-and S-wave P
p
= 10 MPa
170

a) b)
e) f)
c) d)
Liquid Liquid
Gas Gas
Supercritical
Fluid
Liquid Liquid
Supercritical
Fluid
Supercritical
Fluid
Liquid Liquid
Supercritical
Fluid
and b) P
P
P
P
2
phase state is

171

plots also shows an attenuation peak, the attenuation increases with increasing
temperature and then also decreases with increasing temperature, but without any
sudden changes between these two trends. For the P
p
= 25 MPa P-and S-wave
plots, with increasing temperature there is no peak in attenuation. Generally in all
three constant pore pressure runs, wave attenuation is higher for the higher
frequencies and the S-wave attenuation is usually greater than P-wave attenuation.
When comparing the three constant pore pressure runs with one another, no single
general attenuation trend can be established. At low temperatures, it appears the
lower the pore pressure is for the run, the higher the attenuation. At high
temperatures, the P
P
= 7 MPa run shows the lowest attenuation, the P
P
= 10 MPa
run shows the highest attenuation while the P
P
= 25 MPa run is in between.
However this observed complex trend between the different pore pressure runs is
consistent with the attenuation trends observed from the constant temperature
runs; wave attenuation do not vary only positively with pore pressure but instead
varies positively and then negatively with pore pressure where the attenuation
peaks near the phase transition.

7.2.5 Modelled CO
2
The computed P-and S-wave differential attenuation coefficient of the two
constant temperature runs, T = 23C and T = 55C, as a function of temperature
and frequency are shown in figure 7.16. Biots modelled results do show a
significant increase in attenuation in response to the phase transition of gas to
liquid or gas to supercritical fluid around similar pore pressures of that seen with
our observed results. However the peak in attenuation after the phase transition
and its gradual decline with increasing pore pressures seen in the observed data is
not modelled. Instead, Biots modelled results shows a consistent increase in
wave attenuation as pore pressure increases for all frequencies. Furthermore, a
consistent increase in wave attenuation is also modelled with increasing
frequencies for all pore pressures. This attenuation trend with frequency agrees
with our observed attenuation results, although our observed results do show more
variability. Also similar to our observed results, Biots model shows that S-wave
172

attenuation is greater than P-wave attenuation under the same condition.
Comparing the modelled P-and S-wave attenuation of the T = 23C to T = 55C,
the attenuation of T = 23C runs is always greater than the attenuation of T =
55C. This is only consistent with the P-waves and not with the S-waves of the
observed results. Also dissimilar is the magnitude of the differential attenuation
coefficient values; the Biot modelled results are 2 orders of magnitude lower than
the observed results.

a) b)
c) d)
Supercritical
Fluid
Gas Gas
Supercritical
Fluid
Liquid
Gas Gas
Liquid
for a) and b) T = 23C, and c) and d) T = 55C constant temperature runs as a function of
frequency and pore pressure. The prevailing CO
2

Figure 7.17 shows the computed differential attenuation coefficient from
Biots model for the three constant pore pressure runs, as a function of
temperature and frequency. The modelled differential attenuation coefficients do
173

c) d)
e) f)
a) b)
Liquid Liquid
Gas Gas
Supercritical
Fluid
Liquid Liquid
Supercritical
Fluid
Supercritical
Fluid
Liquid Liquid
Supercritical
Fluid
for a) and b) P
P
P
P
= 25 MPa constant
temperature runs as a function of frequency and pore pressure. The expected CO
2
phase
state is indicated at the appropriate physical conditions. The black dashed-line indicates
where the phase transitions are suppose to occur.

174

show large changes in response to the liquid to gas phase transition where as the
modelled attenuation response to the liquid to supercritical phase transition shows
only a smooth and gentle variation, this agrees with our observed results. Within
a given plot, the modelled attenuation increases with increasing frequency and
decreases with increasing temperature. This modelled variation with frequency
agrees with observed results, but the modelled simple decrease with increasing
temperature does not agree. The observed results show a more complex
behaviour between pore pressure and with temperature than that is modelled by
Biot. When the three constant pore pressure runs are contrasted with each other at
a given condition, a general trend can be established where the modelled
attenuation is always higher for the higher pore pressure. This simple trend
however is also not displayed by the observed results. At the same condition of
any pore pressure run, the differential attenuation coefficient of the S-waves is
always greater than that of the P-waves and this is consistent with our observed
data. Similar to the constant temperature runs, the modelled results are two orders
of magnitude smaller than the observed results.
The critical frequencies of Biots theory were computed for the Berea
2
under a range of pressure and temperature
conditions and are shown in figure 7.18. In the figure, the white arrows denote
the actual experimental conditions considered in our study. For the range of
conditions applied in our study, the critical frequency ranges from 0.01 to 0.05
MHz where a large majority of the critical frequencies are below 0.03 MHz.
Therefore theoretically all our laboratory data are in the high frequency regime of
Biots formulation and large discrepancies with Gassmanns results should be
anticipated. Instead overall Gassmanns modelled velocity results provide a much
better description of the observed results than that of Biots modelled velocity
results, where Biot modelled velocities are usually significantly higher than the
observed velocities. Furthermore, Biots modelled differential attenuation
coefficients are two orders of magnitude smaller than the observed results. Biots
results suggest that minimal changes in the waveform should occur over the
investigated conditions for the saturated sample, which is far from what is
175

observed in the laboratory data. In short, Biots modelled results do not provide
an adequate description of the observed wave behaviour for the Berea sandstone.

Figure 7.18: Biots critical frequency for the CO
2
saturated Berea sandstone. The
superimposed white arrows represent the conditions investigated in the study.

7.3 Water
Water saturated measurements were conducted on the Berea sandstone
after the series of CO
2
saturated measurements. The sample was subjected to full
water saturation at T = 23C while pore pressure varied from 2 to 25 MPa and a
constant differential pressure of 15 MPa was maintained.

7.3.1 Water Saturated Waveforms
The P- and S-waveforms obtained under full water saturated conditions of
the Berea sandstone are shown in figure 7.19. Both the P-and S-waveforms
travel times and signal strength show minor changes over this pressure interval,
where the P-waves show more change. With higher pore pressure, the P-waves
appear to be arriving earlier and stronger in amplitude.

176

a) b)
Figure 7.19: Normalized P- (a) and S-wave (b) water saturated waveforms as a function of
pore pressure. A constant differential pressure of 15 MPa was maintained throughout the
measurement run.

7.3.2 Observed Water Saturated Wave Velocities
Figure 7.20 displays the water saturated P-and S-wave velocities with
error as a function of pore pressure determined from the waveforms shown in
figure 7.19. The dry P-and S-wave velocity of the second dry run obtained under
a confining pressure of 15 MPa is shown at P
P
= 0 MPa. Similar to that observed
with CO
2
, with water in the pore space the P-and S-wave velocities decrease with
respect to the dry wave velocities where the decrease in wave velocity is much
greater for S-waves than P-waves. As pore pressure of the saturated sample
increases, the P-wave velocity also increases. Over this 2 to 25 MPa pore
pressure interval, the P-wave velocities increased linearly of about 0.8%. Oddly,
for the S-waves as pore pressure increases the wave velocity also increases, even
though the bulk density of the saturated sample should increase. However the
overall increase in wave velocity is small; over the entire pore pressure interval,
the velocity increase is 0.2% which is only 5 m/s.

7.3.3 Modelled Water Saturated Wave Velocities
Gassmanns and Biots theoretical wave velocity predictions are computed
here over the same pressure and temperature conditions as applied in the

177

a) b)
Figure 7.20: Ultrasonic P- wave (a) and S-wave (b) velocities with their respective errors
measured of the sample saturated with water under various pore pressures at T = 23C and
when the sample is not saturated (black dot at Pp = 0 MPa).

laboratory under water saturation. The fluid properties of water essential to the
two models were obtained from NISTs online chemistry webBook.

Gassmann Model
Figure 7.21 shows the modelled and observed P-and S-wave water
saturated velocities. Both the observed and modelled behaviour of the water
saturated Berea sandstone displays a linear change in wave velocities with

a) b)
Figure 7.21: Observed and Gassmanns modelled P-wave (a) and S-wave (b) water saturated
velocities.

178

changing pore pressure. For P-waves similar to the porous ceramic rods results,
the observed velocity increase with pore pressure is modelled by Gassmanns
equation; however the rates of increase are different. As a result at high pore
pressures the results are in agreement but at low pore pressures the results are in
disagreement. The largest disagreement is at the lowest pore pressure, 2 MPa,
where the discrepancy is 1.0%. For S-waves, dissimilar to the porous ceramic
rods results, the observed slight velocity increase with pore pressure is not
modelled by Gassmanns equation; instead the modelled results show a slight
decrease in wave velocity. The largest discrepancy between S-wave velocity
results is 2.4% and again occurs at the lowest pore pressure. For both water
saturated P-and S-wave velocities, the Gassmann modelled response is higher
than the observed. Furthermore once water is introduced into the dry sample, the
modelled wave velocity shows an increase in P-wave velocity and a decrease in S-
wave velocity with respect to the dry wave velocities. The observed results
instead show that both P-and S-wave velocity are lower with respect to the dry
wave velocities.

Biot Model
Figure 7.22 displays the water saturated modelled and observed P-and S-

a) b)
Figure 7.22: Biots modelled and observed P- wave (a) and S-wave (b) water saturated
velocities.

179

wave velocities. Biots modelled velocity curves show the same trend with pore
pressure as that shown by Gassmanns modelled results. Similar to that seen with
the CO
2
saturated results, Biots modelled velocities for the water saturated Berea
sandstone are all higher than the observed velocities, and have a poorer fit to the
observed results than Gassmanns modelled velocities.

7.4 Results and Discussion
In this section, the pore fluid effects of CO
2
and water on the wave
behaviours observed and modelled are reviewed and discussed. The similarities
between the results of the Berea sandstone and the porous ceramic rod are given,
and also the differences between the results of these two samples will be
discussed. Moreover, the unexpected behaviours and issues with the Berea
sandstone will be examined.
Many of the wave behaviours and responses of the CO
2
saturated porous
ceramic rod, observed and modelled, are also seen with the CO
2
saturated Berea
sandstone. It is perhaps best to begin by noting these similarities:
1.) the dominance of the bulk density over bulk modulus in controlling
the wave velocity is seen in most of the observed and modelled
wave responses of the Berea sandstone. Deviations from this
general wave behaviour with bulk density will be discussed later;
2.) for the constant temperature runs the wave responses (velocity and
attenuation) to the gas to liquid phase transition is more immediate
and occurs at a lower pore pressure than the wave responses to the
gas to supercritical fluid phase transition. Furthermore the
levelling of wave responses after the phase transition is also seen to
occur at a higher pore pressure for a higher temperature run;
3.) the gas-liquid phase transition always results in a larger change in
wave response than a gas-supercritical or liquid-supercritical phase
transition;
4.) in this study the gas-liquid and gas-supercritical phase change is
detectable from sudden changes to both wave velocity and wave
180

attenuation. However the liquid-supercritical phase change is less
detectable due to the lack of sudden and large wave velocity and
wave attenuation changes.
The reasons for all these observed and modelled wave behaviours of CO
2

saturated measurements have been given in section 6.4.1 for the discussion on the
porous ceramic rod.
Although many wave behaviours of the porous ceramic rod are seen with
the Berea sandstone, there are some differences between the two materials. As
mentioned earlier some of the Berea sandstone observed wave velocities deviate
from the general behaviour of being controlled by bulk density. An example of
this exception to this general behaviour appears at high pore pressures for the P-
waves of the two low constant temperature runs of T = 23C and T = 28C, in
figure 7.8. The T = 23C run shows an obvious increase in P-wave velocity right
after the CO
2
transitions into the liquid state where as for the T = 28C run, the
overall increase is slighter. This increase in P-wave velocity for these two low
constant temperature runs are also predicted by both the Gassmann and the Biot
models, however their predicted P-wave velocity increase is minor in comparison
to that observed. For the T = 23C run, the observed increase from 6 MPa to 25
MPa is 0.6% (24 m/s) whereas the modelled increase is less than 0.1% (3 m/s).
This increase in P-wave velocity from both observed and modelled results
suggests that the change in bulk modulus effect is greater than the change in bulk
densitys effect. Referring to figure 7.3 the two lower temperature runs do
undergo a greater change in bulk modulus than in bulk density over this pressure
interval. Only past the critical point of CO
2
does bulk density vary more than
bulk modulus over the same interval. Therefore with CO
2
in the pore space,
depending on the conditions the P-wave velocity may increase or decrease with
increasing bulk density and bulk modulus.
Another exception to the wave velocities being controlled by bulk
density is demonstrated by the observed velocities of the constant pore pressure
runs. For all three constant pore pressure runs as the temperature increases the
bulk density of the sample should decrease and would yield an increase in wave
181

velocity if the bulk density is the controlling factor. This wave response is
demonstrated by both Gassmann and Biots modelled results. For our observed
results however, apart from the obvious jump in P-and S-wave velocity seen for
the P
P
= 7 MPa run in response to the liquid to gas phase transition, for the rest of
the P
P
= 7 MPa run and for the entire P
P
= 25 MPa run, both P-and S-wave
velocity decreases with decreasing bulk density. The decrease in S-wave velocity
suggests that as temperature increases the bulk density is not the only changing
factor, but instead the shear modulus of the sample is also changing. This implies
that the nature of the sample is varying with temperature. The nature of the
sample with increasing temperature changes as a result of frame softening,
therefore wave velocities would decrease with increasing temperature if all other
conditions were held constant. This frame weakening with increasing
temperature was addressed in the velocity analysis for the porous ceramic rod but
was not included for the Berea sandstone because the amount of velocity change
caused by temperatures effect alone is difficult to quantify because of the
hysteresis effects. For the Berea sandstone, the nature of the sample not only
changes with temperature but also with the number of measurements conducted
on the sample as discussed in section 6.2.2. Therefore, the velocity changes
observed during the heated dry measurements cannot be used to correct for frame
weakening since temperature is not the lone cause in affecting the dry wave
velocities.
Furthermore, the waveform changes (velocity and attenuation) in response
to the varying pressure and temperature conditions for the Berea sandstone are
reduced in comparison to the porous ceramic rod. For example, the gas-liquid
phase transition usually yields about a 4-5% observed wave velocity change for
the porous ceramic rod, whereas for the Berea sandstone an observed wave
velocity change of 1-2% is seen. Also when comparing the velocity plots of the
high constant temperature runs of the Berea sandstone to the velocity plot of the
porous ceramic rod under the same conditions, the velocity curves of the Berea
sandstone show less curvature. These reduced wave velocity responses of the
Berea sandstone is a result of its significantly lower porosity compared to the
182

porous ceramic rod. A lower porosity means lower pore fluid content in the
sample which leads to a reduced sensitivity of the overall sample to pore fluid
changes. For wave attenuation, when compared under similar conditions the
observed differential attenuation coefficients of the Berea sandstone are usually a
few times less than the porous ceramic rod. In fact the modelled differential
attenuation coefficients suggest that the attenuation of the Berea sandstone should
actually be two orders of magnitude smaller than of the porous ceramic rod. This
large difference between the modelled wave attenuation of the two samples is due
to the combined effect of Berea sandstones lower porosity and higher tortuosity.
When the porosity and tortuosity values of the porous ceramic rod was used
instead in the modelling of the Berea sandstone, similar magnitudes in differential
attenuation coefficients to that of the porous ceramic rod were computed.
The differential attenuation coefficients observed for the Berea sandstone
show more variability and less consistency than the observed results for the
porous ceramic rod. Theoretically the Berea sandstone should have better data
than porous ceramic rod because its waveforms are less attenuated from the
effects of pore fluid. However the transducer used to acquire the Berea sandstone
data unfortunately is weaker, therefore the resulting signals are smaller and are
more susceptible to noise.
Although the above dissimilar observed behaviours of the Berea sandstone
to the porous ceramic rod can be rationalized, there are other dissimilar observed
wave behaviours that are not understood. First, all the observed CO
2
and water
saturated wave velocities have a closer fit with Gassmanns than Biots predicted
wave velocities, even though the computed critical frequency for the Berea
sandstone suggests our data is in the high frequency regime for both pore fluids.
Furthermore, the Gassmanns modelled wave velocities are actually higher than
the observed wave velocities. Gassmanns formulation strictly describes low
frequency waves (more accurately, zero frequency static deformation) and based
on physical velocity dispersion it is expected that the observed ultrasonic wave
velocities be greater than Gassmanns predicted velocities. However this is not
seen.
183

Second, the observed water saturated P-and S-wave velocities both have
behaviours that are unexpected. Starting with P-waves, the observed wave
velocities compared to the dry wave velocities are lower (figure 7.20). However
both Gassmann and Biots modelled P-wave velocities do not agree with this
observation (figure 7.21 and 7.22). Typically real rocks do exhibit a higher P-
wave velocity in a water saturated state than in its dry state. In fact we conducted
another set of dry and water saturated measurements on a different Berea
sandstone sample that yielded results that agrees with this common observation.
Unfortunately in the set of measurements central to this study the water saturated
results are atypical. For S-waves, the common and reasonable lower water
saturated wave velocity compared to its dry state is observed in the laboratory
data. However, as the pore pressure increased (water density increases too) the S-
wave velocities unexpectedly increased. Since the differential pressure stays
constant while the pore pressure changed, this increase in S-wave velocities is not
explained by the Gassmann-Biot formulations.
Third, regardless of the phase states involved the observed attenuation of
the Berea sandstone tends to peak around the phase transition and usually occurs
in the denser CO
2
phase state. This behaviour is not displayed in the porous
ceramic rod and nor is it modelled by Biots formulation. Furthermore on
observed wave attenuation, as mentioned before negative differential attenuation
coefficients are seen in the plots. This is because the signal obtained at T = 23C
and 2 MPa that is used as the reference signal is not the strongest of all signals
acquired and its signal amplitude spectrum is not consistently the greatest for all
frequencies. This observed behaviour deviates from that seen with the porous
ceramic rod and that of Biots modelled results for the Berea sandstone.
All these unexpected behaviours suggest the nature of the Berea sandstone
sample used in this study is far more complex than the porous ceramic rod and
what Gassmann and Biots formulation can describe. The complex nature of the
Berea sandstone has already been suggested based on the observed sample
consolidation. Due to the presences of micro-cracks in the sample, the integrity of
the sample has changed over the series of measurements where higher wave
184

velocities are observed for later measurements than earlier measurements under
the same conditions. Also the length of the sample has changed over the course
of the measurements, where a length shortening of 0.02 mm has been measured.
Apart from varying the nature of the sample between and possibly during the
measurements, the thorough impact and implications of using a cracked sample
on the resulting wave responses are not fully understood in this study.
Another factor that may contribute to the altering of sample integrity is
chemical effect. The dissolution of minerals is expected to occur when CO
2
is
introduced into an environment that contains water. The mixture of CO
2
and
water will form an acidic solution where this solution will dissolve rock minerals.
Our CO
2
saturated measurements were conducted with full CO
2
saturation and
should not have contained water. However, there may have been a reaction
during the full water saturated measurements. Water saturated measurements
were conducted after the CO
2
saturated measurements where the previously CO
2

saturated sample was vacuumed for 12 hours before being saturated with water.
Even though the pore space is considered to be under vacuum, this vacuum is
not perfect where the lowest pressure reached was 0.0024 KPa (refer to Chapter
4). Therefore a very minor trace of CO
2
may have been left in the sample when
saturated with water. The amount of impact or mineral dissolution from such a
small trace of CO
2
or from full CO
2
saturation was not investigated. However the
dry weight of the sample measured after the full series of measurements did show
a decrease from its original starting weight of 43.70 g to 43.24 g. The porosity of
19.0% of the sample remained unchanged and could be a result of the opposing
effects of sample consolidation and mineral dissolution.

7.5 Summary
Ultrasonic pulse transmission experiments were performed on the Berea
sandstone under dry, CO
2
saturated, and water saturated conditions. The P-and S-
waveforms and the corresponding wave velocities of all three saturated states and
the differential attenuation coefficient of just the CO
2
saturated state were
presented. As a function of applied pressure the dry wave velocities of the Berea
185

sandstone showed a substantial change. This observed wave velocity dependency
on pressure is due to the presence of micro-cracks in the Berea sandstone, which
have been shown to complicate wave responses.
CO
2
saturated measurements were conducted for a variety of temperatures
and pressures where the gas, liquid, and supercritical phase state of CO
2
were
explored. Many trends and behaviours observed for the porous ceramic rod were
also observed here for the Berea sandstone. Such similarities seen are:
- the P-and S-wave velocities of the CO
2
saturated sample are always
lower than the dry wave velocities;
- changes to bulk density appears to be the controlling factor on the
response of the P-and S-wave velocities;
- the wave response to the gas-liquid transition is more immediate and
yields larger changes than the other phase transitions;
- the liquid-supercritical fluid transition is less detectable because there
are no disruption to the smooth and continuous change in wave
velocity.
Despite these similarities between the two samples, there are some
differences and deviations in the wave behaviours observed. Bulk density as the
controlling factor on the resulting wave response was seen for most wave
responses, but not for the constant temperature runs at T = 23C and T = 28C.
Over the gas to liquid and gas to supercritical phase transition, a sudden drop in
wave velocity and wave amplitude was observed however the wave velocity drop
was of less than 2%. Furthermore, unlike that seen with porous ceramic rod
results, the observed differential attenuation coefficients tends to show a peak in
attenuation around the phase transition interval and then drops off subsequently.
This peak in attenuation was also observed for the constant pore pressure runs,
even though there were no distinct changes to the wave velocities to signify the
liquid-supercritical fluid transition.
Fluid substitution using Gassmanns and Biots equations were applied to
the same CO
2
saturated conditions as were conducted in the laboratory for the
Berea sandstone. Both Gassmann and Biots modelled wave responses generally
186

did not agree with the observed waves behaviour; the observed wave velocities
were always lower than both models wave velocities, Biots modelled differential
attenuation coefficient values were two orders of magnitude larger than the
observed values, and the peaks in wave attenuation near the phase transition seen
in the observed data were not seen in Biots model.
Water saturated measurements on the Berea sandstone were only
conducted as a function of pore pressure while under a constant differential
pressure. Minor linear changes in wave velocities with pore pressure were
observed. Gassmanns predicted wave velocities showed a slightly closer fit in
terms of absolute wave velocities to the observed results than Biots predicted
results. The observed S-wave velocity increase with pore pressure is unexpected
in both models results.
Overall the Berea sandstone displays wave behaviours that are less
understood than the porous ceramic rod. This is due to the more complex nature
of the sample. The larger presence of micro-cracks and chemical change in the
sample could be possible factors in influencing and complicating the observed
wave behaviour.

187

Chapter 8
Conclusion

8.1 Summary of Work
The important role of seismic methods in geological CO
2
sequestration is
to locate the subsurface movement of the injected CO
2
and to determine the
amount of CO
2
in place. Given the range of possible temperature and pressure
conditions in an upper sedimentary basin, CO
2
can be a gas, liquid, or
supercritical fluid. The physical properties of the pore fluid in a saturated rock
can have a large impact on the overall seismic wave response. The purpose our
work is to improve the understanding of the effects CO
2
has on the overall wave
response of a saturated medium, under the possible geological sequestration
conditions. Moreover, our work provides an end member study by conducting
measurements under full CO
2
saturation to manifest the possible extreme pore
fluid effects. In this thesis, a series of ultrasonic pulse transmission measurements
were performed on a synthetic and a non-synthetic sample subjected to various
conditions. The samples were subjected to dry, full CO
2
saturation and full water
saturation. Under CO
2
saturated conditions, various pressures and temperatures
were applied in order to yield a gas, liquid, or supercritical fluid phase. All
measurements pertaining to fluid saturation were conducted under constant
differential pressure to ensure pressure effects are eliminated and the resulting
wave variations observed are solely due to pore fluid effects. The samples were
also subjected to petrophysical characterization to assist the understanding and
analysis of the observed wave behaviours.
In this study the synthetic sample used is a porous ceramic rod derived
from an alumina body and the non-synthetic sample is a Berea sandstone that is
188

composed predominantly of quartz grains with minor amounts of clay. The
petrophysical properties of these two samples are quite different with respect to
one another. The porous ceramic rod is extremely porous, very fine-grained, and
contains a small, negligible amount of micro-cracks. The Berea sandstone is two-
thirds less porous than the porous ceramic rod, larger grained, and contains a
significant amount of micro-cracks. The extremely high porosity of the porous
ceramic rod allows for larger pore fluid effects to occur, which reinforces our
intent of providing an end member CO
2
study.
The presence of CO
2
in the pore space is notable from wave velocity
changes in both samples. The introduction of CO
2
into a dry sample yields a
reduction in both P-and S-wave velocities regardless of the prevailing phase state.
Of the three possible CO
2
phases examined over the conditions explored, the
wave velocity reduction is the least for the gas phase. Furthermore, the variations
in CO
2
phase states, or more specifically the variations in CO
2
s physical
properties are also reflected in waveform changes. The behaviour of the wave
velocity change shows a greater dependence on the change in the bulk density of
the sample, which in turn is largely dependent on CO
2
s density, rather than on
the change in bulk modulus. The only exception to this general behaviour is when
CO
2
is in the liquid phase for the Berea sandstone where the change in bulk
modulus effect dominates. Considering the prevailing type of conditions in
geological sequestration projects, the dominating effect of bulk modulus changes
on P-wave velocity is less likely because large changes in bulk modulus only
occur at lower temperatures. These temperatures are still plausible in
sequestration projects but are less likely to occur. Therefore, for rock samples or
any porous media of similar properties as the samples used in our study, a P-wave
velocity decrease indicates that both the pore fluids density and bulk modulus has
increased. Other than the wave velocity of a waveform, the wave amplitude is
also responsive to the variations in CO
2
s physical properties. It appears
whenever the wave velocity decreases, the wave amplitude also decreases.
Therefore, the wave amplitude strength or wave attenuation is also sensitive to
changes in CO
2
physical properties.
189

Moreover, over the range of pressure and temperature conditions
investigated, the difference in waveforms (characterized by wave velocity and
wave amplitude) between a gaseous-CO
2
saturated sample and a
liquid/supercritical fluid-CO
2
saturated sample is larger than that displayed
between a liquid-and supercritical fluid saturated sample. Furthermore, the phase
transition from a gas to a liquid/supercritical fluid phase is more distinctive
through greater waveform changes than the liquid to supercritical fluid phase
transition. The latter transition lacks any distinguishing waveform changes to
signify a phase change. These varying wave responses to the different phase
transitions are dependent on the amount of density change encountered; the larger
the CO
2
density change, the larger the signal variation.
For the porous ceramic rod, the comparison of observed CO
2
saturated
results to Biots modelled results shows a good agreement in P-and S-wave
velocities and only a qualitative agreement in P-and S-wave differential
attenuation coefficients. Biots modelled attenuation is always lower than what is
observed. In contrast, Gassmanns modelled wave velocities usually fit poorly
with the observed wave velocities and the formulation does not account for wave
attenuation. For the Berea sandstone the comparison between observed results
and modelled results generally show large discrepancies, indicating that both
models are inadequate in describing the rock behaviour. The observed wave
velocities are always lower than predicted by both Biots and Gassmanns models.
Furthermore, the predicted wave attenuation by Biots model is substantially less,
two orders of magnitude smaller than observed.
Between the porous ceramic rod and the Berea sandstone results, the
waveforms response to CO
2
physical property changes pertaining to the porous
ceramic rod is much greater and shows more sensitivity than the Berea sandstone.
This is a consequence of the much higher porosity of the porous ceramic rod. The
amount of waveform variation of the porous ceramic rod is unrealistic. However
these obtained results provides an end member knowledge of what extremities can
be expected with CO
2
in the pore space. The waveform variations around the
amount displayed by the Berea sandstone are more realistic; however the changes
190

do pertain to full CO
2
saturation. Also it should be remembered that the observed
wave responses of the Berea sandstone are less consistent and show more
complexity than of the porous ceramic rod. The integrity or nature of the sample
have changed over the series of measurement runs and maybe even during a
measurement run. The mechanisms behind the observed wave behaviour are not
fully understood. Therefore interpretations of the lone effect of CO
2
s physical
properties on Berea sandstones observed wave velocity or wave attenuation
should be preceded with caution.

8.2 Contributions of this work
The main intent of our work is to add to the existing laboratory studies
central to understanding the seismic effects CO
2
has as a pore fluid. To our
knowledge, this is the first systematic laboratory study conducted to inspect the
wave velocity and wave attenuation for the purpose of geological CO
2

sequestration as a result of:
1.) full CO
2
saturation,
2.) the different phase states of CO
2
,
3.) definitively separating pore fluid property effects from pore pressure
build-up effects.
The use of the porous ceramic rod sample in conjunction to full CO
2

saturation has allowed us to present possible end member wave behaviours with
CO
2
in the pore space. This information will help constrain the more complex,
but more probable, partially saturated cases in sequestration projects. The
variations in wave velocity in response to a CO
2
phase transition between a gas
and liquid/supercritical fluid is ~5% from ultrasonic frequency laboratory data and
is ~8% from seismic frequency data determined from Gassmanns prediction.
Further away from the phase transition intervals, the difference between a gas and
liquid/supercritical fluid phase is even greater, 6-8% for ultrasonic frequency
waves and more than 12% for seismic frequency waves. It should be noted that
these percentage changes are solely due to pore fluid effects where pressures
effects are eliminated by maintaining a constant differential pressure.
191

Realistically, the overburden pressure (confining pressure) would stay constant
while the pore pressure of the in-situ fluid varies. Therefore as pore pressure
increases with a constant overburden the differential pressure would decrease. As
shown by the measurements on the dry sample, wave velocity decreases with
decreasing differential pressure. Therefore the percentage changes observed and
modelled for the phase transitions and for distinguishing between phase states
would be even greater if pressure effects are considered as well. From these
considerations, differentiating in situ CO
2
phase states from seismic, namely a
gaseous CO
2
from a liquid or a supercritical fluid CO
2,
should be resolvable.
The identification of the different phase states using changes in seismic
reflectivity, travel times, or wave attenuation are important to the quantification of
CO
2
in situ. In order to quantify the amount in place the density of the CO
2
needs
to be known, where the density is dependent on the phase state. Therefore the
ability to identify the existing phase state implies the ability to quantify the
subsurface CO
2
. Furthermore from our study, it was shown that over the
conditions investigated and with the samples we used the difference in waveforms
between a liquid-and a supercritical fluid-CO
2
saturated sample is minimal. This
would not be a great concern to CO
2
quantification since this minor difference in
waveforms is a consequence of the minor difference in the density between these
two phase states.

8.3 Future Work
As mentioned, the purpose of our work is to present an end member study
to observe the extreme wave behaviour possible with CO
2
in the pore space. In
order for pore fluid effects to manifest in the wave responses, an extremely porous
sample was used and full CO
2
saturation was applied. Due to the properties of the
sample, both P-and S-wave velocity decreased with increasing pore pressure. A
different extreme wave response to CO
2
s behaviour change could be attained if a
less stiff sample is used. A sample with a lower bulk modulus will allow the
change in the overall bulk modulus to be greater in response to the varying pore
fluid properties. This change in bulk modulus could be greater than the change in
192

bulk density. Therefore the resulting P-wave velocity would vary directly
proportionally to the variations in bulk modulus. In addition perhaps the
differentiation between a liquid and supercritical fluid CO
2
phase state could
occur since the difference between their respective bulk modulus is greater than
their respective bulk density.
Full CO
2
saturation was applied in our study, however realistically this
situation is unlikely under in situ conditions. To move away from this end
member and to better represent the conditions encountered in geological
sequestration, other possible in-situ fluids such as brine and oil should be
considered in the pore space. Various saturations of CO
2
in various phase states
should be attempted to determine if the different phases of CO
2
or phase transition
can be identified through waveform changes and the resulting implications on
CO
2
quantification.
The wave frequencies applied in our laboratory work are much higher than
those employed in seismic surveys over geological sequestration projects. As a
consequence of physical velocity dispersion, a direct comparison between
laboratory results and field results should not be carried out. Instead the high
frequency results need to be scaled down to low frequencies by rock physics
models. The accuracy in this approach is restricted by how adequate the model is
in describing the actual wave behaviours with the different frequency ranges. If
low frequency laboratory measurements can be conducted then the laboratory
results can be directly compared with surface seismic, VSP, and crosswell
seismic. Therefore without the need of a model to mediate between results, the
possibility of additional errors introduced by the model is reduced.
The same set of ultrasonic pulse transmission measurements should be
conducted on the Berea sandstone again. However, prior to performing the
measurements the sample should be cycled through a series of high pressures
several times. By applying this procedure, the troublesome wave velocity
hysteresis prevalent in our data would be reduced and the resulting data should be
more consistent. Furthermore, this new and improved data should be compared to
193

our results to see whether the characteristic peak in wave attenuation near CO
2

phase transitions seen in our data is still observed.

194

References
Adam, L., Batzle, M., and Brevik, I., 2006, Gassmanns fluid substitution and
shear modulus variability in carbonates at laboratory seismic and ultrasonic
frequencies: Geophysics, 71, F173-F183.
Aki, K., and Richards, P. G., 1980, Quantitative Seismology: Theory and
Methods: W. H. Freeman and Co., San Francisco, USA.
Attia, A. M., 2005, Effects of petrophysical rock properties on tortuosity factor:
Journal of Petroleum Science and Engineering, 48, 185-198
Arts, R., Elsayed, R., Van Der Meer, L., Eiken, O., Ostmo, O., Chadwick, A.,
Kirby, G., and Zinszner, B., 2002, Estimation of the mass of injected CO2 at
Sleipner using time-lapse seismic data: EAGE 64
th
Annual Conference, Paper
H-16.
Bachu, S. and Shaw, J. C., 2004, CO
2
storage in oil and gas reservoirs in Western
Canada: Effects of aquifers, potential for CO
2
- flood enhanced oil recovery
and practical capacity: Proceedings of the 7
th
International Conference on
Greenhouse Gas Control Technologies. Volume 1: Peer-Reviewed Papers and
Plenary Presentations, Vancouver, BC, September 5-9, 2004.
Bakhorji, A. M., 2010, Laboratory Measurements of Static and Dynamic Elastic
Properties in Carbonate: PhD Thesis, University of Alberta.
Bass, J. D., 1995, Elasticity of Minerals, Glasses, and Melts, in T. J. Ahrens, ed.,
Mineral Physics and Crystallography: A Handbook of Physical Constants:
American Geophysical Union, Washington, USA., pp 45-63.
Batzle, M. L., and Wang, Z., 1992, Seismic properties of pore fluids: Geophysics,
57, 1396-1408.
Benson, S. M., and Myer, L., 2002, Monitoring to ensure safe and effective
geologic sequestration of carbon dioxide: IPCC Workshop for Carbon
Capture and Storage 2002, Regina, Canada.
195

Berryman, J. G., 1999, Origin of Gassmanns equation: Geophysics, 64, 1627-
1629.
Biot, M. A., 1956a, The theory of propagation of elastic waves in a fluid-saturated
solid, I lower frequency range: Journal of Acoustical Society of America, 28,
168-178.
Biot, M. A., 1956b, The theory of propagation of elastic waves in a fluid-saturated
solid, II higher frequency range: Journal of Acoustical Society of America,
28, 179-191.
Biot, M. A., and Willis, D. G., 1957, The elastic coefficients of the theory of
consolidation, Journal of Applied Mechanics, 24, 594-601.
Birch, F., 1961, The velocity of compressional waves in rocks to 10 kilobars, Part
2: Journal of Geophysical Research, 66, 2199-2224.
Blair, D. P, 1990, Seismic pulse assessment of cracked and jointed rock:
Engineering Fracture Mechanics, 35, 447-455.
Blencoe, J. G., Cole, D. R., Horita J., and Moline, G.R., 2001, Experimental
Geochemical Studies Relevant to Carbon Sequestration: Proceedings of the
First National Symposium on Carbon Sequestration, National Energy
Technology Laboratory, Washington DC.
Bourbie, T., Coussy, O., and Zinszner, B., 1987, Acoustics of porous media:
Editions Technip, Paris, France.
Bouzidi, Y., 2003, The Acoustic Reflectivity and Transmissivity of Liquid
Saturated Porous Media: Experimental Tests of Theoretical Concepts: PhD
Thesis, University of Alberta.
Bouzidi, Y., and Schmitt, D. R., 2009, Measurement of the speed and attenuation
of the Biot slow wave using a large ultrasonic transducer: Journal of
Geophysical Research, 114, B08201.
Cadoret, T., Marion, D., and Zinszner, B., 1995, Influence of frequency and fluid
distribution on elastic-wave velocities in partially saturated limestones:
Journal of Geophysical Research, 100, 9789-9803.
Chadwick, R. A., Arts, R., Bentham, M., Eiken, O., Holloway, S., Kirby, G. A.,
Pearce, J. M., Williamson, J. P., and Zweigel, P., 2009, Review of monitoring
196

issues and technologies associated with the long-term underground storage of
carbon dioxide: Geological Society, London, Special Publications, 313, 257-
275
Chadwick, R. A., Williams, G., Delepine, N., Clochard, V., Labat, K., Sturton, S.,
Buddensiek, M. L., Lima, A. L., Arts, R., Neele, F., and Rossi, G., 2010,
Quantitative analysis of time-lapse seismic monitoring data at the Sleipner
CO
2
storage operation: The Leading Edge, 29, 170-177.
Cheng, A., Huang, L., and Rutledge, J., 2010, Time-lapse VSP data processing for
monitoring CO
2
injection: The Leading Edge, 29, 196-199.
Christensen, N. I, and Wang, H. F., 1985, The influence of pore pressure and
confining pressure on dynamic elastic properties of Berea Sandstone:
Geophysics, 50, 207-213.
Cole, D. R., Chialvo, A. A., Rother, G., Vlcek, L., and Cummings, P. T., 2010,
Supercritical fluid behaviour at nanoscale interfaces: Implications for CO
2

sequestration in geologic formations: Philosophical Magazine, 90, 2339-
2363.
Daley, T. M., Myer, L. R., Peterson, J.E., Majer, E.L., and Hoversten, G.M.,
2008, Time-crosswell seismic and VSP monitoring of injected CO2 in a brine
aquifer: Environ Geol, 54, 1657-1665.
De Vilbess, J. W., 1980, Wave dispersion and absorption in partially saturated
rocks: PhD. Thesis, Stanford University, California.
Dewar, J., 2001, Rock Physics for the rest of us An informal discussion: CSEG
Recorder, 26, 42-49.
Diallo, M. S., Prasad, M., and Appel, E., 2003, Comparison between experimental
results and theoretical predictions for P-wave velocity and attenuation at
ultrasonic frequency: Wave Motion, 37, 1-16.
Ely, J. F., Haynes, W. M., and Bain, B. C., 1989, Isochoric (P, V
m
, T)
measurements on CO
2
and on (0.982CO
2
+ 0.018N
2
) from 250 to 330 K at
pressures to 35 MPa: Journal of Chemical Thermodynamics, 21, 879-894.
197

Fabry, V. J., Seibel, B. A., Feely, R. A., and Orr, J. C., 2008, Impacts of ocean
acidification on marine fauna and ecosystem processes: ICES Journal of
Marine Science, 65, 414-432.
Falorni, G., Tamburini, A., Novali, F., Ferretti, A., and Young, B., 2010, Multi-
interferogram InSAR Techniques for Monitoring Surface Deformation in
CO
2
Sequestration, AAPG Annual Convention and Exhibition, New Orleans,
Louisiana, April 11-14, 2010.
Fujimoto, Y., Kono, Y., Hirajima, T., Kanagawa, K, Ishikawa, M., and Arima,
M., 2010, P-wave velocity and anisotropy of lawsonite and epidote
blueschists: Constraints on water transportation along subducting oceanic
crust: Physics of the Earth and Planetary Interiors, 183, 219-228.
Gardner, G. H. F., Wyllie, M. R. J., and Droschak, D. M., 1965, Hysteresis in the
velocity-pressure characteristics of rocks: Geophysics, 30, 111-116.
Garrouch, A. A., Ali, L., and Qasem, F., 2001, Using diffusion and electrical
measurements to access tortuosity of porous media: Ind. Eng. Chem. Res, 40,
4363-4369
Gasperikova, E., and Hoversten, G.M., 2006, A feasibility study of nonseismic
geophysical methods for monitoring geologic CO
2
sequestration, The
Leading Edge, 25, 1282-1288
Gasperikova, E., and Hoversten, G.M., 2008, Gravity monitoring of CO
2

movement during sequestration: Model studies: Geophysics, 73, WA105-
WA112.
Gassmann, F., 1951, Elasticity of porous media: Uber die Elastizitat poroser
Medien: Vierteljahrsschrift der Naturforschenden Gesselschaft in Zurich, 96,
1-23.
Geertsma, J., 1961, Velocity-log interpretation: the effect of rock bulk
compressibility: Society of Petroleum Engineers Journal, 1, 235-248.
Gladwin, M. T., and Stacey, F. D, 1974, Anelastic degradation of acoustic pulses
in rocks: Physics of the Earth and Planetary Interiors, 8, 332336.
Gordon, R. B., and Davis, L. A., 1968, Velocity and attenuation of seismic waves
in imperfectly elastic rock: Journal of Geophysical Research, 73, 3917-3935.
198

Gritto, R., Daley T. M., and Myer, L.R., 2004, Joint cross-well and single-well
seismic studies at Lost Hills, California: Geophys. Prospect, 52, 323-339.
Gunter, W. D., and Perking, E., 2001, Geochemical Monitoring of CO
2
Enhanced
Oil Recovery: Proceedings of the NETL Workshop on Carbon Sequestration
Science.
Hare, J.L., Ferguson, J.F., and Aiken, C. L. V., 1999, The 4-D microgravity
method for waterflood surveillance: A model study from the Prudhoe Bay
reservoir, Alaska: Geophysics, 64, 78-87.
Hart, D. J., and Wang, H. F., 1995, Laboratory measurements of a complete set of
poroelastic moduli for Berea sandstone and Indiana limestone, Journal of
Geophysical Research, 17, 741-751
Hart, D. J., and Wang, H. F., 2010, Variations of unjacketed pore compressibility
using Gassmanns equation and an overdetermined set of volumetric
poroelastic measurements, Geophysics, 75, N9-N18.
Hemsing, D. B., 2007, Laboratory determination of seismic anisotropy in
sedimentary rock from the Western Canadian Sedimentary Basin: M.Sc.
Thesis, University of Alberta.
Hill, R., 1952, The elastic behaviour of crystalline aggregate: Proceedings of the
Physical Society, 65, 349-354.
Hovem, J. M., and Ingram, G. D., 1979, Viscous attenuation of sound in saturated
sand: Journal of the Acoustical Society of America, 66, 1807-1812.
Hoversten, G. M., Gritto, R., Washbourne, J., and Daley, T., 2003, Pressure and
fluid saturation prediction in a multicomponent reservoir using combined
seismic and electromagnetic imaging: Geophysics, 68, 1580-1591.
Hovorka, S.D., Collings, D., Benson, S., Mayer, L., Bryer, C., and Cohen, K.,
2005, Update on the Frio Brine Pilot: eight months after injection: National
Energy Technology Laboratory Fourth Annual Conference on Carbon
Capture and Sequestration, Alexandria, Virginia, May 2-5, 2005. GCCC
Digital Publication Series #05-04i, pp. 1-31. retrieved from
http://www.beg.utexas.edu/gccc/bookshelf/Final%20Papers/05-04-
Final%20(NETL%20Papers)/05-04i-Final.pdf on December 2010.
199

Huang, F. H., Li, M. H, Lee, L. L., Starling, K. E., and Chung F. T. H., 1985, An
accurate equation of state for carbon dioxide, Journal of Chemical
Engineering of Japan, 18, 490-496.
IPCC, 2005, In: Metz B., Davidson, O., de Coninck, H.C., Loos, M., and Mayer,
L. A. (eds) IPCC Special Report on CO
2
Capture and Storage, prepared by
Working group III of the International Panel on Climate Change: Cambridge
University Press.
IPCC, 2007, In: Pachauri, R. K., and Reisinger, A., (eds) Climate Change 2007:
Synthesis Report, Contribution of Working Groups I, II and III to the Fourth
Assessment Report of the Intergovernmental Panel on Climate Change,
IPCC, Geneva, Switzerland, pp 109. Retrieved from
http://www.ipcc.ch/publications_and_data/ar4/syr/en/contents.html
Jaya, M. S, Shapiro, S. A., Kristinsdottir, L. H., Bruhn, D., Milsch, H., and
Spangenberg, E., 2010, Temperature dependence of seismic properties in
geothermal rocks at reservoir conditions, Geothermics, 39, 115-123.
Johnson, D. L., and Plona, T. J., 1982, Acoustic slow waves and the consolidation
transition: Journal of Acoustical Society of America, 72, 556-565.
Johnston, D. H., and Toksoz, M. N., 1980, Ultrasonic P- and S-waves attenuation
in dry and saturated rocks under pressure, Journal of Geophysical Research,
85, 925-936.
Jones, T., and Nur, A., 1983, Velocity and attenuation in sandstone at elevated
temperatures and pressures: Geophysical Research Letters, 10, 140-143.
Jones, T., 1986, Wave Propagation in Porous Rocks and Models for Crystal
Structure: PhD. Thesis, Stanford University, California.
Kesselman, P. M., Kotlyarevskii, P. A., and Afanasev, M. M., 1965, The
equation of state of carbon dioxide in the temperature range 273- 4000 K, for
pressures up to 1000 10
5
N/m
2
: Journal of Engineering Physics, 9, 527-532.
Kilmer, N. H., Morrow, N. R., and Pitman, J. K, 1987, Pressure sensitivity of low
permeability sandstones: Journal of Petroleum Science and Engineering, 1,
65-81.
200

Kim, Y., 2007, Equation of State for Carbon Dioxide: Journal of Mechanical
Science and Technology, 21, 799-803.
Klinkenberg, L. J., 1941, The permeability of porous media to liquids and gases:
Drilling and Production Practice, 200-213.
Kongsjorden, H., Karstad, O., and Torp, T. A., 1998, Saline aquifer storage of
carbon dioxide in the Sleipner project: Waste Management, 17, 303-308
Lakes, R., 2009, Viscoelastic materials: Cambridge University Press, New York,
USA.
Lay, T., and Wallace, T. C., 1995, Modern Global Seismology in International
Geophysics Series, Academic Press, San Diego, USA., vol 58.
Lei, X., and Xue, Z., 2009, Ultrasonic velocity and attenuation during CO
2

injection into water-saturated porous sandstones: Measurements using
difference seismic tomography: Physics of the Earth and Planetary Interiors,
176, 224-234
Lemmon, E. W., McLinden, M. O., and Friend, D. G., Thermophysical Properties
of Fluid Systems in NIST Chemistry WebBook, NIST Standard Reference
Database Number 69, Eds. Linstrom, P. J. and Mallard, W. G., National
Institute of Standards and Technology, Gaithersburg MD, 20899,
http://webbook.nist.gov, (retrieved 2008-2011).
Liang, J., Robertson, E. P., and Raterman, K. T., 2003, A Mechanistic Model for
CO
2
sequestration in Tiffany Coal Bed Methane Field: The 2003
International Coal Bed Methane Symposium, USA, May 5-9, 2003.
Lowrie, W., 1997, Fundamentals of Geophysics: Cambridge University Press
Mashinkskii, E., 2005, Experimental study of the amplitude effect on wave
velocity and attenuation in consolidated rocks under confining pressure:
Journal of Geophysics and Engineering, 2, 199-212.
Mavko, G. M., and Nur, A., 1979, Wave attenuation in partially saturated rocks,
Mavko G., Mukerji, T., and Dvorkin, J., 2009, The Rock Physics Handbook:
Cambridge University Press.
201

McCann, C., and Sothcott, J., 1992, Laboratory measurements of the seismic
properties of sedimentary rocks: The Geological Society, London, Special
Publications, 65, 285-297.
McKavanagh, B., and Stacey, F. D.,1974, Mechanical hysteresis in rocks at low
strain amplitudes and seismic frequencies, Physics of the Earth and Planetary
Interiors, 8, 246-250.
Mobarek, S. A. M., 1971, The effect of temperature on wave velocities in porous
rocks: M.Sc. Thesis, University of California.
Molyneux, J. B., and Schmitt, D. R., 2000, Compressional-wave velocities in
attenuating media: A laboratory physical model study: Geophysics, 65, 1162-
1167.
Murphy, W. F. III, 1982, Effects of partial water saturation on attenuation in
Massilon sandstone and Vycor porous glass: Journal of Acoustical Society of
America, 71, 1458-1468.
New England Research, Inc., Tiny Perm II: Portable Air Permeameter: Users
Manual.
Newmark, R., Ramirez, A., and Daily, W., 2002, Monitoring carbon dioxide
sequestration using electrical resistance tomography (ERT): A minimally
invasive method: Proceedings of the Sixth International Conference on
Greenhouse Gas Control Technologies, 353-358.
Nooner, S. L., Eiken, O., Hermanrud, C., Sasagawa, G.S., Stenvold, T., and
Zumberge, M. A., 2007, Constrains on the in situ density of CO2 within the
Utsira formation from time-lapse seafloor gravity measurements:
International Journal of Greenhouse Gas Control, 1, 198-214.
Nur, A.,1971, Viscous phases in rocks and the low velocity zone: Journal of
Geophysical Research, 76, 1270-1277.
Nur, A., and Murphy, W., 1981, Wave velocities and attenuation in porous media
with fluids, Proceedings of the 4
th
International Conference on Continuum
models of discrete systems, Stockholm, 311-327
202

Nur, A., Mavko, G., Dvorkin, J., and Gal, D., 1995, Critical porosity: the key to
relating physical properties to porosity in rocks. In Proceedings of the 65
th

Annual International Meeting Society of Exploration Geophysics, 878, Tulsa.
OConnell, R. J., and Budiansky, B., 1977, Viscoelastic properties of fluid-
saturated cracked solids: Journal of Geophysical Research, 76, 2022-2034.
OHara, S. G., 1985, Influence of pressure, temperature, and pore fluid on the
frequency-dependent attenuation of elastic waves in Berea sandstone,
Physical Review A, 32, 473-488.
Palmer, I. D., and Traviolia, M. L., 1981, Attenuation by squirt flow in under-
saturated gas sands, Geophysics, 45, 1780-1792.
Park, K. G., Choi, H., Park, Y. C., and Hwang, S., 2009, Ultrasonic laboratory
measurements of the seismic velocity changes due to CO
2
injection: EGU
General Assembly 2009, Expanded Abstract.
Pitzer, K. S., and Sterner, S. M., 1994, Equations of state valid continuously from
zero to extreme pressures for H
2
O and CO
2
: Journal of Chemical Physics,
101, 3111-3116.
Purcell, C., Mur, A., Soong, Y., Mclendon, R., Haljasmaa, I. V., and Harbert, W.,
2010, Integrating velocity measurements in a reservoir rock sample from the
SACROC unit with an AVO proxy for subsurface supercritical CO
2
: The
Leading Edge, 29, 192-195.
Qi, X., 2008, Simulation and laboratory measurements of velocity and attenuation
in viscoelastic body: A frequency dependent study with respect to viscosities:
M.Sc. Thesis, University of Alberta.
Rackley, S. A., 2010, Carbon Capture and Storage: Elsevier.
Reuss, A., 1929, Berechnung der fliessgrense von mischkristallen auf grund der
plastizitatbedingung fur einkristalle: Zeitschrift fur Angewandte Mathematik
aus Mechnik, 9, 49-58.
Rickett, J., 2006, Integrated estimation of interval-attenuation profiles:
Geophysics, 71, A19-A23.
Ronstron, B., White, D., Johnson, J., Hawkes, C., Gardner, C., Chalaturyk, R.,
and Whittaker, S., 2009, CO
2
storage monitoring efforts at the Weyburn-
203

Midale Fields, Canada: AAPG/SEG/SPE Hedberg Conference Expanded
abstract, August 16-19.
Saito, H., Nobuoka, D., Azuma, H., Xue, Z., and Tanase, D., 2006, Time-lapse
crosswell seismic tomography for monitoring injected CO
2
in an onshore
aquifer, Nagaoka, Japan: Exploration Geophysics, 37, 30-36
Santos, C. A., Urdaneta, V., Jaimes, G., and Trujillo, L., 2009, Ultrasonic Spectral
and Complexity Measurements on Brine and Oil Saturated Rocks: Rock
Mechanics and Rock Engineering, 43, 351-359.
Sarma, L. P., and Ravikumar, N., 2000, Q-factor by spectral ratio technique for
strata evaluations: Engineering Geology, 57, 123-132.
Schmitt, D. R., and Zoback, M. D., 1989, Poroelastic effects in the determination
of the maximum horizontal principal stress in hydraulic fracturing tests-A
proposed breakdown equation employing a modified effective stress relation
for tensile failure: International Journal of Rock Mechanics and Mining
Sciences & Geomechanics Abstracts, 26, 499-506.
Seibel, B. A., and Walsh, P. J., 2001, Potential Impacts of CO
2
Injection on Deep
Sea Biota: Science, 294, 319-320
Shankland, T. J., and Johnson, P. A., 1993, Elastic wave attenuation and velocity
of Berea sandstone measured in the frequency domain: Geophysical Research
Letters, 20, 391-394
Shi, J., Xue, Z., and Durucan, S., 2007, Seismic monitoring and modelling of
supercritical CO
2
injection into a water-saturated sandstone: Interpretation of
P-wave velocity data: International Journal of Greenhouse Gas Control, 1,
473-480.
Spetzler J., Xue, Z., Saito, H., and Nishizawa, O., 2008, Case story: time-lapse
seismic crosswell monitoring of CO2 injected in an onshore sandstone
aquifer: Geophys. J. Int, 172, 214-225.
Springer, D. S., and Loaiciga, H. A., Cullen, S. J., and Everetta, L. G., 1998, Air
Permeability of Porus Materials Under Controlled Laboratory Conditions:
Ground Water, 36, 558-565.
204

Stein, W. A., 1972, Die Zustandsgleichung fr reine fluide Stoffe: Chemical
Engineering Science, 27, 1371-1382.
Sweatman, R., and McColpin G. R., 2009, Monitoring technology enables long-
term CO
2
geosequestration, E &P Magazine, November issue.
Telford, W. M., Geldart, L. P., and Sheriff, R. E., 1990, Applied Geophysics
Second Edition : Cambridge University Press.
Thurman, E. S, Abousleiman, Y., and Zaman, M., 2002, Acoustical Imaging and
Mechanical Properties of Soft Rock and Marine Sediments: Quarterly
Technical Progress Report #15302R06
Timur, A., 1968, Velocity of compressional waves in porous media at permafrost
temperatures: Geophysics, 33, 584-595.
Timur, A., 1977, Temperature dependence of compressional and shear wave
velocities in rocks: Geophysics, 42, 950-956.
Toksz, M, N., and Johnston, D. H., and Timur, A., 1979, Attenuation of seismic
waves in dry and saturated rocks, I, Laboratory measurements, Geophysics,
44, 681-690.
Tonn, R., 1991, The determination of seismic quality factor Q from VSP data: A
comparison of different computational techniques: Geophysical Prospecting,
45, 87-109

Tosaya, C., and Nur, A., 1982, Effects of diagenesis and clays on compressional
velocities in rocks: Geophysical Research Letters, 9, 5-8.
Tutuncu, A. N., Podio, A. L., and Sharma, M. M., 1994, An experimental
investigation of factors influencing compressional-wave and shear wave
velocities and attenuations in tight gas sandstones, Geophysics, 59, 77-86.
Walsh, J. B, 1966, Seismic wave attenuation in rock due to friction: Journal of
Geophysical Research, 71, 2591-2599.
Wang, Z., and Nur, A. M., 1989, Effects of CO
2
flooding on wave velocities in
rocks with hydrocarbons: Society of Petroleum Engineers. Reservoir
Engineering, 3, 429-436.
205

Wang, Z., Hirsche, W. K., and Sedgwick, G., 1991, Seismic monitoring of water
floods-A petrophysical study: Geophysics, 56, 1991.
Wang, Z., Cates, M. E., and Langan, R. T., 1998, Seismic monitoring of a CO
2

flood in a carbonate reservoir: A rock physics study, Geophysics, 63, 1604-
1617.\
Wark, K. Jr., and Richards, D. E., 1999, Thermodynamics: McGraw-Hill.
Wepfer, W. W., and Christensen, N. I., 1991, Q structure of the Oceanic Crust:
Marine Geophysical Researches, 13, 227-237.
White, R. E., 1992, The accuracy of estimating Q from seismic data: Geophysics,
57, 1508-1511.
White, D., 2009, Monitoring CO
2
storage during EOR at the Weyburn-Midale
field: The Leading Edge, 28, 838-842.

Wilt, M., 2003, Oil reservoir characterization and CO
2
injection monitoring in the
Permian basin with crosswell electromagnetic imaging: Final report DOE
Award Number: DE-FC26-00BC15307, retrieved from
http://www.netl.doe.gov/KMD/cds/disk22/G-
CO2%20&%20Gas%20Injection/BC15307.pdf on Dec 2010
Winkler, K., 1979, The effects of pore fluids and frictional sliding on seismic
attenuation: PhD. Thesis, Stanford University, California.
Winkler, K., and Nur, A., 1982, Seismic attenuation: effects of pore fluids and
frictional sliding: Geophysics, 47, 1-15.
Winkler, K., and Plona, T., 1982, Technique for measuring ultrasonic velocity and
attenuation spectra in rocks under pressure: Journal of Geophysical Research,
87, 776-780.
Winkler, K. W., 1985, Dispersion analysis of velocity and attenuation in Berea
sandstone: Journal of Geophysical Research, 90, 183
Wolf, K., 2010, Laboratory Measurements and Reservoir Monitoring of Bitumen
Sand Reservoirs: PhD. Thesis, Stanford University, California.
Wu, Y., Pruess, K., and Persoff, P., 1998, Gas Flow in Porous Media with
Klinkenberg Effects: Transport in Porous Media, 32, 117-137.
206

Wyllie, M. R. J., Gregory, A. R., and Gardner, G. H. F, 1958, An experimental
investigation of factors affecting elastic wave velocities in porous media:
Geophysics, 23, 459-493
Verwer, K., Braaksma, H., and Kenter, J. A. M., 2008, Acoustic properties of
carbonates: Effects of rock texture and implications for fluid substitution:
Geophysics, 73, B51-B66
Voigt, W., 1928, Lehrbuch der Kristallphysik: Teubner.
Xue, Z., Oshumi, T., and Koide, H., 2002, Laboratory measurements of seismic
wave velocity by CO
2
injection in two porous sandstones: Proceedings of the
Sixth International Conference on Greenhouse Gas Control Technologies,
359-364.
Xue, Z., and Oshumi, T., 2004, Seismic wave monitoring of CO
2
migration in
water-saturated porous sandstone: Exploration Geophysics, 35, 25-332.
Xue, Z., Oshumi, T., and Koide, H., 2005, An experimental study on seismic
monitoring of CO
2
flooding in two sandstones: Energy, 30, 2352-2359.
Xue, Z., Tanase, D., Saito, H., Nobuoka, D., and Watanabe, J., 2005, Time-lapse
crosswell seismic tomography and well logging to monitor the injected CO2
in an onshore aquifer, Nagaoka, Japan: 75
th
Annual International Meeting of
the Society of Exploration Geophysicists Annual Meeting, Expanded
Abstracts 24:1433
Xue, Z., and Lei, X., 2006, Laboratory study of CO
2
migration in water-saturated
anisotropic sandstone, based on P-wave velocity imaging: Exploration
Yin, C. S., Batzle, M. L., and Smith, B. J., 1992, Effects of partial liquid gas
saturation on extensional wave attenuation in Berea sandstone: Geophysical
Research Letters, 19, 1399-1402.
Yin, H., 1993, Acoustic velocity and attenuation of rocks, isotropy, intrinsic
anisotropy, and stress-induced anisotropy: Ph.D. thesis, Stanford University,
California.
207

Zadler, B. J., Le Rousseau, J. H. L., Scales, J. A., and Smith, M. L., 2004,
Resonant ultrasound spectroscopy: theory and application: Geophysical
Journal International, 156, 154-169.
Zhan, X., Schwartz, L. M., Toksoz, M. N., Smith, W. C., and Morgan, F. D.,
2010, Pore-scale modeling of electrical and fluid transport in Berea
sandstone: Geophysics, 75, F135
Zhang, Z. G., and Duan, Z. H., 2005b, An optimized molecular potential for
carbon dioxide; Journal of Chemical Physics, 122, 214507.
Zhu, X., and Meng, Z. (Schwartz, M. (eds)), 2002, Actuators, piezoelectric
ceramic, functional gradient: Encyclopedia of Smart Material: John Wiley
and Sons.

208

Appendix A
Observed and Modelled Results

For each sample examined in this study, the samples were measured under
dry, CO
2
saturated and water saturated conditions. Here, the full set of
normalized P- and S-wave waveforms, the observed wave velocities, modelled
Gassmann and Biots wave velocities from all measurement conditions, and the
observed and modelled differential attenuation coefficient under CO
2
saturation
are shown including the conditions that were mentioned but not shown in chapter
6 and chapter 7. In addition, the computed amplitude spectra used in determining
the differential attenuation coefficients for the CO
2
saturated P-and S-waves are
also given here.

A.1 Porous Ceramic Rod

Sample
Mass
(g)
Bulk
volume
(cm
3
)
Bulk
density
(g/cm
3
)
Grain
volume
(cm
3
)
Grain
density
(g/cm
3
)
Porosity
(%)
Modal
Pore
size
(m)
Air
permeability
(mD)
Tortuosity
Porous
Ceramic
Rod
39.36 25.33 1.55 10.63 3.70 58.4 2.06 96.94 1.73
Table A.1: Petrophysical properties determined of the porous ceramic rod.

209

A.1.1 Dry Condition Results
The porous ceramic rod under a dry condition (pore space under
vacuum), was measured at room temperature, T = 23C, while confining pressure
varied through both pressurization and depressurization. The dry sample was also
measured while being heated where a constant confining pressure of 10 MPa was
applied. Here, the observed waveforms and the wave velocities from these two
dry measurement runs are given.
a) b)
Figure A.1: Normalized (a) P-waves (b) and S-waves waveforms obtained under dry
conditions at T = 23C plotted as a function of confining pressure displayed as coloured
amplitudes for the porous ceramic rod.

a) b)

Figure A.2: Normalized (a) P-waves (b) and S-waves waveforms obtained under heated dry
conditions at a constant confining pressure of 10 MPa, plotted as a function of temperature
and displayed as coloured amplitudes for the porous ceramic rod.

210

a)
b)
Figure A.3: Dry P-wave (a) and S-wave (b) velocities of the waveforms obtained under dry
conditions at T = 23C as a function of confining pressure with their respective errors for the
porous ceramic rod.

211

a)
b)

Figure A.4: (a) P-waves (b) and S-wave velocities obtained under heated dry conditions at a
constant confining pressure of 10 MPa, plotted as a function of temperature with their
respective errors.

212

T = 23C
Confining
Pressure
(MPa)
Vp (m/s) Vs (m/s)
5 3661 26 2297 11
7.5 3664 26 2303 12
10 3666 26 2306 12
12.5 3669 26 2309 12
15 3672 26 2310 12
17.5 3672 26 2312 12
20 3673 26 2312 12
22.5 3674 26 2313 12
25 3679 26 2315 12
27.5 3682 26 2315 12
30 3685 26 2316 12
32.5 3688 26 2316 12
35 3685 26 2316 12
37.5 3688 26 2318 12
40 3690 26 2318 12
37.5 3685 26 2319 12
35 3688 26 2319 12
32.5 3688 26 2318 12
30 3690 26 2318 12
27.5 3685 26 2318 12
25 3685 26 2316 12
22.5 3685 26 2316 12
20 3682 26 2316 12
17.5 3682 26 2316 12
15 3680 26 2314 12
12.5 3682 26 2313 12
10 3682 26 2313 12
7.5 3674 26 2311 12
5 3672 26 2308 12
Table A.2: Dry P-and S-wave velocities of the porous ceramic rod at T = 23C under various
confining pressures.

213

Pc = 10 MPa
Temperature
(C)
Vp (m/s) Vs (m/s)
23.3 3657 26 2295 11
25.2 3657 26 2294 11
29.2 3652 26 2292 11
30.1 3652 26 2291 11
33.1 3649 26 2289 11
34.5 3649 26 2288 11
36.7 3644 26 2287 11
38.5 3641 25 2285 11
39.7 3641 25 2283 11
41.1 3639 25 2282 11
42.1 3636 25 2281 11
Table A.3: Dry P-and S-wave velocities of the porous ceramic rod at Pc = 10 MPa while
under various temperatures.

A.1.2 CO
2
saturated Results
In this section, provided are the observed waveforms, the observed wave
velocities, Gassmann and Biots modelled wave velocities, the observed
amplitude spectra for all waveforms, the observed differential attenuation
coefficient, and the Biots modelled differential attenuation coefficient for all CO
2

saturated measurements that were conducted in this study for the porous ceramic
rod. The observed and modeled results pertaining to the constant temperature
runs will be given first, and is followed by the results of the constant pore
pressure runs.

Constant Temperature Runs
Constant temperature measurements for the porous ceramic rod were
conducted at 23 C, 28C, 40C and 45C while pore pressure varied.
214

a) b)
c) d)

e) f)
g) h)

Figure A.5: Normalized P- and S-wave waveforms of the CO
2
saturated porous ceramic rod
displayed as a function of pore pressures collected during the constant temperature runs of
a) and b) T = 23C, c) and d) T = 28 C, e) and f) T = 40C, and g) and h) T = 45C. A
215

a)
b)

Figure A.6: P-wave (a) and S-wave (b) velocities with their respective errors measured of the
porous ceramic rod saturated with CO
2
for all constant temperature runs and when the

216

T = 23 C
Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 3653 26 2295 11
3 3638 25 2286 11
4 3630 25 2279 11
5 3612 25 2268 11
6 3568 25 2239 11
7.5 3427 24 2141 11
10 3417 24 2131 11
12.5 3408 24 2123 11
15.05 3403 24 2120 11
17.5 3398 24 2111 11
20 3395 24 2108 11
22.5 3394 24 2102 11
25 3392 24 2100 10
Table A.4: CO
2
saturated P-and S-wave velocities of the porous ceramic rod for the T = 23
C constant temperature run, while under a constant differential pressure of 15 MPa.

T = 28 C
Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 3632 25 2281 11
3 3619 25 2274 11
4 3612 25 2267 11
5 3594 25 2255 11
6 3516 25 2206 11
7 3431 24 2139 11
8 3426 24 2133 11
9 3424 24 2125 11
10 3418 24 2123 11
12.5 3401 24 2120 11
15 3392 24 2112 11
17.5 3392 24 2109 11
20 3389 24 2103 11
22.5 3389 24 2100 10
25 3387 24 2096 10
Table A.5: CO
2
saturated P-and S-wave velocities of the porous ceramic rod for the T = 28C
constant temperature run, while under a constant differential pressure of 15 MPa.

217

T = 40 C
Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 3627 25 2274 11
3 3619 25 2268 11
4 3606 25 2262 11
5 3596 25 2257 11
6 3586 25 2247 11
8 3529 25 2234 11
9 3468 24 2169 11
10 3433 24 2147 11
12.5 3417 24 2132 11
15 3412 24 2127 11
17.5 3412 24 2122 11
20 3410 24 2115 11
22.5 3410 24 2111 11
25 3405 24 2111 11
Table A.6: CO
2

T = 45 C
Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 3627 25 2273 11
5 3599 25 2257 11
7.5 3563 25 2229 11
10 3490 24 2180 11
12.5 3433 24 2142 11
15 3415 24 2129 11
17.5 3410 24 2123 11
20 3405 24 2120 11
22.5 3405 24 2115 11
25 3403 24 2111 11
Table A.7: CO
2
218

a)
b)
Figure A.7: Observed and Gassmanns modelled P- wave (a) and S-wave (b) CO
2
saturated
velocities for T = 23C, 28C, 40C, 45C constant temperature runs. The observed wave
velocities are shown with their respective errors.

219

a)
b)
Figure A.8: Observed and Biots modelled P- wave (a) and S-wave (b) CO
2
saturated
velocities for T = 23C, 28C, 40C, 45C constant temperature runs. The observed wave
velocities are shown with their respective errors.
220

a) b)
e) f)
c) d)
g) h)

Figure A.9: The amplitude spectra of the collected P- and S-waves of the CO
2
saturated
porous ceramic rod for a) and b) T = 23C, c) and d) T = 28C, e) and f) T = 40C, and g)
and h) T = 45C constant temperature runs as a function of frequency. The prevailing CO
2

phase states of the signals are determined from the pressure and temperature conditions
applied. The reference spectrum is of the signal collected at T = 23C and Pp = 2 MPa of
CO
2.

221

a) b)
c) d)

e) f)
g) h)

Figure A.10: Measured P- and S-wave differential attenuation coefficient of the CO
2

saturated porous ceramic rod determined for a) and b) T = 23C, c) and d) T = 28C, e) and
f) T = 40C, and g) and h) T = 45C constant temperature runs as a function of frequency
and pore pressure.
222

a) b)
c) d)
e) f)
g) h)

Figure A.11: Biots modelled P- and S-wave differential attenuation coefficient of the CO
2

saturated porous ceramic rod determined for a) and b) T = 23C, c) and d) T = 28C, e) and
f) T = 40C, and g) and h) T = 45C constant temperature runs as a function of frequency
and pore pressure.
223

Constant Pore Pressure Runs
Constant pore pressure runs for the porous ceramic rod were conducted at
P
p
= 7 MPa, 10 MPa, and 25 MPa while temperature varied.

a) b)
c) d)
e) f)
Figure A.12: Normalized P- and S-wave waveforms as a function of temperature collected
P
P
= 10 MPa and e)
and f) P
P
224

P
P
= 7 MPa
Temperature
(C)
Vp (m/s)
Temperature
(C)
Vs (m/s)
22.6 3426 24 22.4 2133 11
23 3426 24 22.9 2133 11
23.9 3426 24 24.5 2135 11
25.2 3426 24 25.5 2137 11
26.1 3426 24 26.4 2139 11
27.1 3428 24 27.5 2140 11
27.8 3431 24 28.1 2140 11
28.8 3426 24 29.2 2142 11
30 3435 24 29.6 2142 11
30.5 3442 24 30.8 2154 11
32.1 3456 24 31.8 2159 11
33.3 3490 24 33.7 2190 11
35.6 3551 25 35.2 2216 11
37.3 3568 25 37.6 2236 11
37.9 3568 25 39.3 2236 11
39.1 3566 25 41.2 2236 11
41.5 3563 25 43 2237 11
42.8 3568 25 44.4 2237 11
44.5 3568 25 45.8 2236 11
45.6 3568 25 47.2 2235 11
47.3 3566 25 48.9 2235 11
48.8 3571 25 50.8 2235 11
50.9 3566 25 - - -
Table A.8: CO
2
saturated P-and S-wave velocities of the porous ceramic rod for the Pp = 7
MPa constant pore pressure run, while under a constant differential pressure of 15 MPa.

225

P
P
= 10 MPa
Temperature
(C)
Vp (m/s)
Temperature
(C)
Vs (m/s)
24.3 3421 24 24 2126 11
25.1 3421 24 24.8 2126 11
25.8 3421 24 26.3 2126 11
26.7 3421 24 27.2 2127 11
27.7 3424 24 28.2 2128 11
28.8 3424 24 29.3 2130 11
29.8 3421 24 30.4 2130 11
30.9 3424 24 31.4 2130 11
32.2 3424 24 32.6 2131 11
33.3 3424 24 33.8 2135 11
34.3 3421 24 35.7 2135 11
35.1 3424 24 37 2136 11
37.1 3424 24 37.7 2143 11
37.7 3428 24 40.3 2146 11
40.3 3433 24 - - -
Table A.9: CO
2

226

P
P
= 25 MPa
Temperature
(C)
Vp (m/s)
Temperature
(C)
Vs (m/s)
22.3 3398 24 22.3 2101 11
23 3394 24 22.7 2099 10
23.4 3394 24 23.8 2100 10
24.4 3394 24 24.9 2098 10
25.5 3394 24 26.1 2098 10
26.6 3392 24 27.1 2098 10
27.5 3392 24 28 2099 10
28.6 3394 24 29.1 2100 10
29.7 3389 24 30.3 2101 11
31 3389 24 31.7 2101 11
32.2 3394 24 32.9 2100 10
33.4 3398 24 33.9 2100 10
34.5 3398 24 35.2 2103 11
35.7 3401 24 36.4 2104 11
36.8 3401 24 37.4 2105 11
37.9 3398 24 38.5 2108 11
39 3401 24 39.5 2111 11
39.9 3398 24 40.4 2110 11
40.9 3398 24 41.4 2110 11
41.8 3405 24 42.1 2111 11
42.5 3401 24 42.9 2110 11
43.3 3405 24 43.6 2111 11
44 3403 24 44.3 2110 11
44.6 3405 24 44.9 2110 11
45.2 3403 24 45.5 2109 11
Table A.10: CO
2

227

a)
b)
Figure A.13: P- wave (a) and S-wave (b) velocities with their respective errors measured of
the sample saturated with CO
2
for all constant pore pressure runs and when the sample is
not saturated (black dot at Pp = 0MPa).
228

a)
b)
Figure A.14: Observed and Gassmanns modelled (a) P-wave and (b) S-wave CO
2
saturated
velocities for P
P
= 7 MPa, P
P
= 10MPa, and P
P
= 25 MPa constant pore pressure runs. The
observed wave velocities are shown with their respective errors.

229

a)
b)
Figure A.15: Observed and Biots modelled (a) P-wave and (b) S-wave CO
2
saturated
velocities for P
P
=7 MPa, P
P
= 10MPa, and P
P

230

a) b)
e) f)
c) d)
Figure A.16: The computed amplitude spectra of the collected P- and S-wave of the CO
2

saturated porous ceramic rod for a) and b) P
P
=7 MPa, c) and d) P
P
=10 MPa, and e) and f)
P
P
=25 MPa constant temperature runs as a function of frequency. The prevailing CO
2

phase states of the signals are determined from the pressure and temperature conditions
applied. The reference spectrum is of the signal collected at T = 23C and Pp = 2 MPa of
CO
2
.

231

c) d)
e) f)
a) b)
2

saturated porous ceramic rod determined for a) and b) P
P
P
= 10 MPa,
and e) and f) P
P
= 25 MPa constant temperature runs as a function of frequency and pore
pressure.

232

c) d)
e) f)
a) b)
2

saturated porous ceramic rod determined for a) and b) P
P
P
= 10 MPa,
and e) and f) P
P
pressure.

233

A.1.3 Water Saturated Results
The water saturated porous ceramic rod was only measured at T = 23C
as a function of pore pressure while under a constant differential pressure of 5
MPa. The resulting observed waveforms and the wave velocities are provided
here.

a) b)
Figure A.19: Normalized P- (a) and S-wave (b) water saturated waveforms as a function of
pore pressure of the porous ceramic rod. A constant differential pressure of 5 MPa was
maintained throughout the measurement run.
234

a)
b)
Figure A.20: P- (a) and S-wave (b) velocities with their respective errors measured of the
water saturated porous ceramic rod under various pore pressures at T = 23C and when the

235

Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
5 3358 24 1812 9
10 3369 24 1811 9
15 3371 24 1810 9
20 3376 24 1810 9
25 3381 24 1811 9
30 3385 24 1809 9
35 3387 24 1808 9
40 3392 24 1810 9
45 3396 24 1809 9
50 3401 24 1807 9
Table A.11: Water saturated P-and S-wave velocities of the porous ceramic rod at T = 23C
for varying pore pressures, while under a constant differential pressure of 5 MPa.

A.2 Berea Sandstone

Sample Mass
(g)
Bulk
volume
(cm
3
)
Bulk
density
(g/cm
3
)
Grain
volume
(cm
3
)
Grain
density
(g/cm
3
)
Porosity
(%)
Modal
Pore
size
(m)
Air
permeability
(mD)
Tortuosity
Berea
Sandstone
43.24 22.02 2.15 17.79 2.45 19.0 11.33 237.65 3.9
a

Table A.12: Petrophysical properties determined of the Berea sandstone.
a
tortuosity value
taken from Garrouch et al (2001)

A.2.1 Dry Condition Results
The Berea sandstone under a dry condition (pore space under vacuum),
was measured twice at room temperature, T = 23C, while confining pressure
varied through both pressurization and depressurization. Here, the observed
waveforms and the wave velocities from these two dry measurement runs are
given.

236

a) b)
Figure A.21: Normalized (a) P- wave and (b) S-wave waveforms obtained under dry
amplitudes for the Berea sandstone. This is the first dry measurement run and was obtained
prior to running the series of CO
2
measurements.

a) b)

Figure A.22: Normalized (a) P- wave and (b) S-wave waveforms obtained under dry
amplitudes for the Berea sandstone. This is the second dry measurement run and was
obtained after the running the series of CO
2
measurements.

237

a)
b)
Figure A.23: Dry P-wave (a) and S-wave (b) velocities of the waveforms obtained under dry
conditions at T = 23C as a function of confining pressure with their respective errors of the
two runs conducted for the Berea sandstone.

238

T = 23C First run
Confining
Pressure
(MPa)
Vp (m/s) Vs (m/s)
5 3202 22 2041 10
7.5 3379 24 2138 11
10 3510 25 2207 11
12.5 3616 25 2268 11
15 3693 26 2310 12
17.5 3754 26 2344 12
20 3801 27 2375 12
22.5 3835 27 2397 12
25 3870 27 2418 12
27.5 3888 27 2433 12
30 3901 27 2445 12
32.5 3918 27 2460 12
35 3931 28 2472 12
37.5 3941 28 2479 12
40 3951 28 2489 12
37.5 3948 28 2485 12
35 3947 28 2479 12
32.5 3939 28 2472 12
30 3930 28 2469 12
27.5 3919 27 2459 12
25 3898 27 2443 12
22.5 3883 27 2435 12
20 3849 27 2413 12
17.5 3826 27 2398 12
15 3783 26 2373 12
12.5 3727 26 2340 12
10 3649 26 2296 11
7.5 3512 25 2224 11
5 3360 24 2144 11
Table A.13: Dry P-and S-wave velocities of the Berea sandstone at T = 23C under various
confining pressures of the first measurement run.

239

T = 23C Second run
Confining
Pressure
(MPa)
Vp (m/s) Vs (m/s)
5 3832 27 2381 12
7.5 3932 28 2441 12
10 3969 28
12.5 4003 28 2507 13
15 4033 28 2529 13
17.5 4052 28 2544 13
20 4068 28 2555 13
22.5 4076 29 2563 13
25 4085 29 2573 13
27.5 4089 29 2577 13
30 4094 29 2582 13
32.5 4094 29 2587 13
35 4095 29 2591 13
37.5 4096 29 2593 13
40 4097 29 2596 13
37.5 4100 29 2595 13
35 4103 29 2594 13
32.5 4102 29 2592 13
30 4102 29 2589 13
27.5 4101 29 2584 13
25 4096 29 2580 13
22.5 4092 29 2573 13
20 4083 29 2565 13
17.5 4075 29 2555 13
15 4055 28 2542 13
12.5 4036 28 2529 13
10 4002 28 2504 13
7.5 3961 28 2465 12
5 3882 27 2417 12
Table A.14: Dry P-and S-wave velocities of the Berea sandstone at T = 23C under various
confining pressures of the second measurement run.

A.2.2 CO
2
Saturated Results
The set of observed and modeled CO
2
saturated results that were
provided in section A.1.2 for the porous ceramic rod, will be provided here for the
Berea sandstone. Again, the observed and modeled results pertaining to the
constant temperature runs will be given first, and is followed by the results of the
constant pore pressure runs.

240

Constant Temperature Runs
Constant temperature measurements for the Berea sandstone were
conducted at 23 C, 28C, 40C, 45C, and 55C while pore pressure varied.

c) d)
e) f)
a) b)

241

i) j)
g) h)
Figure A.24: Normalized P- and S-wave waveforms of the CO
2
saturated Berea sandstone
displayed as a function of pore pressures collected during the constant temperature runs of
a) and b) T = 23C, c) and d) T = 28C, e) and f) T = 40C, g) and h) T = 45C, and i) and j)
T = 55 C. A constant differential pressure of 15 MPa was maintained throughout each
measurement run.
242

a)
b)
Figure A.25: Corrected P-wave (a) and S-wave (b) velocities with their respective errors
2
for all constant temperature runs and
when the sample is not saturated from the second dry measurement run (black dot at Pp = 0
MPa).

243

T = 23 C
Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 3996 28 2511 13
3 3977 28 2499 12
4 3952 28 2484 12
5 3927 27 2468 12
6 3854 27 2423 12
7 3854 27 2421 12
8 3857 27 2418 12
9 3858 27 2419 12
11 3861 27 2417 12
13 3864 27 2417 12
15 3868 27 2416 12
17 3868 27 2416 12
19 3871 27 2414 12
21 3878 27 2414 12
23 3878 27 2412 12
25 3875 27 2407 12
Table A.15: CO
2
saturated P-and S-wave velocities of the Berea sandstone for the T = 23C

T = 28 C
Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 4007 28 2522 13
3 3993 28 2515 13
4 3979 28 2505 13
5 3965 28 2492 12
6 3952 28 2484 12
7 3875 27 2432 12
8 3865 27 2416 12
9 3862 27 2409 12
11 3855 27 2397 12
13 3856 27 2394 12
15 3856 27 2392 12
17 3856 27 2390 12
19 3859 27 2389 12
21 3869 27 2392 12
23 3872 27 2393 12
25 3869 27 2390 12
Table A.16: CO
2
244

T = 40 C
Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 4013 28 2522 13
3 3995 28 2514 13
4 3981 28 2506 13
5 3967 28 2500 13
6 3952 28 2490 12
7 3938 28 2480 12
8 3924 27 2471 12
9 3900 27 2459 12
11 3883 27 2441 12
13 3880 27 2436 12
15 3880 27 2435 12
17 3880 27 2434 12
19 3880 27 2432 12
21 3877 27 2428 12
23 3877 27 2427 12
25 3873 27 2422 12
Table A.17: CO
2

T = 45 C
Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 4018 28 2523 13
3 4005 28 2516 13
4 3993 28 2508 13
5 3981 28 2501 13
6 3968 28 2489 12
7 3959 28 2486 12
8 3947 28 2480 12
9 3934 28 2472 12
11 3904 27 2449 12
13 3890 27 2434 12
15 3887 27 2431 12
17 3885 27 2426 12
19 3884 27 2425 12
- - - 2421 12
22 3888 27 2422 12
25 3883 27 2413 12
Table A.18: CO
2
245

T = 55 C
Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 4013 28 2522 13
3 3998 28 2514 13
4 3990 28 2511 13
5 3978 28 2505 13
6 3968 28 2498 12
7 3956 28 2490 12
8 3945 28 2485 12
9 3933 28 2480 12
11 3913 27 2466 12
13 3896 27 2450 12
15 3889 27 2446 12
17 3886 27 2441 12
19 3885 27 2438 12
21 3883 27 2436 12
23 3885 27 2434 12
25 3880 27 2428 12
Table A.19: CO
2

246

a)
b)
Figure A.26: Observed and Gassmanns modelled P- wave (a) and S-wave (b) CO
2
saturated
velocities for T=23C, 28C, 40C, 45C, 55C constant temperature runs. The observed
wave velocities are shown with their respective errors.

247

a)
b)
Figure A.27: Observed and Biots modelled P- wave (a) and S-wave (b) CO
2
saturated
velocities for T = 23C, 28C, 40C, 45C, 55C constant temperature runs. The observed
wave velocities are shown with their respective errors.
248

a) b)
e) f)
c) d)
g) h)

249

i) j)
2

saturated Berea sandstone for a) and b) T = 23C, c) and d) T = 28C, e) and f) T = 40C, g)
and h) T = 45C , and i) and j) T = 55C constant temperature runs as a function of
frequency. The prevailing CO
2
phase states of the signals are determined from the pressure
and temperature conditions applied. The reference spectrum is of the signal collected at T =
23 C and Pp =2 MPa of CO
2
.

c) d)
a) b)
e) f)
250

c) d)
a) b)
e) f)
i) j)
g) h)
2

saturated Berea sandstone determined for a) and b) T = 23C, c) and d) T = 28C, e) and f)
T = 40C, g) and h) T = 45C, i) and j) T = 55C constant temperature runs as a function of
frequency and pore pressure. The T = 28C S-wave display and index is quite different than
the other S-wave plots and this is attributed to the deterioration of the transducer used
during the run.

251

c) d)
a) b)
e) f)

252

i)
j)
g) h)
2

saturated Berea sandstone determined for a) and b) T = 23C, c) and d) T = 28C, e) and f)
T = 40C, g) and h) T = 45C, and i) and j) T = 55C constant temperature runs as a function
of frequency and pore pressure.

253

Constant Pore Pressure
Constant pore pressure measurements for the Berea sandstone were
conducted at P
P
= 7 MPa, 10 MPa, and 25 MPa while temperature varied.
a) b)
c) d)
e) f)
Figure A.31: Normalized P- and S-wave waveforms as a function of temperature collected
P
=7 MPa, c) and d) P
P
=10 MPa and e)
and f) P
P
254

a)
b)

Figure A.32: Corrected P- wave (a) and S-wave (b) velocities with their respective errors
2
for all constant pore pressure runs and
when the sample is not saturated from the second dry measurement run (black dot at Pp = 0
MPa).

255

P
P
= 7 MPa
Temperature
(C)
Vp (m/s)
Temperature
(C)
Vs (m/s)
23.2 3854 27 23.3 2420 12
24.1 3856 27 24.4 2420 12
25.2 3854 27 24.9 2420 12
26.1 3854 27 25.9 2420 12
27.1 3853 27 26.9 2420 12
28.9 3851 27 28.6 2420 12
29.3 3852 27 29.1 2420 12
30.3 3851 27 29.9 2420 12
31.4 3852 27 31.1 2420 12
32.1 3853 27 31.9 2421 12
34.4 3894 27 34.5 2448 12
36.1 3898 27 35.1 2449 12
37.2 3901 27 36.3 2449 12
38 3901 27 36.9 2450 12
39.4 3898 27 37.9 2448 12
40.6 3899 27 39 2448 12
41.2 3900 27 40 2448 12
41.8 3899 27 41 2447 12
43.2 3897 27 42.1 2446 12
44 3898 27 43 2446 12
45.7 3895 27 44.5 2443 12
46.9 3893 27 45.3 2443 12
49 3891 27 47.1 2441 12
- - - 49 2440 12
Table A.20: CO
2
saturated P-and S-wave velocities of the Berea sandstone for the Pp = 7

256

P
P
= 10 MPa
Temperature
(C)
Vp (m/s)
Temperature
(C)
Vs (m/s)
23.1 3858 27 23.1 2417 12
23.9 3857 27 24.1 2419 12
25.3 3858 27 25.1 2419 12
25.9 3857 27 26.2 2418 12
28.3 3856 27 28.1 2419 12
29 3856 27 29.2 2418 12
30.3 3856 27 30 2417 12
31.1 3855 27 31.2 2418 12
32.3 3855 27 32 2417 12
32.9 3856 27 33.2 2418 12
34.3 3854 27 34 2417 12
35.1 3855 27 35.4 2418 12
36.5 3857 27 36.2 2417 12
37.2 3856 27 37.4 2417 12
38.2 3857 27 37.9 2418 12
38.9 3857 27 39.2 2417 12
40.1 3858 27 39.9 2418 12
41.2 3859 27 41.3 2419 12
42.2 3858 27 41.9 2419 12
43 3858 27 43.2 2419 12
44.1 3859 27 43.9 2419 12
45.1 3863 27 46 2420 12
46.4 3863 27 47.4 2420 12
47.2 3863 27 48.2 2421 12
48.4 3863 27 49.6 2422 12
49.4 3865 27 50.3 2421 12
50.5 3866 27 51 2422 12
Table A.21: CO
2

257

P
P
= 25 MPa
Temperature
(C)
Vp (m/s)
Temperature
(C)
Vs (m/s)
23.2 3868 27 24.1 2403 12
23.6 3871 27 23.4 2403 12
24.4 3868 27 24.5 2403 12
24.9 3868 27 25.2 2404 12
26.7 3867 27 26.5 2404 12
27.9 3864 27 28 2401 12
29 3861 27 28.8 2400 12
30 3854 27 29.9 2398 12
30.9 3854 27 31 2397 12
32.5 3847 27 31.9 2394 12
33.7 3844 27 34.2 2392 12
34.8 3844 27 34.9 2390 12
36 3840 27 35.8 2390 12
36.9 3834 27 37.1 2387 12
39.3 3827 27 39.2 2383 12
41 3820 27 41.3 2382 12
42.5 3820 27 42.3 2379 12
43 3817 27 43.1 2378 12
44.2 3817 27 44.1 2379 12
45.1 3813 27 45.3 2375 12
46 3810 27 46.3 2375 12
47.1 3807 27 47.3 2374 12
47.9 3807 27 48.1 2374 12
49.3 3803 27 49 2373 12
50.1 3800 27 50.3 2371 12
Table A.22: CO
2

258

a)
b)

Figure A.33: Observed and Gassmanns modelled (a) P-wave and (b) S-wave CO
2
saturated
velocities for P
P
=7 MPa, P
P
=10MPa, and P
P
=25 MPa constant pore pressure runs. The
259

a)
b)

Figure A.34: Observed and Biots modelled (a) P-wave and (b) S-wave CO
2
saturated
velocities for P
P
= 7 MPa, P
P
= 10MPa, and P
P
260

a) b)
e) f)
c) d)
2

saturated Berea sandstone for a) and b) P
P
P
P
= 25 MPa constant temperature runs as a function of frequency. The prevailing CO
2
phase
states of the signals are determined from the pressure and temperature conditions applied.
The reference spectrum is of the signal collected at T = 23C and Pp = 2 MPa of CO
2
.

261

a) b)
e) f)
c) d)
2

saturated Berea sandstone determined for a) and b) P
P
P
= 10 MPa, and
e) and f) P
P
pressure.

262

a) b)
c) d)
e) f)
2

saturated Berea sandstone for a) and b) P
P
P
P
= 25 MPa constant temperature runs as a function of frequency and pore pressure.

263

A.2.3 Water Saturated Results
The water saturated Berea sandstone was only measured at T = 23C as a
function of pore pressure while under a constant differential pressure of 15 MPa.
The resulting observed waveforms and the wave velocities are provided here.

a) b)
Figure A.38: Normalized P- (a) and S-wave (b) water saturated waveforms as a function of
pore pressure of the Berea sandstone. A constant differential pressure of 15 MPa was
maintained throughout the measurement run.

264

a)
b)
Figure A.39: P- (a) and S-wave (b) velocities with their respective errors measured of the
water saturated Berea sandstone under various pore pressures at T = 23 C and when the

265

Pore
Pressure
(MPa)
Vp (m/s) Vs (m/s)
2 3996 28 2344 12
3 3999 28 2345 12
4 3999 28 2345 12
5 3999 28 2346 12
6 4003 28 2346 12
7 4003 28 2346 12
8 4007 28 2347 12
9 4007 28 2347 12
11 4010 28 2348 12
13 4014 28 2348 12
15 4014 28 2348 12
17 4018 28 2349 12
19 4022 28 2349 12
21 4022 28 2350 12
23 4025 28 2350 12
25 4025 28 2349 12
Table A.23: Water saturated P-and S-wave velocities of the Berea sandstone at T = 23C for
varying pore pressures, while under a constant differential pressure of 15 MPa.

Schmitt CMC-NCE Contributions May2013

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Schmitt CMC-NCE Contributions May2013

Загружено:

Авторское право:

Доступные форматы

Please cite this article in press as: Njiekak, G., et al.

, Douglas R. Schmitt, Helen Yam, Randolf S. Kofman

C). The sam-

C. Considering all the different

C) with the pore pressure varying

C) and high temperature (T =50

C) while the curve

C. For each curve, blue open symbols represent velocities in

C, the largest drop

C: (a) sample M1, (b) sample M2.

C: (a) sample V1, (b) sample V2.

C (Fig. 17b), both R

C example (Fig. 17a), both reection co-efcients retain

. This analysis, known as the

. This yields the following:

Specific heat capacity of the sample in

Area of the sample

is the initial temperature of the conductor.

Convection thermal resistance

to a target temperature is:

Вам также может понравиться