detect abi lty analysis

Upload: zacarias-suasnabar

Post on 14-Apr-2018

236 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/27/2019 Detect Abi Lty Analysis

    1/16

    DETECTABILITY ANALYSIS OF MISCIBLE CO2 IN A CCS/EOR

    OIL RESERVOIR

    Julin L. Gmez1, Claudia L. Ravazzoli1

    1: CONICET and Facultad de Ciencias Astronmicas y Geofsicas, Universidad Nacional de La Plata.

    [email protected], [email protected]

    RESUMEN

    En este trabajo se analiza la respuesta ssmica de un modelo simplicado para un reservoriocarbontico de petrleo en el cual se inyecta dixido de carbono (CO2) con nes de recuperacin

    asistida (EOR) y almacenamiento de CO2 (CCS). El objetivo esta centrado en analizar la detectabilidad

    ssmica del CO2, considerando efectos de solubilidad del dixido de carbono en petrleo y la

    existencia de un umbral de saturacin por encima del cual se forma una fase de CO2 libre. Para

    este anlisis se consideran en las rocas estados variables de saturacin y espesor de CO2 con

    distribuciones de uidos tipo patchy, usando modelos poroelsticos estraticados y herramientas

    de la fsica de rocas. El mtodo de matrices propagadoras permite calcular la reectividad ssmica

    generalizada del modelo en el dominio de las frecuencias del cual se obtienen los sismogramas

    sintticos correspondientes. Luego, con el n de evaluar cuantitativamente si las variaciones de

    los parmetros mencionados producen cambios detectables en la respuesta ssmica, se utiliza el

    operador clsico nrms como mtrica de repetibilidad. Se muestran resultados obtenidos utilizando

    parmetros del campo Weyburn (Canad), los que son de inters tanto en el monitoreo dealmacenamientos de CO2 como en el campo de la recuperacin asistida.

    INTRODUCTION

    Carbon dioxide (CO2) is a widely recognized global warming agent, and much recent activity

    has focused on nding ways to reduce, eliminate, or sequester CO2 in order to minimize its impact

    on the environment. In addition, miscible CO2 oods have become increasingly important as an

    EOR (enhanced oil recovery) technique for recovering residual or bypassed oil, and as an effective

    mechanism for geologic sequestration of CO2 (CCS), in which the carbon dioxide is stored

    underground (Meadows, 2008). An understanding of the effects of CO2 on rock-uid systems and

    the ability to map carbon dioxide regions during injection are helpful for improving recovery rates

    and optimizing well locations (Meadows, 2008).

    Application of geophysical tools for monitoring sequestered or injected carbon dioxide

    in geologic reservoirs is a matter of growing importance. The use of seismic information

    is possible due to the marked acoustic contrast between the physical properties of natural

    489

    VIII Congreso de Exploracin y Desarrollo de Hidrocarburos

    IAPG Instituto Argentino del Petrleo y el Gas

  • 7/27/2019 Detect Abi Lty Analysis

    2/16

    IAPG Instituto Argentino del Petrleo y el Gas

    490 VIII Congreso de Exploracin y Desarrollo de Hidrocarburos

    reservoir uids, such as formation water (brine) and oil, and those of carbon dioxide. The use

    of appropriate seismic indicators generates a link between seismic attributes and petrophysical

    properties, composition and state of the rock as well as pore-uid type and in situ physical

    conditions. The characterization of carbon dioxide using reection amplitudes is a topic

    of great interest. The method is based on the changes observed between time lapse seismic

    surveys and their correlation with injected volumes of CO2. Given that carbon dioxide is

    highly soluble in oil reservoirs, it is important to compute the fraction of the injected CO2

    absorbed by oil (which is the base of solubility trapping strategies) and to take into account

    that free carbon dioxide phase can exist. The seismic detection and monitoring of the free

    phase is also of particular importance for CCS purposes. As pointed out by Hassanzadeh et

    al. (2009), when CO2 does not completely dissolves into formation uids, nor is trapped as

    a residual phase, a risk of leakage can arise from the buoyancy of the free CO2 phase (in gas

    or supercritical state).

    With these motivations, in this work we analyze the seismic detectability of CO2 within an oil

    reservoir. We use a matrix propagator method for layered elastic media, combined with Gassmann

    and Hill theories, to compute and analyze the generalized amplitude reection coefcients for

    variable saturation degree and a patchy uid distribution type. Following Carcione et al. (2006),

    we introduce dissolution effects in our computations. The generalized reection coefcients

    are also used to obtain synthetic seismograms in the space-time domain using a reectivity

    method for band-limited seismic sources. To quantify whether the variations in CO2 saturationcan be correlated with detectable changes in the time-lapse seismic response, we use a standard

    repeatability metric such as the nrms.

    For the numerical experiments we consider a geologic framework similar to the Weyburn eld

    (Canada), as described by Brown (2002) and Ma and Morozov (2010). In this reservoir crude oil

    is produced from two carbonate beds and supercritical CO2 is being injected since late 2000 for

    enhanced oil recovery and carbon dioxide sequestration purposes. From these experiments our

    aim is to assess whether time-lapse changes in reectivity can be correlated with variations in CO2

    saturation and if these changes are above current repeatability errors.

    This paper is organized as follows. First we describe the physical model for the solid

    and uid phases, the computation of acoustic properties of the saturated rocks for patchy

    uid distributions and the computation of generalized reectivity using a matrix propagator

    method and synthetic traces. Next we briey describe the Weyburn reservoir and then we

    proceed to our analysis. We generate synthetic traces of the reservoir under an EOR carbon

    dioxide and water injection situation in order to analyze if time-lapse amplitude changes can

    be observed. Then we analyze the generalized reectivity of the reservoir model to investigate

    if CO2 content and miscibility effects will modify the seismic response of the reservoir. Finally

    we simulate and quantify the time-lapse response of the model by changing carbon dioxide

  • 7/27/2019 Detect Abi Lty Analysis

    3/16

    IAPG Instituto Argentino del Petrleo y el Gas 491

    Detectability Analysis of Miscible CO2 in a CCS/EOR Oil Reservoir

    content to further investigate whether reectivity changes due to uid injection are expected

    to be measured on time-lapse surveys.

    MODELING PROCEDURE AND MAIN ASSUMPTIONS

    Our aim is to study the seismic response of the porous reservoir rocks in the Weyburn Midale

    beds (Marly and Vuggy units), which are saturated with oil, brine and variable proportions of

    CO2 (Brown, 2002). This requires the use of rock physics tools to properly take into account the

    physical properties of the rock matrix and the pore saturating uids as well.

    For the study of partially saturated rocks we need to compute the bulk density and elastic

    coefcients (bulk and shear modulus) of the uid saturated medium. However, the mechanical

    response of the saturated rock depends not only on the saturation state but also on the wave

    frequency and spatial distribution of the uids (Mavko and Mukerji, 1998). For the purpose of this

    work, we model the mechanical properties of the saturated rocks considering that the rock matrix

    and the pore uids move in phase. Then, dening an effective bulk modulus for the multiphase

    uids, the bulk modulus of the saturated rock can be obtained using the well-known formulation

    given by Gassmann (1951).

    As pointed out by Mavko and Mukerji (1998), this previous approach is strictly valid only

    when the pore uids are mixed uniformly at very small scales so that the different wave-inducedpressure increments in each uid have time to diffuse and equilibrate during a seismic period. The

    characteristic distance associated to this equilibration process is the so-called diffusion length. The

    presence of uid heterogeneities over scales greater than this length (in which wave induced pore

    pressure gradients do not have time to equilibrate), give rise to a patchy saturation distribution. It

    has been shown that for low frequencies, such as those used in seismic exploration, the effective

    modulus of rock with patches of brine, oil and CO2 of arbitrary geometry can be estimated using

    the Hill (1963) modulus in the form given by Mavko and Mukerji (1998).

    Following these ideas, we assume in this work that the mixture of carbon dioxide, brine and

    reservoir oil at the pore scale can be treated as a single phase effective uid. The dissolution of a

    fraction of CO2 in oil (which turns oil into saturated live oil) and the presence of a remaining free

    part is taken into account in these computations, following the ideas presented by Carcione et al

    (2006) and recently used by Ravazzoli and Gmez (2011). As solubility effects of CO2 in brine are

    negligible (Batzle and Wang, 1992) they are not considered within the scope of the present paper.

    The seismic response of the porous rocks for the different saturation states are computed using a

    uid substitution procedure, assuming that no chemical reaction takes places between the rock

    matrix and its pore uids.

    The density and bulk modulus of the live oil for the given in situ temperature and pressure

  • 7/27/2019 Detect Abi Lty Analysis

    4/16

    IAPG Instituto Argentino del Petrleo y el Gas

    492 VIII Congreso de Exploracin y Desarrollo de Hidrocarburos

    conditions are computed using the empirical relations given by Batzle and Wang (1992). The

    properties of CO2 (density and bulk modulus) for variable temperature and pressure are estimated

    using the Duan et al. (1992) fteen -parameter equation of state.

    For monitoring purposes, it must be bear in mind that the injection and production of uids

    from a reservoir changes the pressure from its initial state. It is well known that most of the

    mechanical and transport properties of cracked or porous rocks generally depend on both the

    conning Pc and pore pressures Pp. To a rst approximation, for low differential pressures (and

    assuming negligible capillary pressure effects), such properties depend on the so-called differential

    pressure Pd (Gangi and Carlson 1996, Zimmerman 1991, Ravazzoli 2001) given by Pd = Pc - Pp.

    Given that the elastic properties of the rock matrix are sensitive to variations in differential

    pressure, we take this effect into account using the polynomial experimental laws given by Brown

    (2002) which are displayed in the next section. This allows us to model the variations in the

    seismic velocities of the uid saturated reservoir rocks before CO2 injection (baseline state) and at

    post-injection states (monitor states) and to analyze the signicance of these effects.

    The computation of the complex P-wave generalized reection coefcient (hereafter denoted

    as Rpp), is performed using propagator matrices. The method, formulated in the space frequency

    domain for monochromatic plane waves, is based on the continuity of the stress components and

    particle displacements at the interfaces within a layered medium considering the set of reected

    and refracted waves at each contact and their interference effects (Kennett, B.L.N., 2009; Tsvankin,

    I., 1995). The matrix propagator method gives a simple and easy-to-implement algorithm to obtainthe resulting generalized reection response versus frequency, incidence angle, layer thickness and

    uid saturation. It must be stressed that, although the modeling procedure is formulated for

    vertically layered media, in this work it is applied to a laterally heterogeneous medium assuming

    that each receiver is located above a set of plane layers of variable thickness according to the

    scheme shown in Figure 3 which is based after Brown (2002).

    Next, following the ideas in Liu and Schmitt (2003), the reection coefcients are also used to

    compute synthetic seismograms in this kind of media using a reectivity method for band limited

    seismic sources. These seismograms are the base of the repeatability analysis performed later using

    the standard nrms operator (Kragh and Christie, 2002). We remark that with the Ricker source

    function employed and the Midale beds velocities, the dimensions of the water, oil and carbon

    dioxide zones are below or near the tunning thicknesses at Marly and Vuggy units.

    THE ROCK PHYSICS MODEL FOR WEYBURN FIELD

    Located in the northern part of the Williston Basin in Saskatchewan, Canada, the Weyburn

    Field was discovered in 1954 and has been water-ooded since the 1960s. More recently, CO 2

  • 7/27/2019 Detect Abi Lty Analysis

    5/16

    IAPG Instituto Argentino del Petrleo y el Gas 493

    Detectability Analysis of Miscible CO2 in a CCS/EOR Oil Reservoir

    injection was started in 2000 to increase the oil recovery rate (Ma and Morozov, 2010). This EOR

    oriented carbon dioxide injection has turn the Weyburn eld the largest land-based CO2 project

    in the world, where technologies and protocols for long term CO2 storage are also been tested and

    developed (White, 2009). In this sense, there is a double interest in Weyburn eld for the academia

    and the industry: as an EOR project and for CCS purposes.

    Production from the Weyburn eld is from Mississippian age carbonates at depths of 1300

    to 1500 meters. The reservoir beds are divided into two main zones, an upper Marly dolostone

    zone and the lower Vuggy limestone zone. The Marly and Vuggy beds constitute a transgression-

    regression sequence deposited in a shallow carbonate shelf environment and form the Midale

    beds of the Mississippian Charles Formation. The petroleum trap is both hydrodynamic and

    stratigraphic (Brown 2002). For more geological description and depositional settings of the

    Weyburn Field, see Churcher and Edmunds (1994), Bunge (2000), and Reasnor (2001).

    The Weyburn carbonate reservoir has been subdivided into the Marly (10m thick) dolostone

    and the Vuggy (20 m thick) limestone. The vertical water injector at the Vuggy unit and the

    horizontal CO2 injector at the Marly layer are depicted in Figure 1-a. The disposition of the water

    and carbon dioxide injectors shown in Figure 1-a are responsible of the nal uid distributions

    that we will simulate later. Figure 1-b presents the geologic model used in this work (without the

    EOR/CCS uid distributions), which is formed of two halfspaces and two plane layers, the Marly

    and Vuggy beds. The halfspaces have the same physical properties of the layers next to them, to

    focus on the response from the 30m thick Midale Beds only (that is, no reection is generated atthe halfspace/layer contacts).

    Figure 1: Sketch of the Midale beds at Weyburn reservoir with the water and CO2 injectors (a) and the geologic units used inthis work (b).

  • 7/27/2019 Detect Abi Lty Analysis

    6/16

    IAPG Instituto Argentino del Petrleo y el Gas

    494 VIII Congreso de Exploracin y Desarrollo de Hidrocarburos

    For the P-wave velocities and densities of the layers, we used the rock parameters listed in

    Table 1.

    LayerDensity (g/

    cc)

    Porosity

    (%)

    Mineral bulkmodulus

    (GPa)

    Matrix bulkmodulus(GPa)

    Matrix shearmodulus

    (GPa)

    Marly 2.87 29 83 16 11.58

    Vuggy 2.71 10 72 36.18 20

    Table 1: Rock parameters for the Midale beds. Porosity values are taken from Ma and Morosov (2010).

    We are aware of the effective pressure polynomial dependence obtained for the sandstones

    bulk ( Kdry ) and shear ( Gdry ) moduli of the dry frame of the Marly and Vuggy beds,

    Kdry=a3Pd

    3+a2Pd2+a1Pd+a0,Gdry=b3Pd

    3+b2Pd2+b1Pd+b0,

    where the ai,b

    iare empirical constants. These relations were studied by Brown (2002)

    from a set of ultrasonic measurements of P and S wave velocities made on different dry rock

    samples under a wide range of confining pressures. For this reason we have considered aconfining pressure of 33 MPa, a baseline fluid pressure equal to 14 MPa and fluid pressures

    of 23 MPa (near injection point) and 8 MPa (near the producing well) in agreement with

    Brown (2002). Next, using the polynomial laws shown above we computed the matrix bulk

    and shear moduli for both pressures and formations. For the numerical applications that

    follow, and since no significant changes in the elastic properties of the rock matrix respect to

    the baseline state were found, we used the averaged moduli for the Marly and Vuggy units,

    which are given in Table 1. A pore pressure of 14 MPa and a temperature of 63 degrees Celsius

    were taken as representative of both baseline and monitor states. At this thermodynamical

    conditions, CO2 is at a supercritical state. More information about the reservoir is found at

    Table 2.

    Oil API CO2 gravity Reservoir Temperature (Celsius)Reservoir pore Pressure

    (Mpa)

    29 1.51 63 14

    Table 2: Oil API , CO2 gravity and in situ pressure and temperature.

  • 7/27/2019 Detect Abi Lty Analysis

    7/16

    IAPG Instituto Argentino del Petrleo y el Gas 495

    Detectability Analysis of Miscible CO2 in a CCS/EOR Oil Reservoir

    Using these parameters, and taking into account that the pore space is occupied by two

    or three phase fluid mixtures, in Figures 2 and 3 we show the modulus of the relative change

    of the compressional wave velocities, densities and P-wave and S-wave impedances, with

    respect to the state of 0% CO2 saturation, versus carbon dioxide saturation within the Marly

    and Vuggy layers. For the brine-oil-CO2 mixture the relative saturations of oil and brine were

    maintained at the ratio of 1:4 following Ma and Morozov (2010).

    The presence of a free gas phase manifests itself as a sharp increase in the relative change

    of P-wave velocity, density and impedances after the critical saturation (Sc). The changes are

    greater in the Marly zone due to its greater porosity. For saturation above Sc, a free gas phase

    starts to develop in the pore space of the host rock, while for saturations below Sc, all the

    injected CO2 goes into solution in oil. For our reservoir model the critical saturation is close

    to 36%. As we will see later, these effects will impact on the reflectivity and detectability of

    the model. In Table 3 we list the physical pore fluid properties.

    We have also calculated the reservoir carbon dioxide properties with the classical Van der

    Waals and Peng and Robinson (1976), equations of state (EOS). The impact of the EOS on

    the final results of this paper was found to be unimportant.

    Figure 2: Relative changes on P-wave velocity and density in the Marly (left column) and Vuggy zones (right column) with respectto the 0% CO2 state as functions of and carbon dioxidesaturation. Miscible and immiscible cases are considered.

  • 7/27/2019 Detect Abi Lty Analysis

    8/16

    IAPG Instituto Argentino del Petrleo y el Gas

    496 VIII Congreso de Exploracin y Desarrollo de Hidrocarburos

    Figure 3: Relative changes in P-wave and S-wave impedances in the Marly (left column) and Vuggy zones (right column) withrespect to the 0% CO2 state as functions of carbon dioxide saturation. Miscible and immiscible cases are considered.

    Figure 4: Depth prole based on Brown (2002) of the EOR taking place in our model of the Weyburn Midale beds. The waterand carbon dioxide injectors are to the left border of the model.

  • 7/27/2019 Detect Abi Lty Analysis

    9/16

    IAPG Instituto Argentino del Petrleo y el Gas 497

    Detectability Analysis of Miscible CO2 in a CCS/EOR Oil Reservoir

    The depth prole of the EOR operations in Weyburn is shown in Figure 4, which is based

    after Figure 1.3 of Brown (2002). In the brine and oil region, oil has a saturation of 53% at Marly

    and 35% at Vuggy units (Ma and Morozov, 2010). In the brine-oil-CO2 zone (red region), brine

    and oil saturations are related by 1:4 as stated previously, with variable CO2 saturation. The

    bottom water ooded zone (blue region) is modeled with 20% oil. The rest of the pore space is

    occupied by brine. The contact between the Marly and Vuggy units, placed at a depth of 10 m, is

    also indicated in Figure 4.

    In the baseline pre-injection state, there is no CO2 yet in the pore space and the Midale beds

    are an oil-brine system where oil has a saturation of 53% at Marly and 35% at Vuggy units.

    Pore Fluid Phase Density (g/cc) Bulk Modulus (GPa)

    Brine 1.05 2.88CO

    20.5357 0.0136

    Oil 0.86 1.54

    Table 3: Physical parameters of the in situ pore uid phases: Brine (as Brown, 2002), CO2 (according to Duan et al., 1992) andOil (according to Batzle and Wang, 1992).

    GENERALIZED REFLECTIVITY: AMPLITUDE VERSUS ANGLE (AVA) AND CO2

    SATURATION (AVS)

    In this section we investigate the generalized reectivity resulting from the combined

    effect of the uid-uid contacts (brine-oi/brine-oil-CO2) and the Marly/Vuggy contact. We

    will show that even though the uid-uid contacts, for the depth EOR/CCS prole given in

    Figure 4, are not visible on the synthetic traces they still give some signal of their presence.

    With this in mind we analyze the generalized reectivity that results from the Weyburn model.

    Figure 5 shows the normal incidence seismic response of the model at the baseline (Figure

    5-a) and monitor (Figure 5-b) stage for the uid geometry of Figure 4. Their difference is also

    shown in Figure 5-c. As we can see, in the monitor case, there is almost no evidence of the

    uid-uid contacts except from an increase of amplitude in the Marly/Vuggy reection near

    the CO2 injector at the left border. The traces corresponding to the difference, between the

    baseline and monitor states, make more evident the changes on pore uids distributions due

    to the injection of CO2 and water into the reservoir. In the monitor case carbon dioxide

    saturation is 80%.

  • 7/27/2019 Detect Abi Lty Analysis

    10/16

    IAPG Instituto Argentino del Petrleo y el Gas

    498 VIII Congreso de Exploracin y Desarrollo de Hidrocarburos

    Figure 5: Baseline (a), monitor (b) and difference (c) traces for the depth EOR prole given in Figure 4.

    Figure 6: (a) Amplitude variation with angle (AVA) and, (b), amplitude variation with CO2 saturation (AVS) formiscible and immiscible cases. In (a) the saturation of carbon dioxide is 80% and in (b) normal incidence is

    considered. In both cases the modulus of the complex potential P-wave reection coefcient (Rpp) is shown.

    The AVA response (Figure 6-a), for a plane wave of 50 Hz, differentiates between miscible

    and immiscible mixtures. But given that the near-offset AVA trend of the Rpp modulus is not

    signicant, and that we are mainly interested in the response due to variable carbon dioxide

    content, we focus our analysis on the normal incidence reectivity as representative of the near-

  • 7/27/2019 Detect Abi Lty Analysis

    11/16

    IAPG Instituto Argentino del Petrleo y el Gas 499

    Detectability Analysis of Miscible CO2 in a CCS/EOR Oil Reservoir

    offset range reection.

    Figure 6-b shows that Rpp is very sensitive to the solubility of CO 2 in oil and to carbon

    dioxide saturation. For saturations above the critical (Sc = 36%), the difference between miscible

    and immiscible cases becomes small. The last also indicates that for CO2 levels greater than 40%

    a simple immiscible approach could be used for modeling purposes. For carbon dioxide levels

    below the Sc, the immiscible case has a stronger reectivity due to the fact that the injected CO2

    in the pore space is in free form, not being absorbed by the saturated in situ oil.

    The plots analyzed point in the direction that a correlation between some measure of

    reectivity and CO2 saturation content can be expected, and possibly retrieved, from seismic

    data. This last result is further investigated in what follows.

    DETECTABILITY ANALYSIS USING THE NRMS METRIC

    Figure 5 shows that a time-lapse response (given by the difference between the baseline and

    a monitor stage) is expected from the uid-uid contacts of the water and carbon dioxide oods

    at Weyburn. We address in this section, whether this time-lapse signal carries some information

    about of the state of the injected CO2, and also, if this information can be expected to be detected

    with current time-lapse registration and processing techniques.

    As a measure of the time-lapse seismic data, we adopt the standard nrms repeatability metric.The nrms metric is generally employed in the industry to measure the post processing of the

    baseline and monitor data. It is a common routine to calculate the nrms above the objective

    reservoir to assess the impact of the post processing stages (such as migration velocity estimation,

    amplitude equalization and residual corrections) in the seismic datum. The nrms has also been

    used in numerical simulations as a way to measure detectability of CO2 migration (Picotti et al.,

    2009). In this work, we will use the nrms metric in a fashion similar to Picotti et al. (2009). We

    will compute the nrms in a time window centered at the objective reservoir as a measure of the

    variation in the seismic data, trace by trace, due solely to the injection of carbon dioxide.

    The nrms is dened as (Kragh and Christie, 2002):

    where n is the number of samples in the time window t2 t

    1and m(t),b(t),x(t) are discrete time

    series. It is usual to call b(t) the baseline trace and m(t) the monitor trace (Miller and Helgerud,

  • 7/27/2019 Detect Abi Lty Analysis

    12/16

    IAPG Instituto Argentino del Petrleo y el Gas

    500 VIII Congreso de Exploracin y Desarrollo de Hidrocarburos

    2009). In case the traces were completely identical (perfect repeatability) the nrms would be

    0%. With the current 4D acquisition and processing techniques, nrms values less than 20% are

    considered excellent (Lumley, 2010). Current development aims to reduce this nrms value, in

    what is beginning to be known as precision 4D seismic (Lumley, 2010).

    Considering the EOR/CCS depth prole already shown in Figure 4, we simulated the nrms

    response of a set of 4 traces (which correspond to receivers located at distances of 0, 3.3, 6.6 and

    10 km from the injectors) for the complete simulation time and for CO2 saturations ranging from

    0% (baseline) to 100% and for immiscible and miscible situations. Normal incidence traces were

    generated with a Ricker time source function with peak frequency at 50 Hz. The registration point

    is located in the rst half space (50m above the Marly bed); more information is shown in Table 4.

    Incidence angle(deg)

    Ricker centralfrequency (Hz)

    Sample interval(ms)

    Simulation time (s)

    0 50 4 0.2

    Table 4: Modeling options for synthetic traces.

    Interestingly, as we see in Figure 7, the nrms curves as a function of saturation display a similar

    behavior to the AVS curve (Figure 6-b). Detectability of the immiscible mixture is greater than

    the miscible case as expected from the reectivity analysis. The presence of free gas in the porespace after the critical saturation makes a jump in the nrms curve which increases the detectability

    of the saturation content of the brine-oil-CO2 region in the reservoir. As measured by the nrms,

    detectability is compromised below the critical saturation, which could make CO2 seismic

    monitoring more difcult to perform in highly miscible situations.

    Near the horizontal carbon dioxide injector (Figures 7-a and 7-b) the nrms values are

    above the 20% level for a wide range of saturations. As we move apart from the CO2 injection

    point, and for the EOR/CCS geometry considered in this work, overall nrms curves starts to

    decrease (Figures 7-c, 7-d) to nally be under the detectability limit of 20% (shown in the plots

    in Figure 7 as a shaded area). However, some further increase in nrms can be observed where

    the brine-oil-CO2 region increases its thickness (Figure 7-d). This later increase in nrms is due,

    in part, to the amplitude changes produced by the time shift induced by the thicker carbon

    dioxide region on the monitor state. This indicates that, possibly, some volume estimation of

    the carbon dioxide zone could also be monitored with high precision 4D seismic data at high

    CO2 saturations.

  • 7/27/2019 Detect Abi Lty Analysis

    13/16

    IAPG Instituto Argentino del Petrleo y el Gas 501

    Detectability Analysis of Miscible CO2 in a CCS/EOR Oil Reservoir

    CONCLUSIONS

    Miscibility of CO2 in oil and CO2 saturation both produce reectivity trends which in turn

    can generate a medium-to-weak time-lapse signal that can possibly be detected on low noise time-

    lapse surveys. This time-lapse reectivity effects, as measured by the nrms operator, can be related

    with CO2 saturation content and possibly thickness of the carbon dioxide zone, even when the

    modeled reservoir constitute a thin layer to the eyes of the seismic data. CO2 monitoring for

    EOR/CCS can be compromised when miscible carbon dioxide at low saturations are injected

    in the pore space of the reservoir. The presence of a free gas phase gives rise to an increase on

    the time-lapse signal facilitating monitoring objectives. We expect that with future improvement

    in time-lapse acquisition and processing, an enhanced characterization of the reservoirs uid

    distribution, as we have shown in this paper, can be expected.

    Future work will consider the presence of noise in the data and the time-shift effects produced

    by the injection of carbon dioxide.

    Figure 7: Time lapse effects on the nrms metric due to varying CO2 content in the brine-oil-CO2 zone depicted inFigure 3. The shaded zone below 20% represents the 4D noise level in precision time-lapse seismic data. From (a) to(d) the traces move to the right of the injection zone (see Figure 1-a). Immiscible and miscible cases are considered.

  • 7/27/2019 Detect Abi Lty Analysis

    14/16

    IAPG Instituto Argentino del Petrleo y el Gas

    502 VIII Congreso de Exploracin y Desarrollo de Hidrocarburos

    ACKNOWLEDGEMENTS

    The results presented here have been obtained in the CO2ReMoVe project, which envisages

    the development of technologies and procedures for monitoring and verifying geological CO2

    storage. The nancial support of the European Commission and the industrial consortium

    involved is greatly appreciated. Partial nancing was also received from CONICET PIP 112-

    200801-00952.

    We greatly acknowledge the reviews and constructive comments of Juan Soldo, which helped

    to increase the quality of the present manuscript.

    BIBLIOGRAPHY

    Batzle M., Wang Z., 1992. Seismic properties of pore

    fluids. Geophysics 57, 1396-1408.

    Brown L. T., 2002, Integration of rock physics and reservoir

    simulation for the interpretation of time-lapse seis-

    mic data at Weyburn Field, Saskatchewan. Masters

    thesis, Colorado School of mines, Golden, CO.

    Bunge R.J., 2000. Midale reservoir fracture character-ization using integrated well and seismic data,

    Weyburn Field, Saskatchewan. Masters thesis,

    Colorado School of Mines, Golden, CO, 218p.

    Carcione J. M., Picotti S., Gei D. and Rossi G., 2006,

    Physics and seismic modeling for monitoring CO2

    storage. Pure and App. Geophys., 175-207.

    Churcher P.L. and A.C. Edmunds, 1994. Reservoir charac-

    terization and geologic study of the Weyburn Unit,

    southeastern Saskatchewan. PanCanadian Petro-

    leum Ltd., August 1994.

    Duan Z., Moller N., Weare J., 1992. An equation for the

    CH4-CO2-H2O system: I. Pure systems from 0 to

    1000 C and 0 to 8000 bar. Geochimica et Cosmo-

    chimica Acta 56, 2605-2617.

    Gangi A. and Carlson R. 1996. An asperity deformation

    model for effective pressure. Tectonophysics, 256,

    241-251.

    Gassmann F., 1951, Uber die elastizitat poroser medi-

    en Vierteljahrschrift der Naturforschenden Ges-

    sellshaft in Zurich, 96, 1-23.

    Hassanzadeh H., Pooladi-Darvish M. and Keith D.

    W., 2009, Accelerating CO2 Dissolution in Sa-

    line Aquifers for Geological Storage - Mechanis-

    tic and Sensitivity Studies. Energy & Fuels, 23,3328-3336.

    Hill R., 1963, Elastic properties of reinforced solids:

    some theoretical principles. J. Mech. Phys. of

    Solids 11, 357-372.

    Kennett B.L.N., 2009, Seismic wave propagation in strati-

    ed media. ANU E-Press, http://epress.anu.edu.au.

    Kragh E. and Christie P., 2002, Seismic repeatability,

    normalized rms, and predictability. The Leading

    Edge, 21, 640-647.

    Liu Y. and Schmitt D.R., 2003. Amplitude and AVO

    responses of a single thin bed. Geophysics, 68(4),

    1161-168.

    Lumley D., 2010, 4D seismic monitoring of CO2 se-

    questration. The Leading Edge, 29 (2), 150-155.

    Ma J. and Morozov I., 2010, AVO modeling of pres-

    sure-saturation effects in Weyburn CO2 seques-

    tration. The Leading Edge, 29 (2), 178-183.

  • 7/27/2019 Detect Abi Lty Analysis

    15/16

    IAPG Instituto Argentino del Petrleo y el Gas 503

    Detectability Analysis of Miscible CO2 in a CCS/EOR Oil Reservoir

    Mavko, G., and Mukerji T., 1998, Bounds on low-frequen-

    cy seismic velocities in partially saturated rocks.

    Geophysics, 63 (3), 918-924.

    Meadows M., 2008, Time-lapse seismic modeling and in-

    version of CO2 saturation for storage and enhanced

    oil recovery, The Leading Edge, 27 (4), 506515.

    Miller A. C. and Helgerud M. B., 2009, 4D seismic repeat-

    ability: lessons from Hoover-Madison-Marshall.

    SEG International Exposition and Annual Meeting,

    2884-3888.

    Picotti S., Santos J. E., Carcione J. M., Gei D. and Ravaz-

    zoli C. L., 2009, A nite element method to model

    attenuation and dispersion effects in highly heteroge-

    neous uid-saturated porous media. Theoretical and

    Computational Acoustics 2009, Steffen Marburg

    Ed., ICTCA 09 Conference Proceedings, 235-246.

    Peng D. Y., Robinson D. B., 1976. A new two-constant

    equation of state. Industrial and Engineering Chem-

    istry Fundamentals 15 (1), 59-64.

    Ravazzoli C. L., 2001, Analysis of the reflection and

    transmission coefficients in three-phase sand-

    stone reservoirs. Journal of Comp. Acoustics, 9

    (4), 1437-1454.

    Ravazzoli C. L., and Gmez J. L., 2011, AVA Seismic re-

    ectivity analysis in carbon dioxide accumulations:

    sensitivity to CO2 phase and saturation. Journal of

    App. Geophysics, 72, 93-100.

    Reasnor M. R., 2001, Forward modeling and interpreta-

    tion of multicomponent seismic data for fracture

    characterization, Weyburn Field, Saskatchewan,

    CSM MSc Thesis #5538.

    Tsvankin I., 1995, Seismic waveelds in layered isotropic

    media. Samizdat Press.

    White D., 2009, Monitoring CO2 storage during EOR at

    the Weyburn-Midale eld. The Leading Edge, 28

    (7), 838-842.

    Zimmerman R. W., 1991, Compressibility of Sandstones.

    Elsevier Science Publ.

  • 7/27/2019 Detect Abi Lty Analysis

    16/16