Introduction

Carbon capture and storage (CCS) are key components in achieving the goal to significantly reduce anthropogenic emissions of greenhouse gases. Safe deployment of large-scale CCS requires monitoring, verification, and accounting (MVA) tools to assure that injectivity and capacity predictions are reliable, fate and configuration of the carbon dioxide (CO2) plume is acceptable, regulatory permitting and for compliance, public assurance, risk reduction, and stakeholder assurance1. In particular, accurate accounting of subsurface CO2 requires an understanding of how to determine the elastic properties, which control seismic wave velocities, of rocks and the corresponding changes because of CO2 injection and migration.

Time-lapse seismic monitoring methods (e.g., surface seismic surveys, vertical seismic profiling, and cross-well seismic monitoring) provide indirect measurements of CO2 and reservoir properties in the deep subsurface of geologic carbon storage (GCS) sites2,3,4. Rock physics models are crucial for interpreting seismic monitoring data to enable accounting of subsurface changes associated with CO2 injection and migration, as 4D (the 3D space and time) anomalies using seismic data are converted to CO2 saturation maps. These CO2 saturation maps based on seismic observations are used to update CO2 fluid migration pathways, which are necessary to assess the movement of CO2 within the storage reservoir, plume formation/migration, and detecting leaks.

Traditionally, observed seismic anomalies are completely attributed to CO2 saturation distribution with sustained phases of CO2 injection over long-time periods (on the scale of tens to hundreds of years). However, seismic interpretation requires considerations into other phenomena that may change the seismic response5. Typically, calculating seismic velocity variations required for forward seismic modeling are calculated as a function of CO2 saturation using the Gassmann equations6. The Gassmann-Biot fluid substitution theory6,7 has been commonly employed in estimating changes in the seismic response caused by fluid injection or leakage4,8,9 and is widely used in seismic reflection data analyses in hydrocarbon exploration and CO2 enhanced oil recovery10. Since fluids have no shear modulus, the predicted changes in the S-wave velocity is very minimal (on the order of 1% change) because there is only a change in the density of the fluid phase with the injected CO2. P-wave velocities are expected to change by 5–15% with CO2 injection because the bulk modulus of the fluid phase changes significantly because of injected CO211,12,13.

Previous studies show that sandstone that is exposed to free-phase CO2 and brine for several months can demonstrate large changes in seismic velocities ( ~ 15% decrease for P-wave velocity [VP] and ~ 16% decrease for S-wave velocity [VS]) caused by chemical reactions between the rock and CO2 (see more in Supplementary Note 1)14,15. In other experimental studies, seismic wave speeds can change negligibly or even more so that discussed here; however, these studies vary widely with the experimental setup (e.g., state of injected CO2, the residence time, porosity, temperature, mineralogy, etc16). Recent geophysical observations of GCS have illustrated that the S-wave velocity can also change as much as the P-wave velocity during CO2 injection and migration because of possible mineral dissolution17,18,19. However, estimating effects of CO2 saturation with the Biot-Gassmann equation cannot replicate these large changes in seismic velocities because this approach only considers the impact of supercritical CO2 on the fluid properties and does not consider the justifiably proven evidence that rocks under these conditions are stress-dependent. More specifically in this study, we refer to stress as effective/differential pressure, which is compressive pressure minus pressure within the pore space. When referring to stress, we do not mean the entire stress tensor but rather the average of the trace of the stress tensor (pressure). As a result of long-term exposure to CO2, the reservoir storage porosity, permeability, mineralogy, and the fluid within the pores are affected by these chemical reactions resulting in changes in the overall elastic moduli (e.g., shear and bulk moduli) of the saturated rock. A misrepresentation of seismic-wave velocities would cause an inaccurate accounting for subsurface CO2 storage, where changes in seismic-wave velocities could be inaccurately ascribed solely to CO2 saturation.

Theoretical approaches and experiments have demonstrated that rocks are stress-dependent and behave like nonlinear elastic bodies20. A series of studies21,22,23,24 demonstrated that the elastic non-linearity of rocks is related to compliant porosity (caused by grain contacts or very thin cracks, etc.), which is often a small part of the total porosity that consists of stiff (spherical, stiff) and compliant pores. As an important note, the formation of small, subcritical cracks or dissolution along grain contacts as a result of exposure to CO2 do have different processes depending on the mineralogy, fluid interactions, porosity, grain size, etc.16. The stress-dependence of the compliant porosity was derived from the theory of poroelasticity under empirically-based assumptions. Closing compliant porosity with increasing differential stress explains the experimentally observed stress dependencies of seismic velocities (exponential relationship).

We develop a new CO2 rock physics model to account for the stress-dependent relationship and the rock frame weakening effect because of chemical reactions with CO2. We address many considerations needed for accurate CO2 accounting in GCS25. Our new rock physics model takes into account changes in seismically important variables in relation to rock properties (i.e., changes in elastic properties, porosity because of chemical reactions and pressure changes), and fluid properties (i.e., fluid saturation) during CO2 injection and migration25. We validate our new CO2 rock physics model using CO2 reservoir simulation results of the Kimberlina-1.2 model (see Section 2.226,27 for more details). Our new CO2 rock physics model enables reliable accounting of CO2 in GCS sites.

Stress dependence of Seismic-wave velocities because of compliant porosity

To construct a stress-dependent rock physics model, we consider compliant porosity in the construction of the bulk and shear moduli of the rock framework to account for changes in differential pressure. To calculate the resulting VP and VS (Fig. 1), we first calculate the bulk modulus of the fluid using published equations of state for brine and CO228,29. More details can be found in the Methods section.

Fig. 1: Workflow of CO2 rock physics modeling.
figure 1

Workflow for calculating seismic velocities (VP, VS) and density (ρsat) of the saturated rock. Black arrows and gray boxes represent methods typically incorporated into rock physics modeling of GCS. Blue boxes and arrows represent the additions to this workflow explored in this paper. The first step is to calculate the fluid composition, which finds the bulk modulus (K), density (ρ), and velocity (V) of the fluid. Next is the mineral grain composition, which finds the bulk modulus (K), density(ρ), and shear modulus (μ) of the mineral grains. Third is the framework properties, such as Krief or Hertz Mindlin approaches, which calculate the framework properties. The following step is to include our proposed workflow by adding compliant porosity and CO2 effects. Lastly, the Gassmann equation is applied to find the seismic wave speeds and density.

The rock framework properties (Kframe and μframe) can be determined with various methods, but we show two possible options: Hertz-Mindlin contact theory30 and Krief modeling31. The Hertz-Mindlin contact theory considers the effects of pore pressure changes in Eqs. (3) and (4) in the general case. The construction of Kframe with Krief modeling is empirically derived31 based on the relationship between the properties of the rocks’ composite grains and the rock frame (through the Biot poroelastic coefficient32,33). Therefore, if sonic log data are available, this approach is ideal in determining rock framework properties. Either approach can be used to calculate rock framework properties.

To incorporate compliant porosity, we update the bulk and shear frame moduli for compliant porosity22 and eqs. (13) and (14). Compliant porosity data require experimental data either from the site or similar rock types. The various parameters in Eqs. (13) and (14) are important to take into account how stiff porosity (θs) and compliant porosity (θc) parameters affect the bulk modulus as a function of stress. As a result, we can use these parameters to also model porosity and incorporate compliant pores, Eq. (24). We resolve these important porosity parameters using the empirical relationship between stress and seismic-wave velocities, as described by Eqs. (18) and (19)22,34. Each coefficient helps to determine the various relevant parameters and update the bulk and shear frame moduli using Eqs. (20) to (26). Any velocity data as a function of stress can be used for site specific lithology. Figure 2 illustrates the linear least-squares fit to two drained Berea rock samples35. Table S1 in the supplementary information provides the resolved coefficients. Figure 2b demonstrates the changes on theoretically-derived seismic-wave velocities using the original approach versus updating the velocities with the compliant porosity parameters from the Berea sandstone. As a result, lower effective pressures have major effects on seismic-wave velocities, where more accurate modeling agrees well with experimental studies on sandstones.

Fig. 2: Effects of compliant porosity on seismic velocities.
figure 2

a Dry Berea sandstone measurements45 for two different experiments (green squares and blue triangles). Lines represent linear least-square fits to the data with eqs. (18) and (19). b Estimated velocities (top set of lines = Vp and bottom set of lines = Vs) using the Hashin-Strikman from Eq. (7) and using the c Krief modeling from Eq. (9) by applying the compliant porosity parameters from the top image. Black lines are the reference for the methods without including compliant porosity. The gray lines are results obtained using compliant porosity parameters from24. All compliant porosity parameters are included in Table S1 in the supplementary information. Assumed parameters: Poisson’s ratio = 0.2, μmineral = 44 GPa, Kmineral = 33 GPa, ϕc = 0.4, Kfluid = 2.5 GPa, ϕ = 0.2.

Results: rock frame changes caused by exposure to CO2 and mineral dissolution

Previous experiments show that rock frame weakening or strengthening can occur14,36, caused by swelling clays, dissolution of feldspars, chlorite, and carbonates37. One implication of rock frame alterations is an increase in microseismicity caused by increasing formation permeability and elevated pore pressure. Changes in porosity, mineral composition, or permeability result in changes in seismic-wave velocities that impact accounting of subsurface CO2. However, it can be challenging to predict changes in seismic-wave velocities as there are complicated chemical reactions that depend on temperature, pressure, mineralogy, and brine composition. Rock samples also undergo a porosity decrease (rock strengthening) or increase (rock weakening) because of these chemical reactions14.

We simulate a scenario under which rock frame weakening could influence seismic-wave velocities, which is associated with an increase in porosity and permeability. We use VP and VS ultrasonic measurements14 on a Mt. Simon sandstone sample that underwent rock frame weakening (Fig. 3). The study measured ultrasonic VP and VS over a range of effective pressures (0-40 MPa) at 50 °C in a drained Mt. Simon sandstone that was pre- and post-exposed to supercritical CO2 and brine for four weeks, resulting in VP reduction by ~ 14%, VS reduction by ~ 15%, and porosity increase by 8%. Since the seismic velocities of pre-exposure samples were not measured over a large range of effective pressures, we use our compliant porosity parameters from the previous section for the two Berea sandstones as reasonable compliant porosity parameters (as shown in Fig. 3), but we adjust the AP and AS (Eqs. (18) and (19)) parameters to fit the Mt. Simon experimental data. In Table S1 in the Supplementary Information, we use the Kp, Ks, Bp, Bs, and D parameters from the Berea sandstones to fit the pre-exposure sample data (triangles in Fig. 3). This approach is not needed when more data are available at higher pressures, where just a standard fitting approach can be used. We employ the same fitting approach in Eqs. (18) and (19) to the post-exposure samples (Fig. 3) to accommodate changes in compliant porosity and those effects on seismic velocities.

Fig. 3: Effects of CO2 weakening on seismic velocities.
figure 3

Two different models of CO2 weakening: a using green-labeled Berea sandstone data (Sample #121-141) and b using blue-labeled Berea data (Ref. 17). We fit pre-exposure (triangles) and post-exposure (squares) to CO2 samples14. Since there are limited data for pre-exposure samples, we use the two different Berea sandstone fitted parameters (see fitted parameters in Table S1) from Fig. 2.

Ideally, additional pre-exposed measurements should be made in Fig. 3 over a larger range of effective pressures; therefore, the exact changes in compliant porosity may not be accurate from the pre- to post-exposure samples. If there is dissolution of minerals, then the contribution of compliant porosity would likely increase, but it is more challenging to estimate how derivative parameters would change, such as θc (\({K}_{drys}\frac{\partial \frac{1}{Kframe}}{\partial {\phi }_{c}}\)). Parameters θs and θsμ are related to stiff porosity; these values could change somewhat between pre- and post-exposure samples because they are mainly dependent on the mineral grain bulk and shear moduli21. If the bulk modulus of the mineral grains decrease after CO2 exposure (possibly caused by chemical interactions), then θs would decrease. This value controls the slope of the Vp curve (Fig. 3) when stiff porosity is dominated at higher effective pressures. As a result, Vp wave speeds would be lower at higher effective pressures. θc (how the bulk modulus changes with compliant porosity) is also dependent on the bulk and shear moduli of the mineral and the inverse of the aspect ratio of the pores. As the aspect ratio decreases, this derivative would increase. Therefore, it is challenging to predict how the aspect ratio of pore space changes as a result of CO2 exposure without lab measurements. In general, if the aspect ratio decreases, Vp and Vs would decrease at smaller effective pressures compared with the pre-exposure samples (see Eq. (15)).

For each model point with supercritical CO2 in the Kimberlina-1.2 model, we update the elastic moduli and porosity for long-term exposure to CO2 and use the new fitted compliant porosity parameters from Fig. 3. All main parameters in Eqs. (18) and (19) decrease between the pre- and post-exposure sample, where compliant porosity ϕc0 increases, which aligns with the observation showing porosity increase (see Table S1 in the Supplementary Information). To update the modeling for rock frame weakening, we recalculate the increase in porosity in Eq. (16) and calculate the new compliant porosity parameters and Kdrys/μdrys values in Eqs. (13) and (14). Modifying the Kdrys/μdrys values is not trivial because we have limited information on how the moduli of the minerals has changed; therefore, we simply apply a similar reduction in the elastic moduli (Kdrys/μdrys) as if the sandstones in our Kimberlina-1.2 model underwent rock frame changes similar to the Mt. Simon sandstones. We employ empirical changes in seismic velocities and porosity as a result of rock frame weakening but do not model the actual changes in mineralogy as a result of these chemical reaction. This method demonstrates how to consider the non-linearity of seismic velocity changes as a function of effective pressure. As a result, there are two mechanisms for porosity to change: non-reactive, where Eq. (14) describes how porosity is updated with the effective pressure, which is a result of the pore pressure changes; and, reactive, which occurs when there is mineral dissolution because of chemical reactions. These porosity changes then propagate throughout the model, changing the seismic velocities even more.

Discussion: accuracy of CO2 accounting with seismic data

Dry rock frame changes because of compliant porosity or weakening/strengthening impact interpretation of time-lapse seismic monitoring of GCS and accounting of subsurface storage of CO2. Table 1 demonstrates all the changes on different properties and how they changes as a result of compliant porosity and CO2 weakening. Additionally, Figs. S2 and S3 in the Supporting Information show the quantitative changes of bulk and shear moduli and porosity as a function of time over the timescale of the model. Figure 4 illustrates a demonstration of CO2 weakening on seismic-wave velocities using GCS time-lapse Kimberlina-1.2 flow simulations (see Supplementary Note 4 for more details).

Table 1 Changes of rock physical parameters caused by compliant porosity and CO2 weakening
Fig. 4: Shear-wave velocities in the Kimberlina-1.2 model.
figure 4

Kimberlina-1.2 model for Vs for three time steps using the a Hashin-Strikman approach and b Hashin-Strikman with the added CO2 rock frame weakening, using the workflow in Fig. 1. The top rows illustrate two time steps (0 years and 20 years) for Vs (m/s). The bottom rows show changes in Vs relative to the baseline Vs model at 0 years in percent. The initial injection of CO2 is at X = 0 m. The corresponding Vp and CO2 gas saturation are described in the supplementary information.

We assume the entire model is sandstone for simpler modeling since CO2 is only injected into sandstone, but shale layers are also included in the Kimberlina-1.2 model (see Supporting Note 3). We calculate seismic velocities and density of the saturated rock matrix (workflow in Fig. 1) using the compliant porosity parameters of the Berea sandstone model from Fig. 2 and the CO2 weakening introduced in Fig. 3.

As predicted, because of CO2 fluid substitution only, only Vp changes are significant, whereas VS does not significantly change. However, we find that rock frame weakening and compliant porosity can also change VP. Significant VS changes are mainly caused by the CO2 weakening effect as compliant porosity has only a small effect on VS. Compliant porosity increases the porosity, which mostly influences saturated density because the pore space increases, leading to a reduction in the rock’s saturated density because of the stronger influence of brine and/or supercritical CO2.

Shear waves should be used in monitoring of geologic carbon storage because VS changes significantly because of chemical reactions between CO2 and the rock matrix. Modeling with compliant porosity (such as Eqs. (18) and (19)) provide a first order approximation in how porosity and permeability have changed because of CO2 exposure; therefore, changes in VP and VS can be used for accounting of CO2 and changes in porosity.

We provide a workflow to model changes in VP and VS with compliant porosity that takes into account the nonlinear relationship of seismic velocities with stress. The largest impact of this kind of modeling is that it accounts for pore pressure changes and more accurate seismic modeling of the entire reservoir, especially at lower stresses when the nonlinear relationship dominates.

More experiments are needed on chemical interactions between reservoir/cap rocks and CO2 to constrain timing, temperature, pressure, and mineralogical conditions and the resulting changes in seismic-wave velocities and porosity. The concurrent chemical reactions are complicated and vary with a number of factors, but modeling compliant porosity opens the opportunity to include more experimental results on chemical reactions between CO2 and the reservoir and cap rocks. If shear-wave changes exceed ~1% because of CO2 injection, then rock framework changes have likely occurred and the resulting seismic velocity changes should not be estimated solely based on fluid substitution.

Lastly, accurate accounting of the amount of CO2 relies on geophysical monitoring, where changes in seismic velocities cannot just be attributed to changes in gas saturation, but also changes in stress and chemical reactions between the rock framework and CO2. However, seismic velocities provide additional information on porosity and permeability changes that improve flow imaging of CO2 plume migration. We suggest two approaches to improving CO2 accounting: (1) other geophysical techniques (e.g., gravity, electromagnetic methods, etc.38) should be integrated with seismic velocity changes to fully account for the mass of CO2 in the reservoir rock (these methods would be needed at high CO2 saturations, when seismic velocity sensitivities asymptote), or (2) experiments should be performed on the different reservoir and cap rock types to evaluate the seismic velocity changes because of CO2 injection/migration and the chemical reactions therein.

Methods

All variables defined

The following lists all variables relevant in this paper.

Ksat = the saturated bulk modulus

Kframe = the bulk modulus of the dry rock matrix

Kmineral = the bulk modulus of the mineral grains

Kfluid = the bulk modulus of the fluid

Kbrine = the bulk modulus of the brine

\({K}_{C{O}_{2}}=\) the bulk modulus of supercritical CO2

Kdrm = the bulk modulus of the drained rock frame with compliant porosity included

ρsat = the saturated rock density

ρfluid = the density of the fluid in the pore space

ρbrine = the density of the brine in the pore space

\({\rho }_{C{O}_{2}}=\) the density of supercritical CO2 in the pore space

ρmineral = the density of mineral grains

ρmatrix = the density of the dry rock matrix

Sg = volume fraction of gas saturation in pore space

Sw = volume fraction of brine saturation in pore space (Sw = 1 − Sg)

ϕ = porosity

ϕc = critical porosity

μsat = the shear modulus of the saturated bulk modulus

μframe = the shear modulus of the drained rock matrix

μmineral = the shear modulus of the mineral grains

μdrm = the shear modulus of the drained rock frame with compliant porosity included

C = average number of contacts per spherical grain, C = 2.8/ϕc

pd = effective pressure (also known as differential pressure), pd = pc − pp, where pc is the confining pressure and pp is pore pressure

ν = Poisson ratio

n = exponent derived empirically for Krief modeling, n = 3 for most sandstones

ϕc0 = compliant porosity (thin, oblate cracks) in the unloaded case (pd = 0)

ϕs0 = stiff porosity in the unloaded case

θc/θs = how bulk modulus changes with compliant/stiff porosity, dimensionless

θcμ/θsμ = how shear modulus changes with compliant/stiff porosity, dimensionless

ϕ0 = initial porosity or assumed stiff porosity

VPdry/VSdry = seismic velocities of a dry rock matrix

VPdrys/VSdrys = seismic velocities of a dry rock matrix with compliant pores closed

Kdrys/μdrys = bulk/shear modulus of a dry rock matrix with compliant pores closed

Calculation of fluid and mineral properties

The bulk modulus and density of effective pore fluid (Kfluid and ρfluid) can be estimated (Eqs. [(1)] and [(2)]) using inverse bulk modulus averaging39 and arithmetic averaging of densities of the separate fluid phases (brine phase and supercritical CO2 phase), respectively40. The fluid composition requires salinity, temperature, pressure, and gas saturation to determine the bulk modulus of the fluid (Kfluid), density of the fluid (ρfluid), and thus the velocity of the fluid (Vfluid). Within this formation of calculating sound speed of fluid mixtures/suspensions, the heterogeneities are assumed to be small compared to the seismic wavelength. This step is the first step in the workflow in Fig. 1:

$$\frac{1}{{K}_{fluid}}=\frac{{S}_{w}}{{K}_{brine}}+\frac{{S}_{g}}{{K}_{C{O}_{2}}},$$
(1)
$${\rho }_{fluid}={S}_{w}{\rho }_{brine}+{S}_{g}{\rho }_{C{O}_{2}}.$$
(2)

The bulk and shear moduli of mineral grains (Kmineral and μmineral) are estimated by taking Voigt-Reuss-Hill averaging41 of the mineral constituents with published experimental moduli8,42. The mineral grain composition requires the mineralogy (such as quartz, clay ratio for sandstone) and literature values of bulk (K) and shear (μ) moduli to determine the density of the mineral grains (ρmineral) and the bulk (Kmineral) and shear (μmineral) moduli of the grains. We assume 70% quartz and 30% clay. Mineral composition at a GCS site can be estimated from wireline log data acquired during site characterization9. We assume the bulk and shear moduli of the mineral grains (Kmineral) has little dependency on stress because of the small range of stresses within CO2 reservoirs, while the bulk and shear moduli of the rock framework are very sensitive to stress, which we incorporate into the CO2 rock physics model. This step is the second step in Fig. 1.

Calculation of K frame with Hertz-Mindlin theory and Hashin-Strikman approaches

The first step is to determine Kframe = KHSframe using the Hertz-Mindlin model with the Hashin-Strikman lower bound. The Hertz-Mindlin model combines the normal and tangential forces of precompacted identical spheres. The Hertz-Mindlin contact theory considers the effects of pore pressure changes in Eqs. (3) and (4) in the general case. The modified Hashin-Strikman lower bound43,44 estimates the effective bulk and shear moduli of the dry rock frame at a different porosities ϕ in Eqs. (5)–(6). The Hashin-Shtrikman bounds define a range of plausible elastic moduli for a mix of two phases, such as sand and cement/clay. The lower bound assumes the soft-sand model, which calculates moduli based on dry sand where the cement is deposited away from grain contacts. The upper bound constitutes the stiff-sand model, where cement is deposited at grain contacts. The Hertz-Mindlin theory requires elastic moduli of the mineral grains, Poisson’s ratio, critical porosity, and porosity. Then, we determine Ksat with the Gassmann equation and the calculated Kframe (Eq. [(7)]).

The Hertz-Mindlin theory expression for the effective bulk and shear moduli of a dry, dense, random pack of identical spheres at the critical porosity:

$${K}_{mc}={\left(\frac{{C}^{2}{(1-{\phi }_{c})}^{2}{\mu }_{mineral}^{2}{p}_{d}}{18{\pi }^{2}{(1-\nu )}^{2}}\right)}^{1/3},$$
(3)
$${\mu }_{mc}=\frac{5-4\nu }{10-5\nu }{\left(\frac{3{C}^{2}{(1-{\phi }_{c})}^{2}{\mu }_{mineral}^{2}{p}_{d}}{2{\pi }^{2}{(1-\nu )}^{2}}\right)}^{1/3}.$$
(4)

The effective moduli at different porosities are found with the modified Hashin-Shtrikman lower bound:

$${K}_{HS \, frame}={\left[\frac{\frac{\phi }{{\phi }_{c}}}{{K}_{mc}+\frac{4}{3}{\mu }_{mc}}+\frac{1-\frac{\phi }{{\phi }_{c}}}{{K}_{mineral}+\frac{4}{3}{\mu }_{mc}}\right]}^{-1}-\frac{4}{3}{\mu }_{mc},$$
(5)
$${\mu }_{HS \, frame}= {\left[\frac{\frac{\phi }{{\phi }_{c}}}{{\mu }_{mc}+\frac{{\mu }_{mc}}{6}\left(\frac{9{K}_{mc}+8{\mu }_{mc}}{{K}_{mc}+2{\mu }_{mc}}\right)}+\frac{1-\frac{\phi }{{\phi }_{c}}}{{\mu }_{mineral}+\frac{{\mu }_{mc}}{6}\left(\frac{9{K}_{mc}+8{\mu }_{mc}}{{K}_{mc}+2{\mu }_{mc}}\right)}\right]}^{-1} \\ -\frac{1}{6}{\mu }_{mc}\left(\frac{9{K}_{mc}+8{\mu }_{mc}}{{K}_{mc}+2{\mu }_{mc}}\right),$$
(6)
$${K}_{sat}={K}_{HS \, frame}+\frac{{(1-\frac{{K}_{HS \, frame}}{{K}_{mineral}})}^{2}}{\frac{\phi }{{K}_{fluid}}+\frac{1-\phi }{{K}_{mineral}}-\frac{{K}_{HS \, frame}}{{K}_{mineral}^{2}}}.$$
(7)

The density of the saturated rock matrix is determined using:

$${\rho }_{sat}=\phi * {\rho }_{fluid}+(1-\phi )* {\rho }_{matrix}.$$
(8)

Calculation of K frame with Krief modeling

Another method to estimate Ksat is using the Krief equation that employs a different method to estimate Kframe = KKrframe.

The Krief modeling only requires porosity and elastic moduli of the grains to determine the framework elastic moduli. The workflow is similar to that for updating Ksat, Eqs. ((9)-(12)):

$${K}_{sat}={K}_{Kr \, frame}+\frac{{\alpha }^{2}}{\frac{\phi }{{K}_{fluid}}+\frac{(\alpha -\phi )}{{K}_{mineral}}},$$
(9)
$$\alpha =1-{(1-\phi )}^{\frac{n}{1-\phi }},n=3,$$
(10)
$${K}_{Kr \, frame}=(1-\alpha ){K}_{mineral},$$
(11)
$${\mu }_{Kr \, frame}=(1-\alpha ){\mu }_{mineral}.$$
(12)

Incorporating compliant porosity

The bulk and shear moduli and porosity of the dry rock frame are updated from a series of papers21,23,24. Any approach can be used to estimate drained rock bulk or shear moduli, such as Hashin-Strikman or Krief modeling (denoted with Kdrs), but it must assume stiff porosity, where compliant pores are closed. We assume that DS = DP = D in Eqs. (18) and (19). We employ a grid search for D. For each guess for D, we use a linear least-squares fit to resolve AP, AS, KP, KS, BP, and BS. We select the D value, which provides the minimum least-squares misfit. Each modified modulus is denoted with ’m’ to consider compliant porosity with the following equations:

$${K}_{drm}={K}_{drys}* \left[1+{\theta }_{s}\left(\frac{1}{{K}_{drys}}-\frac{1}{{K}_{mineral}}\right){p}_{d}-{\theta }_{c}{\phi }_{c0}\exp \left(-{\theta }_{c}{p}_{d}\frac{1}{{K}_{drys}}\right)\right],$$
(13)
$${\mu }_{drm}={\mu }_{drys}* \left[1+{\theta }_{s\mu }\left(\frac{1}{{K}_{drys}}-\frac{1}{{K}_{mineral}}\right){p}_{d}-{\theta }_{c\mu }{\phi }_{c0}\exp \left(-{\theta }_{c}{p}_{d}\frac{1}{{K}_{drys}}\right)\right],$$
(14)
$$\phi ={\phi }_{s0}-\left(\frac{1}{{K}_{drys}}-\frac{1}{{K}_{mineral}}\right){p}_{d}+{\phi }_{c0}\exp \left(-{\theta }_{c}{p}_{d}\frac{1}{{K}_{drys}}\right),$$
(15)
$${\phi }_{m}({p}_{d})={\phi }_{0}+{\phi }_{c0}\exp \left(-{\theta }_{c}{p}_{d}\frac{1}{{K}_{drys}}\right).$$
(16)

The density of the saturated rock matrix with compliant porosity is obtained using:

$${\rho }_{sat}={\phi }_{m}* {\rho }_{fluid}+(1-{\phi }_{m})* {\rho }_{matrix}.$$
(17)

Rock measurements of VP and VS as a function of differential pressure can be measured and fit with a linear least-squares to the following equations. We seek the parameter, D, for the minimum misfit:

$${V}_{Pdry}({p}_{d})={A}_{P}+{K}_{P}{p}_{d}-{B}_{P}\exp (-{p}_{d}D),$$
(18)
$${V}_{Sdry}({p}_{d})={A}_{S}+{K}_{S}{p}_{d}-{B}_{S}\exp (-{p}_{d}D),$$
(19)

where,

$$\begin{array}{rclrcl}{A}_{P}&=&{V}_{Pdrys} \hfill &{A}_{S}&=&{V}_{Sdrys} \hfill \\ {K}_{P}&=&0.5{V}_{Pdrys}{H}_{s}{\theta }_{s\mu }\left(\frac{1}{{K}_{drys}}-\frac{1}{{K}_{mineral}}\right)&{K}_{S}&=&0.5{V}_{Sdrys}{\theta }_{s\mu }\left(\frac{1}{{K}_{drys}}-\frac{1}{{K}_{mineral}}\right)\\ {B}_{P}&=&0.5{V}_{Pdrys}{H}_{c}{\theta }_{c\mu }{\phi }_{c0} \hfill &{B}_{S}&=&0.5{V}_{Sdrys}{\theta }_{c\mu }{\phi }_{c0} \hfill \\ D&=&\frac{{\theta }_{c}}{{K}_{drys}} \hfill\\ {V}_{Sdrys}&=&\sqrt{\frac{{\mu }_{drys}}{\rho }} \hfill &{V}_{Pdrys}&=&\sqrt{\frac{{K}_{drys}+\frac{4}{3}{\mu }_{drys}}{\rho }} \hfill \end{array}$$

We use the following steps to solve for the compliant porosity parameters in Eqs. (13), (14), and (16). First, we determine the shear and bulk moduli of the dry rock matrix with close compliant pores:

$${\mu }_{drys}={A}_{S}^{2}\rho ,$$
(20)
$${K}_{drys}={A}_{P}^{2}\rho -4/3{\mu }_{drys}.$$
(21)

Second, we compute θc:

$${\theta }_{c}=D{K}_{drys}.$$
(22)

Third, we determine θcμ:

$$\frac{{B}_{P}}{{B}_{S}}=\frac{{A}_{P}{H}_{c}}{{A}_{S}}\quad \quad {H}_{c}=\frac{{K}_{drys}\frac{{\theta }_{c}}{\theta c\mu }+4/3{\mu }_{drys}}{{K}_{drys}+4/3{\mu }_{drys}}$$
$${\theta }_{c\mu }={K}_{drys}{\theta }_{c}{\left(\frac{{B}_{P}{A}_{S}}{{B}_{S}{A}_{P}}({K}_{drys}+4/3{\mu }_{drys})-4/3{\mu }_{drys}\right)}^{-1}.$$
(23)

Then, we obtain ϕc0:

$${\phi }_{c0}=\frac{2{B}_{S}}{{A}_{S}{\theta }_{c\mu }}.$$
(24)

For the stiff porosity components in Eqs. (13) and (14), we can also obtain the following:

$${\theta }_{s\mu }\left(\frac{1}{{K}_{drys}}-\frac{1}{{K}_{mineral}}\right)=\frac{2{K}_{S}}{{A}_{S}}$$
(25)

and

$${H}_{s}=\frac{{K}_{drys}\frac{{\theta }_{s}}{\theta s\mu }+4/3{\mu }_{drys}}{{K}_{drys}+4/3{\mu }_{drys}}$$
$${\theta }_{s}\left(\frac{1}{{K}_{drys}}-\frac{1}{{K}_{mineral}}\right)={K}_{drys}^{-1}* \left[2{K}_{P}{A}_{P}\rho -8/3{\mu }_{drys}{K}_{S}* {A}_{S}^{-1}\right].$$
(26)