SCA2023-025

Prediction of wetting condition in a reservoir rock using coupled digital

rock and molecular dynamics simulation
Ashraful Islam, Joshua D. Moore, Rafael Salazar-Tio, Guangyuan Sun, Andrew Fager, Bernd Crouse, Sabine Schweizer, Kwan
Skinner, Lalitha Subramanian

Dassault Systémes, Johnston, RI

Abstract. Wettability is one of the key inputs for digital rock simulations to calculate multiphase fluid transport
properties. Reservoir rocks have spatially varying wettability that depends on local pressure, temperature, surface
roughness, and mineral and fluid compositions. Here, a multiphase lattice Boltzmann method (LBM) is used to
simulate a capillary pressure drainage simulation in an initially water-wet Berea sandstone, utilizing wettability
inputs derived from fully atomistic molecular dynamics simulations. When oil invades the pore space as the
capillary pressure is increased, a fraction of the pore-grain surface is altered from a water-wet to an oil-wet condition.
The eligibility for alteration takes into account local capillary pressure, disjoining pressure, and micro-scale surface
roughness. Wettability inputs to the multiphase simulation are contact angles, calculated from molecular dynamics
simulations of oil droplets on mineral surfaces with hydrophilic and hydrophobic surface chemistry. The multiphase
workflow demonstrates an iterative process where wettability is altered along with quasi-static fluid flow resulting
in spatially varying wetting conditions of the simulated rock sample. The study investigates the slow dynamics of
oil invasion and wettability alteration, taking into account the micro-scale pore shape dependence of the rock. The
wettability distributions are compared with an image-based geometric intrusion analysis which mimics mercury
intrusion porosimetry. The resulting wettability distribution holds the potential to enhance the realistic prediction of
transport properties under in-situ reservoir conditions.

reservoir to test. Obtaining representative rock samples from

the reservoir for wettability testing can be a time-consuming
1 Introduction process. It involves drilling multiple core samples, preserving
Wettability is a measure of a fluid’s tendency to adhere to a their integrity, and preparing them for laboratory testing.
solid surface in the presence of other fluids. In the context of Consequently, obtaining wettability data in a timely manner
reservoir rocks, wettability plays a crucial role in determining can sometimes be challenging for oil and gas operators.
fluid behavior and consequently the efficiency of oil and gas
extraction. Reservoir rocks are usually composed of various With the advancement of imaging, such as x-ray micro-
minerals, such as quartz, clays, carbonates etc., that have tomography (microCT) and computational fluid dynamics, it
various affinities toward fluids like oil, brine and others. The is now possible to perform special core analysis studies such
wettability of the rock determines how these fluids are as relative permeability or capillary pressure analysis
distributed within the pore space and their ability to flow digitally for the rock samples. Digital rock simulations offer
through the rock. As a result, wettability can significantly certain benefits such as non-destructive analysis, allowing the
impact multiphase fluid displacement in a reservoir as well as user to perform what-if studies for the same sample,
capillary pressure and relative permeability experiments [1- uncertainty quantifications, and above all else, reduced
5]. turnaround time. However, wettability is still needed as an
input for relative permeability and capillary pressure
Typically, the wettability of a reservoir rock can be simulations.
determined through various laboratory tests such as contact
angle measurements, the Amott-Harvey method, the USBM Digital imaging of rock samples by microCT or FIB-SEM can
method, and spontaneous imbibition [6]. Wettability tests are also help to determine the wettability of a reservoir rock
conducted in a controlled laboratory environment to replicate without relying solely on laboratory testing. By imaging rock
reservoir conditions and require access to facilities equipped samples in the presence of various fluid phases and analyzing
with specialized instruments and expertise. Reservoir rocks, those using 3D image processing tools, one can calculate
unlike man-made materials, tend to have spatially varying contact angles [10-15]. Observations of the relative fluid
wettability, i.e., the contact angle varies from location to coverage of rock mineral surfaces can also be used to create
location on the surface of the grains in contact with the fluids a 3D map of the wetting state of the rock [7-8]. However,
in the pore space of a rock [7-9]. Thus, proper wettability these procedures still require several weeks of laboratory
characterization requires multiple samples across the “aging”, which involves cleaning samples, followed by a
“drainage” procedure where oil is pushed into the rock

*
Corresponding author: [email protected]
 The 36th International Symposium of the Society of Core Analysts

sample filled with brine and aged for few weeks. Recently, a mechanism for wettability alteration is considered by
complementary numerical procedure to mimic aging was substituting functional groups capping the mineral surface,
proposed [21]. However, the underlying molecular from hydroxyl groups (-OH) in the presence of water to alkyl
mechanisms for obtaining spatially variant wettability groups (-CnH2n+1) as oil intrudes into the pores. As the oil-
distributions was lacking in that initial approach. water interface moves along the mineral surface, this
functional group substitution ultimately triggers the
Determining wettability and the initial fluid distribution using wettability alteration, as contact angles change from water-
digital rock simulations is a complicated and challenging wet to oil-wet, shown later in this paper.
problem that spans the molecular to micro-scales. When oil
intrudes the brine-filled pore space for the first time, the The following section describes the molecular dynamics and
grains are usually coated with a thin water film. For the oil to multiphase fluid flow simulation methodologies, followed by
be in contact with the grains, this film needs to be ruptured, a description of the workflow for the coupling of these two
which can be described by the disjoining pressure theory [16- scales, along with details of the wettability alteration method.
18]. At the micro-scale, pore shape, curvature and surface Results from molecular dynamics and multiphase fluid
roughness are critical to determining whether the water film simulations are then discussed, followed by the conclusions
can be ruptured. Afterwards, based on the pore fluid pressure, and a discussion of future research directions.
temperature, and fluid and mineral compositions, oil can
create another layer on the initially water-wet surface and
make the surface oil-wet. At the molecular scale, surface
chemistry, temperature, pressure, and salinity play a major 2 Methodology
role in the wettability of mineral surfaces [19]. For example,
the concentration of silanol (-SiOH) functional groups on 2.1 Molecular dynamics simulation
quartz mineral surfaces can determine whether the wettability
conditions are water-wet or oil-wet, and surface Molecular dynamics (MD) simulations model atoms and
hydrophobicity may be increased through alkylation by molecules as interacting particles. The basic approach to all-
organics present in oil reservoirs and rock formations [20]. atom MD involves solving Newton’s second law for every
Thus, both molecularly driven physicochemical properties atom in the simulation, where atomic forces are determined
and micro-scale multiphase fluid flow play an important role by an intermolecular force field, which describes van der
in reservoir rock wettability. Waals forces, electrostatic interactions, and chemical
bonding between different atoms in the simulation [23].
In a previous study [21], a numerical methodology for Pressure and temperature can be controlled through
wettability alteration was implemented using a multiphase thermostats and barostats [24]. Since the force field
lattice Boltzmann method, utilizing a given disjoining inherently takes into account the chemistry of the atoms and
pressure isotherm and assuming certain contact angles for the molecules, its development typically involves quantum
oil-wet and water-wet conditions. The initial workflow mechanical calculations as well as rigorous benchmarking
showcased an iterative process for wettability alteration, against experimental data, and these force fields can be used
focusing on the force balance between curvature-based, for a range of materials and conditions. For an MD run, the
applied, and disjoining pressures. That study [21] also chemistry of the materials involved, their specific
discussed the possibility of extending the workflow using formulation and processing conditions, if applicable, is all
molecular dynamics simulations. Building upon this that is needed. Experimental data is not utilized for
framework, the current work integrates the wettability performing MD simulations. By comparing system
alteration CFD workflow with molecular dynamics behaviour or properties (such as density, viscosity,
simulations of three-phase wettability models [22] of oil diffusivity, miscibility, moduli, and stress-strain curves)
droplets in brine on mineral surfaces. In particular, we calculated from MD simulations against experimentally
simulated the change in contact angle of an oil/brine/quartz measured lab data, molecular dynamics results are rigorously
system when the quartz surface transforms from a water-wet validated. With over six decades [25] of research and
to an oil-wet condition under in-situ reservoir conditions development on applications of MD simulations, it has
using molecular dynamics. The computed contact angles are become a mainstay in geochemistry, catalyst design,
provided as inputs to the multiphase LBM simulation, polymerization reaction optimization, and sustainable
whereas in previous work [21] they were assumed. innovation.

In this work, we describe a numerical digital rock workflow Since this technique is applicable to systems made up of any
starting from pore-scale 3D image acquisition to fluid flow element in the periodic table, it applies to a wide range of
simulation, including a previously demonstrated wettability topics. In the energy industry, specifically in the oil and gas
alteration process. Wettability alteration is carried out in a upstream sector, molecular dynamics simulations are used to
digitally reconstructed 3D pore-grain model, considering study extraction, transport and processing of hydrocarbons.
micro-scale interactions. In parallel, molecular dynamics Design and optimization of fracturing fluids, gaining key
simulations are performed at the reservoir pressure and insights into scale and hydrate formation [26], studying the
temperature for oil droplets in brine on mineral surfaces to aggregation behaviour of asphaltenes [27], and optimizing
calculate contact angles. The resulting contact angles are used sustainable and green catalysts [28] for oil refining are a few
in multiphase pore-scale fluid flow simulations to obtain examples where molecular dynamics simulations have been
spatial distributions of fluid wettability. A molecular
 The 36th International Symposium of the Society of Core Analysts

successfully applied. In the area of sustainability, molecular These surfels precisely define the boundaries between pores
dynamics simulations have been used to investigate CO2 and grains, resulting in accurate flux calculations near
absorption in porous materials such as MOFs [29, 30] and boundaries and better accuracy at lower numerical
zeolites [30], as well as in rocks and cement [31]. Critical resolutions. Moreover, the utilization of surfels enhances the
phenomena such as CO2 dissolution in brine present in calculation of local curvature and curvature-based pressure as
reservoir rocks [32], supercritical CO2 impact on degradation required for the wettability alteration workflow used in the
and ageing of elastomer [33] completion tools can also be current work.
simulated. Complex systems, such as the formation of oil
droplets on mineral surfaces [22], can be simulated and
2.3 Workflow description
tracked in both time and space, and thermodynamic and
dynamic properties [23] can be calculated, which can be
2.3.1 Sample Imaging
directly compared with experimental measurements.
As for the digital rock multiphase workflow, a 3D image of a
In this study, all MD simulations [34] were performed using
Berea sandstone sample is obtained using x-ray micro-
the quantum mechanics parameterized COMPASS
tomography (microCT). Figure 1 shows a sub-volume of the
(Condensed-phase Optimized Molecular Potentials for
3D microCT scan oriented perpendicular to the fluid
Atomistic Simulation Studies) III force field [35]. The
displacement direction. The 3D image went through image
COMPASS III force field is applicable to a broad range of
pre-processing, including artifact corrections, filtering and
molecules and polymers and accurately describes multiphase
subsequent segmentation, with details available in [42].
systems such as liquid-liquid equilibrium and was used in this
study because we are simulating three-phase wettability
models to predict oil in brine on mineral surface contact
angles.

2.2 Multiphase LBM simulation

Multiphase numerical simulations are conducted using a

lattice Boltzmann method (LBM) solver in a 3D segmented
pore space of the chosen rock domain. Unlike the traditional
Navier-Stokes equations, the LBM is based on mesoscopic
kinetic theory and solves a discrete form of the Boltzmann Fig. 1. Sub-volume of a 3D microCT scan for the Berea
transport equations [36-40]. The LBM operates locally on a Sandstone used in this work: Left, original greyscale image,
cubic lattice, allowing for highly parallel computational right, segmented image.
performance [41]. This makes it an excellent choice for
simulating porous media applications, which often involve The 3D micro-CT image of a Berea sandstone was acquired
large domains with intricate pore structures. The LBM has at 2.0 𝜇m per voxel resolution to ensure good mesh quality
been successfully used in a range of porous media for subsequent fluid flow simulation. At this resolution, the
applications, including hydrocarbon-bearing rocks [42, 43], resulting 3D model shows sufficient pore connectivity in all
gas diffusion layers in fuel cells [44, 45] and porous three directions, with a total resolved porosity around 14%.
electrodes in lithium-ion batteries [46, 47]. The typical range of porosity for this Berea sample is
18~20%. The micro-CT image resolved mostly the relevant
In this study, the LBM implementation is based on the connected porosity, while the missing porosity (crevices and
multiphase model introduced by Shan and Chen [48], which clay porosity) does not contribute much to the flow. The pore
has undergone several improvements to achieve accurate bottleneck radius, which is defined by the radius of the largest
simulations for porous media applications even at coarse rigid sphere that can be percolated across the 3D pore
resolutions (pore throat of as low as 3 voxels radii) and high structure without getting stuck, needs to be large enough to
viscosity ratios (up to 100) scenarios [41, 49]. The multiphase ensure sufficient accuracy of the fluid flow simulation and
model considers different particle species to represent that requirement is solver specific. In a previously published
separate fluid phases, and the interaction forces between these work [41], it was demonstrated that the LBM solver used for
species determine the inter-component interfacial tension. this study can provide accurate results at 3 voxel resolution.
Consequently, the interfaces between different components The bottleneck radii for this Berea sample model were
are automatically determined once the species’ interactions calculated to be 3.0, 3.6 and 3.2 voxels in the x, y and z
are defined. This eliminates the need for explicit interface directions, respectively.
tracking methods like volume-of-fluid or level-set
approaches, simplifying the simulation process. Afterwards, a representative elementary volume (REV)
analysis is performed on the captured image and a subdomain
To accurately capture the physics of multi-component fluid of 500 voxels cubic size is selected for subsequent multiphase
flow, even at low resolution, the current LBM analysis. The details of the REV analysis can be found in [21].
implementation incorporates triangulated surface elements,
known as surfels, in addition to voxel elements [50-52]. 2.3.2. Multiphase Simulation Setup
 The 36th International Symposium of the Society of Core Analysts

In the selected rock domain, a primary drainage simulation

driven by capillary pressure is conducted. Initially, the
simulated pore space is fully saturated with water and has a
uniform water-wet condition (hydroxylated surface as
modeled in the MD simulation) applied throughout. During
the primary drainage simulation, the pressure of the oil is
gradually increased, leading to the oil displacing the water in
the pore space. At each pressure step, once the saturation is
converged, capillary pressure is calculated as the pressure
difference between the bulk oil and water phases. By bulk we
mean the average value in each phase. Concurrently, the
wetting condition of the grains dynamically transitions from
a water-wet (hydroxylated surface) to an oil-wet (alkylated
surface) condition over the course of the simulation. The
wetting condition of the grain surface is altered iteratively, Fig. 2. Molecular representations of components in the three-
where each iteration leads to further oil invasion and phase wettability model: n-octane, water, NaCl, and unit cell
incremental oil-wet surface fraction. An illustration of this of 𝛼-quartz crystal.
process is provided in the results section. The simulation
setup closely resembles the laboratory process for capillary The quartz surfaces were prepared by converting the
pressure displacement using a porous plate. In this study, a primitive unit cell of -quartz to its orthorhombic cell
unique feature of the current lattice Boltzmann method is equivalent as described in [53]. A 10x6x6 supercell with
used – a volume-free, massless numerical membrane that dimensions of 4.9 nm x 5.1 nm x 3.8 nm was created, and the
serves as a substitute for the porous plate in laboratory (001) surface (the (001) -quartz surface is known to be the
analysis. This numerical membrane effectively addresses most stable [54]) was cleaved, capping the surface with
computational expenses and potential numerical issues hydroxyl groups to create the theoretical maximum (9.4
associated with modeling a finely resolved porous plate. For OH/nm2) [55] hydroxylated surface coverage. However, in
further information regarding the porous plate simulation nature [56], the hydroxyl surface coverage on -quartz is 4.6
setup, refer to [21]. OH/nm2, about half of the theoretical maximum, and
experimental measurements indicate that alkylation [57]
2.3.3. Molecular dynamics model reduces the hydroxylation. Thus, to compare the two
extremes, we have alkylated a second -quartz surface to 4.8
Molecular dynamics (MD) simulations are utilized to pentyl groups per nm2, with the remaining coverage being
compute oil/brine contact angles on -quartz, the main hydroxylated at 4.6 OH/nm2. For oil droplets, a fully
component of Berea sandstone. Contact angles are highly hydroxylated surface is hydrophilic, which will tend to result
affected by the surface chemistry of the mineral. When water in a water-wet contact angle (<90°) [55], while a fully
is more prevalent, the surface of -quartz will tend to alkylated surface will tend to be oil-wet (>90°). The resulting
hydroxylate, which results in hydroxyl groups (-OH) fully hydroxylated and alkylated surfaces were replicated 8
covalently bonding to the surface, but in the presence of oil, times in the x-direction to create a 39.3 nm x 5.1 nm x 3.8 nm
the surface may undergo alkylation, with alkyl groups (- surface in which the oil and brine phases were added on top.
CnH2n+1) covalently bonding to the surface. The exchange of The oil phase contained 617 n-octane molecules. The brine
these functional groups at the oil/water interface while the phase contained 55,824 water and 36 NaCl molecules. The
fluids displace each other is the molecular mechanism for hydroxylated -quartz mineral phase contained 66,884
wettability alteration in reservoir rocks. atoms, and the alkylated -quartz mineral phase contained
71,520 atoms. The total system sizes were 259,470 atoms for
Fully-atomistic three-phase wettability models [22] were
the fully hydroxylated surface model and 264,106 atoms for
used to calculate the contact angles of cylindrical oil droplets the alkylated surface model. The initial box dimensions were
in brine on -quartz surfaces. The oil phase was modelled as approximately 39.3 nm x 5.1 nm x 16.4 nm for both the
n-octane. The second phase was the brine phase which was hydroxylated and alkylated three-phase wettability surface
modelled as 0.2 wt % NaCl in water. The third phase was the models.
mineral phase, modelled as fully hydroxylated or alkylated -
quartz. The molecular components of the three-phase
wettability model are shown in Figure 2. 2.3.4. Molecular dynamics simulation setup

Contact angle simulations were performed using all-atom

molecular dynamics (MD) simulations of the three-phase
wettability models for cylindrical oil droplets on the fully
hydroxylated and alkylated -quartz surfaces. MD
simulations were performed in the isobaric isothermic (NPT)
ensemble at a referential reservoir condition: 6800 psi and
240 °F. Temperature and pressure were maintained using a
 The 36th International Symposium of the Society of Core Analysts

Berendsen thermostat and barostat, respectively. Simulations angles are calculated by simulating oil droplet on top of a
were run for 5 ns using a 1 fs time step with atomic hydroxyl and alkyl treated quartz surface from the molecular
coordinates recorded every 50 ps. The oil droplets are fully dynamics simulation. We assumed that the increase in
formed within 2 ns, but an additional 3 ns of dynamics were capillary pressure during the drainage simulation is not
run to confirm that no significant changes occurred to the sufficient enough to warrant a significant change in contact
droplet shape. All intermolecular and intramolecular angles. Indeed, as evidenced from [59], the degree of pressure
interactions were modeled using the COMPASS III force change required to see any discernible change in surface
field [35], with van der Waals interactions truncated with a tension is a few of orders higher than the capillary pressure
cubic spline and a 1.25 nm cut-off. Electrostatic interactions change observed in the CFD simulation. Instead of
were summed using the Particle-Particle Particle-Mesh calculating the contact angle variations due to pressure
(PPPM) method [58]. change, we simulated the change in contact angle by
modelling changes of functional groups (hydroxyl for water-
2.3.5. Wettability alteration wet and alkyl for oil-wet) capping the quartz surface. Thus,
the MD simulations are performed only once and the
The current workflow involves the integration of molecular computed contact angle values are used as inputs in various
and micro-scale fluid simulation data to achieve wettability pressure stages of CFD simulation. At step-2, the curvature
alteration. Firstly, molecular dynamics simulations are of the pore/grain boundary is calculated at each voxel of the
employed to calculate the oil/water contact angle for a fully 3d micro-CT image, as
hydroxylated (water-wet) and alkylated quartz surface (oil-
wet). Secondly, utilizing the contact angle inputs obtained

Fig. 4. Oil droplet formation on (a) hydroxylated and (b)

alkylated -quartz surfaces from fully-atomistic three-phase
wettability molecular dynamics simulations. Brine (water
and NaCl) is explicitly modeled in the simulation but is not
shown in the images for better clarity of the oil droplets,
which are represented in all black. The blue background is for
visualization purposes and has no physical meaning.

described in [21], to incorporate microscale roughness. The

curvature based pressure, 𝑃𝑣 , is calculated as:

̅|
𝑃𝑣 = 𝜎|𝐻 (1)

Fig. 3. Molecular dynamics and lattice Boltzmann workflow Here, 𝜎 and |𝐻 ̅| represent the interfacial tension and mean
for wettability alteration. curvature of the pore/grain boundary. Note that 𝑃𝑣 varies
spatially with the curvature of the pore/grain boundary and it
from the MD simulation, a primary drainage displacement is pre-computed before performing the multiphase LBM
simulation is conducted using the multiphase LBM method. simulation. In the multiphase LBM simulation (step-3), as the
pressure at the oil buffer increases, oil displaces water in the
Figure 3 provides a detailed overview of the workflow
implementation. At step-1, water-wet and oil-wet contact
 The 36th International Symposium of the Society of Core Analysts

1st quadrant 2nd quadrant

4th quadrant 3rd quadrant

Fig. 5. Various stages of wettability alteration simulation. The 1st quadrant in each figure shows an isosurface of oil, representing
oil pore volume (beige) and a volume rendering of water (blue), 2 nd quadrant shows the water saturation profile along the flow
direction, 3rd quadrant shows capillary pressure versus water saturation curve and 4th quadrant shows oil-wet (red) and water wet
(blue) surface within the pore space in 3D.

pore space, where the water-wet contact angle was Here, 𝑃𝑐 and Π 𝑚𝑎𝑥 represent the capillary pressure and
determined by the MD simulation for the hydroxylated maximum disjoining pressure, respectively. The capillary
surface. During the first iteration, surface fraction, 𝑆, defined pressure corresponds to the pressure difference between bulk
as the ratio of oil-wet and total surface area, is set to zero. oil/water phases within the pore space and is calculated from
the various stages of the CFD simulation as the pressure in oil
Once saturation has reached convergence for a given buffer is increased. The disjoining pressure indicates the
pressure, the following augmented Young-Laplace equation pressure needed to rupture the water film for the oil to make
(in step -5) is used to determine the eligibility of the grain contact with the grain surface. According to [16], when oil is
surface in the vicinity of the oil (step-4) for wettability pushed against a water-wet flat surface, the capillary pressure
alteration: resistance caused by pore curvature is absent, but a certain
level of pressure is still required for film breakage, known as
𝑃𝑐 > 𝑃𝑣 + Π max (2) the disjoining pressure. However, for a curved interface, both
 The 36th International Symposium of the Society of Core Analysts

the capillary pressure caused by curvature and the disjoining

pressure need to be taken into account. In the current work,
the value for the maximum disjoining pressure is assumed to
be 5 (104 ) Pa, from the available literature [16].

At step-6, the eligible surface fraction for the wettability

alteration (𝑆𝑖+1 ) is calculated by taking a subset of the surface
in step-4 and 5, which is later assigned to an oil-wet contact
angle as computed from the MD simulations for the alkylated
surface (step-7). The simulation is then resumed maintaining
the same buffer pressure (step-8). This process allows further
redistribution of the invading fluid, and steps 4 to 8 in Fig. 3
are iteratively repeated until convergence in the surface
fraction (𝑆) is achieved. Afterwards, the oil pressure is
increased to 𝑃𝑐,𝑖+1 (step-9) and the workflow starts over from
step-3.

3 Results and Discussions

3.1 Contact angle calculation

Oil droplet contact angles were measured from the final

configurations of the all-atom molecular dynamics three-
phase wettability model simulations shown in Figure 4.
Contact angles were measured [60] by fitting a circle to where
the oil-brine interface meets the solid surface. Measurements
were repeated 5 times to calculate averages and standard
deviations. The oil droplet contact angle was 53 ± 2° on the
fully hydroxylated -quartz mineral surface, indicating a
water-wet condition due to the high concentration of
hydroxyl groups (9.4 OH/nm2) which leads to a hydrophilic
interaction between the brine and -quartz surface. As
described in 2.3.3, the alkylated surface contains 4.8 pentyl
groups per nm2 which are covalently bonded to the -quartz.
The alkylated surface has a hydrophobic interaction with the
brine, which causes the oil droplet’s wettability on the
alkylated surface to increase compared to the fully
hydroxylated hydrophilic surface. This increased wettability
on the alkylated surface results in an oil-wet condition with a
contact angle of 150 ± 4°. These results are summarized in
Table 1.

Table 1. Contact angles calculated from molecular dynamics

simulations

3.2 Wettability alteration

Figure 5 displays the results of capillary pressure drainage

(oil displacing water) simulation in a water-wet Berea Fig. 6. Saturation change in the iterative wettability alteration
sandstone filled with water. General trends of the observed process. Blue, red and orange colors represent the water
capillary pressure curve are very similar to previously phase, the oil phase before wettability alteration and the
reported experimental and numerical studies. As the oil newly intruded oil phase due to wettability alteration,
pressure is increased, oil starts to invade the pore space from respectively.
the oil buffer. Initially, the water saturation changes are
 The 36th International Symposium of the Society of Core Analysts

methodology described in section 2.3.5, oil further infiltrates

the pore space, which is identified by the orange color in
Figure 6b. With the updated fluid distribution, another round
of wettability alteration is conducted, resulting in further
changes in fluid saturation observed in Figure 6c. In the
current workflow, the wetting condition is iteratively
modified until both saturation and the surface fraction are
converged. Thus, Figure 6 shows that oil can get into smaller
(a) (b) (c) pores through an iterative wettability alteration process more
effectively than without alteration.
Fig. 7. Determination of wettability alteration eligibility. (a)
a pore space partially filled with oil (red), (b) pore/grain This wettability alteration criteria can be described as an
boundary that satisfies the Young-Laplace criteria (red) and implementation of the augmented Young-Laplace equation,
that does not (blue), and (c) pore/grain boundary after Eq (1), that also accounts for quasi-static changes in fluid
wettability alteration with oil-wet (red) and water-wet (blue) saturation. Note that the presence of an oil phase alone is not
surface. sufficient for wettability alteration, as it also requires the
relationship between local capillary pressure, curvature-
minimal in response to the increase in oil pressure. However, based pressure, and maximum disjoining pressure in Eq (1)
once the pressure surpasses the capillary pressure imposed by to be satisfied. Figure 7 illustrates a pore-scale event where a
the critical pore throat radius, oil rapidly invades the pore certain pore is partially filled with oil, but depending on the
space behind these pore throats, resulting in a phenomenon Young-Laplace equation, only a fraction of the oil-contacted
known as “Haines jump” [61]. It should be noted that not all grain is altered towards the oil-wet condition.
of the pore space is filled during Haines jumps, particularly
during a slow convergence to static equilibrium. In this stage, The current workflow incorporates the consideration of
the phase pressure difference is not significant enough for the microscale pore shape for wettability alteration by calculating
non-wetting phase to rapidly invade, and the flow of oil can the 3D curvature throughout the entire pore space. As
be described as quasi-static and capillary dominant. In such a discussed in [63], reservoirs can be classified as mixed-wet
scenario, oil droplets can be detached when passing through large (MWL), mixed-wet small (MWS), or fractional-wet
narrow pore throats, similar to the “Roof snap off” described (FW). MWL reservoirs mainly consist of concave pores, with
in [62]. More details of these phenomena can be found in mixed wet conditions observed in the larger pores. In
[21]. Once the dominating percolating flow paths are invaded contrast, interestingly, MWS reservoirs consist of convex-
by oil, it starts to displace water in smaller pores at higher shaped pores, where smaller pores exhibit mixed wet
pressure slowly. However, most of the remaining water is characteristics. That study [63] analysed 13 reservoirs from
either trapped in the pore space or coating the grains as a the North Sea oil fields database, with five being classified as
higher pressure thin water film by this time. As a result, the MWL, four as MWS, and four as FW.
slope of the capillary pressure curve becomes steep again.

Illustrating the current workflow, Figure 5 displays the

capillary pressure curve, highlighting the static convergence
points after each pressure increment (black) and wettability
alteration (red). As described in section 2.3.5, the process
ensures that fluid saturation changes due to pressure ramp-up
converges in the time domain before a modification to the
wetting condition is made. Wettability alteration induces
further changes to the overall saturation. In accordance with
[21], this phenomenon can be attributed to the slow dynamic
impact of wettability alteration, allowing the non-wetting
fluid to intrude into very small pores without needing to
overcome capillary pressure. Moreover, our proposed
molecular mechanism of hydroxyl to alkyl functional group
replacement when oil displaces water is consistent with the
slow dynamics observed experimentally.

An example of this wettability alteration process is presented

in Figure 6, illustrating the iterative convergence of
wettability and the resulting fluid saturation. Figure 6 depicts Fig. 8. Calculated mean curvature at pore/grain boundary in
the fluid distribution within a specific plane perpendicular to one of the simulation subdomains.
the overall flow direction. In Figure 6a, oil (red) and water
(blue) are present after saturation convergence at a given
pressure. Following the alteration of wettability using the
 The 36th International Symposium of the Society of Core Analysts

Laplace equation. During the multiphase simulation,

approximately 23% of the water becomes trapped at the end
of the drainage process. In contrast, the geometric intrusion
analysis relies on the connectivity of the oil phase and the size
of the pore space, neglecting the connectivity of the water
phase. Consequently, this method is unable to predict the
presence of the trapped water phase, leading to a discrepancy
observed during the late drainage process. However, it is
important to note that the capillary entry pressures observed
in both methods are very similar.

Figure 10 shows the oil-wet surface fraction obtained from

multiphase analysis and the oil-contacted surface fraction
predicted by the MIP analysis as a function of water
saturation. Note that, even for saturation values where the two
Fig. 9. Comparison of the capillary pressure curve in the capillary pressure curves (Figure 9) are similar, between
multiphase simulation and geometric intrusion process. 40%-80%, the MIP-predicted oil-wet surface fraction (Figure
10) is considerably lower than that obtained from the
multiphase simulation. This difference suggests that, in the
multiphase simulation that incorporates the Young-Laplace
equation (Eq (1)) and a slow dynamic interface advancing
scheme during wettability alteration, supported by the
molecular mechanism of functional group substitution, oil
can penetrate deeper into the smaller pores at a given
saturation. In contrast, the oil saturation in the MIP analysis
strictly represents the pore space that can be invaded
according to the Young-Laplace equation. In the absence of
dynamic wettability alteration, the MIP process cannot
predict oil intrusion into smaller pores without overcoming
capillary pressure first.

4 Conclusion

A numerical wettability alteration process was presented that

integrates molecular-driven physicochemical properties with
Fig. 10. Oil-wet surface fraction as predicted from the pore-scale capillary flow in order to predict rock wettability
multiphase drainage and geometric MIP analysis. under in-situ reservoir conditions. We proposed a molecular
mechanism for wettability alteration, which involves the
alkylation of mineral surfaces as oil displaces water within
In the current study, the curvature-based pressure, 𝑃𝑣 , is
̅|, of the the pore space of rock. An iterative workflow for multiphase
computed as a function of the mean curvature, |𝐻 CFD simulation was presented, which determines the spatial
pore/grain boundary (Eq. 1). |𝐻 ̅| values are positive or
rock wettability distribution as a function of applied capillary
negative for concave or convex pores, respectively. Figure 8 pressure and saturation. Inputs to the workflow were contact
showcases a mean curvature map in a fraction of the angles calculated from fully atomistic molecular dynamics
simulated pore space. 𝑃𝑣 , calculated from equation (1), is used models of oil droplets on hydroxylated and alkylated quartz
in equation (2) for surface wettability candidacy. According surfaces. Detailed pore-scale events during wettability
to equation (2), for a given maximum disjoining alteration were examined and compared with an image-based
pressure (Π 𝑚𝑎𝑥 ), convex pores are more likely candidates for geometric intrusion analysis, which mimics mercury
wettability alteration. Thus, using the augmented Young- intrusion porosimetry (MIP) virtually.
Laplace equation and corresponding curvature calculation in
the current workflow can potentially explain the wettability
characterization of the MWS reservoirs. The developed virtual wettability alteration process achieves
a force balance by considering local capillary pressure,
disjoining pressure and curvature-based pressure. A detailed
Figure 9 compares the drainage capillary pressure curve description of the slow oil invasion mechanism during
obtained from the multiphase simulation and an image-based wettability alteration has been provided, taking into account
geometric intrusion analysis, which resembles the laboratory pore/grain curvature and its influence on the augmented
procedure of mercury intrusion porosimetry (MIP). The Young-Laplace equation. Comparison with the virtual MIP
geometric intrusion process employs the inscribed sphere analysis reveals that iterative wettability alteration allows oil
method, where the radius of the sphere corresponds to the to invade smaller pores without overcoming capillary
applied capillary pressure determined using the Young- pressure. Additionally, the overall oil-wet surface fraction
 The 36th International Symposium of the Society of Core Analysts

was found to be significantly higher than that obtained from 5. M. Rücker, W.-B. Bartels, K. Singh, N. Brussee, A.
virtual MIP analysis at a given water saturation. The Coorn, H. A. van der Linde, A. Bonnin, H. Ott, S.
combination of slow dynamic wettability alteration and M. Hassanizadeh, M. J. Blunt, H. Mahani, A.
curvature effects offers an explanation for the observed Georgiadis, S. Berg, “The effect of mixed
wettability conditions in mixed-wet small reservoirs. wettability on pore-scale flow regimes based on a
flooding experiment in Ketton limestone,” Geophys.
Res. Lett. 46, 3225-3234 (2019).
Through a synergistic combination of atomistic molecular
dynamics and multiphase CFD fluid flow simulations, the 6. C. McPhee, J. Reed, I. Zubizarreta, “Chapter 7-
presented workflow enables wettability predictions based on wettability and wettability tests,” Developments in
first principles numerical simulations. The developed process Petroleum Science, 64, 313-345, (2015).
can serve as a valuable addition to laboratory procedures for
determining wettability. Simultaneously, it offers realistic 7. G. Garfi, C.M. John, M. Rücker, Q. Lin, C. Spurin,
inputs for a range of digital rock workflows, including S. Berg, S. Krevor, “Determination of the spatial
capillary pressure, relative permeability, and enhanced oil distribution of wetting in the pore networks of
recovery. Consequently, time-constrained projects can now Error! Bookmark not defined.”, J Colloid
embrace a combination of virtual and laboratory-based Interface Sci., 613, 786 (2022).
wettability assessment, resulting in enhanced uncertainty
quantification of the reservoirs. The present work forms the 8. G. Garfi, C.M. John, Q. Lin, S. Berg, S. Krevor,
basis for a more detailed analysis of disjoining pressure “Fluid surface coverage showing the controls of
isotherms. This will be the subject of future work. rock mineralogy on the wetting state,” Geophys.
Res. Lett. 47 (8) (2020).

List of Symbols and Notations 9. S. Foroughi, B. Bijeljic, Q. Lin, A.Q. Raeini, M.J.
Blunt, “Pore-by-pore modeling, analysis, and
prediction of two-phase flow in mixed-wet rocks,”
Phys. Rev. E 102, 02330 (2020).

10. Q. Lin, B. Bijeljic, S. Berg, R. Pini, M.J. Blunt, S.

Krevor, “Minimal surfaces in porous media: Pore-
scale imaging of multiphase flow in an altered
wettability Bentheimer sandstone”, Phys. Rev. E, 99,
6 (2019).

11. A. Scanziani, K. Singh, M.J. Blunt, A. Guadagnini,

“Automatic method for estimation of in situ
effective contact angle from X-ray micro
tomography images of two-phase flow in porous
media”, J. Colloid Interface Sci. 496, 51 (2017).

References 12. W.-B. Bartels, M. Rücker, M. Boone, T. Bultreys,

H. Mahani, S. Berg, S.M. Hassanizadeh, V. Cnudde,
1. W.G. Anderson, “Wettability literature survey- “Imaging spontaneous imbibition in full Darcy-scale
part4: Effects of wettability on capillary pressure,” J samples at pore-scale resolution by fast X-ray
Pet Technol 39, 1283 (1987) tomography”, Water Resour. Res., 55 7072 (2019).

2. W.G. Anderson, “Wettability literature survey-part 13. A.M. Alhammadi, A. AlRatrout, K. Singh, B.
5: The effects of wettability on relative Bijeljic, M.J. Blunt, “In situ characterization of
permeability,” J Pet Technol 39, 1453 (1987) mixed-wettability in a reservoir rock at subsurface
conditions”, Sci. Rep., 7, 10753 (2017).
3. Q. Lin, B. Bijeljic, S. Berg, R. Pini, M.J. Blunt, S.
Krevor, “Minimal surfaces in porous media: Pore- 14. M. Andrew, B. Bijeljic, M.J. Blunt, “Pore-scale
scale imaging of multiphase flow in an altered contact angle measurements at reservoir conditions
wettability Bentheimer sandstone”, Phys. Rev. E, 99, using X-ray microtomography”, Adv. Water
6 (2019). Resour., 68, 24 (2014).

4. M. Rücker , W.-B. Bartels , G. Garfi, M. Shams, T. 15. M. Prodanović, W.B. Lindquist, R.S. Seright,
Bultreys, M. Boone, S. Pieterse, G.C. Maitland, S. “Residual fluid blobs and contact angle
Krevor, V. Cnudde, H. Mahani, S. Berg, A. measurements from X-ray images of fluid
Georgiadis, P.F. Luckham, “Relationship between displacement,” XVI International Conference on
wetting and capillary pressure in a crude oil/brine/ Computational Methods in Water Resources,
rock system: From nano-scale to core-scale”, J (2006).
Colloid Interface Sci., 562, 159(2020)
 The 36th International Symposium of the Society of Core Analysts

16. C.J. Radke, A.R. Kovscek, H. Wong, “A pore-level 30. B. Liu and B. Smit, “Comparative Molecular
scenario for the development of mixed wettability in Simulation Study of CO2/N2 and CH4/N2 Separation
oil reservoirs,” SPE 24880, SPE Ann. Tech Conf. in Zeolites and Metal-Organic Frameworks,”
and Exhib., Washington, D.C. USA, Oct, 1992. Langmuir, 25, 5818-5926, 2009.

17. B.V. Derjaguin, E.V. Obukhov, "Anomalous 31. A. Al-Ostaz, W. Wu, A.H.-D. Cheng, C.R. Song, “A
properties of thin polymolecular films V.,” Acta molecular dynamics and microporomechanics study
Physicochim. URSS, 10(1), 25-44 (1939a). on the mechanical properties of major constituents
of hydrated cement” Composites Part B:
18. B.V. Derjaguin, E.V. Obukhov, "Anomalous Engineering, 41, 543-549, 2010.
properties of thin polymolecular films V.,” Acta
Physicochim. URSS, 10(2), 153-174 (1939b). 32. G. Javanbakht, M. Sedghi, W. Welch, L. Goual,
“Molecular Dynamics Simulations of
19. S. Iglauer, “CO2-water-rock wettability: variability, CO2/Water/Quartz Interfacial Properties: Impact of
influencing factors, and implications for CO2 CO2 Dissolution in Water,” Langmuir, 31, 5812-
geostorage“ Acc. Chem. Res., 50, 1134−1142, 2017. 5819, 2015.
20. J.S. Buckley, C. Bousseau, Y. Liu, “Wetting 33. M. Luo, Z.A. Putman, J. Incavo, M.Y. Huang, J.B.
alteration by brine and crude oil: from contact angles McLaughlin, S. Krishnan, “Molecular Simulations
to cores,” SPE J., 1, 341-350, 1996. and Experimental Characterization of Fluorinated
Nitrile Butadiene Elastomers with Low H2S
21. A. Islam, R. T. Salazar, B. Crouse, “Digital rock
Permeability,” Ind. Eng. Chem. Res., 58, 14823-
workflow to calculate wettability distribution in a
14838, 2019
reservoir rock,” SCA 2022-024, Int. Symp. Soc.
Core Analysts, Austin, TX, USA, (2022). 34. BIOVIA, Dassault Systèmes, Materials Studio,
2022, San Diego: Dassault Systèmes, 2022.
22. S. Xu, J. Wang, J. Wu, Q. Liu, C. Sun, B. Bai, “Oil
contact angles in a water-decane-silicon dioxide 35. R.L.C. Akkermans, N.A. Spenley, S.H. Robertson,
system: effects of surface charge,” Nanoscale “COMPASS III: Automated Fitting Workflows and
Research Letters, 13, 1-9, 2018. Extension to Ionic Liquids,” Mol. Sim., 47, 540-551,
2021.
23. K.E. Gubbins and J.D. Moore, “Molecular modeling
of matter: Impact and prospects in engineering,” Ind. 36. S. Chen and G. Doolen, “Lattice Boltzmann method
Eng. Chem. Res., 49, 3026-3046, 2010. for fluid flows,” Annu. Rev. Fluid Mech., 30, 329
(1998).
24. G.J. Martyna, D.J. Tobias, M.L. Klein, “Constant
pressure molecular dynamics algorithms,” J. Chem. 37. S. Chen, H. Chen, D. Martinez, and W.
Phys., 101, 4177-4189, 1994. Mattheus,“Lattice Boltzmann model for simulation
of magnetohydrodynamics,” Phys. Rev. Lett., 67,
25. B.J. Alder and T.E. Wainwright, “Molecular
3776 (1991).
dynamics by electronic computers,” In International
Symposium on Statistical Mechanical Theory of 38. H. Chen, S. Chen, W. Matthaeus, “Recovery of the
Transport Processes, Brussels, 1956; Prigogine, I., Navier - Stokes equations using a lattice-gas
Ed.; Interscience: New York, 1958. Boltzmann method,” Phys. Rev. A, 45, R5339
(1992).
26. C. Moon, P.C. Taylor, P.M. Rodger, “Molecular
dynamics study of gas hydrate formation,” J. Am. 39. Y. Qian, D. d'Humieres, P. Lallemand, “Lattice
Chem. Soc., 125, 4706-4707, 2003. BGK models for Navier-Stokes equation,”
Europhys. Lett., 17, 479 (1992)
27. T.F. Headen, E.S. Boek, G. Jackson, T.S. Totton,
E.A. Müller, “Simulation of asphaltene aggregation 40. H. Chen, “Volumetric formulation of the lattice
through molecular dynamics: Insights and Boltzmann method for fluid dynamics: Basic
limitations” Energy Fuels, 31, 1108-1125, 2017. concept,” Phys. Rev. E, 58, 3955(1998)
28. A. Kate, K.K. Sahu, J. Pandey, M. Mishra, P.K. 41. H. Otomo, H. Fan, R. Hazlett, Y. Li, I. Staroselsky,
Sharma, “Green catalysis for chemical R. Zhang, R., and H. Chen, H., “Simulation of
transformation: The need for the sustainable residual oil displacement in a sinusoidal channel
development” Curr. Res. Green. Sustain. Chem., 5, with the lattice Boltzmann method,” Comptes
100248, 2022. Rendus Mécanique, 343, 559 (2015).
29. T. Düren, Y.-S. Bae, R.Q. Snurr, “Using molecular 42. G.R. Jerauld, J. Fredrich, N. Lane, Q. Sheng, B.
simulation to characterise metal–organic Crouse, D.M. Freed, A. Fager, and R. Xu,
frameworks for adsorption applications,” Chem. “Validation of a workflow for digitally measuring
Soc. Rev., 38, 1237-1247, 2009. relative permeability,” SPE 188688, SPE Abu Dhabi
 The 36th International Symposium of the Society of Core Analysts

Int. Pet. Exhib. & Conf., Abu Dhabi, U.A.E., Nov., dynamics simulations,” ChemPhysChem, 9, 994-
2017. 1002, 2008.
43. H. Andrä, N. Combaret, J. Dvorkin, E. Glatt, J. Han, 54. V.V. Murashov, “Reconstruction of pristine and
M. Kabel, Y. Keehm, F. Krzikalla, M. Lee, C. hydrolyzed quartz surfaces,” J. Phys. Chem. B., 109,
Madonna, M. Marsh, T. Mukerji, E. H. Saenger, R. 4144-4151, 2005.
Sain, N. Saxena, S. Ricker, A. Wiegmann, X. Zhan,
“Digital rock physics benchmarks—part II: 55. A. Abramov, A. Keshavarz, S. Iglauer, “Wettability
Computing effective properties,” Comp. Geosci., of fully hydroxylated and alkylated (001) α-quartz
50, 33 (2013) surface in carbon dioxide atmosphere,” J. Phys.
Chem. C, 123, 9027-9040, 2019.
44. J. Wu, J.J. Huang, “Dynamic behaviors of liquid
droplets on a gas diffusion layer surface: 56. L.T. Zhuravlev, “The surface chemistry of
Hybrid lattice Boltzmann investigation,” J. App. amorphous silica. Zhuravlev model,” Colloids Surf.
Phys., 118, 044902 (2015) A, 173, 1-38, 2000.

45. M.A. Safi, N.I. Prasianakis, J. Mantzaras, A. 57. M. Ali, S. Al-Anssari, M. Arif, A., Barifcani, M.
Lamibrac, F.N. Buchi, “Experimental and pore-level Sarmadivaleh, L Stalker, M. Lebedev, S. Iglauer,
numerical investigation of water evaporation in gas “Organic acid concentration thresholds for ageing of
diffusion layers of polymer electrolyte fuel cells,” carbonate minerals: Implications for CO2
Int J Heat and Mass Trans.,115, 238 (2017). trapping/storage,” J. Colloid Interface Sci., 534, 88-
94, 2019.
46. L. Martin, P. Benedikt, K. Benjamin, W. Julius, D.
Timo, S. Volker, L. Arnulf, “Understanding 58. R.W. Hockney, J.W. Eastwood, Computer
electrolyte filling of Lithium-ion battery electrodes simulation using particles, IOP Publishing: Bristol
on the pore scale using the Lattice Boltzmann (1988).
Method,” Batteries and Supercaps, e202200090, (to
59. B.Y. Cai, J.T. Yang, T.M. Guo, “Interfacial tension
be published)
of hydrocarbon+ water/brine systems under high
47. D.H. Jeon, “Wettability in electrodes and its impact pressure,” J. Chem. Eng. Data, 41, 493-496 (1996).
on the performance of Lithium-ion batteries,”
60. C.A. Schneider, W.S. Rasband, and K.W. Eliceiri,
Energy Storage Materials, 18, 139 (2019)
“NIH Image to ImageJ: 25 years of image analysis,”
48. X. Shan and H. Chen, “Lattice Boltzmann model for Nature Methods, 9, 671-675, 2012.
simulating flows with multiple phases and
61. W.B. Haines, “Studies in the physical properties of
components,” Phys Rev E, 47, 1815 (1993).
soil : the hysteresis effect in capillary properties and
49. H. Otomo, H. Fan, Y. Li, M. Dressler, I. Staroselsky, the modes of moisture distribution.” J Agric Sci, 20,
R. Zhang, and H. Chen, “Studies of accurate multi- 97 (1930)
component lattice Boltzmann models on benchmark
62. J.G. Roof, “Snap-Off of oil droplets in water-wet
cases required for engineering applications,” Jour.
pores,” SPE J., 10, 85 (1970)
Comp. Sci., 17, 334 (2016).
63. A. Skauge, K. Spildo, L. Høiland, B. Vik, B.
50. H. Otomo, B. Crouse, M. Dressler, D.M. Freed, I.
Ottesen, “ Experimental evidience of different
Staroselsky, R. Zhang, H. Chen, “Multi-component
intermediate wetting states,” SCA 2004-04, Int.
lattice Boltzmann models for accurate simulation of
Symp. Soc. Core Analysts, Abu Dhabi, UAE, (2004).
flows with wide viscosity variation,” Computer
Fluids, 172, 674 (2018)

51. H. Chen, C. Teixeira, and K. Molvig, “Realization

of fluid boundary conditions via discrete Boltzmann
dynamics,” Int. J. Mod. Phys. C, 9, 1281 (1998).

52. B. Crouse, D. M. Freed, N. Koliha, G.

Balasubramanian, R. Satti, D. Bale, and S. Zuklic,
“A Lattice-Boltzmann based method applied to
Digital Rock characterization of perforation tunnel
damage,” SCA 2016-058, Int. Symp. Soc. Core
Analysts, Snow Mass, CO, USA., (2016).

53. W.A. Adeagbo, N.L. Doltsinis, K. Klevakina, J.

Renner, “Transport processes at α‐Quartz–water
interfaces: insights from first‐principles molecular