Obtaining capillary pressure curves from resistivity measurements in low-permeability sandstone

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Journal of Petroleum Science and Engineering 221 (2023) 111297

Contents lists available at ScienceDirect

Journal of Petroleum Science and Engineering


journal homepage: www.elsevier.com/locate/petrol

Obtaining capillary pressure curves from resistivity measurements in


low-permeability sandstone
Muhammad Saafan *, Mysara Mohyaldinn, Khaled Elraies
Department of Petroleum Engineering, Universiti Teknologi PETRONAS (UTP), 32610, Seri Iskandar, Perak Darul Ridzuan, Malaysia

A R T I C L E I N F O A B S T R A C T

Keywords: The increasing global demand for energy necessitates exploring and developing low-quality prospects, e.g., low-
Tight reservoirs permeability reservoirs, which contain substantial hydrocarbon resources and have the potential to fill the gap in
Partially-saturated media the energy markets. Typically, modeling fluid flow using reservoir simulators requires capillary pressure curves
Fractal theory
as an input. Nonetheless, laboratory capillary pressure measurements in low-permeability samples are time-
Genetic algorithm
Resistivity modeling
consuming and challenging. On the contrary, resistivity measurements are easier to perform in the laboratory
Capillary pressure modeling and offer a different prospect for obtaining capillary pressure curves. This paper proposes a new approach for
obtaining capillary pressure curves from resistivity measurement in low-permeability sandstone using fractal
theory and genetic algorithm. First, the fractal pore system is characterized as tortuous square and triangular
capillaries to account for angular pores. Afterward, the drainage process is simulated to develop an innovative
electrical resistivity model in fully and partially saturated porous media. Next, the genetic algorithm matches
laboratory-measured resistivity data and obtains the developed model parameters. Afterward, the matched pa­
rameters are adopted in the drainage capillary pressure model to generate capillary pressure curves. The pro­
posed model’s reliability is verified by analyzing the prediction results of eighteen sandstone core samples.
Furthermore, the developed model performance is compared with different models from the literature, and the
results indicated its superiority in predicting capillary pressure curves.

1. Introduction ultralow-permeability (0.1–1 md), and non-permeable (<0.1 md).


Moreover, reservoirs containing gas are divided into low-permeability
The global energy demand is continuously increasing as the world’s (0.1–10 md) and ultralow-permeability (<0.1 md) (Lin et al., 2020).
population grows and the average living standard rises. For instance, Characterizing petrophysical properties, e.g., capillary pressure, of
global energy consumption is forecasted to rise by approximately 50% low permeability reservoirs is necessary for reservoir modeling and
by 2050 compared to the baseline year of 2020. Moreover, oil and gas resource evaluation. Capillary pressure is often acquired experimentally,
will remain the dominant energy sources, and the petroleum industry yet, the complex pore structure of low-permeability samples with sig­
will be required to continue boosting hydrocarbon fuel supply to fulfill nificant heterogeneity renders laboratory measurements time-
global energy demand (EIA, U.S.E.I., 2021). Nevertheless, the amount of consuming and challenging. On the other hand, resistivity measure­
recoverable oil from conventional reservoirs declines yearly. As a result, ments in the laboratory are easier, and real-time resistivity data in wells
petroleum engineering is increasingly focused on discovering and are available through resistivity logging tools (Li and Williams, 2007).
exploiting low permeability reservoirs. Generally, hydrocarbon re­ Consequently, obtaining capillary pressure from resistivity measure­
sources are plentiful enough to meet the demand as the ment would be valuable if a reliable connection could be established
low-permeability reservoirs contain substantially more oil and natural (Hou et al., 2016; Li et al., 2019). Generally, both are functions of fluid
gas than conventional reservoirs (Holditch and Chianelli, 2008). saturation, and a relationship should be found between them (Li and
Low-permeability reservoirs are characterized by low productivity, Williams, 2007). Nevertheless, there is little literature relating capillary
complex pore structure, and significant heterogeneity (Lin et al., 2020; pressure and resistivity.
Zhang et al., 2020). Sandstone oil reservoirs are categorized as Szabo (1974) suggested a linear model correlating capillary pressure
low-permeability (10–50 md), very low-permeability (1–10 md), (Pc) and resistivity index (Ir), assuming their relationships with water

* Corresponding author.
E-mail address: [email protected] (M. Saafan).

https://doi.org/10.1016/j.petrol.2022.111297
Received 9 May 2022; Received in revised form 27 October 2022; Accepted 26 November 2022
Available online 14 December 2022
0920-4105/© 2022 Elsevier B.V. All rights reserved.
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

saturation have equal power-law exponents. Li and Williams (2007) modeling the mixed-wet conditions. Although the capillary bundle
demonstrated that Szabo’s model is inaccurate in several cases, espe­ model is a simplistic approach to modeling realistic reservoir rocks, it
cially for low permeability core samples. Moreover, the authors derived allows studying the influence of different parameters on the resulting
a power-law fractal model that correlates capillary pressure and re­ capillary pressure behavior and deriving analytical expressions to be
sistivity index. Their model outperformed in low-permeability samples. included in reservoir simulators. The drainage process is simulated, and
Mahmoud and Abdelgawad (2013) proposed a method for determining the electrical resistivity of partially saturated fractal porous media is
the capillary pressure profile in the transition zone using resistivity log modeled. Then, analytical fractal models are derived for characterizing
data. The predicted capillary pressure curves provided a good match capillary pressure and electrical resistivity. The genetic algorithm is
with the laboratory-measured curves in medium permeability carbon­ utilized to adjust the parameters of the developed resistivity model by
ate, i.e., 100–250 md, and low permeability sandstone, i.e., less than 60 minimizing the error between the model results and laboratory-
md. Tan (2018) introduced a Pc/Rt versus Sw plot defined by an expo­ measured resistivity. Afterward, the adjusted parameters are utilized
nential function with four fitting coefficients. The fitting coefficients for in the drainage capillary pressure model to obtain the capillary pressure
each core are determined and related to permeability. Afterward, the curve. Finally, experimental measurements of eighteen samples are
established relationships are employed to compute the log-based capil­ utilized to validate the proposed method and compare its performance
lary pressure using resistivity, water saturation, and permeability logs. with different models from the literature.
Lately, Li et al. (2019) performed additional experiments at room tem­ The paper is organized as follows. Section 2 summarizes the sample
perature to support Li and Williams’ power-law model. The findings preparation and the different laboratory measurements. Moreover, it
showed that power-law model works effectively in many tested core details the pore system representation and simulates the drainage pro­
samples. Moreover, the authors correlated the power-law model expo­ cess to derive fractal models for capillary pressure and electrical re­
nent with rock permeability. sistivity. Also, the suggested procedures for obtaining drainage curves
Many theoretical investigations have been limited to predicting from resistivity measurements are detailed. Section 3 examines the
porous media’s flow characteristics in the homogeneous microstructures developed models’ outcomes and illustrates the validation process.
while failing in heterogeneous microstructures. The limitation is Furthermore, the developed model performance is compared with
attributable to the complex pore structure with tortuous flow paths, different models from the literature. Section 4 provides the conclusions
creating uncertainty in the flow distribution throughout the media (Yu, and recommendations for future works.
2008). On the other hand, fractal theory was introduced to characterize
complex and irregular geometries in nature and represent the middle 2. Samples and methods
ground between linear geometry and complete chaos (Mandelbrot,
1982). Subsequently, experimental measurements demonstrated that 2.1. Samples
the porous media microstructures are fractal (Katz and Thompson,
1985). Recently, several researchers adopted the fractal theory to Eighteen low-permeability sandstone samples are cut from cores
characterize the pore structure of low permeability sandstone (Dong obtained from the Jurassic lower Safa sandstones, a member of the
et al., 2021; Lu and Liu, 2021; Shao et al., 2017). Moreover, fractal Khatatba formation in Faghur Basin located in the Western Desert of
theory can help analyze porous media’s petrophysical properties, i.e., Egypt. Fig. 1 shows the generalized stratigraphic column of the Faghur
permeability, capillary pressure, and electrical properties (Saafan et al., Basin (Edress et al., 2022). Generally, Lower Safa reservoir consists of
2022b; Saafan and Ganat, 2020). For example, Wei et al. (2015) derived fine to medium-grained moderate to well-sorted sandstones character­
a fractal expression from modeling electrical conductivity in fully ized by large reservoir quality differences.
saturated porous media and compared it to Archie’s equation. Cai et al.
(2018) suggested a leaf vein network and studied the electrical prop­ 2.2. Experiments
erties of porous media. Later, Rembert et al. (2020) derived a fractal
model to evaluate fully saturated porous media’s formation factor and The experimental work starts with cleaning core plugs by hot soxhlet
electrical conductivity. Moreover, they establish a relationship between using toluene and drying in a conventional oven to remove any fluids.
electrical conductivity and transport characteristics, e.g., permeability. Then the samples’ dry weights are recorded. After that, the samples are
The available methods for obtaining capillary pressure from re­ saturated by putting them in a vacuum desiccator with brine. The
sistivity measurements have the following limitations: samples’ fully saturated weight is recorded, and the porosity is calcu­
lated. Afterward, the liquid permeameter is used to measure the
1) They correlated the capillary pressure and resistivity using equations permeability of the core plugs. Then, the fully saturated sample re­
with fitting coefficients. Those coefficients must be determined in sistivity, R0, is measured at ambient conditions. The porous plate
advance by fitting laboratory data of capillary pressure and re­ apparatus is used to obtain the drainage Pc - Sw curve. At each desa­
sistivity, then correlating them with petrophysical properties such as turation step, the plugs are removed and weighed to calculate the water
permeability. Consequently, those correlations are not universal and saturation. Also, the value of resistivity, Rt, is measured. The plugs are
applicable only to cores from the same formation. returned to the porous plate, and the air pressure is increased. Fig. 2
2) Li et al. (2019) only verified the power-law relation between capil­ illustrates the detailed work flow of the experimental steps.
lary pressure and resistivity. Although the authors correlated the
model exponent with permeability, they never verified its ability to 2.3. Pore system representation
predict capillary pressure from resistivity measurements.
3) The fractal electrical conductivity models are limited to fully- Irregular pore walls characterize the pore structure of porous media.
saturated porous media and should be developed in partially- As a result, polygons such as triangles and squares are more appropriate
saturated media. for modeling pore geometry than circles (Mason and Morrow, 1991).
Moreover, polygons permit the retention of the wetting phase in invaded
This study proposes a new method to obtain capillary pressure curves capillary corners. The dimensionless pore shape factor, G, is used to
from resistivity measurements based on fractal theory and genetic al­ define pore morphology and is obtained from the pore’s area (A) and
gorithm. The pore system is represented using the capillary bundle perimeter (P) as:
model with separate bundles of tortuous fractal square and triangular
A
capillaries. Using capillaries with polygon cross-sections allow the G= (1)
P2
wetting phase to be retained in the corners of invaded tubes, hence

2
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Fig. 1. Generalized stratigraphic column of the Faghur Basin (Edress et al., 2022).

This work utilizes the capillary bundle model proposed by Saafan Patzek equations (Patzek and Silin, 2001). Suppose the corner-half an­
et al. (2022a, 2022b), which represents the pore system as tortuous gles satisfy 0 < β1 ≤ β2 ≤ β3 ≤ π/2, then the minimum and maximum
triangular and square capillaries (Fig. 3). values of β2 for a given G is:
The radii of capillary inscribed circles change sinusoidally (Fig. 4) to { [ ( √̅̅̅ ) ]}
account for the existence of pores and throats. The inscribed circle’s 2 cos− 1 − 12 3 G 4π
β2,min = tan − 1
√̅̅̅ cos + (4)
radius at a distance x is given as: 3 3 3
( ) [ ( )]
2π x 2π x { [ ( √̅̅̅ )]}
r(x) = R + r sin = R 1 + a sin (2) 2 cos− 1 − 12 3 G
λ λ β2,max = tan − 1
√̅̅̅ cos (5)
3 3
Where R is the average capillary inscribed radius, r is the oscillation
In this paper, β2 is set to:
amplitude, λ is the wavelength, and a is the cross-sectional restriction
factor (0 ≤ a ≤1). The capillary’s cross-sectional area at a distance x is β2,min + β2,max
β2 = (6)
represented by: 2

r2 (x) R2
[ ( )]2
2π x Then, β1 and β3 are calculated as:
A(x) = = 1 + a sin (3) ( )
4G 4G λ 1 1 tan β2 + 4G
β1 = − β2 + sin− 1 sin β2 (7)
Each corner-half angle of a square capillary is π/4. For a triangular 2 2 tan β2 − 4G
capillary, the corner-half angles (β1, β2, and β3) are obtained using the

3
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Fig. 2. The detailed workflow of the experimental steps.

dimension. The number of capillaries with radii in the range R and R +


dR is obtained through the differentiation of Eq. (9) (Yu, 2008):

dN = − Df RDmax
f
R− Df − 1
dR (10)

Based on the work of Wheatcraft and Tyler (1988) and (Yu and
Cheng, 2002), Saafan and Ganat (2021) proposed a tortuous length
equation to eliminate the fractal models’ reliance on straight capillary
length, L0. The tortuous length, Lt, is expressed as:

Lt = 21− Dt
R1− Dt
L0 (11)

where Dt is the tortuosity fractal dimension. The porosity of the pore


system represented in Fig. 3, comprising two separate bundles of cap­
illaries, is expressed as (Saafan and Ganat, 2021):
⎧ ( ) ⎫
Df i 3− D − D 3− D − D
(1 + 0.5a2 ) ∑2 ⎨ Df Rmax
i i Rmaxi f i t i − Rmini f i t i ⎬
φ= ( ) (12)
Across i=1
⎩ 2Dt i +1 Gi 3 − Df i − Dt i ⎭
Fig. 3. Pore system representation (Saafan et al., 2022a).

where Across is the cross-sectional of the capillary bundle model and Rmin
is the smallest pore radius.

2.4.1. Drainage capillary pressure model


The drainage process is a forced displacement process in which the
nonwetting fluid, e.g., oil, displaces the wetting fluid, e.g., water. Fig. 5
displays the change of fluids configuration in a triangular capillary
during primary drainage.
The threshold capillary pressure is obtained from the Mayer-Stowe-
Fig. 4. The sinusoidally varying inscribed circles of triangular and square Princen method by equating the work associated with surface changes
capillaries (Saafan et al., 2022a). to the net change in surface energies (Mayer and Stowe, 1965; Princen,
1970). For any polygon, the drainage threshold capillary pressure is
π
β 3 = − β1 − β2 (8) expressed as (Saafan et al., 2022a):
2 ( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ )
σ ow cos θr 4GC
Pc = 1+ 1 − (13)
2.4. Fractal models (1 − a)R cos2 θr

Mandelbrot (1982) developed the fractal theory to describe complex Where σow is the oil-water interfacial tension, θr is the receding contact
geometries in nature and represent the middle ground between linear angle, and C is:
geometry and complete chaos. In a fractal porous medium, the number ∑ n {
cos(θr + βi ) (π )}
of capillaries (N) with radii larger than or equal to R is expressed as (Yu C= cos θr − − θ r − βi (14)
sin βi 2
and Li, 2002): i=1

( )D
Rmax f where n is the number of corners satisfying:
N= (9)
R π
βi < − θ r (15)
2
where Rmax is the largest pore radius, and Df is the pore fractal

4
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Fig. 5. Change of fluids configuration during primary drainage (Saafan and Ganat, 2021).

The first integration in Eq. (18) calculates the water volume in


uninvaded capillaries as:
( )
∫R Df 3− D − D
(1 + 0.5a2 ) L0 Df Rmax R3− Df − Dt − Rmin f t
vp (R) (− dN) = ( ) (21)
2Dt +1 G 3 − Df − Dt
Rmin

The second integration in Eq. (18) defines the water volume left in
Fig. 6. A simplified view of the residual water on the walls of
invaded capillary.
the invaded capillaries’ corners and is evaluated as:

Rmax ∫
Rmax

The cross-sectional area occupied by the remaining water is calcu­ vwr (− dN) = Awr Lwall (− dN) (22)
lated by: R R

Awr = C r2 (16) where Lwall is the length of the capillary wall, which represents the
length of a sinusoidally varying curve. Fig. 6 shows a simplified repre­
where r is the radius of the arc meniscus between the two phases, which
sentation of the remaining water on a straight capillary wall. Lwall is
is computed as follows:
calculated for tortuous capillary as (Appendix A):
σ ow √̅̅̅̅̅̅̅̅̅̅̅̅̅ (√̅̅̅̅̅̅̅̅̅̅̅̅̅)
r= (17) √̅̅̅̅̅̅̅̅̅̅̅̅̅ ( )
Pc a2 + 1 a2 2− Dt a2 + 1 a
Lwall = E Lt = 2 E √ ̅̅̅̅̅̅̅̅̅̅̅̅̅ R1− Dt L0
π/2 a2 + 1 π a2 + 1
The water volume remaining in the pore system at a given Pc is
calculated as (Saafan and Ganat, 2021): (23)

∫R ∫
Rmax
Where E is the complete elliptical integral of the second kind and is
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Vw = vp (R) (− dN) + vwr (− dN) (18) calculated at a/ a2 + 1. Using Eq. (23), Eq. (22) is evaluated as:
Rmin R
√̅̅̅̅̅̅̅̅̅̅̅̅ ( a ) (σ )2 ( 1− Df − Dt )

Rmax
22− Dt a2 +1 E √̅̅̅̅̅̅̅ ̅ C Pow Df L0 RDmax
f
Rmax − R1− Df − Dt
where R is the average radius of the smallest capillary that will be a2 +1 c
vwr (− dN)= ( )
invaded at a given Pc and is obtained from re-arranging Eq. (13) as: π 1− Df − Dt
R
( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) (24)
σ ow cos θr 4GC
R= 1+ 1 − (19)
(1 − a)Pc cos2 θr The water saturation remaining in the pore structure, which com­
prises two separate bundles of capillaries (m = 2), at given Pc is:
R determined by Eq. (19) satisfies:
{
Rmin if R < Rmin
R= (20)
Rmax if R > Rmax

( )
⎧ ⎡ 3− D f i − Dt i 3− D − Dt i ⎤⎫
⎪ Ri − Rmini f i ⎪

⎪ ⎪

⎪ ⎢ ( ) + ⎥⎪


2 ⎨
⎢ 4 Gi 3 − Df i − Dt i ⎥⎪

∑ Df i ⎢ ⎥⎬
1− Dt i
2 Df i Rmax ⎢
i⎢ √̅̅̅̅̅̅̅̅̅̅̅̅̅ ( ) ( )2 ( ⎥
)⎥
i=1⎪
⎪ ⎢ 2 a2 + 1 E √̅̅̅̅̅̅̅̅̅̅̅̅̅a σ ow 1− D − D 1− Df i − Dt i ⎥⎪


⎪ ⎣ Ci Rmaxi f i t i − Ri ⎦⎪


2 ⎪
⎪ a2 + 1 Pc ⎪

Vwi ⎩ ( ) ⎭
π (1 + 0.5a2 ) 1 − Df i − Dt i
Sw = i=1 = ⎧ [ ]⎫ (25)
∑2
2 ⎨Df i Rmaxi i Rmaxi i
Df 3− Df − Dt i 3− Df − Dt i
− Rmin i ⎬
Vp i ∑ i
i=1 2Dt i +1 Gi (3− Df i − Dt i )
i=1 ⎩ ⎭

5
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

2.4.2. Electrical resistivity model ( )2


D f ( Dt − D f − 1 )
In this work, the surface conductivity σs is neglected; hence the ∫
Rmax
ΔV πκw C σPowc Df Rmax Rmax − R Dt − Df − 1
bundle model shown in Fig. 3 is analogous to a circuit of parallel con­ iwr (R)(− dN) = √̅̅̅̅̅̅̅̅̅̅̅̅̅ ( ) ( ) (34)
a ̅
22− Dt a2 + 1 E √̅̅̅̅̅̅̅ L0 Dt − Df − 1
ductances. The electrical conductance of a single sinusoidal capillary is R 2 a +1
calculated as:
The total electric current flowing in the pore system at a given Pc is
∫ Lt
∑ Aw (x) then:
= κw (26) ⎧
capillary 0 dx ⎡ ( )3 ( 1+D − D ) ⎤⎫
⎪ 1+D − D ⎪

⎪ 1 − a2 2 Ri t i f i − Rmini t i f i ⎪


⎪ ⎢ ( ) + ⎥⎪⎪
⎪ ⎪
Where κw represents the pore-water conductivity, and Aw is the cross- ⎪



⎢ 2 G i 1 + D ti − D fi






⎨ ⎢ ⎥ ⎪

sectional area filled by water. For fully saturated capillary, Aw is ΔV κw ∑ − Dt i
2
Df i ⎢ ( )2 ( ) ⎥
I= 2 Df i Rmaxi ⎢ σow ⎥
given by Eq. (3); hence Eq. (26) is expressed as:
Dt i − Df i − 1 Dt i − Df i − 1
L0 i=1 ⎪
⎪ ⎢ π Ci Rmaxi − Ri ⎥⎪

⎪ ⎢ Pc ⎥⎪⎪

⎪ ⎢ √̅̅̅̅̅̅̅̅̅̅̅̅̅ ) ⎥⎪⎪
∑ ∫ [ ( )]2 ⎪
⎪ ⎣ ( a ( ) ⎦⎪⎪

κw R2 Lt 1 + a sin 2πλ x ⎪ 2 ⎪
= (27) ⎪
⎩ a + 1 E √ ̅̅̅̅̅̅̅̅̅̅̅̅̅
2
D t i − D f i − 1 ⎪

capillary
4G 0 dx a +1
(35)
Considering Lt as multiple integers of λ and using Eqs. (11), Eq. (27)
can be expressed as: The resistivity of the pore system is then represented as:

3 ΔV Across Across Rw
∑ κw (1 − a2 )2 R1+Dt Rt = = ( )3 ( )
= (28) I L0 ⎧ ⎡ 1− a2 2 R1+Dt i − Df i − R1+Dt i − Df i
i mini
⎤⎫
capillary
23− Dt G L0 ⎪
⎪ ( ) + ⎪


⎪ ⎢ ⎥⎪


⎪ ⎢ 2Gi 1+Dt i − Df i ⎥⎪

⎪ ⎥⎪
For partially saturated capillary, i.e., invaded capillary, Aw is given ∑⎨ −
2
D f
⎢ ( )
⎢ 2 ( ) ⎥

by Eq. (16), and the electrical flow path length is Lwall. Recalling Awr is 2 Dt i D R i
f i maxi ⎢ σ ow D ti − D fi − 1 D ti − D fi − 1 ⎥
i=1⎪
⎪ ⎢ π Ci Rmaxi − Ri ⎥⎪
⎪ ⎢ ⎥⎪

independent on R, Eq. (26) is expressed as: ⎪


Pc
⎣ √̅̅̅̅̅̅̅̅̅̅̅ ( a )( ⎦⎪



⎩ ) ⎪

∫ Lwall 2
∑ a +1 E √̅̅̅̅̅̅̅̅̅̅̅ Dt i − Df i − 1
1 κw Cr2
(29)
2
= κw Cr2 = a +1
capillary 0 dx Lwall (36)
Using Eqs. (17) and (23), Eq. (29) is represented as: Where Rw is the pore-water resistivity (Rw = 1/κw ). The developed re­
( )2 sistivity model in Eq. (36) can be written in terms of porosity using Eq.
∑ πκw C σPowc (12) as:
= √̅̅̅̅̅̅̅̅̅̅̅̅̅ ( a ) (30)
capillary 22− Dt a2 + 1 E √̅̅̅̅̅̅̅ ̅ R1− Dt L0 ⎧ ( )⎫
2 ⎨ Df i Rmaxi i Rmaxi i ⎬
2 Df 3− Df − Dt i 3− Df − Dt i
a +1 − Rmin i
2
∑ i
(1 + 0.5a )
From Ohm’s law, the electric current flowing through each capillary Rw i=1 ⎩ 2Dt i +1 Gi (3− Df i − Dt i ) ⎭
due to a voltage difference ΔV is defined as: Rt = × ( ) 3
( )
φ ⎧ ⎡ 1 − a2 2 R1+Dt i − Df i − R1+Dt i − Df i ⎤⎫
∑ ⎪ i mini

i(R) = ΔV (31) ⎪


⎪ ⎢
(
2 Gi 1 + Dt i − Df i
) + ⎪
⎥⎪


capillary ⎪
⎪ ⎢ ⎥⎪

∑2 ⎨ ⎢ ( ) ⎥⎬
D fi ⎢ 2( ) ⎥
2− D Df i Rmaxi ⎢ σow ⎥
The total current flowing through the pore system at a given pressure
t i Dt i − Df i − 1 Dt i − Df i − 1
i=1 ⎪
⎪ ⎢ π Ci Rmaxi − Ri ⎥⎪
⎪ ⎢ Pc ⎥⎪

is obtained by adding the electric currents through all the capillaries as: ⎪

⎪ ⎣ √̅̅̅̅̅̅̅̅̅̅̅̅ ) ⎪
⎦⎪

⎩ 2
( a ( ) ⎪


a + 1 E √̅̅̅̅̅̅̅̅̅̅̅̅ Dt i − Df i − 1
∫R ∫
Rmax 2
a +1
I= i(R) (− dN) + iwr (R)(− dN) (32) (37)
Rmin R

The first term in Eq. (32) defines the current flowing through unin­ 2.5. Proposed method
vaded capillaries and is evaluated using Eqs. (10), (28), and (31) as:
( ) This section illustrates the method of obtaining drainage capillary
pressure curves from resistivity measurements. The pore structure is
3 Df 1+D − D
∫R ΔVκw (1 − a2 )2 Df Rmax R1+Dt − Df − Rmin t f
i(R) (− dN) = ( ) (33) characterized as two bundles of tortuous capillaries with square and
23− Dt G L0 1 + Dt − Df
Rmin triangular geometries. Generally, the shape factor of a triangle varies
from zero for a slit-shaped triangle to 0.0481 for the equilateral triangle
The second term in Eq. (32) defines the current flowing through the
(Oren et al., 1998). The shape factor of triangular pores is set to 0.024, i.
corners fluids of invaded pores and is evaluated using Eqs. (10), (30),
e., the median value of slit-shaped and equilateral triangles. Moreover,
and (31) as:
the pore fractal dimension of each bundle is calculated as (Yu and Li,
2002):
ln(φ)
Df = 2 − (38)
ln(Rmin /Rmax )

Table 1 Table 1 provides a rule of thumb for estimating Rmin/Rmax using the
Empirical range of Rmin/Rmax as function permeability of the rock sample. The lower the permeability, the
permeability. broader range of the pore size.
K (md) Rmin/Rmax Thomas et al. (1968) correlated the threshold pressure with porosity,
<0.5 10− 6 - 10− 5 permeability, surface tension, and formation resistivity factor as:
0.5–5 10− 5 - 10− 4 √̅̅̅̅̅̅̅̅̅̅
5–50 10− 4 - 10− 3 0.1461 × 10− 3 σ 1000
PT = √̅̅̅̅̅ (39)
>50 10–3 ko F φK

6
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

3. For the square capillaries:


Table 2
a. G1 = 0.0625 and β = [π/4, π/4, π/4, π/4].
Search space limits of the eight parameters.
b. Calculate C1 from Eq. (14).
Parameter Minimum Maximum Parameter Minimum Maximum 4. For the triangular capillaries:
PT1, psi 0.3 × PT 1.5 × PT Dt1 1 2 a. G2 = 0.024.
PT2, psi Dt2 b. Calculate the corner half angles from Eqs. (6)-(10).
0.25 × φ 0.75 × φ Rmin1/Rmax1 Form Table 1
φ1
c. Calculate C2 from Eq. (14).
a 0 1 Rmin2/Rmax2
5. Employ genetic algorithm as discussed earlier and determine the
model parameters.
6. Obtain the drainage Pc - Sw curve from Eq. (25).

where PT is the threshold pressure, F is the formation resistivity factor, 3. Results and discussions
and ko is the shape factor that varies between 2 and 3. A little error is
introduced by using ko value between 2 and 3 since it is raised to the one- This section shows a comparative analysis of validating the proposed
half power, and in this work, ko is set to 2.5. Also, Eq. (39) yielded results model. The model results are compared with the experimental data and
between +50% and − 70% error lines compared to laboratory mea­ the results of other literature models. Table 3 displays the measured
surements (Thomas et al., 1968). porosity, permeability, and fully saturated resistivity of the eighteen
There are nine unknowns in Eqs. (37), which are Rmax1, Rmax2, Rmin1, core samples. The porosities range from 6.8% to 15.8% and perme­
Rmin2, Dt1, Dt2, Df1, Df2, and a. Rmax1 and Rmax2 are determined using the abilities from 0.36 to 63 mD. Also, the samples’ drainage capillary
threshold pressure of each pore set, i.e., PT1 and PT2, using Eq. (19). pressure curves and resistivity measurements are displayed in Fig. 8 and
Afterward, Rmin1 and Rmin2 are obtained using Rmin1/Rmax1 and Rmin2/ Fig. 9.
Rmax2 from Table 1. Then, the pore fractal dimensions, i.e., Df1 and Df2,
are calculated from Eq. (38), which require the porosity of each bundle. 3.1. Comparisons with laboratory measurements
As the sample porosity is measured, the porosity of one set is utilized to
calculate the other. Therefore, eight parameters are required to calculate The steps shown in Fig. 7 are employed to obtain the capillary
resistivity, which are PT1, PT2, Rmin1/Rmax1, Rmin2/Rmax2, Dt1, Dt2, φ1, and pressure curves for the eighteen core samples using resistivity mea­
a. surements. The MAPE is calculated and displayed in Table 4. The MAPE
The genetic algorithm (GA) is employed to adjust the eight param­ of the eighteen core samples ranges between 2.4% and 39.4%, with a
eters and minimize the error between the results of the fractal resistivity median of 11.3%. Moreover, the mean MAPE is 13.3%, with one stan­
model and laboratory-measured resistivity. The GA optimization steps dard deviation of 9.8%.
are summarized as follows: Fig. 10 depicts the calculated versus laboratory-measured drainage
water saturation of the eighteen core samples with an error factor of
1. Generate a random initial population of individuals, utilizing the ±0.1. From the preceding analysis, the proposed method of obtaining
search space limits of the eight parameters shown in Table 2. drainage capillary pressure curves using resistivity measurements yields
2. Calculate the fitness of each individual. The fitness function is rep­ a good match with laboratory measurements.
resented by the mean absolute percentage error (MAPE) between the
calculated and laboratory-measured resistivity.
3.2. Comparisons with literature models
N ⃒ ⃒
100 ∑ ⃒Rt model − Rt laboratory ⃒
MAPE = ⃒


⃒ (40) The suggested approach for obtaining capillary pressure curves from
N i=1 Rt laboratory
resistivity measurements is compared with three different models from
the literature, i.e., Kwon and Pickett (1975) (Appendix B), Pittman
3. Select two individuals for crossover, and with a crossover probability (1992) (Appendix C), and Li et al. (2019). Li et al. correlated the
of 0.7, exchange strings and create two offsprings. dimensionless capillary pressure to the resistivity index. However, they
4. With a mutation probability of 0.01, randomly change genes in the did not introduce a way to calculate the capillary entry pressure.
offspring. Therefore, we utilize Eq. (39) to obtain the entry pressure and calculate
5. Place the resulting individuals in the new population. Li et al.’s capillary pressure curves. Fig. 11 graphically compares the
6. Repeat steps 3–5 until the new population size is obtained. calculated drainage curves of the eighteen core samples from our pro­
7. Replace the existing population of chromosomes with the new posed and literature models.
population. Known and Pickett’s model calculates the capillary pressure for
8. Repeat steps 2–7 until the stopping condition is reached. water saturation from 0.3 to 0.9. Consequently, we only calculate the
9. Determine the eight parameters, i.e., PT1, PT2, Rmin1/Rmax1, Rmin2/ MAPE in the mentioned water saturation range to have a representative
Rmax2, Dt1, Dt2, φ1, and a, of the fittest individual. comparison between different models. The MAPE of the computed
capillary pressure from the various models is summarized in Table 5.
The optimized model parameters are then utilized in the capillary The best-performing model, i.e., the lowest MAPE, is highlighted in bold
pressure model, i.e., Eq. (25), to generate the drainage Pc - Sw curve. for each sample. The proposed model outperforms the literature models
Fig. 7 displays the detailed methodology for obtaining capillary pressure for fourteen samples of the eighteen, i.e., S1, S4, S6 – S12, and S14 – S18.
curves from resistivity measurements, and the steps are as follows: On the other hand, the Known and Pickett model outperforms for
samples S2, S3, and S13. While Pittman’s model only outperforms for
1. Calculate PT from Eq. (39). sample S5. Moreover, Li et al.’s model has a high MAPE compared to
2. Determine the range of Rmin/Rmax from Table 1. other models, attributed to the dependency of their model on the

7
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Fig. 7. Flow chart of obtaining the drainage capillary pressure curve from resistivity data.

8
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Table 3
Porosity, permeability, and fully saturated resistivity of the eighteen core plugs.
Sample φ (%) K R0 (Ω. Sample φ (%) K R0 (Ω.
(md) m) (md) m)

S1 9.5 6.4 4.37 S10 8.4 1.6 8.38


S2 13 56 2.44 S11 10.2 25 3.77
S3 11.5 11 5.06 S12 9.8 39 3.23
S4 8.9 5.3 5.47 S13 10.2 16 4.24
S5 15.8 5.6 2.42 S14 7.4 5.1 10.53
S6 8.4 0.85 7.91 S15 12.3 63 3.4
S7 10.4 2.5 5.29 S16 6.8 0.36 8.38
S8 12.6 11 3.61 S17 12.6 1.3 3.29
S9 7.6 0.61 9.96 S18 7.8 3.3 10.07

Fig. 10. The calculated versus laboratory-measured drainage water saturation


of the eighteen core samples. The black dashed lines indicate an error factor
of ±0.1.

capillary entry pressure.


Fig. 12 illustrates the box and whisker box plot for MAPE of the
different models. Our model has the lowest median and mean MAPE of
8.65% and 10.33% for the eighteen core samples included in the study.
Also, the MAPE for 75% of the samples is lower than 13.75%, and the
lowest and highest MAPE are 2.30% and 25%. On the other hand,
Known and Pickett’s model yields median and mean MAPE of 16.20%
Fig. 8. Drainage capillary pressure curves of the eighteen core plugs.
and 17.39%, with the lowest and highest MAPE of 6% and 29.8%. Also,
the MAPE for 75% of the samples is lower than 21.23%. For Pittman’s
model, the median and mean MAPE are 18.90% and 20.92%, with the
lowest and highest MAPE of 12.9% and 39.3%. Also, the MAPE for 75%
of the samples is lower than 23.09%. Finally, Li et al.’s model renders
median and mean MAPE of 50.90% and 61.79%, with the lowest and
highest MAPE of 13.3% and 157.6%. Also, the MAPE for 75% of the
samples is lower than 76.33%.
Fig. 13 shows the calculated MAPE from different models versus the
permeability. The proposed model of obtaining drainage capillary
pressure from resistivity measurements is superior in the very low
permeability region, i.e., K < 5 md. This indicates the necessity of uti­
lizing resistivity measurements in obtaining capillary pressure for low
permeability samples. For higher permeabilities, the proposed model
outperforms for six out of ten samples. However, Known and Pickett’s
model still provides good results for higher permeabilities and is rec­
ommended in the absence of resistivity measurements.
Finally, we should emphasize that the capillary bundle model does
Fig. 9. Resistivity measurements of the eighteen core plugs.
not consider Haines jumps due to the lack of interconnectivity. The ef­
fect of multiple pore filling during a Haines jump is more marked early
Table 4 in the displacement when most pores are unfilled. This explains why the
MAPE of the calculated versus laboratory-measured capillary. measured capillary pressures gradually increase with decreasing wetting
phase saturation initially before rising more steeply. On the other hand,
Sample MAPE (%) Sample MAPE (%)
the surface conductivity of the rock grains was neglected in the proposed
S1 17.6 S10 2.4 model. Consequently, the model is not applicable to predict capillary
S2 39.4 S11 16.5
S3 33.7 S12 15.1
pressure curves of the shaly sample.
S4 10.4 S13 17.4
S5 16.6 S14 8.3 4. Conclusions and future work
S6 7.9 S15 12.2
S7 13.6 S16 6.1
This paper proposes a new approach to obtain drainage capillary
S8 6.5 S17 7
S9 4.2 S18 4.4 pressure curves in low-permeability sandstone from resistivity mea­
surements utilizing fractal modeling of porous media. A proper

9
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Fig. 11. Calculated capillary pressure curves from our proposed model and literature models versus laboratory measurements.

10
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Table 5 representation of the pore structure is used to characterize pore angu­


MAPE of the calculated capillary pressure from the different models (0.3 ≤ Sw ≤ larity. The drainage process is simulated to derive fractal models for
0.9). capillary pressure and electrical resistivity. Also, the genetic algorithm is
Sample Our Kwon and Pickett Pittman Li et al. employed to tune the resistivity model parameters to match the
Model (1975) (1992) (2019) measured resistivity data. The tuned parameters are then used to
S1 10.1 12.8 39.3 139 construct capillary pressure curves.
S2 25 6.3 18.6 157.6 The developed method was validated as a valuable tool for gener­
S3 20.6 8.6 30.9 128 ating drainage capillary pressure curves by comparing it to laboratory-
S4 11.6 27.7 23.34 22.6
measured data of eighteen low permeability sandstone samples from
S5 15.7 20.3 12.9 53.2
S6 8.1 28 18.8 49.3 the Khatatba formation, located in the western desert of Egypt. The
S7 11.3 19 18.3 50 proposed method outperforms existing literature models in the very low
S8 4.2 15.6 15.9 53.9 permeability zone, i.e., K < 5 md, emphasizing the necessity to use
S9 6.8 29.8 23.9 13.3 resistivity measurements in calculating the capillary pressure of low
S10 2.3 20.6 20.9 43.3
S11 13.1 15.9 18.4 44.7
permeability samples. Moreover, the suggested model outperforms for
S12 9.2 9.9 16.1 77 six out of ten samples with K > 5 md. However, Known and Pickett’s
S13 20 10.2 19 55.7 model still yields acceptable results and is recommended in the lack of
S14 5.1 16.2 23 31.5 resistivity data.
S15 5.2 13.2 15.5 76.1
The capillary pressure curves are computed using a capillary bundle
S16 5.1 16.2 23 33.7
S17 6.1 19.6 16.3 51.8 model. The model assumes separate bundles of tortuous square and
S18 6.5 23.1 22.5 31.5 triangular capillaries, which is a simplistic approach to modeling real­
istic reservoir rocks. The capillary bundle model does not account for the
interconnections between pores; therefore, phase entrapment and the
Haines jump effect are absent. Future research should focus on utilizing
real-time resistivity logging data obtained in the transition zone to
calculate the capillary pressure curves. Also, the fractal resistivity model
should be extended to model clay-bearing formations by considering the
surface conductivity of the rock grains. For instance, the rock should be
modeled as a three-component system, i.e., grains, pores, and interfaces,
as detailed by Ruffet et al. (1995). Moreover, the developed modeling
procedures should be implemented for other rocks, such as carbonates
and tight shale.

Credit author statement

Muhammad Saafan: Conceptualization, Methodology, Software,


Validation, Resources, Formal analysis, Writing - Original Draft, Visu­
alization. Mysara Mohyaldinn: Investigation, Review & Editing, Su­
pervision. Khaled Elraies: Investigation, Review & Editing,
Supervision.

Fig. 12. Box and whisker plot of MAPE for the different models.
Declaration of competing interest

The authors declare that they have no known competing financial


interests or personal relationships that could have appeared to influence
the work reported in this paper.

Data availability

No data was used for the research described in the article.

Nomenclature

a capillary cross-section restriction factor


A Pore cross-sectional area
Fig. 13. The calculated MAPE from different models versus permeability of the Awr Remaining water in the capillary corners
eighteen samples. C Function of corner-half angles
d Power-law exponent

11
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Df Pore fractal dimension


Dt Tortuosity fractal dimension
E Complete elliptical integral of the second kind
F Formation resistivity factor
G Dimensionless pore shape factor
i Electric current flowing in a single capillary
I Total current flowing through the pore system
K Permeability
ko shape factor
L0 Straight length of the fluid flow path
Lt Tortuous length of the fluid flow path
Lwall Length of the capillary wall
m Number of separate geometric systems
n Number of corners that will retain water
N Cumulative number of pores
P Pore perimeter
Pc Capillary pressure
PT Threshold pressure
r Arc meniscus radius between oil and water

r Amplitude of the fluctuation
R Average capillary inscribed radius
R0 Fully saturated rock resistivity
Rmax Maximum pore radius
Rmin Minimum pore radius
Rt Partially saturated sample rock resistivity
Rw Pore-water resistivity
Sw Water saturation
Vp Volume of a single capillary
Vw Water volume remaining in the pore system
vwr Water volume remaining in the corners of a single capillary
β Half-corner angle
ΔV Voltage difference
θr Receding contact angle
σ ow Oil-water interfacial tension
φ Porosity
κw Pore-water conductivity
λ Wavelength

Electrical conductance

Appendix A. Length of a sinusoidally varying curve

Consider a curve (Fig. A1) that varies sinusoidally according to the following equation:
y = a sin x (A.1)

Where a is the amplitude of the sine curve.

Fig. A.1. A sinusoidal varying curve.

The length of the arc, ds, connecting two points as in Fig. A2 is approximated using Pythagoras theorem:
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( )2
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
dy
ds ≈ (dx)2 + (dy)2 = 1 + dx (A.2)
dx

where dx and dy represent the horizontal distance and the vertical height of the right-angled triangle.

12
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Fig. A.2. Length of the arc connecting two points.

The integration of the infinitely small distances for the distance between x = 0 and x = X is:
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( )2
∫X ∫X
dy
L= ds = 1+ dx (A.3)
0 0 dx

Differentiating Eq. (A.1) yields

dy
= a cos x (A.4)
dx
Using Eq. (A.4), the length of the arc given by Eq. (A.3) is expressed as:
∫ X √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
L= 1 + a2 cos2 x dx (A.5)
0

Using cos2 x = 1 − sin2 x, Eq. (A.5) is expressed as:


√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( )2
√̅̅̅̅̅̅̅̅̅̅̅̅̅ ∫ X a
L= 1+ a 2 1 − √̅̅̅̅̅̅̅̅̅̅̅̅̅ sin2 x dx (A.6)
0 1 + a2
The integration of Eq. (A.6) yields:
√̅̅̅̅̅̅̅̅̅̅̅̅̅ ( )
1 + a2 a
L= × E √̅̅̅̅̅̅̅̅̅̅̅̅̅ X (A.7)
π/2 1 + a2

where E represents a complete elliptical integral of the second kind evaluated at k as:
∫ π2 √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
E(k) = 1 − k2 sin2 θ dθ (A.8)
0

MATLAB is used to evaluate E using the following command line, [~,E] = ellipke (k). As shown in Fig. A3, the value of E changes from E(0) = π/2 to
E(1) = 1. Hence, for a = 0, E(1) = 1 and the length of the elliptical curve L equals X.

Fig. A.3. Evaluation of the complete elliptical integral of the second kind.

13
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Appendix B. Kwon and Pickett capillary pressure equations

For in the Sw in the range from 30% to 90%, a relation between Pc and k/φ was presented as follows:

Pc = A(K/φ)− B
(B.1)

where Pc is mercury-air capillary pressure from measurement on cores, K is air permeability (md), and φ is porosity (%). The empirical
values for A and B are shown in Table B1.
Table B.1
Empirical values of A and B (Kwon and Pickett, 1975).

Sw (%) Log A B

30 2.18 − 0.407
40 1.97 − 0.455
50 1.81 − 0.468
60 1.67 − 0.452
70 1.56 − 0.446
80 1.47 − 0.440
90 1.36 − 0.428

Appendix C. Pittman capillary pressure equations

The empirical equations for determining pore aperture radii in μm corresponding to various mercury saturations are shown in Table C1.

Table C.1
Pore aperture radii corresponding to various mercury satura­
tions (Pittman, 1992).

SHg (%) Log(r) (μm)

10 0.459 + 0.500 Log(K) – 0.385 Log(φ)


15 0.333 + 0.509 Log(K) – 0.344 Log(φ)
20 0.218 + 0.519 Log(K) – 0.303 Log(φ)
25 0.204 + 0.531 Log(K) – 0.350 Log(φ)
30 0.215 + 0.547 Log(K) – 0.420 Log(φ)
35 0.255 + 0.565 Log(K) – 0.523 Log(φ)
40 0.360 + 0.582 Log(K) – 0.680 Log(φ)
45 0.609 + 0.608 Log(K) – 0.974 Log(φ)
50 0.778 + 0.626 Log(K) – 1.205 Log(φ)
55 0.948 + 0.632 Log(K) – 1.426 Log(φ)
60 1.096 + 0.648 Log(K) – 1.666 Log(φ)
65 1.372 + 0.643 Log(K) – 1.979 Log(φ)
70 1.664 + 0.627 Log(K) – 2.314 Log(φ)
75 1.880 + 0.609 Log(K) – 2.626 Log(φ)
Where K is permeability (md) and φ is porosity (%).

References Li, K., Hou, B., Bian, H., Liu, H., Wang, C., Xie, R., 2019. Verification of model for
calculating capillary pressure from resistivity using experimental data. Fuel 252,
281–294. https://doi.org/10.1016/j.fuel.2019.04.083.
Cai, R.J., Tang, S.W., He, Z., 2018. The modeling of electrical property in porous media
Li, K., Williams, W., 2007. Determination of capillary pressure function from resistivity
based on fractal leaf vein network. Int. J. Eng. Sci. 123, 143–157. https://doi.org/
data. Transport Porous Media 67, 1–15. https://doi.org/10.1007/s11242-006-0009-
10.1016/j.ijengsci.2017.11.006.
9.
Dong, F., Liu, N., Sun, Z., Wei, X., Wang, H., Nan, J., Ren, D., 2021. Quantitative
Lin, X., Zeng, J., Wang, J., Huang, M., 2020. Natural gas reservoir characteristics and
characterization of heterogeneity in different reservoir spaces of low-permeability
non-Darcy flow in low-permeability sandstone reservoir of sulige gas field, ordos
sandstone reservoirs and its influence on physical properties. Adv. Civ. Eng. 2021
basin. Energies 13, 1–17. https://doi.org/10.3390/en13071774.
https://doi.org/10.1155/2021/2399016.
Lu, Y., Liu, K., 2021. Pore structure characterization of eocene low-permeability
Edress, N.A.A., Fagelnour, M.S., Hassan, M.H.M., 2022. Subsurface geology and
sandstones via fractal analysis and machine learning: an example from the Dongying
geochemical evaluation of the middle jurassic-lower cretaceous organic-rich
Depression, Bohai Bay basin, China. ACS Omega 6, 11693–11710. https://doi.org/
intervals, west Kalabsha area, Western Desert, Egypt. Arabian J. Geosci. 15 https://
10.1021/acsomega.1c01015.
doi.org/10.1007/s12517-022-10686-2.
Mahmoud, M.A., Abdelgawad, K.Z., 2013. New insights into the prediction of capillary
EIA, U.S.E.I., 2021. International Energy Outlook 2021 with Projections to 2050, 20585.
pressure from resistivity measurements. Soc. Pet. Eng. - SPE Saudi Arab. Sect. Tech.
Washington, DC.
Symp. Exhib. 29–44. https://doi.org/10.2118/168059-ms, 2013.
Holditch, S.A., Chianelli, R.R., 2008. Factors that will influence oil and gas supply and
Mandelbrot, B., 1982. The Fractal Geometry of Nature. Wiley-VCH Verlag GmbH & Co.
demand in the 21st century. MRS Bull. 33, 317–323. https://doi.org/10.1557/
KGaA, Weinheim, Germany.
mrs2008.65.
Mason, G., Morrow, N.R., 1991. Capillary behavior of a perfectly wetting liquid in
Hou, B., Liu, H., Bian, H., Wang, C., Xie, R., Li, K., 2016. Experimental study on
irregular triangular tubes. J. Colloid Interface Sci. 141, 262–274. https://doi.org/
calculating capillary pressure from resistivity. In: All Days. SPE. https://doi.org/
10.1016/0021-9797(91)90321-X.
10.2118/179529-MS.
Mayer, R.P., Stowe, R.A., 1965. Mercury porosimetry-breakthrough pressure for
Katz, A.J., Thompson, A.H., 1985. Fractal sandstone pores: implications for conductivity
penetration between packed spheres. J. Colloid Sci. 20, 893–911. https://doi.org/
and pore formation. Phys. Rev. Lett. 54, 1325–1328. https://doi.org/10.1103/
10.1016/0095-8522(65)90061-9.
PhysRevLett.54.1325.
Oren, P.-E., Bakke, S., Arntzen, O.J., 1998. Extending predictive capabilities to network
Kwon, B.S., Pickett, G.R., 1975. A New Pore Structure Model and Pore Structure
models. SPE J. 3, 324–336. https://doi.org/10.2118/52052-PA.
Interrelationships.

14
M. Saafan et al. Geoenergy Science and Engineering 221 (2023) 111297

Patzek, T.W., Silin, D.B., 2001. Shape factor and hydraulic conductance in noncircular Shao, X., Pang, X., Li, H., Zhang, X., 2017. Fractal analysis of pore network in tight gas
capillaries. J. Colloid Interface Sci. 236, 295–304. https://doi.org/10.1006/ sandstones using NMR method: a case study from the ordos basin, China. Energy
jcis.2000.7413. Fuel. 31, 10358–10368. https://doi.org/10.1021/acs.energyfuels.7b01007.
Pittman, E.D., 1992. Relationship of porosity and permeability to various parameters Szabo, M.T., 1974. New methods for measuring imbibition capillary pressure and
derived from mercury injection-capillary pressure curves for sandstone. Am. Assoc. electrical resistivity curves by centrifuge. Soc. Pet. Eng. AIME J. 14, 243–252.
Petrol. Geol. Bull. https://doi.org/10.1306/bdff87a4-1718-11d7- https://doi.org/10.2118/3038-pa.
8645000102c1865d. Tan, W., 2018. A new method to predict FWL based on porous plate capillary pressure
Princen, H., 1970. Capillary phenomena in assemblies of parallel cylinders. J. Colloid and resistivity relationship. In: Day 2 Wed. SPE, pp. 1–21. https://doi.org/10.2118/
Interface Sci. 34, 171–184. https://doi.org/10.1016/0021-9797(70)90167-0. 192086-MS. October 24, 2018.
Rembert, F., Jougnot, D., Guarracino, L., 2020. A fractal model for the electrical Thomas, L.K., Katz, D.L., Tek, M.R., 1968. Threshold pressure phenomena in porous
conductivity of water-saturated porous media during mineral precipitation- media. Soc. Petrol. Eng. J. 8, 174–184. https://doi.org/10.2118/1816-PA.
dissolution processes. Adv. Water Resour. 145, 103742 https://doi.org/10.1016/j. Wei, W., Cai, J., Hu, X., Han, Q., 2015. An electrical conductivity model for fractal
advwatres.2020.103742. porous media. Geophys. Res. Lett. 42, 4833–4840. https://doi.org/10.1002/
Ruffet, C., Darot, M., Guéguen, Y., 1995. Surface conductivity in rocks: a review. Surv. 2015GL064460.
Geophys. 16, 83–105. https://doi.org/10.1007/BF00682714. Wheatcraft, S.W., Tyler, S.W., 1988. An explanation of scale-dependent dispersivity in
Saafan, M., Ganat, T., 2021. Inferring capillary pressure curve from 2D rock images based heterogeneous aquifers using concepts of fractal geometry. Water Resour. Res. 24,
on fractal theory in low-permeability sandstone: a new integrated approach. Fractals 566–578. https://doi.org/10.1029/WR024i004p00566.
29, 2150149. https://doi.org/10.1142/S0218348X21501498. Yu, B., 2008. Analysis of flow in fractal porous media. Appl. Mech. Rev. 61 https://doi.
Saafan, M., Ganat, T., 2020. A new capillary pressure model from fractal characterization org/10.1115/1.2955849, 0508011–05080119.
of porous medium : a case study from Malaysia. Solid State Technol. 63, 936–946. Yu, B., Cheng, P., 2002. A fractal permeability model for bi-dispersed porous media. Int.
Saafan, M., Ganat, T., Mohyaldinn, M., Chen, X., 2022a. A fractal model for obtaining J. Heat Mass Tran. 45, 2983–2993. https://doi.org/10.1016/S0017-9310(02)00014-
spontaneous imbibition capillary pressure curves based on 2D image analysis of low- 5.
permeability sandstone. J. Pet. Sci. Eng. 208, 109747 https://doi.org/10.1016/j. Yu, B., Li, J., 2002. ERRATUM: “some fractal characters of porous media. Fractals 10,
petrol.2021.109747. 365–372. https://doi.org/10.1142/S0218348X02001300.
Saafan, M., Mohyaldinn, M., Elraies, K., 2022b. An improved capillary pressure model for Zhang, K., Lai, J., Bai, G., Pang, X., Ma, X., Qin, Z., Zhang, X., Fan, X., 2020. Comparison
fractal porous media: application to low-permeability sandstone. J. Eng. Technol. of fractal models using NMR and CT analysis in low permeability sandstones. Mar.
Sci. 54, 220507 https://doi.org/10.5614/j.eng.technol.sci.2022.54.5.7. Petrol. Geol. 112, 104069 https://doi.org/10.1016/j.marpetgeo.2019.104069.

15

You might also like