International Geomechanics Conference (IGS) ARMA-IGS-2024–665

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
3D Structural Geomechanics and Machine Learning Based
Petrophysical Evaluation of Proven and Mature Oilfield in Upper
Indus Basin, Pakistan.

Najwa Fayyaz.
Institute of Geology, University of the Punjab, Quaid-e-Azam Campus, Lahore, Punjab, Pakistan.

Copyright 2024 ARMA, American Rock Mechanics Association

This paper was prepared for presentation at the International Geomechanics Conference (IGS) on 18 -20 November 2024 in Kuala Lumpur,
Malaysia. This paper was selected for presentation at the conference by an IGS Technical Program Committee based on a technical and critical
review of the paper by a minimum of two technical reviewers. The material, as presented, does not necessarily reflect any position of the partner
societies ARMA/DGS/SEG/SPE/AAPG/SPWLA/CSRME/EAGE/SRMEG/MOGSC, its officers, or members. Electronic reproduction,
distribution, or storage of any part of this paper for commercial purposes without the written consent of the partner societies is prohibited.
Permission to reproduce in print is restricted to an abstract of not more than 200 words; illustrations may not be copied. The abstract must contain
conspicuous acknowledgment of where and by whom the paper was presented.

ABSTRACT: Machine learning (ML) has revolutionized petrophysical analysis by providing advanced tools to efficiently interpret
well-log data and predict reservoir properties. In this study, key well logs, including Gamma Ray (GR), Resistivity (LLD, LLS, and
MSFL), Neutron Porosity (NPHI), Bulk Density (RHOB), and Sonic (DT), were utilized to evaluate reservoir characteristics in the
Meyal Oilfield. Supervised ML algorithms, such as Random Forest Regressor (RFR), Extra Trees Regressor (ETR), and Support
Vector Machines (SVM), were deployed to predict critical properties, including porosity, permeability, water saturation, and shale
volume. Dimensionality reduction via Principal Component Analysis (PCA) and clustering techniques like K-Means further
enhanced feature selection and geological interpretation. The application of ensemble learning and artificial neural networks (ANNs)
demonstrated exceptional accuracy in automating well-log interpretation, surpassing traditional methods in efficiency and precision.
In addition, seismic data analysis was conducted using 2D seismic lines, integrating ML-predicted petrophysical properties with
structural interpretation. Horizons corresponding to the Sakesar and Chorgali formations were delineated, revealing structural traps
and fault systems crucial for hydrocarbon accumulation. This study underscores the transformative role of ML in subsurface
reservoir characterization, highlighting its potential to optimize hydrocarbon exploration and production strategies in complex
geological environments like the Potwar Basin.

Keywords: Artificial Intelligence, Machine Learning, Random Forest Regression, ETR, Clustering, Neural Networks.

1. INTRODUCTION
Artificial neural networks (NNs) have been used in a other industries (Akkurt et al., 2018). A comparative
study to forecast how well CO2 foam flooding will work analysis of the integration of different machine-learning
to improve oil recovery on a lab scale. Using methods used to estimate the energy efficiency of
petrophysical data, created a model that uses artificial buildings' heating loads for smart city design. ML-based
intelligence (AI) to forecast the porosity and permeability models to forecast permeability impairment due to scale
of petroleum reservoirs in addition to reservoir deposition were studied, along with a comparison of
characteristics modeling, many scientists have created various ML techniques for estimating the permeability
data-driven methods for wax deposition prediction and and porosity of oil reservoirs using petrophysical logs
applied advanced machine learning techniques to the (Al-Khalifa et al., 2020). When compared to traditional
problems encountered in engineering, construction, and methods, it was found that the Extra Trees Regressor

1
performed exceptionally well in estimating the volume of 2. GEOLOGICAL SETTINGS AND
shale and porosities, while RFR and DTC were the most STRATIGRAPHY OF MEYAL OILFIELD
effective in modeling Sw and facies. This is ascribed to The Meyal Oil Field is situated in an active foreland fold
its capacity to accurately detect patterns in the training and thrust zone in District Attock, in the Potwar sub-basin
data and hence model reservoir features (Zhang et al., of the Upper Indus Basin (Hasany & Saleem, 2012). One

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
2021). With a time-efficient approach and optimized of Pakistan's most established petroleum exploration
results, suggested machine learning algorithms have locations is the Potwar sub-basin, a significant
hydrocarbon province. The discovery of the Potwar sub-
effectively addressed the drawbacks of traditional
petroleum basin began in 1915 when the forerunner of the
methods, including generalization and data range, for Attock Oil Company (AOC) found commercial amounts
petrophysical prediction without requiring extensive use of oil close to Khaur (District Attock), Punjab (Kadri,
of geological or lithological characteristics of the 1995). Pakistan Oilfields Limited (POL) discovered the
reservoir formation as shown in Figure 1 By adding more Meyal Oil Field in 1968. One of the most significant oil-
realistic data (core) and comprehensive data sets from all producing fields in the Potwar is the 16-well Meyal Oil
over the world with difficult reservoirs, this strategy can Field. Since it sheds light on the formation and structural
arrangement of a basin, accurate visualization of
be further enhanced (Najwa et al., 2023) With a variety of subsurface geological features and structures is essential
inputs, machine learning algorithms can gain experience for hydrocarbon exploration. The Potwar Basin yielded
without being specifically designed to do so. the first successful oil discovery in 1915. The Eocene
Classification, continuous value prediction, and limestones are known to be a large hydrocarbon-
performance or event forecasting are among the producing reservoir. Furthermore, Miocene shales are
predictions that may be made using the built model (Bader classified as seal rocks (Jadoon et al., 2015).
et al., 2019).

In order to predict petrophysical properties like shale

volume, porosity, water saturation porosity, and
permeability, among others, machine learning (ML) and
statistics have been widely used as regression tools [e.g.,
random forests (RF), one-class support vector machine
(SVM), support vector regression (SVR), genetic
algorithms, etc.]. The Random Forest Regressor and Extra
Trees Regressor are two of these machine-learning
algorithms that are quite popular and efficient for
estimating reservoir parameters from well-log variables
(Banas et al., 2021). To assess the permeability of the
reservoirs, core data and two hybrid machine learning
techniques have been used such as least-squares support
vector machines (SVM) and PSO-trained neural
networks. A comparison of machine-learning-based and
physics-based methods for permeability prediction in
cemented sandstones. A neural network (NN) and
nonparametric regression have been used with
multivariate analysis to forecast porosity and permeability
Figure 1. Illustration of unsupervised learning techniques, including dimensionality reduction (PCA) and clustering (K-Means),
used for identifying patterns and reducing noise in well log data.

2
The Sub-Himalaya division's Potwar Basin contains Thick-skinned tectonics are depicted by pop-up structures
significant hydrocarbons that are trapped in both and faults terminating from the basement caused by
compressional and transgressional formations. Numerous transpressional forces (Riaz et al., 2018; Moghal et al.,
scientists have postulated that the geologic structure has a 2007). The entire region is affected by Neogene buckling,
general east-west tendency at the Eocene level. Among and outcrops and faults show an ENE-SSW trend in
the notable formations are pop-ups, salt cores, and several locations. The Meyal area's subsurface structure is

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
reversal faults that limit the incline. The trapping not directly beneath the surface structure. Fold structures
mechanism of the Meyal Field is largely structural, and in this area are often sub-latitudinal in orientation, and
the eastern fold is marginally tighter than the western fold. structural complexities grow as one moves northward.
This study's objective is to assess the Meyal Field's The majority of wells in the Potwar sub-basin are drilled
petroleum play utilizing well and seismic data that will be based on structural interpretation are shown in Figure 2.
helpful for future development in the Potwar Basin and The stratigraphic formations and Lithostratigraphic and
the surrounding areas. Thick-skinned tectonics are Chronostratigraphic correlations of Meyal 6P and Meyal-
depicted by pop-up structures and faults terminating from 5P Wells are shown in Table 1. In Figure 3 Schlumberger
the basement caused by transpressional forces (Riaz et al., chart shows Rmf, Rmfeq Rwe, and Rw values at formation
2018; Moghal et al., 2007). temperature for Lockhart Limestone of Meyal-05P.

Figure 2. Geological section highlighting the structural framework, including anticlines, synclines, and thrust
faults, as interpreted from seismic data in the Potwar Basin.

3
Table 1. Lithostratigraphic and Chronostratigraphic Correlation of Meyal 5P and Meyal 10P Wells modified
after Awais et al., 2019.

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
Studied Formations

Figure 3. Schlumberger chart showing Rmf, Rmfeq and Rwe and Rw values at
formation temperature for Lockhart Limestone of Meyal-05P (modified after
Schlumberger 2009)

4
3. DATA AND METHODOLOGY Supervised learning consists of two subtypes:
classification and regression. Classification tasks require
3.1. Data Analysis trained data to convert discrete input values into discrete
This study used two sources that offered useful data, such output categories or labels, such as facies classification
as logs and other information used in the petrophysical and reservoir rock typing (Gudivada et al., 2017).
analysis. The fundamental purpose of this research is to

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
establish the most effective ML model for efficient 3.4. Seismic Interpretation
quality control of well-log data sets and accurate The data available for this inquiry include five 2D seismic
prediction of petrophysical attributes and also a good lines and wireline logs from two wells, Meyal-05 and
comparative analysis with the standard ways of Meyal-06. Dip lines are GO-MYL-801-02, GO-MYL-
petrophysical interpretation. ML algorithms can generally 801-03, and GO-MYL-801-04, whereas strike lines are
be classed into supervised learning, unsupervised GO-MYL-801-06 and GO-MYL-801-07. Table 3
learning, and reinforcement learning (Bukar et al., 2019). Displays the orientation of seismic lines. This information
was obtained by Pakistan's Directorate General of
3.2. Unsupervised Machine Learning Petroleum Concessions (DGPC). The Seismic Micro
Unsupervised learning, also known as clustering or Technology Kingdom program was used to analyze
dimensionality reduction algorithms, is a type of machine seismic data and conduct petrophysical analyses. The next
learning (ML) algorithm that works on data without step was to mark horizons, and two of them were marked:
explicit labels or supervision. Its purpose is to discover Chorgali (Green) and Sakesar (Red) as in Table 2
hidden patterns in the data that are difficult to detect. Contour maps were created on top of the specified
Figure 1 illustrates how these algorithms project high- horizons using the two-way time (TWT) and depth to
dimensional data onto a lower-dimensional space while locate promising hydrocarbon prospects.
retaining as much of the original structure as possible.
There are several unsupervised algorithms such as one- 3.5. Petrophysical Interpretation
class SVM, K-means, principal component analysis Petrophysical analysis is crucial for characterizing
(PCA), etc. reservoirs, giving valuable information on underlying
formations and resource evaluation. Petrophysical
3.3. Supervised Machine Learning metrics, including shale/clay volume, effective porosity,
Supervised Machine Learning is widely utilized for its water saturation, and facies, are crucial for the gas and oil
ability to perform numerous functions on input and output industry (Taner, 2001). However, these approaches are
data. Supervised machine learning datasets are labeled, time-consuming, labor-intensive, and require specific
allowing algorithms to identify attributes and make knowledge and equipment. Table 4 Shows the calculated
predictions on test data accurately. In supervised learning, depth and velocity parameters of wells at different
a feature and target dataset is required for model fitting horizons In contrast, ML-based petrophysical
and training (Dramsch, 2020). Supervised learning interpretation uses AI and ML algorithms to analyze and
techniques can handle both regression and classification interpret data. This approach efficiently analyzes large
issues. Supervised learning is the most common machine amounts of data and identifies patterns and relationships
learning task. The function is learned by analyzing that may not be apparent with traditional methods.
example input-output pairs.
Table 2. Workflow Chart for Geophysical Seismic Interpretation.

5
 Table 3. Summary of 2D seismic line orientations (dip and strike) and their correlation with wells
in the Meyal Oilfield highlighting key control lines
.

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
SEISMIC LINES ORIENTATION LINE WELLS
NAMES

801-MYL-02 North-South Dip line ----------

801-MYL-03 North-South Dip line ----------

801-MYL-04 North-South Dip line MEYAL-5P

801-MYL-06 Northeast Southwest Strike line ----------

MEYAL-6P
801-MYL-07 Northwest-Southeast Strike line

Table 4. Overview of seismic reflectors (Chorgali, Sakesar, and Nammal formations), including calculated
depths, velocities, and time correlations from seismic and well-log data (Moghal et al., 2007; Jadoon et al., 2015).

Reflector’s Name Time on Velocity (ms-1) Calculated Depth in well

seismic
Depth(m) (m)
section(s)
S=Vt/2

Chorgali Formation 2.301 3175 3652.8 3681.4

Sakesar Limestone 2.365 3175 3754.43 3746

Nammal Formation 2.435 3175 3865.56 3828.3

6
3.6. Conventional Petrophysical Analysis 3.7. Petrophysical Analysis by Machine Learning
The rock characteristics and fluid content of reservoir Methods
rocks are often ascertained using conventional Determining the properties of reservoirs and creating new
procedures, which is crucial for assessing the fields need the application of petrophysics. The
hydrocarbon reservoirs' commercial feasibility. Since application of machine learning (ML) has improved
shale impacts the reservoir's effective porosity and drilling efficiency, data restoration, reservoir property

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
permeability, its distribution is one of the most important prediction, reservoir rock type, and other petrophysical
factors to take into account when evaluating a formation. tasks. To determine petrophysical features, strong
Petrophysical interpretations are performed through machine learning is used in place of conventional
conventional techniques in both training and testing methods. A range of machine learning models are used to
wells. Reservoir facies have been interpreted in the facies identify the best model for accurately interpreting well
column after applying petrophysical cutoffs logs at different well sites. The most accurate model for
systematically. After examining the petrophysical blind well prediction has been identified utilizing the ML
parameters of the selected well, zones of interest were algorithm performance heat map (Hampson et al., 2001).
defined. The shale volume has been determined by Following quality control with supervised and
utilizing the Gamma-ray log response as shown in unsupervised learning algorithms, the petrophysical
Equation (1). parameters, including water content, effective porosities,
and the quantity of shaly material, Raw logs are used to
(GRlog − GRmin) (1) predict saturations. The ETR is an ensemble-supervised
ML approach that employs decision trees as shown in
Figure 4.

Porosity can be determined from bulk density. If the fluid

density is ρf, Ф is the porosity (1- Ф) is the remaining
volume fraction and ρma is the density of the rock matrix,
then the bulk density ρb can be determined as in
Equations 2, 3, and 4.

(2)

(3)

(4)

Saturation of water in flushed zone Sxo can be

calculated as in Equation (5).

Figure 4. Comprehensive depiction of supervised machine

(5) learning methodologies, including Random Forest Regressor,
Extra Trees Regressor, and Support Vector Machines,
employed for predicting petrophysical properties such as
porosity, permeability, and water saturation. The workflow
encompasses data preprocessing, model optimization, and
validation, as discussed by Gudivada et al. (2017).

7
4. RESULTS
4.1. Box plots of Data Sets The whiskers extend to include data points within 1.5
times the IQR, and the median, represented by a line
inside the box, highlights the dataset's central value.
Box plots were used to visualize the data distribution Individual points outside the whiskers, or outliers,
through their quartiles, as illustrated in Figure 5. Box plots

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
frequently point to important features or anomalies in the
are a crucial tool for displaying and evaluating the data. In well-log analysis, box plots have been applied to
distribution of petrophysical parameters in statistical a variety of parameters, such as gamma-ray, resistivity,
analysis of well-log data. By offering a thorough porosity, and permeability logs. These visualizations help
overview of the quartile distributions of big datasets, box delineate key reservoir characteristics and geological
plots facilitate the identification of patterns, trends, and heterogeneity. For instance, gamma-ray box plots reveal
anomalies (Zhang and Zhan, 2017). Because they provide variations in shale content, aiding in lithological
information on geological variability and facilitate differentiation. Porosity box plots highlight zones of
subsurface decision-making, their use is especially higher reservoir quality, while resistivity box plots assist
beneficial in reservoir characterization. The minimum, in identifying hydrocarbon-bearing intervals and
first quartile (Q1), median (Q2), third quartile (Q3), and differentiating them from water-saturated zones as stated
maximum are the five essential summary statistics used to in McDonald, 2021. Figure 5.
create box plots. The center of the picture is the
interquartile range (IQR), which runs from Q1 to Q3 and
reflects the central 50% of the data.

Figure 5. Box plots illustrating the distribution of key well log parameters from the MEYAL-06 well across the
depth interval of 106.7 to 4072.08 meters. The parameters include Gamma Ray (GR) in API units, Caliper (CALI)
measurements, Neutron Porosity (NPHI) in volume/volume decimal units, Bulk Density (RHOB) in g/cm³, and
Spontaneous Potential (SP) in millivolts. The box plots provide a statistical overview of the data, highlighting
median values, interquartile ranges, and potential outliers for each parameter, offering insights into reservoir
heterogeneity and log data variability.

8
 Figure 6. (a) Seismic Lines Base Map of Meyal Oilfield showing Meyal 5P and Meyal 6P locations. (b) Seismic
interpretation showcasing major thrust faults (F1, F2, F3) and pop-up structures that act as hydrocarbon traps in the Meyal
Oilfield. Three prominent reflectors are marked on seismic section e.g., Nammal, Sakesar and Chorgali Formations

4.2. Structural Interpretation

Hydrocarbon-capturing structural traps are shown by

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
structural interpretation. One vital step in seismic
interpretation is the well-to-seismic tie as in Error!
Reference source not found. (a). To adjust the datum, the
formation tops were subtracted from the seismic reference
datum (SRD) and well Kelly bushing (KB). In Meyal-05
well the Kelly bushing (KB) is 421.50m, respectively,
while the
seismic
reference
datum is
400m above Chorgali Sakesar Nammal
Formation Formation Formation
sea level.
Three
reflectors are
noted in the
seismic
F1 and F2,
section,
which are
indicating the
Chorgali, major thrust
Sakesar, and faults dipping
Nammal towards the
formations southeast, and
(Riaz et al., F3, which is a
2019). rear thrust
fault dipping
southwest,
were
indicated on
the control
line. The pop-
up structure is
produced by
faults that dip
in opposite
F1 F3 directions.
Line GO-
MYL-801-04
is the control
line because
F2 well-5 exists
on this line.
Three faults
were
indicated on
the control
line: F1 and
F2, which are main thrust faults dipping north, and F3,
which is a rear thrust fault dipping south as shown in
Error! Reference source not found. (b). Some younger
thrust faults can also be seen to the south. The formation

9
 Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025

10
of a popup structure is caused by faults dipping in

(b)
(a)
different directions.
4.3. Time Contour Maps and 3D Visualization of The anticline is indicated in multiple locations where the
Grids contours close with the lowest times in the cores
A plunging anticline surrounded by the thrust from both Investigating these anticlines for hydrocarbons can be
the north and the south is shown on the temporal structure beneficial if done properly Figure 9. The 3D Grid of all
map of the Eocene reflectors Figure 7. Closed contours the reflectors comprises the Meyal Fault Figure 10. The
with the highest times in the cores are called synclines. precise positions of the structures in the research area are

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
Closed contours with the lowest times in the cores are indicated by the contours (Friedman and Smith, 2011).
called incline lines Figure 8.

Meyal 5P

Figure 7. Two-way Time structure map of Chorgali Formation depicting anticlines, synclines,
and fault-related hydrocarbon traps. The contours reveal significant structural highs associated
with hydrocarbon accumulations in the Potwar Basin.

Meyal 5P

Figure 8. Two-way Time structure map of Sakesar Limestone depicting anticlines, synclines,
and fault-related hydrocarbon traps. The contours reveal significant structural highs associated
with hydrocarbon accumulations in the Potwar Basin

11
 Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
Figure 9. Two-way Time structure map of Nammal Formation depicting anticlines, synclines,
and fault-related hydrocarbon traps. The contours reveal significant structural highs associated
with hydrocarbon accumulations in the Potwar Basin.

Meyal Fault

Figure 10. Combined Three-dimensional model of Chorgali, Sakesar and Nammal Formations
displaying fault systems and reservoir structures interpreted from seismic data

12
4.4. Petrophysical Evaluation of Reservoir 1 The Chorgali formation's gamma-ray and median
Gamma-ray logs with sliced spherical shapes suggest a porosity values were compared. The fractured zone was
distal submerged funnel, while plain spherical-shaped identified by well-log analysis at depths ranging from
logs show constant deposition (Miah et al., 2020). Five 3940m to 4075m. Table 5 Displays the petrophysical
well-log tracks as shown in Figure 11. parameters derived from log responses (Mohadier et al.,
2010).

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
Figure 11. Petrophysical parameters of the Chorgali formation, GR= gamma ray, CAL= caliper, deep lateral-log
(LLD), short lateral-log (LLS), and micro spherical focus log (MSFL) are resistivity logs, Rhob = density, DT =
sonic log, NPHI = neutron log, Sw = water saturation and Shc = hydrocarbon saturation and PEF= Photoelectric
factor.

Table 5. Calculated Petrophysical parameters of Reservoir 1 i.e, Chorgali Formation of Meyal-5P Oil Well.

RESERVOIR 1

INTERVAL Vsh Ф𝑎𝑣𝑔 Net Sw 𝑆ℎ𝑐 LLD LLS MSFL Hydrocarbon

DEPTH Pay (%) (%) (Ωm) (Ωm) (Ωm) Predicted
(%) (m)

3927-3940 m 14% 11% 14 m 22% 80% 142 Ωm 147 Ωm 189 Ωm

3940-3957 m 10% 13% 17 m 21% 75% 171 Ωm 180 Ωm 176 Ωm
3957-3974 m 08% 22% 08 m 14% 72% 136 Ωm 250 Ωm 239 Ωm
3974-3991 m 11% 14% 18 m 20% 55% 202 Ωm 171 Ωm 155 Ωm
Oil Well
3991-4007 m 04% 10% 14 m 25% 47% 180 Ωm 310 Ωm 337 Ωm
4007-4024 m 18% 07% 05 m 18% 86% 220 Ωm 169 Ωm 132 Ωm
4024-4041 m 10% 17% 10 m 17% 79% 290 Ωm 271 Ωm 274 Ωm
4041-4058 m 05% 05% 17 m 29% 63% 154 Ωm 236 Ωm 120 Ωm
4058-4075 m 19% 17% 05 m 21% 90% 137 Ωm 300 Ωm 309 Ωm
4075-4092 m 12% 10% 17 m 12% 84% 190 Ωm 278 Ωm 250 Ωm

13
4.5. Petrophysical Evaluation of Reservoir 2 The Chorgali formation's gamma-ray and median
Sakesar limestone has undergone comparable processing. porosity values were compared. The fractured zone
Figure 12 Depicts a fractured zone between 2126m- was identified by well-log analysis at depths
2137m depth. The Eocene era revealed two hydrocarbon ranging from 3780m to 3940m. Table 6 displays
zones. Low Gamma-Ray values indicate smooth the petrophysical parameters derived from log
development, while high levels indicate the existence of

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
responses (Storey, 2016). zones in the formations
shale (Mulders et al., 2017).

Figure 12. Petrophysical parameters of the Sakesar formation, GR= gamma ray, CAL= caliper, deep lateral-log
(LLD), short lateral-log (LLS), and micro spherical focus log (MSFL) are resistivity logs, Rhob = density, DT =
sonic log, NPHI = neutron log, Sw = water saturation and Shc = hydrocarbon saturation and PEF= Photoelectric
factor.

Table 6. Calculated Petrophysical parameters of Reservoir 2 i.e., Sakesar Limestone of Meyal-5P Oil Well.

RESERVOIR 1

INTERVAL Vsh Ф𝑎𝑣𝑔 Net Sw 𝑆ℎ𝑐 LLD LLS MSFL Hydrocarbon

DEPTH Pay (%) (%) (Ωm) (Ωm) (Ωm) Predicted
(%) (m)

3770-3787 m 09% 08% 14 m 19% 80% 150Ωm 220 Ωm 219 Ωm

3787-3804 m 18% 09% 19 m 15% 71% 180Ωm 149 Ωm 223 Ωm
3804-3821 m 13% 19% 11 m 35% 56% 172Ωm 241 Ωm 165 Ωm
3821-3838 m 08% 15% 09m 40% 78% 200Ωm 124 Ωm 150 Ωm
3838-3855 m 05% 10% 18 m 22% 76% 215Ωm 320 Ωm 271 Ωm Oil Well
3855-3872m 17% 06% 15 m 32% 57% 143 Ωm 189 Ωm 155 Ωm
3872-3889 m 04% 17% 05 m 13% 48% 256 Ωm 171 Ωm 210 Ωm
3889-3906 m 19% 04% 04 m 43% 74% 162 Ωm 192 Ωm 186 Ωm
3906-3923 m 07% 19% 12 m 36% 65% 190 Ωm 202 Ωm 109 Ωm
3823-3940 m 15% 13% 05 m 24% 79% 207 Ωm 250 Ωm 217 Ωm

14
5. CONCLUSIONS (ii) Al Khalifah, H.; Glover, P. W. J.; Lorinczi, P.
Permeability prediction and diagenesis in tight
• The existence of thrusting on all dip lines, which carbonates using machine learning techniques. Mar.
could be the result of compression, is confirmed Pet. Geol. 2020, 112, No. 104096.
by 2D seismic interpretation. https://doi.org/10.1016/j.marpetgeo.2019.104096
(iii) Awais, M., Ullah, F., Khan, N., Ghani, M., Siyar, S.
• the pop-up structure and post-Eocene deposits

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
M., Wadood, B., & Mukhtiar, A. (2019). Investigation
provide a suitable structural trap for hydrocarbon of reservoir characteristics, depositional setting and T–
accumulation. The reversal fault served as a seal, R sequences of the Lockhart Limestone of Meyal Oil
with Eocene shale serving as the caprock. Field, Pakistan: a petrophysical approach. Journal of
Petroleum Exploration and Production Technology,
• The 3D view of temporal structure demonstrated 10, 1423–1441.
the development of subsurface structures in terms https://doi.org/10.1007/s13202-019-0730-x
of time and space.
• In contrast, the Sakesar Limestone, which is (iv) Bader, S.; Wu, X.; Fomel, S. Missing log data
interpolation and semiautomatic seismic well ties
fractured at depths of 2126m to2137m, and the
using data matching techniques. Interpretation 2019,
Chorgali Formation is fractured at a depth of 7, T347−T361.
2021m to 2031 m. https://doi.org/10.1190/INT-2018-0237.1
• There are fracture zones present in Eocene (v) Banas, R.; McDonald, A.; Perkins, T. J. In Novel

limestones, according to Traditional as well as Methodology for Automation of Bad Well Log Data
Identification and Repair, SPWLA 62nd Annual
Machine Learning log responses.
Logging Symposium, OnePetro, 2021.
https://www.onepetro.org/conference-paper/SPWLA-
2021-v62n3a1
ACKNOWLEDGMENTS (vi) Bukar, I.; Adamu, M. B.; Hassan, U. In A Machine
Learning Approach to Shear Sonic Log Prediction;
The authors acknowledge the Data Administration OnePetro, 2019.
division of LMKR (the landmark affiliate), the https://www.onepetro.org/conference-paper/SPWLA-
Directorate General of Petroleum Concessions (DGPC), 2019-v60n3a1
Pakistan, for providing subsurface geophysical data (vii) Dramsch, J. S. 70 years of machine learning in
(seismic and well), and SMT (Seismic Micro geoscience in review. Adv. Geophys. 2020, 61, 1−55.
Technology) for providing license geophysical software https://doi.org/10.1016/bs.agph.2020.08.002
Kingdom Suit 8.8. (viii) Friedman, T.; Smith, M. Measuring the Business
Value of Data Quality; Gartner: Stamford, 2011; Vol.
464.
https://www.gartner.com/en/documents/1730416
DATA AVAILABILITY STATEMENT (ix) Gudivada, V.; Apon, A.; Ding, J. Data quality
The original contributions presented in the study are considerations for big data and machine learning:
included in the article/Supplementary Material, further Going beyond data cleaning and transformations. Int.
J. Adv. Softw. 2017, 10, 1−20.
inquiries can be directed to the corresponding author.
https://arxiv.org/abs/1610.07143
(x) Hampson, D. P.; Schuelke, J. S.; Quirein, J. A. Use of
AUTHOR CONTRIBUTIONS multiattribute transforms to predict log properties
from seismic data. Geophysics 2001, 66, 220−236.
Najwa Fayyaz; conceptualization, methodology, https://doi.org/10.1190/1.1444899
software, visualization, formal analysis, investigation, (xi) McDonald, A. Data Quality Considerations for
data collection, literature review, validations, writing the Petrophysical Machine-Learning Models.
original draft, review, editing, and submission of the final Petrophysics 2021, 62, 585−613.
paper. https://www.onepetro.org/journal-paper/SPWLA-
2021-v62n6a1
(xii) Miah, M. I.; Zendehboudi, S.; Ahmed, S. Log data-
REFERENCES driven model and feature ranking for water saturation
prediction using machine learning approach. J. Pet.
(i) Akkurt, R.; Conroy, T. T.; Psaila, D.; Paxton, A.; Low, Sci. Eng. 2020, 194, No. 107291.
J.; Spaans, P. In Accelerating and Enhancing https://doi.org/10.1016/j.petrol.2020.107291
Petrophysical Analysis with Machine Learning: A (xiii) Mohadjer, S., Bendick, R., Ischuk, A., Kuzikov, S., &
Case Study of an Automated System for Well Log Kostuk, A. (2010). Partitioning of India-Eurasia
Outlier Detection and Reconstruction, SPWLA 59th convergence in the Pamir Hindu Kush from GPS
Annual Logging Symposium, OnePetro, 2018. measurements, Geophysical Research Letters, 37 (4)
https://www.spwla.org/SPWLA/Publications/Publicat 90-98.
ion_Detail.aspx?iProductCode=2018_BB https://doi.org/10.1029/2009GL041737

15
(xiv) Mulders, F.; Lemanczyk, R.; Johnstone, K.; Spencer,
S.; Castillo, E.; Sorrentino, Y.; Nugraha Khadafi, A.
In From 3D Seismic Inversion to Drilling Engineering:
A Multidisciplinary Approach to Resolve Wellbore
Stability Issues Offshore West Madura, SPE/IATMI
Asia Pacific Oil & Gas Conference and Exhibition,

Downloaded from http://onepetro.org/armaigs/proceedings-pdf/IGS24/IGS24/ARMA-IGS-2024-0665/4428066/arma-igs-2024-0665.pdf by Texas A&M University, Kingsville, Tresia Kaanankoto Shifugula on 12 March 2025
OnePetro, 2017.
https://www.onepetro.org/conference-paper/SPE-
187134-MS
(xv) Najwa, F., Naveed, A., & Abid, A. (2023). Advanced
Seismic Interpretation, Multivariate Attribute
Analysis, and Petrophysical Evaluation Techniques
for Holistic Structural and Reservoir Characterization
of the Joya Mair Oilfield, NW-Himalayas, Pakistan},
volume = {International Geomechanics Symposium.
10.56952/IGS-2023-0347, ARMA-IGS-2023-0347.
https://doi.org/10.56952/IGS-2023-0347
(xvi) Riaz, M., Pimentel, N., Ghazi, S., Zafar, T., & Alam,
A. (2018). Lithostratigraphic Analysis of the Eocene
reservoir units of Meyal Area, Potwar Basin, Pakistan,
Himalayan Geology, 39 (2) 72-81.
http://nceg.uop.edu.pk/
(xvii) Riaz, M., Pimentel, N., Zafar, T., & Ghazi, S.
(2019). 2D seismic interpretation of Meyal Area,
Northern Potwar Deform Zone (NPDZ), Potwar Sub-
Basin, Pakistan, Open Geoscience, 11 (1) 1-1
https://doi.org/10.1515/geo-2019-0001
(xviii) Storey, M. C. In Demystifying Log Quality
Control, SPE Asia Pacific Oil & Gas Conference and
Exhibition, OnePetro, 2016.
https://www.onepetro.org/conference-paper/SPE-
183754-MS
(xix) Taner, M. T. (2001). Seismic attributes. CSEG
Recorder, 26(7), 48-5
https://www.cseg.ca/
(xx) Zhang, Z.; Zhang, H.; Li, J.; Cai, Z. Permeability and
porosity prediction using logging data in a
heterogeneous dolomite reservoir: An integrated
approach. J. Nat. Gas Sci. Eng. 2021, 86, No. 103743.
https://doi.org/10.1016/j.jngse.2020.103743
(xxi) Zhang, L.; Zhan, C. Machine Learning in Rock Facies
Classification, Application of XGBoost, International
Geophysical Conference, Society of Exploration
Geophysicists and Chinese Petroleum Society:
Qingdao, China, 2017; pp 1371−1374.
https://doi.org/10.1190/segam2017-17769305.1

16