Reservoir Engineering: SPE-194746-MS

Artificial
Reservoir Lower
Carbon
Radical
Intelligence,
Productivity
Engineering Intensity Improvement
Digital,
Automation
SPE-194746-MS The End of Petroleum Engineering

as We Know It:
The next generation of petroleum engineers will have to address demands for
sustainability, lower carbon intensity and needs for radical productivity
improvements, which only AI and digital can drive
Tasks to do:
People Development, Skill gap assessment

Guidelines, Best Practices
Tools :
Number of Licenses
Scope/Spec (AHM module, OpenServer etc)
100 ARTIFICIAL INTELLIGENCE
USE CASES IN THE
UPSTREAM OIL & GAS INDUSTRY
DISCLAIMER
THE MATERIAL OR INFORMATION CONTAINED ON THIS DOCUMENT IS FOR GENERAL

INFORMATION PURPOSES ONLY. YOU SHOULD NOT RELY UPON THE MATERIAL OR INFORMATION
AS A BASIS FOR MAKING ANY BUSINESS, LEGAL OR ANY OTHER DECISIONS.
MUSTAFA KAOURI MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND, EXPRESS OR

IMPLIED ABOUT THE COMPLETENESS, ACCURACY, RELIABILITY, SUITABILITY OR AVAILABILITY WITH
RESPECT TO THE INFORMATION, PRODUCTS, SERVICES OR RELATED GRAPHICS CONTAINED ON
THE DOCUMENT FOR ANY PURPOSE. ANY RELIANCE YOU PLACE ON SUCH MATERIAL IS
THEREFORE STRICTLY AT YOUR OWN RISK.
MUSTAFA KAOURI WILL NOT BE LIABLE FOR ANY FALSE, INACCURATE, INAPPROPRIATE OR
INCOMPLETE INFORMATION PRESENTED ON THIS DOCUMENT.
100 AI USE CASES IN THE UPSTREAM OIL & GAS INDUSTRY
#32: DETECTION OF NATURAL OIL SEEPS IN OCEAN FLOOR
Activity: Reservoir Engineering
Challenge:
Natural seeps occur when oil escapes from rock found on the ocean floor. An
estimated 60 percent of oil underneath the earth's surface in North America is due
to natural seeps.
Solution:
Robots with the ability to navigate these oceanic regions and detect oil seeps can
protect the ecosystem and serve as indicators for robust energy resources. Apply AI
to boost its natural seep detection capabilities.
Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 5

#33: FORECASTING OF OIL PRODUCTION
Challenge:
Production forecasting helps in the estimation of available reserves and economic
evaluation. Traditionally numerical simulations and decline curve analysis (DCA)
techniques are extensively used for production forecasting. However, these
techniques require knowledge of the reservoir behavior and lack flexibility for
modeling complex physics.
Solution:
AI methodologies using feedforward neural networks and recurrent neural networks
have demonstrated their effectiveness and accuracy in a production forecast of
single and multiple wells. These AI approaches make the forecasting process
efficient and accurate for the assets with or without significant operational history
information.

#34: OPTIMIZATION OF HISTORY MATCHING
Challenge:
The most crucial task in oil and gas development is to predict future development
based on the existing data and make a reasonable development plan. Therefore,
history matching is needed to facilitate the subsequent numerical simulation before
prediction. However, due to the complexity of actual oilfield development, multiple
factors simultaneously interfere, which increases the difficulty of history matching
with conventional methods.
Solution:
With the development of AI, it has been proven feasible to apply a neural network
to the history matching of oilfield development. The history matching and training of
the existing data can effectively capture the nonlinearity of problems with fast
matching speed and good precision and further become a new history matching
method. Artificial neural network (ANN) is commonly utilized to conduct training and
history matching oilfield development. In addition, genetic algorithm (GA) is often
combined to optimize and improve the matching speed and accuracy.

#35: PREDICTION OF ACCURATE PVT PROPERTIES
Challenge:
It is essential to accurately determine the pressure-volume-temperature (PVT)
properties of oil and gas due to their high significance for material balance
calculations. In addition, these parameters directly affect reservoir performance
and evaluation and production operation and design. PVT correlations and equation
of state (EOS) are two approaches to estimate PVT properties when the
experimental PVT data is unavailable.
Solution:
To overcome the shortcomings of empirical correlations, intelligent models were
developed to predict the PVT properties more accurately. Artificial neural networks
(ANNs) to estimate the bubble point pressure (Pb) and oil formation volume factor
(Bo) in the area of PVT properties prediction. Also, support vector machines (SVMs)
could accurately predict the Pb and Bo. Application of genetic algorithms (GAs) can
be employed to predict the viscosity of crude oils.

#36: PREDICTION OF ASPHALTENE PRECIPITATION
Challenge:
Asphaltenes are very complex molecular compounds that exist as a heavy fraction
in petroleum. Asphaltenes have the potential to damage the oil-bearing formation
and production equipment considerably. Asphaltene precipitation and deposition
can lead to catastrophic formation damage problems in terms of pore throat
plugging and wettability alteration during different phases of field development
(drilling, production, and injection). Therefore, it is essential to predict this
phenomenon so that preventive precautions can be made.
Solution:
Employ the MLP-ANN technique to predict the weight percent of precipitated asphaltene considering
P, API gravity, the molecular weight of the precipitant, the dilution ratio, and the ratio of asphaltene
as input parameters. Employ the SVM, FCM, ANFIS, and LSSVM modeling techniques to predict the
precipitation of asphaltene about the conventional input variables temperature, type of precipitant,
and the dilution ratio. Utilize the RBF network to estimate the reduction of asphaltene precipitation
due to the presence of inhibitors as a function of molecular weight and concentration of inhibitors,
API, oil asphaltene content, and the number of carboxylic, hydroxyl and circular structure groups.
Employ the LSSVM and the RBF NNs to predict the amount of asphaltene precipitation considering
paraffin type, crude oil colloidal stability index, and the dilution ratio as the input variables.

#37: PREDICTION OF FORMATION DAMAGE
Challenge:
Formation damage is an inevitable and undesirable problem during the oilfield
development and production processes. Many controllable and uncontrollable
factors that can cause the disturbance of the existing equilibrium inside a porous
medium may potentially be considered a cause of formation damage. In addition,
formation damage can severely reduce reservoir permeability and well
performance. Therefore, accurate prediction of formation damage before its
occurrence can significantly contribute to the petroleum industry.
Solution:
Deploy the CSA-LSSVM model to predict the permeability reduction factor (Kd/Ki )
regarding the injection rate, the volume of injected water, T, ΔP, Ki, and solution ionic
components after seawater injection as input parameters. A predictive model based on
quantum NNs estimates the severity of aqueous phase trapping (a formation damage
mechanism). The input parameters are initial water saturation (Swi), IFT between oil and
water, ϕ, gas permeability, and average pore diameter. Develop a GEP algorithm for
evaluating the formation damage owing to mixed sulfate deposition.

#38: PREDICTION OF INTERFACIAL TENSION (IFT)
Challenge:
Interfacial tension (IFT) and minimum miscibility pressure (MMP) are among the most
critical two-phase properties of reservoir fluids. The significance of IFT is well
understood since it has an essential role in many industrial and engineering
processes.
Solution:
An artificial neural network (ANN) model can predict the IFT value of pure
hydrocarbon and water systems over temperature and pressure. Furthermore, the
Common Scrambling Algorithm (CSA) can optimize the model and apply the radial
basis function (RBF) kernel function.

#39: RECOGNIZING RESERVOIR MODEL THROUGH WELL TESTING

DATA
Challenge:
Identifying well test interpretation models and estimating reservoir parameters are
two essential parts of a Well test analysis process. The information provided by
Well tests is also necessary for estimating the productive reservoir capacity and
average pressure. The analysis of reservoir performance and predicting its future
production is based on having appropriate information about the reservoir
properties and circumstances.
Solution:
Employ a recurrent ANN to categorize the reservoir models based on pressure transient
data. The three optimization algorithms, namely, PSO, DE, and covariance matrix
adaptation evolution strategy, can predict various reservoir parameters such as
drainage radius, wellbore storage coefficient, skin factor, and reservoir permeability in
different reservoir models (fractured, radial composite, and homogeneous reservoirs)
through well test data. Employ an MLP ANN or PNN to recognize the reservoir model
through well-testing data. Also, you can use the GRNN and MLP for predicting the
reservoir permeability, wellbore storage coefficient, skin factor, and storativity ratio.

#40: SPEEDUP RESERVOIR MODELING
Challenge:
Computations done with conventional reservoir modeling tools perform numerical
solutions of partial differential equations describing the physics of reservoir flows.
However, these computations are lengthy, limiting possible runs and the optimization
ability for proper field development planning.
Solution:
Modern reservoir models with a new computation engine based on AI deep neural
networks compress the mathematical problem dimensionality and approximate the
time derivatives promise 100–1000 times the conventional models' speedup while
keeping similar functionality.

#41: SPEEDUP UPSCALING IN RESERVOIR MODELING
Challenge:
Many reservoir engineers use tricks to perform manual upscaling (i.e., bringing
information gained from various scales of geophysical studies to a single geological
and then hydrodynamical reservoir models) in a way that seems correct to
themselves. This upscaling process introduces a strong bias to the reservoir model.
Solution:
Summing up the multiple experiences with a deep learning algorithm trained on
multiple cases of manual upscaling would increase objectiveness and increase the
speed of the upscaling process.

#42: PREDICTION OF MMP AND CO2 STREAMS
Activity: Enhanced Oil Recovery
Challenge:
Minimum miscibility pressure (MMP) is an essential two-phase property of reservoir
fluids. In miscible gas injection processes, the injection pressure should be high
enough (higher than MMP) to make the injected gas miscible in the reservoir fluid.
Otherwise, the injection process is considered as an immiscible gas injection. The
miscibility can be attained through first contact or multiple contact processes. The
lowest pressure at which the injected gas can achieve miscibility with the reservoir
oil is called MMP.
Solution:
To estimate the MMP of pure and impure CO2, use the GMDH scheme to develop
accurate correlations. Consider the T, Ppc, and Tpc of injected gas, the ratio of
volatile-to-intermediate oil fractions, and MW of C51 fraction as input variables. A
new model, namely, AdaBoost SVR, predicts the MMP of pure and contaminated
CO2 streams through reservoir temperature, pseudo critical temperature of injected
gas, mole fractions of volatile and intermediate fractions, and MW of C51 oil
fraction.

#43: PREDICTION OF RECOVERY FACTOR (RF) AND OIL RATE
Activity: Enhanced Oil Recovery
Challenge:
After the primary and secondary oil production phase from reservoirs, two-thirds of
the original oil in place will be left behind inside the porous media. The main
reasons for such recovery efficiency may be macroscopic and microscopic
heterogeneities, different fluid characteristics (densities and viscosities), various
Interfacial tensions (IFTs) and wettability, and other driving forces.
Solution:
Employ AI techniques in enhanced oil recovery (EOR) processes to predict the RF
and oil production rate in CO2-foam flooding processes. Two models were
considered: a multilayer perceptron (MLP) and a radial basis function (RBF) NN. The
RBF and MLP models predicted the recovery factor (RF) and oil rate as functions of
surfactant type, rock porosity, permeability, core pore volume (PV), initial oil
saturation, and injected PV of foam.

#68: DETECTION OF FAULTS IN PIPELINE SYSTEMS
Activity: Production Engineering
Challenge:
Pipelines are the most economical and efficient means of oil and natural gas
transportation over long distances in different environments; however, they are
subjected to corrosion and degradation. Pipeline accidents result in vast economic
losses as well as catastrophic environmental effects such as oil spills. Natural
hazards, mechanical, operational, corrosion, and third-party activities are the most
probable causes of oil pipeline failure.
Solution:
An ANN to locate and detect the leaks in liquified gas pipelines based on the pressure and flowrates
at the inlet and outlet of the pipeline. Apply ANNs to classify the pipe weld defects using magnetic
flux leakage signals. ANNs to detect faults in pipeline systems based on stationary and
nonstationary status. Utilize the FL model to predict the rate of corrosion regarding T, flowrates of
gas and liquid, pH, P, and CO2 partial pressure. Use SVM with Artificial Bee Colony (ABC) to detect
leaks in pipelines. Employ the ANN and FL techniques to predict pipe failure based on the type of
product carried by the pipeline, land use, pipeline age, pipeline location, and pipeline diameter.
Develop an MLP network trained through the ICA algorithm to predict the drag reduction in
petroleum pipelines as a function of temperature, type of pipe, Reynolds number, and type and
concentration of drag-reducing agents (DRAs).

#69: ESP FAILURES PREDICTION USING PCA APPROACH
Challenge:
The electrical submersible pump (ESP) is widely used for artificial lift, making it
critical to ensure continued oil production from the wells. Despite being among
crucial equipment in the upstream operations, ESPs exhibit significantly high failure
rates. These failures are often random and result in lost oil production from the
wells. It is estimated that ESP failures lead to hundreds of millions of barrels of lost
or deferred oil production each year.
Solution:
A principal component analysis (PCA) approach is used to detect ESP failures and
predict the remaining useful life of the equipment before failure by using complete
historical data. Combining these predictions from the AI-based models with
engineering principles to detect problems with ESPs, long before they occur and
prescribing preventive actions can have a significant economic impact.

#70: OPTIMIZATION OF GAS-LIFT PROCESSES
Challenge:
Gas lift works by injecting gas into the Well tubing through gas lift valves to reduce
the hydrostatic pressure on the fluid column below the fluid pressure in the reservoir
formation.
Solution:
Utilize ANNs and GAs to predict the efficiency of long-time gas-lift processes.
Employ GAs to optimize the gas-lift allocation problem considering the instability
phenomenon as a constraint. Employ SVM technique to optimize the natural gas lift
process.

#71: OPTIMIZATION OF OILFIELD DEVELOPMENT PLAN
Challenge:
It is challenging to evaluate an oilfield development plan's performance with one
production index due to the mutual influence of technical, economic, and social
indicators in the adjustment and optimization of that plan.
Solution:
AI applications in developing plan optimization mainly improve production or its
rate with the consideration of economic factors. The most used method for oilfield
developing plan optimization is the Artificial neural network (ANN) and the genetic
algorithm (GA) models. Combining geological background, historical matching,
dynamic monitoring, and economic benefit to optimize dynamic development plan.

#72: OPTIMIZING PRODUCTION OPERATIONS IN REAL-TIME
Challenge:
The objective of production optimization is to manage and ensure asset goals
profitably using all available information up to that point to predict outcomes with
confidence and to make decisions that produce optimal results and implement such
decisions until the next decision-making point in time.
Solution:
The increasing availability of real-time downhole measurements and remotely
activated valves in the oilfields has made field-wide operations optimization in
real-time a distinct possibility. With more real-time data and measurements, it's
possible to build AI models, which could be updated with the availability of new
data to optimize production in real-time.

#73: OPTIMIZING SELECTION OF OPTIMAL COMPLETION

PARAMETERS
Challenge:
Completion is a term to indicate the hardware deployed inside the wellbore to
ensure production lifting from the reservoir to the wellhead. Optimal completion
strategy requires the proper understanding of reservoir potential and the
production requirements (i.e., rate targets and wellhead pressure). Traditionally,
production engineering models, such as nodal analysis Well models, are good
enough to provide accurate predictive performance relationships for successful
completions.
Solution:
In the cases of unknown well failure or uncertain reservoir phenomena, data-driven
models provide much better predictions than pure engineering models. AI methods
assist operators in selecting optimum completion parameters, which have a positive
effect on production and are advantageous to lower unit costs, such as stage
intensity, plug-and-perforate cemented-well designs, injection rate, and proppant
mass per lateral foot, and fluid volume per lateral foot.

#74: PREDICTION OF CASING COLLAPSE OCCURRENCE & DEPTH
Challenge:
Well integrity is a crucial aspect that must be maintained through the lifecycle of a well,
and one component of which, the casing, must be able to withstand all the internal and
external loads. These loads include the invariable factors in the geothermal environment,
such as high-temperature, high-strength rock, highly-fractured formation, corrosive fluid,
and under-saturated pressure. Thus, in the casing construction process of setting depth
and design of a geothermal production well, these factors needed to be taken into
account based on the environment where the Well is located to ensure that casing
collapse does not happen.
Solution:
A back-propagating (BP) network of user-defined number of internal (hidden)
layers can be connected to input and output layers to provide an 'experienced'
estimate on casing collapse depth for the wells to be drilled. The data layer can
have a number of inputs such as location, depth, pore pressure, corrosion rate,
casing strength, etc., to analyze and provide feed on expected collapse depth and
probability of casing collapse.

#75: PREDICTION OF ESP LIFESPAN
Challenge:
Electric submersible pump (ESP) repairs are a costly issue. Identify the key drivers
behind ESP failures and determine if they could predict an ESP's lifespan accurately.
Solution:
Use predictive model techniques such as linear regression, decision trees, and high-
performance random forests (HP Forest). The best prediction model is the HP Forest
model, which can predict ESP lifespans within approximately five days. The top
variables of importance when predicting ESP lifespan are metrics related to ESP
shutdowns.

#76: PREDICTION OF MULTIPHASE FLOW RATES
Challenge:
Several sensors can provide measurements of temperature and pressure downhole
a well. The problem is how oil, gas, and water flow depends on these
measurements: i.e., the function that describes the multiphase flow rates.
Solution:
Capture the thermodynamics and fluid dynamics of the multiphase flow of oil, gas, and,
water from the production well to generate lots of simulated training data for an AI
model. By generating large amounts of training data from the physics-based model, we
can teach the AI model the physics of the problem. A trained AI model can use just the
sensor measurements from the physical Well, i.e., pressures and temperatures, to predict
the oil, gas, and water rates simultaneously. More importantly, it can make these
predictions within a fraction of a second, making it ideal for running on real-time data
from the production wells.

#77: PREDICTION OF PRESSURE GRADIENT IN WATER-OIL

PIPELINES
Challenge:
The precise determination of the pressure gradient parameter is complex because it
results from the frictional effects between fluids and pipe wall and the interfacial
effects between the fluids themselves.
Solution:
Employ LSSVM and the RBFNN models to predict the pressure gradient in water-oil
pipelines considering oil and water slip velocity, pipe diameter and roughness, and
oil viscosity as input variables.

#78: PREDICTION OF PRODUCTION INDEX
Challenge:
Production dynamic analysis methods (oilfield numerical simulation method,
characteristic curve method, production decline method, material balance method,
analogy method, empirical formula method, chart method, etc.) have been applied
in oilfield production for years. However, they still have apparent limitations due to
complex factors affecting the production index dynamic prediction.
Solution:
Combining a neural network with a fuzzy logic algorithm to accurately predict
production index to reach an excellent fitting accuracy, with static and dynamic
production data involved.

#79: PREDICTION OF PRODUCTION POTENTIAL OF NEW WELLS
Challenge:
During the field development planning, reservoir engineers use data from the
existing wells. Based on the analysis of data from existing wells, they may decide
the placement of new wells. Specifically, this is a standard practice in the
development of unconventional oil and gas resources. The challenge is to predict the
production potential of that new proposed wells.
Solution:
One such approach used dimensionality reduction by applying principal component
analysis (PCA), followed by regression methods to predict the production potential of
new proposed wells. The analysis based on AI methods has also demonstrated the
capability of uncovering hidden patterns, which are not easily noticed in a high-
dimensional space. By identifying these patterns, it's been shown that the wells expected
to behave similarly can be identified. Further analysis can reveal common properties of
wells exhibiting similar behavior, the underlying reasons for poor performance in the
Wells not behaving satisfactorily, and measures to avoid similar poor performance in the
new wells.

#80: PREDICTION OF TWO-PHASE IPR IN VERTICAL WELLS
Challenge:
Well Inflow Performance Relationship (IPR) has many applications in both applied
and theoretical sciences, especially in petroleum production engineering. For
example, an accurate prediction of well IPR is crucial to determine the optimum
production scheme, design production equipment, and artificial lift systems.
Solution:
Employ GP and ANN techniques in predicting the two-phase IPR in vertical wells
considering flowing bottom hole pressure, average reservoir pressure, maximum oil
flow rate, bubble point pressure, oil formation volume factor, solution gas-oil ratio,
and gas viscosity as the inputs of the models and oil flow rate as the output.

#81: PREDICTION OF WAX DEPOSITION THICKNESS
Challenge:
Wax deposition is a significant problem in the oil and gas industry that can lead to
catastrophic situations such as oil production reduction, pipeline plugging, and
formation damage. Wax precipitation is a concern in crude oil and gas condensate
fields when the temperature falls below a specified temperature known as wax
appearance temperature.
Solution:
Employ ANN, CSA-LSSVM, RBFNN, or ANFIS techniques toward predicting the amount of
wax precipitation considering oil composition and SG, pressure, and temperature of the
system as input parameters. ANNs to predict the wax disappearance temperature
(WDT) considering molar mass and pressure as input variables. Utilize ANFIS and ANN
techniques to predict the wax deposition thickness as a function of Reynolds number, wax
content, oil and pipeline temperature, and deposition time. Employ RBF-ANN and
LSSVM-CSA techniques, respectively, to predict the wax deposition rate.

#82: PREVENTION OF JUNKED WELLS
Challenge:
Junked wells were mainly caused by maintenance activities that went wrong, leaving
running tools, wires, plugs, and other materials abandoned and often jammed in the
well. Failures often occurred in tubing or casing due to corrosion, erosion, thermal
stress resulted in the collapse of casing which could lead to flooding of wells.
Junked wells also frequently occurred when the wells were converted from
producers to injectors. Mobile rock formations such as salts and shales were some of
the likely causes, too.
Solution:
Data from well logs, tubing stress analysis and service providers profile can be fed
into AI algorithms to produce accurate models to predict the probability of
encountering junked wells before any decisions to convert the wells from producers
to injectors. This approach may help oil companies and service providers to better
arrive at decisions to prevent junked wells from taking place.

#83: PREVENTION OF WELL CLOGGING
Challenge:
Detecting when Wells is about to clog or predict Well events can lead to costly
production deferrals or safety concerns.
Solution:
AI models can be developed by combining physics-based modeling with specific
sensor data to recognize patterns that indicate the beginning of clog in the wells,
allowing engineers to intervene before any real risk occurs.

#84: PUMPS PREDICTIVE MAINTENANCE
Challenge:
Many of the pumps, including electric submersible pumps, pumps for injection wells,
hydraulic fracturing, and other Well treatment pumps, are equipped with a high
number of sensors measuring pressures, temperatures, vibrations, flow rates, etc.
Solution:
AI applications for various pumps to implement predictive maintenance and select
the optimal operation regimes concerning operational costs vs. production. There
are many examples when an entirely data-driven or a hybrid model containing
physics-driven and data-driven math helps optimize the regimes, prevent
unexpected failures, and save on maintenance-on-schedule.

#85: WELL TREATMENT RECOMMENDATION
Challenge:
The investments to the costly well treatment campaigns are always at high risk
because of two things. The first relates that physics-driven models for predicting the
Well treatment produce very rough estimates due to the lack of precise knowledge
of the near-wellbore formation's physical properties. The second relates to the
experts' bias in figuring out the final selection of the Well treatment procedures for
a particular set of wells.
Solution:
There is an excellent opportunity to reduce the investment risks by accumulating
data from already produced well treatment jobs and then predicting the efficiency
of hydraulic fracturing jobs and AI-based analysis of injectivity issues. Also, the
development of algorithms based on optimization math and programming will
enable full-scale recommending systems. The recommending systems will help select
the particular well treatment design for a specific Well and plan the Well treatment
campaigns.

#86: PREDICTING VALVE FAILURES IN RECIPROCATING

COMPRESSORS
Challenge:
The reciprocating compressor is one of the most widely used compressor
technologies in today's oil and gas industries. It can compress various gases and has
a wide range of applications, high compression efficiency, and stable working
pressure. However, due to its complicated structure and many vulnerable parts, it
will bring huge losses once a failure cannot be detected and eliminated in time.
Solution:
A modern Reciprocating compressor has installed sensors. They take periodic readings of the
compressors' physical properties, including motor winding temperatures, compressor vibrations, and
pressure and temperature for both suction and discharge at various compression stages. Utilize
shapelets methodology and multivariate time series classification algorithm to find patterns that
capturing differences between sensor data related to normal valve function versus failed valve
function, considering valve sensor time series data as an input parameter. Also, you can employ a
fault diagnosis model based on a one-dimensional convolutional neural network (1DCNN). This
method takes the differential pressure and differential temperature of each compressor stage as the
input of 1DCNN, using the characteristics of the CNN to extract the features and finally using
Softmax to classify the fault.

REFERENCES
1. Kumba Sennaar. 2019. Artificial Intelligence in Oil and Gas – Comparing the Applications of 5 Oil Giants. Emerj.com
2. Dylan Azulay. 2020. Information Extraction in Oil and Gas – Using AI to Find Oil. Emerj.com
3. Trent Jacobs. 2019. The Oil and Gas Chat Bots Are Coming. Journal of Petroleum Technology
4. Michael. 2021. How AI and Annotation are Changing the Oil and Gas Industry. keymakr.com
5. Jessamyn Sneed. 2017. Predicting ESP Lifespan With Machine Learning. Unconventional Resources Technology Conference, Austin, Texas, 24-26 July 2017. library.seg.org
6. Sadegh Karimpouli & Hadi Fattahi. 2016. Estimation of P- and S-wave impedances using Bayesian inversion and adaptive neuro-fuzzy inference system from a carbonate reservoir in Iran. springer.com
7. Andrei Erofeev, Denis Orlov, Alexey Ryzhov & Dmitry Koroteev. 2019. Prediction of Porosity and Permeability Alteration Based on Machine Learning Algorithms. springer.com
8. Xidong Wang, Shaochun Yang, Ya Wang, Yongfu Zhao & Baoquan Ma. 2018. Improved permeability prediction based on the feature engineering of petrophysics and fuzzy logic analysis in low porosity–
permeability reservoir. springer.com
9. Iqbal H. Sarker, Md Hasan Furhad & Raza Nowrozy. 2021. AI-Driven Cybersecurity: An Overview, Security Intelligence Modeling and Research Directions. springer.com
10. Yogendra Narayan PandeyAyush RastogiSribharath KainkaryamSrimoyee BhattacharyaLuigi Saputelli. 2020. Machine Learning in the Oil and Gas Industry. springer.com
11. Essenreiter R., Karrenbach M., Treitel S. 2003. Identification and Suppression of Multiple Reflections in Marine Seismic Data with Neural Networks. springer.com
12. Hong Li, Haiyang Yu, Nai Cao, He Tian, Shiqing Cheng. 2020. Applications of Artifcial Intelligence in Oil and Gas Development. springer.com
13. Marc Lesage; Peter Hall; John R.A. Pearson; Marc J. Thiercelin. 1991. Pore-Pressure and Fracture-Gradient Predictions. onepetro.org
14. Siong Jong Hang. 2020. Better decision making using machine learning in oil and gas exploration and production. towardsdatascience.com
15. Vegard Flovik. 2018. How do you teach physics to machine learning models?. towardsdatascience.com
16. Vinugayathri. AI and IoT Blended - What It Is and Why It Matters?. clariontech.com
17. Yattish Ramhorry. 2021. Blockchain and AI: The Next Revolution in Data Integrity. datadriveninvestor.com
18. James Lowell, Peter Szafian and Nicola Tessen; Geoteric. 2019. Artificial Intelligence and Seismic Interpretation. geoexpro.com
19. Jake Bouma, Cognite. 2020. Machine Learning in Oil and Gas Exploration. geoexpro.com
20. Fu-yan Guo,Yan-chao Zhang, Yue Wang, Ping Wang, Pei-jun Ren, Rui Guo, and Xin-Yi Wang. 2020. Tianjin Chengjian University, Tianjin 300384, China. hindawi.com
21. Neau, Audrey. 2009. Reservoir characterization based on seismic data with the help of geomodeling methods.
22. Bello, Opeyemi & Teodoriu, Catalin & Yaqoob, Tanveer & Oppelt, Joachim & Holzmann, Javier & Obinwanne, Alisigwe. 2016. Application of Artificial Intelligence Techniques in Drilling System Design and Operations:
A State of the Art Review and Future Research Pathways. 10.2118/184320-MS.
23. Li, Meng & Liu, Zhen & Liu, Minzhu & Zhang, Huilai. 2017. Prediction of residual oil saturation by using the ratio of amplitude attributes of time-lapse seismic data. GEOPHYSICS. 82. IM1-IM12. 10.1190/geo2015-
0453.1.
24. Santos Silva, Lucas David & Marinho, J. & Oliveira, Leonardo. 2017. PRESSURE GRADIENT PREDICTION FOR DIFFERENT FLOW PATTERNS OF HEAVY OIL AND NATURAL GAS IN A HORIZONTAL PIPE USING CFD
TECHNIQUES AND EMPIRICAL CORRELATIONS. Brazilian Journal of Petroleum and Gas. 11. 175-186. 10.5419/bjpg2017-0015.
25. Koteleva, Natalia & Buslaev, George & Valnev, Vladislav & Kunshin, Andrey. 2020. Augmented Reality System and Maintenance of Oil Pumps. International Journal of Engineering, Transactions B: Applications. 33.
1620-1628. 10.5829/ije.2020.33.08b.20.
26. Hamada, Ghareb & Nyein, Chaw & Elsakka, Ahmed. 2020. Prediction of Porosity and Water Saturation Using Neural Networks in Shaly Sand Reservoirs, Western Deseret, Egypt Article Info. Journal of Petroleum and
Mining Engineering. 10.21608/jpme.2020.36116.1040.
27. Abdolhossein Hemmati-Sarapardeh, Menad Nait Amar, Aydin Larestani, and Sassan Hajirezaie. 2020. Book. Applications of Artificial Intelligence Techniques in the Petroleum Industry. sciencedirect.com
28. "Nikita Klyuchnikov, Alexey Zaytsev, Arseniy Gruzdev, Georgiy Ovchinnikov, Ksenia Antipova, Leyla Ismailova, Ekaterina Muravleva, Evgeny Burnaev, Artyom Semenikhin, Alexey Cherepanov, Vitaliy Koryabkin, Igor
Simon, Alexey Tsurgan, Fedor Krasnov, Dmitry Koroteev. 2019.Data-driven model for the identification of the rock type at a drilling bit,Journal of Petroleum Science and Engineering."
29. Evgeny E. Baraboshkin, Leyla S. Ismailova, Denis M. Orlov, Elena A. Zhukovskaya, Georgy A. Kalmykov, Oleg V. Khotylev, Evgeny Yu. Baraboshkin, Dmitry A. Koroteev. 2020. Deep convolutions for in-depth
automated rock typing, Computers & Geosciences.
30. "B.T.H. Marbun, R.H. Ridwan, H.S. Nugraha, S.Z. Sinaga, B.A. Purbantanu. 2020. Casing setting depth and design of production well in water-dominated geothermal system with 330 °C reservoir temperature,Energy
Reports."
31. Dmitry Koroteev, Zeljko Tekic. 2021. Artificial intelligence in oil and gas upstream: Trends, challenges, and scenarios for the future, Energy and AI.
32. ScienceSoft. DEVELOPMENT OF DEFECT RECOGNITION SOFTWARE FOR AN OIL DRILLING EQUIPMENT MANUFACTURER. Oil & Gas Industry. https://www.scnsoft.com/case-studies/development-of-defect-
recognition-software-for-an-oil-drilling-equipment-manufacturer
33. Frederik de Hosson. 2020. Eight Reasons to Use Process Mining in Your Organization. https://www.uipath.com/blog/automation/eight-reasons-to-use-process-mining
34. Mustafa Kaouri, Manoj Bapat, Sai Vijay Krishna Josyula, Bilal Kureshi, Femida Merchant, and Ahmed Abdel Razek. 2021. Project Management Bot. PGP in AI for Leaders Feb 2021, Capstone Project at UTexas at
Austin.
35. Mustafa Kaouri and Abdullah Al-Sairafi, 2020, Feasibility study of Analysis for facilities planning optimization.
FOR MORE INFORMATION ABOUT AI IN THE UPSTREAM INDUSTRY,
PLEASE FOLLOW MUSTAFA KAOURI AT LINKEDIN @MUSTAFAKAOURI

Reservoir Engineering: SPE-194746-MS

Uploaded by

Copyright:

Available Formats

Reservoir Engineering: SPE-194746-MS

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Reservoir Engineering: SPE-194746-MS

Uploaded by

Copyright:

Available Formats

Artificial

SPE-194746-MS The End of Petroleum Engineering

People Development, Skill gap assessment

THE MATERIAL OR INFORMATION CONTAINED ON THIS DOCUMENT IS FOR GENERAL

MUSTAFA KAOURI MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND, EXPRESS OR

#32: DETECTION OF NATURAL OIL SEEPS IN OCEAN FLOOR

Activity: Reservoir Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 5

#33: FORECASTING OF OIL PRODUCTION

Activity: Reservoir Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 6

#34: OPTIMIZATION OF HISTORY MATCHING

Activity: Reservoir Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 7

#35: PREDICTION OF ACCURATE PVT PROPERTIES

Activity: Reservoir Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 8

#36: PREDICTION OF ASPHALTENE PRECIPITATION

Activity: Reservoir Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 9

#37: PREDICTION OF FORMATION DAMAGE

Activity: Reservoir Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 10

#38: PREDICTION OF INTERFACIAL TENSION (IFT)

Activity: Reservoir Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 11

#39: RECOGNIZING RESERVOIR MODEL THROUGH WELL TESTING

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 12

#40: SPEEDUP RESERVOIR MODELING

Activity: Reservoir Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 13

#41: SPEEDUP UPSCALING IN RESERVOIR MODELING

Activity: Reservoir Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 14

#42: PREDICTION OF MMP AND CO2 STREAMS

Activity: Enhanced Oil Recovery

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 15

#43: PREDICTION OF RECOVERY FACTOR (RF) AND OIL RATE

Activity: Enhanced Oil Recovery

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 16

#68: DETECTION OF FAULTS IN PIPELINE SYSTEMS

Activity: Production Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 17

#69: ESP FAILURES PREDICTION USING PCA APPROACH

Activity: Production Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 18

#70: OPTIMIZATION OF GAS-LIFT PROCESSES

Activity: Production Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 19

#71: OPTIMIZATION OF OILFIELD DEVELOPMENT PLAN

Activity: Production Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 20

#72: OPTIMIZING PRODUCTION OPERATIONS IN REAL-TIME

Activity: Production Engineering

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 21

#73: OPTIMIZING SELECTION OF OPTIMAL COMPLETION

Mustafa Kaouri – Email: mkaouri@gmail.com – LinkedIn: @mustafakaouri 22

#74: PREDICTION OF CASING COLLAPSE OCCURRENCE & DEPTH

Activity: Production Engineering