Projects at the University of Texas at Austin
Below is a list of projects done at the University of Texas at Austin as part of my graduate degree in Petroleum Engineering.
Integrated Multi-Well Formation Evaluation using Machine Learning
Advanced Multi-Well Formation Evaluation, Fall 2023.
Integrated workflow for assisted multi-well petrophysical interpretation. Automated data processing and quality control. Dynamic time warping for zonation and unsupervised machine learning for mineralogy classification. Shaly sandstone analysis with nonlinear inversion for parallel- and perpendicular-to-bedding plane resistivity. Estimation of mineral volumetric concentration using linear programming. Rock classification and saturation height analysis using Leverett J-function and machine learning. [link]
Reservoir Inversion via Deep Learning
Inverse Problems in Subsurface Geophysics and Petroleum Engineering, Fall 2022.
Latent-space variational inversion for 2D geomodels from dynamic data with decoupled autoencoders. Development of a novel workflow for the estimation of realistic geologic models from multi-source dynamic well response data. Triple autoencoder structure for reduced-order representations. Transfer learning architecture for latent regression of dynamic to static data. Variational latent space for fake geomodel generation. Comparison of computational efficiency. [link]
Simulation and Monitoring of CO2 Storage via Machine Learning
Carbon Capture and Storage, Fall 2022.
Systematic literature review on recent machine learning-based techniques for the simulation and monitoring of geologic carbon storage. Comparison with classical numerical reservoir simulation and classical seismic monitoring. Focus on reduced-dimensionality models and data assimilation techniques. Proposal of a new method for spatiotemporal simulation and monitoring of CO2 storage.
Parallel Solution Scheme for Laplace's Equation using MPI
High-Performance Computing for Engineers, Spring 2022.
Numerical solution of partial differential equations using MPI and Gridap in Julia. Implemented for the Lorenz equations and Laplace's equation using finite difference and finite element methods. Visualization and sensitivity analysis for various combinations of parameters. Compared computational efficiency for serial versus parallel implementations.
CNN-RNN Forward Proxy Modeling for CO2 Monitoring
Machine Learning Applications in Geoscience, Spring 2022.
Generate 2D permeability realizations by partitioning the SPE10 model, and perform forward simulations using MRST for a small-scale CO2 injection project and collect time-lapse saturation forecasts as a 3D tensor. Import into Python and preprocess data by normalization, augmentation, and shuffling. Build a CNN-RNN encoder-decoder model using Keras as a proxy model for the forward simulations, exploiting latent space dynamics. Obtained a significant speedup for highly-accurate forward predictions. [link]
Gulf Coast Carbon Center Marine Seismic Data Processing
Seismic Data Processing, Spring 2022.
Developed an end-to-end workflow for seismic data processing, from a marine survey in the Gulf of Mexico by the Gulf Coast Carbon Center, Bureau of Economic Geology. Steps included: acquisition geometry analysis, wavelet deconvolution, surface-consistent amplitude correction, velocity analysis, multiple attenuation, and Stolt and phase-shift migration. Used Madagascar software for data processing and visualization.
Subsurface Advanced Dimensionality Reduction
Subsurface Machine Learning, Fall 2021.
Advanced dimensionality reduction using SVD and Dictionary Learning. Workflow in Python to compute a reduced-dimensional model of a generic (MNIST) dataset and a 2D subsurface (SPE10) dataset. Image compression, dimensionality reduction, and image reconstruction using tailored data-driven basis expansion. Demonstration of transfer learning from generic images to subsurface data. [link]
Projects at the University of Southern California
Below is a list of projects done at the University of Southern California as part of my graduate degree in Petroleum Engineering.
Multiscale Coupled Well Placement and Control Optimization
Optimization Methods for Subsurface Energy Resource Development, Spring 2021.
Sequential optimization of well placement and control schedule for 2D and 3D, Gaussian and fluvial geologic reservoir models using the SPE10 benchmark. Multiscale representations of the geologic models are obtained by flow-based upscaling, and optimization is performed in the low-dimensional coarse-scale model, and compared to the initial and optimal setup at the fine scale. Workflow in MRST and MATLAB optimization toolbox.
Multiphase Numerical Reservoir Simulation
Numerical Simulation of Subsurface Flow and Transport Processes, Spring 2021.
Developed entirely a MATLAB numerical reservoir simulator for a 1D, single-phase or two-phase black oil model. Included finite-difference discretization, gridding systems, boundary conditions, and banded matrix solvers. Performed Leak-Proof, Symmetry, and Accuracy tests, as well as compare different pressure and rate control schedules, and compared the numerical solutions for pressure distributions, cumulative production, water breakthrough, and NPV.
Comparative Regression Modeling for the Ames Housing dataset
Machine Learning, Spring 2021.
Implemented and compared the performance of several regression techniques for the Ames, Iowa housing dataset. Data preprocessing, statistical analysis, and in-depth visualization for feature engineering. Clustering using K-means, GMM. Dimensionality reduction using PCA, tSNE. Regression using linear models, LASSO, ridge, random forests, boosting, and artificial neural networks. Programming in Python, including parallelization and GPU computing. [link]
Probabilistic Estimation of Underground Reserves and Economic Output
Engineering and Economic Evaluation of Subsurface Reservoirs, Fall 2020.
Monte Carlo simulation to estimate OIIP and NPV in a subsurface gas reservoir. Probabilistic volumetric and economic modeling; application of the Law of Large Numbers and the Central Limit Theorem. Reference to Petroleum Resources Management System for P10, P50, P90 computation. Project analysis and recommendations under uncertainty.
Recent Approaches Towards Understanding of the Unconventional Reservoir Physics
Fluid Flow and Transport Processes in Porous Media, Fall 2020.
Analysis of the effect and physics of hydraulic fracture cluster spacing for an unconventional reservoir model. Review of unconventional reservoir physics in literature and development of relevant mathematical model. Numerical simulation using CMG GEM. Optimization of NPV based on cluster spacing.
Projects at the University of Tulsa
Below is a list of some selected projects done at the University of Tulsa as part of my undergraduate and graduate degrees in Petroleum Engineering and Applied Mathematics.
Optimization of Multivariate Gaussian Mixture Model
Numerical Optimization, Summer 2020.
Developed a MATLAB code to perform the Expectation-Maximization algorithm for clustering of random data using multivariate Gaussian Mixture Model. Used the Optimization Toolbox to improve estimates and obtain optimal predictors. Further study with the Classification Learner, Distribution Fitter, and Neural Network Clustering toolboxes. [link]
Predicting Production in Multivariate Unconventional Reservoirs
Statistical Learning, Spring 2020.
Natural gas production forecasting from an unconventional formation with multivariate predictors. Programming in R for Decision Trees, LASSO, Boosting, Random Forest, PCA. Feature selection and model comparison. [link]
Dynamic Financial Modeling & Risk Assessment
Financial Modeling, Summer 2019.
Dynamic financial statement modeling for a real-life case study using VBA and @Risk. Selected a major energy company, and analyzed historical data, identified and modeled trends, forecasted future performance, and prepared economic recommendations. Evaluated several key financial metrics, and performed an in-depth risk assessment using Monte Carlo simulation.
Predictive Analytics and Artificial Lift Solutions for Liquid Loading wells using Statistical Learning and Physical Modeling
Capstone Design, Spring 2019
Worked in collaboration with Bravo Natural Resources in developing an algorithm to predict liquid loading in gas wells in the STACK/SCOOP region. Used R/Shiny to deploy a real-time user-friendly GUI for visualization and prediction of production performance. Used ARIMA and modified decline curve analysis to develop an algorithm to predict liquid loading, and cross-validation with mechanistic physical modeling using Turner-Coleman.
Numerical Reservoir Simulation for 2-phase 2D Waterflooding
Reservoir Engineering II, Spring 2019
Simulation of a 2D waterflooding process in an anisotropic media using CMG and Eclipse to predict reservoir performance. Estimation of saturation, WOR, SOR, breakthrough time, and cumulative productin.
Text Mining & Text as Data: NLP Applications to the Oil & Gas Industry
Data Science, Spring 2019
Exploratory analysis on the evolution of data science in the oil and gas industry over time. Data mining the OnePetro database for keywords, and plotted popularity and publications over time. Implementation of natural language processing and text mining in R.
Formation Evaluation of Payne County, OK
Formation Evaluation, Fall 2018
Determination of pay zone thickness through well log analysis and well data reports. Construction of structure and isopach maps for formation of interest. Usage of Weatherford PreView viewer and IHS MarkIt database.
Casing Program Design and Selection Check & Automation
Well Construction and Completion, Fall 2018
VBA program to design casing program given any wellbore trajectory. Conduct von-Mises failure criteria for burst, collapse, and body stress checks, with several casign options to determine most economical and safe casing design.
MonteCarlo Methods in MPI, OpenMP, Pthreads
High Performance Computing, Fall 2018
Built multi-core cluster with Raspberry PI, programming using MPI, OpenMP, and Pthreads. Developed algorithms in C for MonteCarlo simulations for graph theory, numerical integration and estimation, histogram building, and sorting algorithms.
Reservoir Characterization and History Matching based on Static and Dynamic Data
Integrated Reservoir Modeling, Spring 2018
Estimate reservoir properties based on data from three actives wells and two observation wells. Implementation of Kriging and Bayesian methods in FORTRAN and MATLAB. Comparison with simulation results from SGeMS modeling.
Finite-Difference and Finite-Element Methods for IBVP
Numerical Methods for Initial and Boundary Value Problems, Spring 2018
Numerical solution of PDE's using Mathematica for elliptic, parabolic, hyperbolic, homogeneous and non-homogeneous problems. Stability analysis, Crank-Nicolson, method of characteristics. Applications to problem in science and engineering.
3D Directional Drilling Trajectory Survey and Drag & Torque DRC
Drilling Engineering II, Spring 2018
Development of VBA program to trace 3D wellbore trajectory using different methods. Drilling requirement check based on drag and torque calculations. 3D Visualization in Mathematica.
Petroleum Production Optimization with Nodal Analysis for Gas Lift Applications
Production Engineering I, Spring 2018
Design and optimization of production system, and selection of most economical components. Calculations of productivity index and operating points. Validation through nodal analysis in PipeSim and VBA.
Drilling Engineering Toolbox & Drilling Program Prognosis and AFE
Drilling Engineering I, Fall 2017
Development of a drilling engineering toolbox through VBA, including key equations, conversions, and rules-of-thumb. Used toolbox for the design and prognosis of drilling program, including engineering and cost analysis.
Implementation of Spiral Finned Tubes for Heat Exchanger System
Fluid Mechanics, Spring 2017
System analysis for the implementation of spiral finned tubes in a heat exchanger. Comparison and advantages of spiral finned tubes. Effects of tube angle in the pressure drop and system efficiency. VBA calculations for physical and fluid properties, and economic analysis.
Strength vs Weight Optimization for an Acryclic Chain Linkage
Mechanics of Materials, Spring 2017
Design of a 6" by 6" acrylic piece for chain linkage, maximizing strength and minimizing weight. Use SolidWorks for design and strength-strain calculations. Yield strength calculations with safety factors. Laser cutter for piece fabrication. Experimental testing and validation.
Refrigeration of a Boat using the Carnot Cycle
Thermodynamics, Fall 2016
Selection of an ideal refrigerant fluid for a specific boat model. Consideration of the Carnot cycle for physical system requirements. Comparison of fluid properties for final selection. Cost analysis, and recommendations.