Data-Driven Numerical Simulation and Optimization Using Machine Learning, and Artificial Neural Networks Methods for Drilling Dysfunction Identification and Automation

dc.contributor.advisorGildin, Eduardo
dc.contributor.advisorMedina-Cetina, Zenon
dc.contributor.committeeMemberNoynaert, Samuel F.
dc.contributor.committeeMemberSchubert, Jerome
dc.creatorZarate Losoya, Enrique
dc.date.accessioned2023-02-07T16:13:24Z
dc.date.available2023-02-07T16:13:24Z
dc.date.created2022-05
dc.date.issued2022-04-11
dc.date.submittedMay 2022
dc.date.updated2023-02-07T16:13:25Z
dc.description.abstractProviding the necessary energy supply to a growing world and market is essential to support human social development in an environmentally friendly. The energy industry is undergoing a digital transformation and rapidly adopting advanced technologies to improve safety and productivity and reduce carbon emissions. Energy companies are convinced that applying data-driven and physics-based technologies is the economical way forward. In drilling engineering, automating components of the drilling process has seen remarkable milestones with considerable efficiency gains. However, more elegant solutions are needed to plan, simulate, and optimize the drilling process for traditional and renewable energy generation. This work contributes to such efforts, specifically in autonomous drilling optimization, real-time drilling simulation, and data-driven methods by developing: 1) a physics-based and data-driven drilling optimization and control methodologies to aid drilling operators in performing more effective decisions and optimizing the Rate of Penetration (ROP) while reducing drilling dysfunctions. 2) developing an integrated real-time drilling simulator, 3) using data-driven methodologies to identify drilling inefficiencies and improve performance. Initially, a novel drilling control systems algorithm using machine learning methods to maximize the performance of manually controlled drilling while advising was investigated. This study employs feasible non-linear control theory and data analysis to assist in data pre-analysis and evaluation. Further emphasis was spent on developing algorithms based on formation identification and Mechanical Specific Energy (MSE), simulation, and validation. Initial drilling tests were performed in a lab-scale drilling rig with improved ROP and dysfunction identification algorithms to validate the simulated performance. Ultimately, the miniaturized drilling machine was fully automated and improved with several systems to improve performance and study the dynamic behavior while drilling by designing and implementing new control algorithms to mitigate dysfunctions and optimize the rate of penetration (ROP). Secondly, to overcome some of the current limitations faced by the industry and the need for the integration of drilling simulation models and software, in which cross-domain physics are uni-fied within a single tool through the proposition and publication of an initial common open-source framework for drilling simulation and modeling. An open-source framework and platform that spans across technical drilling disciplines surpass what any single academic or commercial orga-nization can achieve. Subsequently, a complementary filter for downhole orientation estimation was investigated and developed using numerical modeling simulation methods. In addition, the prospective drilling simulator components previously discussed were used to validate, visualize, and benchmark the performance of the dynamic models using prerecorded high-frequency down-hole data from horizontal wells. Lastly, machine-learning techniques were analyzed using open, and proprietary recorded well logs to identify, derive, and train supervised learning algorithms to quickly identify ongoing or incipient vibration and loading patterns that can damage drill bits and slow the drilling process. Followed by the analysis and implementation feasibility of using these trained models into a con-tained downhole tool for both geothermal and oil drilling operations was analyzed. As such, the primary objectives of this interdisciplinary work build from the milestones mentioned above; in-corporating data-driven, probabilistic, and numerical simulation methods for improved drilling dysfunction identification, automation, and optimization.
dc.format.mimetypeapplication/pdf
dc.identifier.urihttps://hdl.handle.net/1969.1/197249
dc.language.isoen
dc.subjectenergy
dc.subjectsimulation
dc.subjectcontrol
dc.subjectdrilling automation
dc.subjectdata-driven
dc.subjectnumerical simulation
dc.subjectdysfunction identification
dc.subjectoptimization
dc.subjectpetroleum
dc.subject
dc.titleData-Driven Numerical Simulation and Optimization Using Machine Learning, and Artificial Neural Networks Methods for Drilling Dysfunction Identification and Automation
dc.typeThesis
dc.type.materialtext
local.etdauthor.orcid0000-0001-7763-3349
thesis.degree.departmentMultidisciplinary Engineering
thesis.degree.disciplineInterdisciplinary Engineering
thesis.degree.grantorTexas A&M University
thesis.degree.levelDoctoral
thesis.degree.nameDoctor of Philosophy

Files

Original bundle
Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
ZARATELOSOYA-DISSERTATION-2022.pdf
Size:
17.29 MB
Format:
Adobe Portable Document Format
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
LICENSE.txt
Size:
6.32 KB
Format:
Plain Text
Description: