Prediction of Complications and Accidents during Drilling with Application of Machine Learning Model

Mikail Seynaroev

Research output: ThesisMaster's Thesis

Abstract

One of the severe failure events during drilling is the sticking of the drill string. That results in time loss for freeing the pipe and the risk of losing an expensive portion of tubular and equipment. Therefore, there is huge interest in applying predictive systems to avoid stuck pipe occurrences. Drilling time reduction and, after that, its cost reduction can be achieved when accident signs are detected in advance. An intelligent system, performing automatic analysis of the wells’ electronic passports of the specific field, warns the drilling crew about possible stuck events during drilling. Drilling accidents lead to prolonged, costly downtime and high financial costs for their elimination and liquidation. Early forecasting and prevention of complications is an essential and urgent task requiring modern engineering methods and approaches, for instance, machine learning algorithms. The research target presents an application of machine learning techniques, such as artificial neural network and random decision forest for stuck pipe prediction.
Translated title of the contributionVorhersage von Komplikationen und Unfällen während des Bohrens mit Anwendung des maschinellen Lernmodells
Original languageEnglish
QualificationDipl.-Ing.
Awarding Institution
  • Montanuniversität
Supervisors/Advisors
  • Vita, Petr, Co-Supervisor (internal)
  • Thonhauser, Gerhard, Supervisor (internal)
Award date25 Jun 2021
Publication statusPublished - 2021

Bibliographical note

embargoed until null

Keywords

  • Stuck pipe
  • Machine learning

Cite this