Abstract

Electric Submersible Pumps (ESPs) are one of the important artificial lift methods for sustaining production in mature and high-water-cut wells; but may suffer frequent failures due to mechanical, electrical, hydraulic, chemical, and operational failures. These failures can yield substantial deferred production and intervention costs. Plenty of ESP installations are fitted with downhole sensors. Yet, it is observed that the current industry practice underutilizes the wealth of available sensor and operational data and lacks standardized, explainable failure-type identification and classification.

In this thesis, a comprehensive Machine Learning (ML) and Deep Learning (DL) framework was introduced for ESPs that simultaneously estimates remaining useful life (RUL) while predicting and classifying failure types. The RUL prediction module achieves high accuracy (R² > 0.96; MAE < 10 days) through supervised algorithms specifically designed for the ESPs' dataset dynamics. Concurrently, the failure classification module identifies ten distinct failure types with F1 scores exceeding 0.95, utilizing a suite of complementary algorithms. The datasets were cleaned and preprocessed to address missing values, inconsistencies, and other data quality issues. Various strategies were applied for splitting the data into training, validation, and test sets. Model tuning included parameter adjustment and techniques such as SMOTE-Tomek resampling to address class imbalance and to prevent data leakage. Additionally, a novel two-step Integrated Failure Modes and Root Cause (IFMRC) framework is introduced: it first classifies observable component-level failure modes, then attributes latent root causes—including design deficiencies, operational misalignment, and equipment degradation—thereby significantly improving diagnostic granularity and system interpretability.

Validation on a dataset of over 4,000 ESP installations from Egypt’s Western Desert demonstrates state-of-the-art predictive fidelity under time-aware, cluster-specific cross-validation. Three field case studies validate the model’s performance; multiclass failure prediction in one of the wells, where the model identified recurrent failures; RUL forecasting in another well achieving a $704,000 saving in predicting failure 12 days before it occurs; and IFMRC diagnostics in another specific well case enabled enhanced root cause identification. This work integrates statistical analysis, failure mechanics, and field operations to advance predictive modeling. The resulting framework establishes a scalable methodology—empirically validated through field applications—that extends equipment lifespan and reduces costly interventions in data-rich oilfield environments.

School

School of Sciences and Engineering

Department

Petroleum & Energy Engineering Department

Degree Name

MS in Petroleum Engineering

Graduation Date

Fall 9-8-2025

Submission Date

9-17-2025

First Advisor

Ahmed H. El-Banbi

Second Advisor

Gehad Hegazy

Committee Member 1

Mohamed Ghareeb

Committee Member 2

Abdelaziz Khlaifat

Committee Member 3

Moustafa Oraby

Extent

233p.

Document Type

Master's Thesis

Institutional Review Board (IRB) Approval

Approval has been obtained for this item

Disclosure of AI Use

No use of AI

Download

Share

COinS