This book covers unsupervised learning, supervised learning, clustering approaches, feature engineering, explainable AI and multioutput regression models for subsurface engineering problems. Processing voluminous and complex data sets are the primary focus of the field of machine learning (ML). ML aims to develop data-driven methods and computational algorithms that can learn to identify complex and non-linear patterns to understand and predict the relationships between variables by analysing extensive data. Although ML models provide the final output for predictions, several steps need to be performed to achieve accurate predictions. These steps, data pre-processing, feature selection, feature engineering and outlier removal, are all contained in this book. New models are also developed using existing ML architecture and learning theories to improve the performance of traditional ML models and handle small and big data without manual adjustments. This research-oriented book will help subsurface engineers, geophysicists, and geoscientists become familiar with data science and ML advances relevant to subsurface engineering. Additionally, it demonstrates the use of data-driven approaches for salt identification, seismic interpretation, estimating enhanced oil recovery factor, predicting pore fluid types, petrophysical property prediction, estimating pressure drop in pipelines, bubble point pressure prediction, enhancing drilling mud loss, smart well completion and synthetic well log predictions. This book provides comprehensive research and explores the different applications of data science and machine learning in subsurface engineering.
Author(s): Daniel Asante Otchere
Publisher: CRC Press
Year: 2023
Language: English
Pages: 322
Cover
Title Page
Copyright Page
Dedication
Foreword
Preface
Table of Contents
1. Introduction
2. Enhancing Drilling Fluid Lost-circulation Prediction: Using Model Agnostic and Supervised Machine Learning
1. Introduction
2. Background of Machine Learning Regression Models
3. Data Collection and Description
4. Methodology
4.1 Data Analysis and Visualisation
4.2 Machine Learning Model Application
4.3 Explainable AI
4.3.1 Permutation Feature Importance
4.3.2 Shapley Values
5. Results and Discussion
5.1 Evaluation of Model Performance
5.2 Model Agnostic Results
5.3 Analysis of Features Using Model Agnostic Metrics
5.4 Analysis of Features Using Shapley Values Model Agnostic Metrics
5.5 Evaluation of Top Features
5.6 Model Optimisation
5.7 Sensitivity Analysis
6. Conclusions
Acknowledgement
Data Availability
References
3. Application of a Novel Stacked Ensemble Model in Predicting Total Porosity and Free Fluid Index via Wireline and NMR Logs
1. Introduction
2. Nuclear Magnetic Resonance
2.1 Concept and Application
2.2 Works Related to the Use of Machine Learning in NMR for Reservoir Characterisation
3. Methodology
3.1 Data Collection and Description
3.2 Data Analysis and Feature Engineering
3.3 Machine Learning Model Application
3.3.1 Building Deep Learning Models
3.3.2 Building a Hybrid Stacked Ensemble Model
3.4 Criteria for Model Evaluation
4. Results and Discussion
4.1 Evaluation of Models’ Performances
5. Conclusions
Acknowledgement
References
4. Compressional and Shear Sonic Log Determination: Using Data-Driven Machine Learning Techniques
1. Introduction
2. Literature Review
3. Background of Machine Learning Regression Models
3.1 Decision Tree Conceptual Overview
3.1.1 Attribute Selection Measures
3.2 Random Forest Conceptual Overview
3.3 Extremely Randomised Trees Conceptual Overview
4. Data Collection and Description
5. Methodology
5.1 Data Analysis and Visualisation
5.2 Machine Learning Model Application
6. Results and Discussion
6.1 Evaluation of Model Performance
6.2 Model Optimisation
6.3 Model Deployment
7. Conclusions
Acknowledgement
Data Availability
References
5. Data-Driven Virtual Flow Metering Systems
1. Introduction
2. VFM Key Characteristics
3. Data Driven VFM Main Application Areas
3.1 Virtual Sensing in ESP Wells
3.2 Virtual Sensing for SRP Wells
3.2.1 Virtual Flow Meter on Rod Pumping Systems
3.2.2 Virtual Sensing of the Dynamometer Card
3.3 Virtual Sensing for Gas Lifted Wells
3.4 Virtual Sensing for Gas Wells and Plunger Lifted Wells
3.5 Miscellaneous Applications for Identifying Flow Regimes
4. Methodology of Building Data-driven VFMs
4.1 Data Collection and Preprocessing
4.2 Model Development
5. Field Experience with a Data-driven VFM System
References
6. Data-driven and Machine Learning Approach in Estimating Multi-zonal ICV Water Injection Rates in a Smart Well Completion
1. Introduction
2. Brief Overview of Intelligent Well Completion
2.1 ICV Setting and Determination
2.2 Literature Review of ICV Innovations and Machine Learning Applications
3. Methodology
3.1 A Brief Overview of Models Used in This Study
3.2 Criteria for Model Evaluation
4. Results and Discussion
4.1 Explainable AI
4.2 Model Evaluation
4.3 Sensitivity Analysis
4.4 Model Deployment
5. Conclusions
Code Availability
Acknowledgement
References
7. Carbon Dioxide Low Salinity Water Alternating Gas (CO2 LSWAG) Oil Recovery Factor Prediction in Carbonate Reservoir: Using Supervised Machine Learning Models
1. Introduction
2. Methodology
2.1 Modeling of CO2-LSWAG
2.2 Geochemical Reactions of CO2-LSWAG
2.3 Machine Learning Methods
2.3.1 Multivariate Adaptive Regression Splines (MARS)
2.3.2 Group Method of Data Handling (GMDH)
2.3.3 Performance Metrics
2.3.4 Dataset Standardisation
3. Results and Discussion
3.1 Numerical Model Description
3.2 Input and Target Dataset
3.3 MARS Modeling
3.4 GMDH Modeling
3.5 Numerical Simulator and Machine Learning Computational Time
4. Conclusion
Acknowledgment
References
8. Improving Seismic Salt Mapping through Transfer Learning Using A Pre-trained Deep Convolutional Neural Network: A Case Study on Groningen Field
1. Introduction
2. Method
2.1 Collection and Description of Data
2.2 Deep Convolutional Neural Network in Salt Mapping and Post-processing
2.2.1 Simplified Architecture of Residual U-net
2.3 Transfer Learning Application
2.4 Criteria for Model Evaluation
3. Results and Discussion
3.1 Calculated Salt Body Volume
3.2 Semantic Segmentation – Transfer Learning Application
3.3 Sensitivity Analysis of Model and Expert Interpretations
4. Conclusions
Data and Software Availability
Acknowledgement
References
9. Super-Vertical-Resolution Reconstruction of Seismic Volume Using A Pre-trained Deep Convolutional Neural Network: A Case Study on Opunake Field
1. Introduction
2. Brief Overview
3. Methodology
3.1 Regional Geological Overview of the Opunake Field
3.2 Local Geological Overview of the Opunake Field
3.3 Deep Convolutional Neural Network in Seismic Image Resolution
3.3.1 Simplified Architecture of Residual U-net
3.4 Training and Testing Process
3.5 Criteria for Model Evaluation
4. Results and Discussion
4.1 Conditioned Seismic Volume
4.2 Model Evaluation
5. Conclusions
Data and Software Availability
Acknowledgement
References
10. Petroleum Reservoir Characterisation: A Review from Empirical to Computer-Based Applications
1. Introduction
2. Empirical Models for Petrophysical Property Prediction
2.1 Porosity and Permeability Prediction Models
2.2 Saturation Prediction Models
3. Fractal Analysis in Reservoir Characterisation
4. Application of Artificial Intelligence in Petrophysical Property Prediction
4.1 Artificial Neural Networks (ANNs)
4.1.1 ANN Application in Petrophysical Reservoir Prediction
4.2 Support Vector Machine (SVM)
4.2.1 Machine Learning (ML) Application in Petrophysical Reservoir Prediction
5. Lithology and Facies Analysis
5.1 AI Applications in Lithology and Facies Analysis
6. Seismic Guided Petrophysical Property Prediction
7. Hybrid Models of AI for Petrophysical Property Prediction
8. Summary
9. Challenges and Perspectives
9.1 AI Perspective
9.2 Rock Physics Perspective
10. Conclusions
References
11. Artificial Lift Design for Future Inflow and Outflow Performance for Jubilee Oilfield: Using Historical Production Data and Artificial Neural Network Models
1. Introduction
2. Methodology
2.1 Artificial Lift Screening Techniques
2.2 Inflow Performance Relationship Production Forecast
2.3 Outflow Performance Relationship Production Forecast
2.4 PROSPER Procedure for Well Model Set-Up
2.4.1 Deviation Survey Data Input
2.4.2 Surface Equipment Data Input
2.4.3 Downhole Equipment Data Input
2.4.4 Average Heat Capacities Data Input
2.5 Artificial Neural Networks
2.5.1 Back Propagation Neural Network
2.5.2 Radial Basis Function Neural Network
2.5.3 ANN Procedure
3. Results and Discussion
3.1 Production and Well Data of the Study Area
3.1.1 Base Case Flow Rates
3.2 Artificial Lift Screening
3.3 PROSPER Simulation Results
3.3.1 IPR Curves
3.3.2 Vertical Lift Performance Correlations
3.3.3 Desired Flow Rates
3.4 Gas Lift Results
3.4.1 Optimum Production Rates
3.5 ANN Results
3.5.1 ANN Architecture
3.5.2 Model Visualization
3.6 Discussion
4. Conclusions
Acknowledgment
References
12. Modelling Two-phase Flow Parameters Utilizing Machine-learning Methodology
1. Introduction
2. Data Sources and Existing Correlations
3. Methodology
4. Results and Discussions
4.1 Data Pre-processing
4.2 Model Development and Evaluation
5. Comparison between ML Algorithms and Existing Correlations
6. Conclusions and Recommendations
Nomenclature
References
Index
