Artificial Intelligence and Data Analytics for Energy Exploration and Production

Fred Aminzadeh is an expert in artificial intelligence and energy. He was professor at the University of Houston and University of Southern California. He worked at dGB, Unocal (now part of Chevron) and Bell Laboratories. His work experience includes fossil energy, geothermal energy, and carbon sequestration. He served as the president of Society of Exploration Geophysicists. He has authored over 15 books and holds several patents. He was the editor in chief of The Journal of Sustainable Energy Engineering. Currently, he is president of FACT, an energy services company. He is also a member of technical advisory board of DOE/NETL’s SMART initiative and an adjunct Professor at the University of Wyoming. Cenk Temizel is a Sr. Reservoir Engineer with Saudi Aramco. He has over 15 years of experience in reservoir simulation, data analytics, smart fields, unconventional, enhanced oil recovery with Aera Energy, Schlumberger, and Halliburton in the Middle East, the U.S., and the U.K. He is the recipient of the Aramco Unconventional Resources Technical Contribution Award (2020), 2nd place at SPE Global R&D Competition at ATCE 2014, and the Halliburton Applause Award in Innovation (2012). He holds an MS degree in Petroleum Engineering from University of Southern California and was a research assistant at Stanford University before joining the industry. Yasin Hajizadeh is the founder and CEO of nowos, a boutique Texas based technology consulting firm with a focus on software product management and workforce development planning for industry 4.0. Previously, he was a program and product manager of Azure ML and IoT at Microsoft. Yasin has also worked for Schlumberger as a data scientist and reservoir engineer, University of Calgary as an associate professor of computer science, and CMG as an optimization and uncertainty quantification scientist. He holds a PhD in petroleum engineering from Heriot-Watt University, and a Masters in technology management from Memorial University of Newfoundland.

Foreword xvii

Preface xix

1 Introduction to Modern Intelligent Data Analysis 1

1.1 Introduction 1

1.2 Introduction to Machine Learning 4

1.3 General Example of Machine Learning 8

1.4 E&P Examples of Machine Learning 9

1.5 Objectives of the Book 10

1.6 Outline of Chapters 10

2 Machine Learning and Human Computer Interface 23

2.1 Introduction 23

2.2 Visualization of Machine Learning 24

2.3 Interactive Machine Learning 30

3 Artificial Neural Networks 39

3.1 Introduction 39

3.2 Structure of Biological Neurons 41

3.2.1 Artificial Neurons Structure 42

3.2.2 Integration Function 43

3.2.3 Activation Function 44

3.2.4 Decision Boundaries 46

3.3 Learning and Deep Learning Process for ANN 47

3.3.1 ANN Learning 47

3.3.2 Deep Learning 50

3.4 Different Structures of ANNs 51

3.4.1 Multi-Layer Perceptron (MLP) 53

3.4.2 Radial Basis Function Neural Networks (RBF) 54

3.4.3 Modular Neural Networks (Committee Machines) 55

3.4.4 Self-Organizing Networks 58

3.4.5 Kohonen Networks 61

3.4.6 Generalized Regression (GRNN) and Probabilistic (pnn) 62

3.4.7 Convolutional Neural Network (CNN) 64

3.4.8 Generative Adversarial Network (GAN) 65

3.4.9 Recurrent Neural Network (RNN) 66

3.4.10 Long/Short-Term Memory (LSTM) 67

3.5 Pre-Processing of the ANN Input Data 67

3.5.1 Dimensionality Reduction 69

3.5.2 Artificial Neural Networks (ANN) Versus Conventional Computing Tools (CCT) 70

3.6 Combining ANN with Human Intelligence 70

3.7 ANN Applications to the Exploration and Production (E&P) Problems 73

3.7.1 First Break Picking Seismic Arrivals 74

3.7.2 Porosity Prediction in a CO2 Injection Project 76

3.7.3 CNN for Permeability Prediction 78

3.7.4 Creating Pseudologs 81

3.7.5 Facies Classification with Exhustive PNN 81

3.7.6 Machine Learning for Estimating the Stimulated Reservoir Volume (SRV) 83

4 Fuzzy Logic 85

4.1 Introduction to Fuzzy Logic 85

4.2 Theoretical Foundation and Formal Treatment of Fuzzy Logic 90

4.2.1 Some Definitions in Fuzzy Logic 93

4.2.2 Fuzzy Propositions 94

4.2.3 Thresholding or α-Cut Concept 95

4.2.4 Additional Properties of Fuzzy Logic 96

4.2.5 Fuzzy Extensions of Classical Mathematics 98

4.2.5.1 Fuzzy Averaging 98

4.2.5.2 Fuzzy Arithmetic 99

4.2.5.3 Fuzzy Function and Fuzzy Patches 100

4.2.5.4 Fuzzy K-Means and C-Means or Clustering 103

4.2.5.5 Fuzzy Kriging 105

4.2.5.6 Fuzzy Differential Equations 108

4.2.6 Fuzzy Systems, Fuzzy Rules 109

4.2.6.1 Fuzzy Rules 110

4.2.6.2 Fuzzy Knowledge-Based Systems 112

4.2.7 Type-2 Fuzzy Sets and Systems 114

4.2.8 Computing with Words and Linguistic Variable 116

4.2.8.1 CWW versus Fuzzy Logic 116

4.2.8.2 Linguistic Variables 118

4.2.9 Mining Fuzzy Rules from Examples 120

4.2.10 Fuzzy Logic Software 121

4.3 Oil and Gas Industry Application Domain Discussion 122

4.3.1 Linguistic Goal-Oriented Decision Making (LGODM) to Optimize Enhanced Oil Recovery in the Steam Injection Process 123

4.3.2 Use of Fuzzy Clustering in Perforation Design 124

4.3.3 Stratigraphic Interpretation Using Fuzzy Rules 127

4.3.4 Fuzzy Logic-Based Interpolation to Improve Seismic Resolution 132

4.4 Conclusions 135

5 Integration of Conventional and Unconventional Methods 137

5.1 Strengths and Weaknesses of Different Computing Techniques 137

5.2 Why Integrate Different Methods? 140

5.2.1 Neuro-Fuzzy Methods 141

5.2.1.1 Why Combine NN and FL? 142

5.2.1.2 NN-Based FL Inference 143

5.2.2 Neuro-Genetic Methods 145

5.2.3 Fuzzy-Genetic (FG) 147

5.2.4 Soft Computing - Conventional (SC) Methods 148

5.3 Oil and Gas Applications of NF, NG, FG, CF, and CN 150

5.3.1 NN-CM- Rock Permeability Forecast Using Machine Learning and Monte Carlo Committee Machines 151

5.3.2 (NN-CM) Pseudo Density Log Generation Using Artificial Neural Network 154

5.3.2.1 Well Log Data Preprocessing 155

5.3.2.2 Well Log Data Mining 156

5.3.2.3 Data Postprocessing for Generating Pseudo Density Logs 157

5.3.3 NN-FL- Integrating Neural Networks and Fuzzy Logic for Improved Reservoir Property Prediction and Prospect Ranking 159

5.3.4 (FL-NN-CM) Gas Leak Detection 161

5.3.5 GA-FL for Improving Oil Recovery Factor 162

5.3.6 GA-FL to Improve Coal Mining Process 165

5.4 Conclusions 166

6 Natural Language Processing 167

6.1 Introduction 167

6.2 A Brief History of NLP 168

6.3 Basics of the NLP Method 171

6.3.1 Sentence Segmentation 171

6.3.2 Tokenization 172

6.3.3 Parts of Speech Prediction 172

6.3.4 Lemmatization 173

6.3.5 Stop Words Removal 173

6.3.6 Dependency Parsing 174

6.3.7 Named Entity Recognition 175

6.3.8 Coreference Resolution 175

6.4 Use Cases of NLP 175

6.5 Applications of NLP in the Oil and Gas Industry 177

6.6 Conclusion 193

7 Data Science and Big Data Analytics 195

7.1 Introduction 195

7.2 Big Data 195

7.3 Algorithms and Models in Data Sciences 197

7.3.1 Automated Machine Learning 198

7.3.2 Interpretable, Explainable, and Privacy-Preserving Machine Learning 198

7.4 Infrastructure and Tooling for Data Science 202

7.5 Oil and Gas Focused Issues Associated with Data Science and Big Data High Performance Computing in the Age of Big Data 206

7.5.1 Big Data in Oil and Gas 208

7.5.2 High-Performance Computing for Handling Big Data in Subsurface Imaging 209

7.5.3 Access to Oil and Gas Data 210

8 Applications of Machine Learning in Exploration 213

8.1 Introduction 213

8.1.1 Petroleum System and Exploration Risk Factors 214

8.1.2 Data Acquisition, Processing, and Integration for Exploration 215

8.1.3 Exploration and Appraisal Drilling 217

8.2 AI for Exploration Risk Assessment 218

8.2.1 Petroleum System Risk Assessment 218

8.2.2 Geological Risk Assessment Level of Knowledge and Experience (LoK) 221

8.3 AI for Data Acquisition, Processing, and Integration in Exploration 224

8.3.1 Auto-Picking for Micro-Seismic Data 224

8.3.2 Facies Classification Using Supervised CNN and Semi-Supervised GAN 226

8.3.3 Generating Gas Chimney Cube Using MLP ANN 227

8.3.4 Reservoir Geostatistical Estimation of Imprecise Information Using Fuzzy Kriging Approach 230

8.3.5 Fracture Zone Identification Using Seismic, Micro-Seismic and Well Log Data 232

9 Applications in Oil and Gas Drilling 239

9.1 Real-Time Measurements in Drilling Automation 239

9.2 Event Detection in Drilling 243

9.3 Rate of Penetration Estimations 251

9.4 Estimation of the Bottom Hole and Formation Temperature by Drilling Data 255

9.5 Drilling Dysfunctions 258

9.6 Machine Learning Applications in Well Drilling Operations 262

9.7 Conclusion 269

10 Applications in Reservoir Characterization and Field Development Optimization 271

10.1 Introduction 271

10.1.1 Reservoir Characterization 273

10.1.1.1 Porous Media Characterization 275

10.1.1.2 Porosity 278

10.1.1.3 Permeability 278

10.1.1.4 Permeability-Porosity Relationship 281

10.1.2 Machine Learning Applications for Reservoir Characterization 282

10.1.2.1 Reservoir Modeling 291

10.1.2.2 Capabilities of Data Mining 293

10.1.2.3 Computational Intelligence in Petroleum Application 294

10.1.2.4 Computational Intelligence in Permeability and Porosity Prediction 295

10.1.2.5 Hybrid Computational Intelligence (HCI) 296

10.1.2.6 Ensemble Machine Learning for Reservoir Characterization 297

10.1.2.7 Prediction of Sand Fraction (SF) by Using Machine Learning 300

10.1.2.8 Machine Learning Application in Classification of Water Saturation 301

10.1.2.9 Physics-Informed Machine Learning for Real-Time Reservoir Management 302

10.1.2.10 Well-Log and Seismic Data Integration for Reservoir Characterization 303

10.1.2.11 Machine Learning for Homogeneous Reservoir Characterization 304

10.1.2.12 The Gradient Boosting Method for Reservoir Characterization 305

10.1.2.13 The Parameterizing Uncertainty for Reservoir Characterization 306

10.1.2.14 Geochemistry and Chemostratigraphy for Reservoir Characterization 307

10.2 Conclusions 310

11 Machine Learning Applications in Production Forecasting 313

11.1 Introduction 313

11.2 Analytical Solution 315

11.2.1 Type Curves 316

11.2.2 Limitations 317

11.3 Numerical Solution 317

11.3.1 Limitations 318

11.3.2 Machine Learning Applications 319

11.4 Decline Curve Analysis (DCA) 320

11.4.1 Arps Method 320

11.4.2 Method Modifications of the Arps Method 321

11.4.3 Limitations 326

11.4.4 Machine Learning Applications 327

11.5 Data-Driven Solutions 330

11.5.1 Sensitivity Analysis 331

11.5.2 Machine Learning Applications 331

11.5.3 Limitations 349

11.6 Conclusion 350

12 Applications in Production Optimization, Well Completion and Stimulation 353

12.1 Introduction 353

12.2 Production Optimization 354

12.3 Stimulation 358

12.4 Well Completion 363

13 Machine Learning Applications in Reservoir Engineering and Reservoir Simulation 369

13.1 Introduction 369

13.2 Fluid Properties Estimation with Machine Learning Methods 370

13.2.1 Machine Learning Applications in Reservoir Simulation 376

13.2.2 Machine Learning Applications in Geothermal Reservoir Engineering 390

13.3 Machine Learning Applications in Well Testing 397

13.4 Conclusion 403

14 Machine Learning Applications in Artificial Lift 405

14.1 Introduction 405

14.2 Big Data and Analytical Solutions in Drilling Operations 407

14.3 Machine Learning 408

14.3.1 Using Machine Learning in the Oil and Gas Industry 410

14.3.2 Failure Prediction Frameworks and Algorithms for Artificial Lift Systems 413

14.4 Artificial Lift 415

14.4.1 Brief Overview of Production Systems Analysis 416

14.4.2 Types of the Artificial Lift Systems 418

14.4.2.1 Plunger Lift 418

14.4.2.2 Gas Lift-Continuous and Intermittent 419

14.4.2.3 Pumps 421

14.4.3 Artificial Lift Applications, Monitoring, and Automation Services 424

14.5 Conclusion 429

15 Machine Learning Applications in Enhanced Oil Recovery (EOR) 431

15.1 Introduction 431

15.2 Enhanced Oil Recovery 432

15.2.1 Thermal Methods 437

15.2.2 Chemical Methods 439

15.2.3 Gas Methods 441

15.2.4 Microbial Methods 444

15.3 Enhanced Oil Recovery (EOR) Reservoirs 445

15.4 The Economic Value of EOR 453

15.5 Simulation Models 454

15.6 Machine Learning (ML) 455

15.7 Machine Learning in Enhanced Oil Recovery (EOR) Applications 458

15.8 Machine Learning in Enhanced Oil Recovery (EOR) Screening 462

15.9 Applications 465

15.10 Software 467

15.11 Conclusion 468

16 Conclusions and Future Directions 471

16.1 Technology Advances in Artificial Intelligence and Data Science 471

16.1.1 Technology Advances in Artificial Intelligence and Data Science 471

16.1.2 Future Directions of Machine Learning and Human-Computer Interface 471

16.1.3 Future Directions of Artificial Neural Networks 474

16.1.4 Future Directions of Fuzzy Logic 476

16.1.5 Future Directions of Integrated AI Techniques 477

16.1.6 Future Directions of Natural Language Processing 479

16.1.7 Future Directions of Data Science and Big Data Analytics 480

16.2 Future Trends in the Energy Applications of Artificial Intelligence and Data Science 482

16.2.1 Future Trends in Exploration Applications of AI-DA 483

16.2.2 Future Trends in Drilling Applications of AI-DA 484

16.2.3 Future Trends in Reservoir Characterization Applications of AI-DA 486

16.2.4 Future Trends in Production Forecasting Applications of AI-DA 487

16.2.5 Future Trends in Production Optimization, Well Completion and Stimulation Applications of AI-DA 488

16.2.6 Future Trends in Reservoir Engineering and Simulation Applications of AI-DA 489

16.2.7 Future Trends in Artificial Lift Applications of AI-DA 491

16.2.8 Future Work for Machine Learning Applications in EOR 492

References 495

Index 555