M.Sc Thesis-Peyman Kor (2019)
Preface
Acknowledgment
Thesis Objectives
1
Introduction
1.1
Background and Motivation
1.2
Novelty of the Work
1.3
Outline of the Thesis
2
Development of Proxy Model (Less Rich Model) Using a Machine Learning
2.1
Introduction
2.1.1
Physics-Based Proxies
2.1.2
Data-Driven or Physical Based-Proxy?
2.2
Workflow
2.2.1
Geological Model and Heterogeneity
2.2.2
Generating Geological Realization using Gaussian Sequential simulation (SGS)
2.2.3
Measure of Connectives
2.2.4
Analytical Method
2.2.5
Features and Response
2.2.6
Machine Learning Algorithm
2.2.7
eXtreme Gradient Boosting Model
2.2.8
Model Building and Validation
3
Value of Data Analytics in Field Development Project (VOI Analysis)
3.1
Background
3.2
High-Resolution Probability Tree Method (HRPT) for VOI Analysis
3.3
Sensitivity Analysis of VOI to Prior, Likelihood and CAPEX
4
Robust Field Development Optimization Using the Proxy Model
4.1
Introduction
4.2
Brief Summary of the Proxy-Model
4.3
Robust Optimization of Well Placement and Water Injection Scheme
4.4
Optimization Process
4.5
Results of Optimization
5
Final Remarks on ML Application and Conclusions
5.1
Final Remarks on ML Application
5.2
Conclusions
References
6
Appendix
6.1
Sequential Guassian Simulation (R Code)
6.2
Calculation of Connectivities (FMM) for 5-Spot Pattern (Python Code)
6.3
XGBOOST Machine Learning Model (R Code)
6.4
Modyfing the Eclipse Data File + Running the Eclipse Simulator from R Shell (R Code)
6.5
Net Present Value Calculation - Processing the Eclipse Output Data (R Code)
6.6
Alghorithem of VOI Calculation in HRDP Method (R Code)
6.7
Alghorithem of Sensitivity Analysis of VOI to Mean of Prior (R Code)
6.8
Alghorithem of Sensitivity Analysis of VOI to Standard Deviation of Prior (R Code)
6.9
Optimization Alghorithem, Fittness Function: Machine Learning Model (R Code)
6.10
Optimization Alghorithem, Fittness Function: Eclipse Reservoir Simulator (R Code)
Published with bookdown
Decision-Driven Data Analytics for Well Placement Optimization in Field Development Scenario - Powered by Machine Learning
References