IntelligentEnergy

Innovative algorithms which use statistical, AI/ML, and forecasting methodologies to analyze oil-and-gas data. Note– Client data is not available on Github, and references to such data has beeen de-identified for security purposes

---

Table of Contents:

1. Dependencies
2. Notable File Structure
3. Design Decisions
4. Pipeline
5. Benchmarking and Backtesting
6. Final Implementation
7. Other

---

Part 1 - Dependencies

Dependency Chart for entire repository

---

Part 2 - Notable File Structure

File structure content here

---

Part 3 - Design Decisions

Important design decisions and their motivations

---

Part 4 - Pipeline

Overview

There are seven sequential components in the main pipeline of this repository.

1. Source Table Aggregation
2. Physics Feature Engineering
3. Anomaly Detection and Aggregation
4. Mathematical Feature Engineering
5. Model Creation and Performance
6. Optimization and Producer Level Steam Allocation
7. Injector Level Steam Allocation

Process Flow of Pipeline

Stage 1 - Source Table Aggregation

Joining the production, production testing, injection, and fiber data.

Stage 2 - Physics Feature Engineering

Generating new features based on domain-specific [modified] SAGD knowledge.

Stage 3 - Anomaly Detection and Aggregation

Using custom anomaly detection package to determine weights.

Stage 4 - Mathematical Feature Engineering

Generating new features - solely based on combinatorial explosions that are helpful towards model performance.

Stage 5 - Model Creation and Performance

Running the h2o experiment, assesssing model performance, and picking the best ones.

Stage 6 - Optimization and Producer Level Steam Allocation

Determinining optimization base table and using domain-specific contraints to tune high-level steam allocation.

Stage 7 - Injector Level Steam Allocation

The highest resolution of steam allocation.

---

Part 5 - Benchmarking and Backtesting

Using the top performing models and running incremental backtests to compare recommended and actual steam allocations.

---

Part 6 - Final Implementation

Based on model performance and general design of project, what's the best way to implement it.

---

Part 7 - Other

Miscellaneous.