Pilot AI Drill - Intelligent Rate of Penetration (ROP) Prediction

📋 Project Overview

This project focuses on developing a hybrid machine learning model for predicting Rate of Penetration (ROP) in drilling operations, specifically for geothermal/energy wells. The solution combines mechanistic understanding of drilling physics with advanced machine learning techniques to improve drilling efficiency and reduce operational costs.

🎯 Business Problem

Drilling operations account for a significant portion of well construction costs and timelines. Inefficient drilling rates directly impact project economics through:

• Extended rig time
• Increased operational costs
• Delayed revenue generation

Our solution aims to optimize ROP by 15-25% compared to traditional methods, leading to substantial cost savings and improved operational efficiency.

🛠️ Technical Stack

• Core Libraries:

  • pandas - Data manipulation and analysis
  • numpy - Numerical computing
  • scikit-learn - Machine learning models and utilities
  • welly - Well log data handling and visualization
  • matplotlib & seaborn - Data visualization
  • jupyter - Interactive data exploration

• Development Tools:

  • black - Code formatting
  • pytest - Testing framework
  • pre-commit - Git hooks for code quality

📁 Project Structure

pilot-ai-drill/
│
├── data/                           # Data storage
│   ├── raw/                       # Raw data (immutable)
│   ├── processed/                 # Processed data
│   └── external/                  # External data sources
│
├── notebooks/                     # Jupyter notebooks
│   ├── 01_eda.ipynb              # Exploratory data analysis
│   ├── 02_feature_engineering.ipynb
│   └── 03_model_development.ipynb
│
├── src/                           # Source code
│   ├── data/                      # Data processing
│   ├── features/                  # Feature engineering
│   ├── models/                    # Model development
│   └── visualization/             # Visualization utilities
│
├── models/                        # Trained models
├── config/                        # Configuration files
├── tests/                         # Unit tests
└── docs/                          # Documentation

🚀 Getting Started

Prerequisites

• Python 3.8+
• pip (Python package manager)

Installation

1. Clone the repository:

   git clone https://github.com/luiscm17/pilot-ai-drill.git
   cd pilot-ai-drill

2. Create and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: .\venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

📊 Data

The project uses drilling data from Utah FORGE Well 16-16B(78)-32, including:

• Drilling mechanics (WOB, RPM, Torque)
• Formation characteristics (lithology, mineralogy)
• Environmental conditions (temperature, pressure)
• Gas measurements

🤖 Model Development

Our approach combines:

1. Mechanical Models: Physical understanding of drilling dynamics
2. Machine Learning: Data-driven pattern recognition
3. Transfer Learning: Knowledge application across different well conditions

📈 Performance Metrics

• Prediction Accuracy: MAE < 10 ft/hr
• Model Performance: R² > 0.75
• Error Tolerance: RMSE < 15 ft/hr

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

📬 Contact

For questions or feedback, please contact luiserwinc@gmail.com

🙏 Acknowledgments

• Utah FORGE for providing the drilling data
• Open-source community for the Python data science ecosystem