Artificial Lift Optimization

Machine learning project for optimizing artificial lift systems in oil wells using CatBoost and multi-objective optimization.

Tech Stack

Component	Technology
ML Framework	CatBoost - gradient boosting with categorical features
Multi-objective Optimization	pymoo - evolutionary algorithms
Data Processing	pandas, numpy, joblib
Web Server	FastAPI + uvicorn
Monitoring	prometheus-fastapi-instrumentator
Validation	pydantic v2
Visualization	matplotlib, seaborn

Key Libraries

• CatBoost - Gradient boosting with native categorical support
• pymoo - Multi-objective optimization framework
• FastAPI - Modern async web framework
• pandas / numpy - Data processing

Overview

This project provides ML-based optimization for three types of artificial lift systems:

• ESP (Electric Submersible Pump): Optimizes pump speed, motor current for maximum production.
• Rod Pump: Optimizes rod load, pump speed for maximum production.
• Gas Lift: Optimizes gas injection rate for maximum production.

Models

• Lift Optimizer (CatBoost): Predicts production rate and finds optimal parameters to maximize output using pymoo.
• Failure Predictor (RandomForestClassifier): Classifies equipment failure modes (normal, pump wear, gas lock, rod fatigue, motor overheat).

API Endpoints

Endpoint	Method	Description
/	GET	Web dashboard
/api/health	GET	System health check
/api/models	GET	Model metadata and metrics
/api/optimize	POST	Optimize lift parameters
/api/failure	POST	Predict equipment failure mode

Quick Start

pip install -r requirements.txt
python train.py
python app.py

The API runs on port 5007.

Example API Calls

Optimize

curl -X POST http://localhost:5007/api/optimize -H "Content-Type: application/json" -d '{"lift_type":"ESP","pump_speed_rpm":2500,"well_depth_ft":6000,"water_cut_pct":50,"downhole_pressure_psi":2000,"motor_current_amp":40}'

Predict Failure

curl -X POST http://localhost:5007/api/failure -H "Content-Type: application/json" -d '{"lift_type":"ESP","pump_speed_rpm":3200,"well_depth_ft":8000,"water_cut_pct":80,"downhole_pressure_psi":3000,"motor_current_amp":75}'

Running Tests

python test_api.py

Project Structure

artificial-lift-optimization/
├── artificial_lift_optimization/
│   ├── __init__.py
│   ├── data_generator.py
│   ├── models/
│   │   ├── __init__.py
│   │   ├── lift_optimizer.py
│   │   └── failure_predictor.py
│   └── utils/
│       ├── __init__.py
│       └── preprocessor.py
├── outputs/
│   └── models/
├── templates/
│   └── index.html
├── .github/
│   └── workflows/
│       └── ci.yml
├── train.py
├── app.py
├── test_api.py
├── requirements.txt
├── setup.py
├── .gitignore
└── README.md

---

Elaborado por Ing. Kelvin Cabrera