Managing offshore oil production requires precise telemetry monitoring to maximize hydrocarbon recovery and prevent equipment failure. The Equinor Volve field provides a unique, data-rich environment to deploy next-generation predictive algorithms.

This Deep Dive explores the OMNI-CORE SCADA (Volve Field Edition), an enterprise-grade AI production platform proudly developed by the Magnora team and engineered by Arya Azimi. By integrating Virtual Flow Metering (VFM), deep learning time-series forecasting, and anomaly detection, this system transforms raw well telemetry into actionable reservoir intelligence.

Project Vision

The goal was to build a unified machine learning architecture capable of optimizing the entire production lifecycle. We aimed to predict future production rates, optimize waterflood injection, and replace expensive physical flow meters with highly accurate AI-driven virtual alternatives tailored for brownfield management.

Technical Architecture

Designed for the extreme demands of offshore energy production, the pipeline is divided into specialized intelligent modules:

1 Volve Production Telemetry

Data Ingestion (01_data_inspector.py): The system processes thousands of daily records from the Volve production data.xlsx dataset. Extensive cleaning and scaling are applied to normalize pressures, temperatures, and choke valve metrics across multiple active wells.

2 Deep Learning Forecaster & VFM

LSTM & XGBoost Models: An advanced Long Short-Term Memory (LSTM) neural network (lstm_forecaster.pth) forecasts temporal production trends. Simultaneously, a highly tuned regressor (vfm_regressor.json) acts as a Virtual Flow Meter to estimate multiphase flow rates instantaneously without hardware sensors.

3 Waterflood Optimization & Security

Anomaly Engine (04_anomaly_engine.py): We deployed an anomaly detection model (anomaly_detector.pkl) to monitor well health and detect critical sensor drift. Additionally, the waterflood_optimizer.json algorithm intelligently recommends water injection rates to maintain reservoir pressure and sweep efficiency.

4 Real-Time SCADA API

FastAPI Backend (app_fastapi.py): All predictive components are seamlessly integrated into a high-performance FastAPI service. This RESTful architecture enables instant, asynchronous communication between the AI engine and physical drilling control systems.

Key Takeaways & Tech Stack

OMNI-CORE SCADA demonstrates the immense value of artificial intelligence in energy optimization. By leveraging PyTorch and XGBoost, the Magnora team delivers a robust framework for maximizing offshore energy yields safely and efficiently.

Languages & Frameworks: Python, FastAPI.

Machine Learning: PyTorch (LSTM), XGBoost, Virtual Flow Metering, Anomaly Detection.

Data Engineering: Pandas, NumPy, Scikit-Learn, SCADA Analytics.