Drilling operations rely heavily on physical core samples, which are incredibly expensive and time-consuming to extract. Virtual coring using AI offers a revolutionary alternative by predicting geological properties directly from standard well log telemetry.

This Deep Dive explores the OMNI-CORE SCADA platform, an advanced geophysical machine learning system proudly developed by the Magnora team and engineered by Arya Azimi. By processing massive volumes of industry-standard LAS files, this pipeline forecasts subsurface rock characteristics with remarkable accuracy, serving as a high-speed virtual core generator.

Project Vision

The primary objective was to deploy a scalable AI solution capable of interpreting raw well log data in real-time. We aimed to deliver an intuitive interface where geoscientists can instantly generate predicted core metrics, bridging the gap between traditional geophysics and machine learning while drastically reducing operational costs.

Technical Architecture

Built to handle rigorous industrial workloads, the system is divided into several highly optimized engineering phases:

1 Well Log Telemetry Ingestion

LAS Data Processing (pipeline/01_master_pipeline.py): The pipeline dynamically ingests hundreds of raw .LAS (Log ASCII Standard) files. We applied rigorous exploratory data analysis and feature engineering, utilizing the virtual_core_scaler.pkl to standardize non-linear petrophysical readings.

2 Virtual Coring AI Engine

XGBoost Model (models/xgboost_virtual_core.json): At the heart of the system lies a hyper-tuned XGBoost regression engine. By learning the complex relationships within well log telemetry, the model accurately predicts physical core properties and is serialized into a lightweight JSON format for ultra-fast SCADA integration.

3 Geophysical Operations Dashboard

Interactive Interface (app/app.py): We wrapped the complex machine learning architecture into a highly responsive web application. Energy sector professionals can seamlessly upload new LAS logs to the dashboard, visualizing subsurface profiles and generating virtual core logs without writing a single line of code.

Key Takeaways & Tech Stack

OMNI-CORE SCADA exemplifies how artificial intelligence is transforming the energy sector. By automating petrophysical analysis, the Magnora team provides a blueprint for next-generation, data-driven subsurface exploration.

Languages & Frameworks: Python, Streamlit.

Machine Learning: XGBoost, Predictive Analytics, Virtual Coring.

Geophysical Processing: LAS File Analytics, Pandas, NumPy, Scikit-Learn.