Evaluating deep coal rock gas fracturing sweet spot intervals using PSO-ELM algorithm and petrophysical logging data

Area of Science:

• Petroleum Geoscience
• Artificial Intelligence in Energy
• Reservoir Engineering

Background:

• Deep coal gas reservoirs present challenges like heterogeneity and difficulty identifying fracturing sweet spots, hindering productivity.
• Optimizing fracturing operations is crucial for efficient extraction of deep coal rock gas.

Purpose of the Study:

• To develop and validate a machine learning model for predicting fracturing sweet spot intervals in deep coal gas reservoirs.
• To enhance productivity and operational efficiency in the Daning Jixian block by identifying high-quality resources.

Main Methods:

• An integrated geology-engineering approach was used, combining geological and production data.
• Ensemble optimized machine learning algorithms, specifically the Particle Swarm Optimization-Extreme Learning Machine (PSO-ELM), were employed.
• A graded evaluation model for fracturing sweet spot intervals was established using the PSO-ELM algorithm.

Main Results:

• The PSO-ELM model achieved over 85% prediction accuracy for fracturing sweet spot intervals in coal seams.
• Class I sweet spot intervals were predominant, indicating abundant high-quality exploitable resources.
• Spatial analysis revealed significant Class I and II sweet spots across the block.

Conclusions:

• The developed PSO-ELM model provides reliable support for perforation interval selection and fracturing stimulation optimization.
• This methodology effectively addresses challenges in deep coal reservoir development, improving resource extraction.
• The study confirms substantial exploitable high-quality resources in the Daning Jixian block through validated predictions.