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

Abstract

The Daning Jixian block is situated on the eastern periphery of the Ordos Basin. it faces multiple challenges including strong reservoir heterogeneity and unclear identification of fracturing sweet spot interval interval intervals, which restrict the enhance productivity and operational efficiency of deep coal rock gas. This study adopts an integrated geology-engineering approach and employs ensemble optimized machine learning algorithms to predict fracturing sweet spot interval intervals. The recent exploration and development experiences along with extensive geological and production data in the block are summarized, 10 key controlling factors for reservoir quality and engineering quality are selected and organized into a dataset. The PSO-ELM algorithm was programmed on the MATLAB platform, into which this dataset was imported for iterative prediction. This process established a graded evaluation model for fracturing sweet spot interval interval intervals based on the PSO-ELM algorithm. The model was applied to predict the fracturing sweet spot intervals in the No. 5 and No. 8 coal seams of Well X-11, the overall prediction accuracy exceeds 85%, with Class I sweet spot intervals being predominant. Spatial distribution analysis across the block revealed abundant Class I and II sweet spots, confirming substantial exploitable high-quality resources. These predictions were validated by subsequent perforation and fracturing operation data, demonstrating that this methodology provides reliable logging technology support for perforation interval selection and fracturing stimulation optimization in deep coal reservoir development.