Post-fracturing evaluation of fractures by interpreting the dynamic matching between proppant injection and fracture propagation

This study presents a new data-driven approach to interpreting fracture evolutions by analyzing the dynamic matching between proppant injection and fracture propagation. Proppant settling in low-viscosity fluids accumulates at the bottom of the fracture, compresses the flowing channel, and increases the net pressure in fractures due to increasing flow friction, which is the inherent correlation between proppant injection and fracture propagation. Evaluating the flow status of injected proppant in fractures may reveal how much fractures have been filled with proppant, and then estimate the fracture propagation indirectly - the fundamental logic of this study (using proppant as an indicator to detect the underground fractures). To accomplish this new conception, a machine-learning-based workflow is established by integrating an ensemble learning algorithm and a newly defined parameter - proppant filling index (PFI). Data from shale gas fracturing wells are collected to train the algorithm for the prediction of PFIs. The variations in PFI curves are then used to reveal the dynamic matching between proppant injection and fracture propagation, based on which the development of underground fracture volumes is estimated.