Evaluating essential features of proppant transport at engineering scales combining field measurements with machine learning algorithms

Abstract

The behaviours of the particle settlement, stratified flow and inception of settled particles are essential features that determine the proppant transport in low-viscosity fracturing fluids. Although great efforts have been made to characterize these features, limited research work is performed at field scales. To test the laboratory outcomes, we propose a machine-learning-based workflow to evaluate the essential features using the measurements obtained from shale gas fracturing wells. Over 430,000 groups of fracturing data (1 s time interval) are collected and pre-processed to extract the particle settlement, stratified flow and inception features during fracturing operations. The GRU and SVM algorithms, trained by these features, are applied to predict fracturing pressure. Error analysis (the root mean squared error, RMSE) is carried out to compare the contributions of different features to the pressure prediction, based on which the features and the corresponding calculations are evaluated. Our result shows that the stratified-flow feature (fracture-level) possesses better interpretations for the proppant transport, in which the Bi-power model helps to produce the best predictions. The settlement and inception features (particle-level) perform better in cases where the pressure fluctuates significantly. The features characterize the state of proppant transport, based on which the development of subsurface fracture is also analyzed. Moreover, our analyses of the remaining errors in the pressure-ascending cases suggest that (1) an introduction of the alternate-injection process, and (2) the improved calculation of proppant transport in highlyfilled fractures will be beneficial to both experimental observations and field applications.