A Data Driven Approach in Screenout Detection for Horizontal Wells

Abstract

Abstract Multi-stage hydraulic fracturing has gained popularity all over the world as more tight geologic formations are developed economically for hydrocarbon resources. However, due to the stages' operating complexity, different kinds of disruptions in fracturing operations may occur and even result in great economic loss. Screenout is one of the issues caused by the blockage of proppant inside the fractures. This paper presents a screenout classification system based on Gaussian Hidden Markov Models (GHMMs), trained on simulated data, that predicts screenouts and provides early warning by learning pre-screenout patterns in the simulated surface pressure signals. The simulated data are generated in a hydraulic fracturing software using a horizontal well with three fracturing stages landing in the Niobrara B shale, Denver-Julesburg (DJ) Basin. During the 270 simulations, various synthetic fracturing treatment data are forward modeled for both screenout and non-screenout scenarios. The classification system consists of two Gaussian Hidden Markov Models (screenout and non-screenout), each of which is fitted and optimized by its respective training set. Both Hidden Markov Models are assigned with two 1D Gaussian probability density functions to represent the distribution of their associated simulated surface pressure signals. During the classification process, once a new surface pressure sequence is observed, the maximum log likelihood is calculated under both fitted models and the model with a greater likelihood will be predicted as the class of this new observation. The classification system is validated with a hold-out testing data set from the simulations and the statistics of the performance is visualized in a confusion matrix. The results indicate the classification system achieves an overall classification accuracy of 81% and an accuracy of 86% for successfully predicting screenout events around 8.5 minutes prior to screenout occurring in the simulation. The described methodology is demonstrated to be a useful tool for early screenout detection and shows its promising feasibility of other fracturing time-series data analysis.