Integrated Optimization System Improves Hydraulic Fracturing Operations

Abstract

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 209127, “Efficiency and Effectiveness—A Fine Balance: An Integrated System To Improve Decisions in Real-Time Hydraulic Fracturing Operations,” by Somnath Mondal, SPE, Ashan Garusinghe, and Sebastian Ziman, Shell, et al. The paper has not been peer reviewed. _ In the complete paper, the authors demonstrate the need to balance optimizing fracture efficiency with effectiveness; present an integrated system for fracturing optimization using real-time, historical data along with organizational knowledge; and discuss the challenges and key considerations of setting up such a system, along with examples of large, untapped potential that can be unlocked with data science. Introduction Currently, most third-party fracture-monitoring solutions do not provide a true fracture-optimization platform that goes beyond fracture monitoring and efficiency or cost analytics. Without consideration for creating effective fracture geometries and good stimulation distribution, this may lead to poor resource recovery. Integrated fiber-optic diagnostics data have also shown that perforation-cluster screenout signatures sometimes can be observed in the surface-treatment data. Historical analysis of surface-treatment data has revealed that perforation-cluster screenout signatures are widely observed in other basins (Fig. 1). Cluster screenouts lead to severe nonuniformity in stimulation distribution. These examples demonstrate the need for including effectiveness considerations in any real-time fracture-optimization algorithm. The Completions Automation and Remote Technology (CART) Center was set up in 2017 by the operator to monitor and optimize all shale-fracturing operations remotely across multiple basins. The Center is staffed by consultants whose main role is to provide input directly to the engineer on site to optimize each treatment. Currently, this remote model relies on the experience of the consultant to know when and where to provide the right input. Data Preprocessing Modularity is a key component of the authors’ framework. The first step is to process new data, calculating important statistics used in historical analysis. Calculations to generate statistics used within the models also are performed. Models can be applied sequentially within this framework, posing no problems with regard to interdependency. As data are fed into the framework, automated analyses are performed to detect the start and end of a stage in real time. Instead of opting for a complex deep-learning algorithm, the authors implemented a simpler rule-based model to improve operational efficiency of the tool. At each new time stamp, the injection rate and pressure are evaluated to determine if fracturing operations have commenced; should the flow rate fall back to near zero, the framework will reset itself if this was simply a testing period of the stage or counts the stage as ended if the total sand pumped is more than the minimum required for the stage. Easy or Hard Stage A fracturing stage can be classified as “easy” or “hard” depending on how easily injectivity can be achieved and maintained. Engineers try to classify a stage as early as possible to decide how readily a stage will accept sand. The complete paper presents a work flow to assist engineers in the early classification of these stages.