Summary This paper provides the technical details of developing models to enable automated stage-wise analyses to be implemented within the real-time completion (RTC) analytics system. The models—two of which use machine learning (ML), including the convolutional neural network (CNN) technique (LeCun et al. 1990) and the U-Net architecture (Ronneberger et al. 2015)—detect the hydraulic fracture stage start and end, identify the ball seat operation, and categorize periods of pump rate. These tasks are performed on the basis of the two reliably available measurements of slurry rate and wellhead pressure, which enable the models to run automatically in real time, and also lay the foundation for further hydraulic fracturing advanced analyses. The presented solution provides real-time automated interpretations of hydraulic fracture events, enabling auto-generation of key performance indicator (KPI) reports, dispelling the need for manual labeling, and eliminating human bias and errors. It replaces the manual tasks in the RTC workflow/data pipeline and paves the way for a fully automated RTC system.

This paper provides the technical details of developing the automated stage-wise KPIs report generator, which is to be implemented as a module in the Real-Time Completion (RTC) analytics system. The generator is constructed with three models, two of which use Machine Learning (ML), that detect the stage start and end, identify the ball pumpdown and seat operation, and segments of a single stage of time series data into operationally similar sections. These tasks are performed based on the reliably available measurements of slurry rate and wellhead pressure, which enable the real-time automated stage-wise KPI analysis, and also lay the foundation for further advanced analysis regarding operational decision making. The stage start and end detection is treated as a classification task. The slurry rate and wellhead pressure along with their first and second order derivatives are extracted by a fixed-length sliding window and structured as matrices, which resemble the data structure of ML inputs. A Convolutional Neural Network (CNN) is trained for the classification, and each data point is classified as either within a stage's pumping time or otherwise as the data is received, with minimal latency. Data of eight wells with 648 total stages were labeled for the stage start and end detection model. The five-fold cross-validation technique was used to evaluate the performance of the model, and a 15-second-window was used to extract data from the time series data. The model achieved an accuracy of ∼99.7% with a tolerance of 25 seconds in all blind tests, meaning the predicted start or end of the stage was fewer than 25 seconds before or after the actual flag. After the start of stage classification is made, the ball pumpdown/seat recognition tasks take place. The ball pumpdown/seat event detection is a two-step strategy. The first step is to detect if there is a ball pumpdown/seat event in a stage, and the second step is to locate the end of that event. The first step is accomplished via a deep learning model with U-Net architecture, which detects the ball pumpdown and seat pattern within the slurry rate and pressure time series data. 179 samples are used to train the U-Net model. The transfer learning technology is used, as the dataset is small, and the U-Net is materialized with the pre-trained ResNet-34. The blind test's F1 score for the U-Net models is 0.97, which indicates excellent performance on the prediction. The second step can be achieved by a rule-based selection given the information from the first step. The third model analyzes a single stage and splits the stage into differently categorized segments. The model takes a three-step strategy. First, the stage data is smoothed by the sequential application of three different filters. The smoothed data is used in the second step, which detects points of interest and categorizes the segments in between. Segments are categorized by the slope of a linear fit or the mean first order derivative along the segment. Finally, a rule-based method is applied to agglomerate segments, which leads to a more interpretable categorization. The solution presented by this paper, provides real-time automated interpretations of hydraulic fracture events, enabling auto-generation of KPI reports, dispelling the need for manual labeling and eliminating human bias and errors. It fills the manual task gaps in the RTC workflow/data pipeline and paves the way for a fully automated RTC system.

This paper presents an analysis of the drag in extended- and mega-reach wells. An extended-reach well has been defined as any well with a measured depth to true vertical depth ratio (MD/TVD) greater than 2.0. A mega-reach well has a MD/TVD ratio greater than 3.0. The drag while running casing in the mega-reach well can be excessive. This paper investigates various applications of selective floatation devices to reduce the drag on the casing. First, the theoretical foundations of drag analysis for directional wells are reviewed. A methodology for analyzing buoyancy-assisted casing designs is presented. This methodology is used to analyze casing run fully filled with mud, partially filled with mud, and completely floated in mud. In addition, the friction forces during cementing operations are studied. While previous papers were presented on selectively floated casing designs for extended-reach wells, this study goes further to include an analysis of fully and selectively floated liners. In addition, a liner/tie-back combination is compared with the alternative long string casing design. The study concludes that completion of some high angle wells, previously thought unattainable due to high drag, can be achieved using buoyancy-assisted casing programs.

The paper provides a technical overview of an operator's Real-Time Drilling (RTD) ecosystem currently developed and deployed to all US Onshore and Deepwater Gulf of Mexico rigs. It also shares best practices with the industry through the journey of building the RTD solution: first designing and building the initial analytics system, then addressing significant challenges the system faces (these challenges should be common in drilling industry, especially for operators), next enhancing the system from lessons learned, and lastly, finalizing a fully integrated and functional ecosystem to provide a one-stop solution to end users. The RTD ecosystem consists of four subsystems as shown in architecture Figure 1. (I) The StreamBase RTD streaming system, which is the backbone of the ecosystem. It takes the real-time streaming log data as well as other contextual well data (for example, OpenWells), processes it through analytical models, generates results, and delivers them to the web-based user interface; (II) The analytics models, which include the Machine Learning (ML)/Deep Learning (DL) models, the physics-based models and the stream analytical/statistical models; (III) The digital transformation solution, which wasdesigned to address contextual well data digitization issues to enable real-time physics-based modeling. Contextual well data like bottom hole assemblies (BHAs) and casing programs are challenging to aggregate and deliver to models, as this data is often stored in locations across multiple systems and in various formats. The digital transformation applications are designed to fit into the drilling teams' workflows and collect this information during the course of normal engineering processes, enhancing both the engineering workflow and the data collection process; (IV) the cloud based ML pipeline, which streamlines the original ML workflows, as well as establishes an anomaly detection and re-training mechanism for ML models in production. Figure 1 RTD ecosystem architecture All of these subsystems are fully integrated and interact with each other to function as one system, providing a one-stop solution for real-time drilling optimization and monitoring. This RTD ecosystem has become a powerful decision support tool for the drilling operations team. While it was a significant effort, the long term operational and engineering benefits to operators designing such a real-time drilling analytics ecosystem far outweighs the cost and provides a solid foundation to continue pushing the historical limitations of drilling workflow and operational efficiency during this period of rapid digital transformation in the industry.

The automated real-time torque and drag (RT-T&D) analysis compares real-time measurements to evergreen models to monitor and manage downhole wellbore friction, improving drilling performance and safety. Enabling RT-T&D modeling with contextual well data, rig-state detection, and RT-interval event filters poses significant challenges. To address these challenges, this paper presents a solution that integrates a physics-based T&D stiff/soft string model with a real-time drilling (RTD) analytics system using a custom built ETL-Translator and digital transformation applications to automate the T&D modeling workflow. The overall RT-T&D solution consists of four parts (see Figure 1): the digital transformation apps/ETL-Translator, the T&D model API, pre-existing data infrastructure, and the RTD analytics system. The pre-existing data infrastructure and workflows have time lag (updated daily) and data gap issues which are not acceptable for the RTD analytics system. To overcome this obstacle, digital transformation applications (PWP Digitizer and DPAT) were designed to fit into the drilling team's workflow, enabling automated digitization of contextual well data during normal engineering processes. The PWP Digitizer is a data pipeline that fits within the existing Planned Well Path (PWP) approval workflow and digitizes the PWP automatically. The Drilling Program Automation Tool (DPAT) is an application that automates the drilling program preparation process and digitizes related well information (BHAs, casing program, etc.) automatically. The commercial T&D model is deployed as a REST API and serves any applications for T&D modeling via a Translator. The current use cases are the RTD analytics system inferencing the T&D REST API via the Translator for real-time analysis, as well as a desktop application for pre-job/post-job analysis. The automated Extract, Transform and Load (ETL) module was developed to interact with various databases, pulling and storing all data needed for the model. The Translator module is designed to communicate with the T&D model via a REST API, and act as a coordinator to link all elements together. The RT-T&D modeling workflow executes as follows: once a real-time directional survey arrives in the RTD system, the RTD system will send the necessary data to the Translator and trigger a T&D model calculation. The Translator maps all inputs for the commercial model using a lookup table, prepares a JSON payload for the T&D model API, and, finally, returns the status to the RTD system. Once the calculation is ready, the RTD system will request the results so that they can be pushed to the RTD user interface. This automated RT-T&D workflow is plan to be integrated into the RTD analytics system to serve Delaware Basin operations. Business value is derived from both reduced time and cost to generate and analyze modeled and actual torque and drag data as well as operational risk reduction during the drilling process. The RT-T&D workflow can accommodate any commercial or proprietary T&D model without impacting the overall architecture. Drilling engineers can leverage the T&D workflow online or offline to perform recalculations, comparative analyses, and friction calibrations, enabling optimized operations, pre-job planning, and post-job analysis. The digital transformation apps not only digitize and aggregate contextual well data for RTD system in an automated and sustainable way, but they also streamline the drilling team's workflow by reducing manual data aggregation tasks during the drilling program preparation process, thereby saving engineering hours. This paper proposes an automated and sustainable digital transformation solution to address a common well data digitization issue, which enables automated RT-T&D modeling. The presented solution architecture is not limited to RT-T&D modeling; it lays the foundation for any real-time physics-based modeling, including real-time bottom-hole assembly (BHA) and equivalent circulating density (ECD) modeling. After the proof of concept of the RT-T&D modeling workflows, more physics-based models will be integrated into the RTD analytics system for real-time analysis using the same architecture. End user endorsement is the key to success for any digital transformation solution. This paper shares the lessons learned in obtaining the end users’ buy-in for use of digital transformation apps: instead of imposing new workflows on end users, these apps are designed to fit within end users’ existing workflows, streamlining and optimizing the processes.