Real-Time Fracturing Sleeve Monitoring Nearly as Accurate as Human Experts
Real-time detection of fracturing sleeve open and closed states for an oilfield services company
About the customer
Our client is a leading US oilfield services company working to make its fracturing operations more efficient. A key task is knowing whether each fracturing sleeve, a downhole valve, is open or closed, something subject matter experts (SMEs) had to work out by analyzing data by hand. The goal was to detect these states automatically in real time, reducing the workload on SMEs and simplifying the fracturing process.
What needed solving
Manual expert review
SMEs manually analyzed data to determine whether sleeves were open or closed
Slow and error-prone
Manual analysis was time-consuming and prone to human error, affecting operational efficiency
Accurate automation
The client needed an automated system to predict sleeve states accurately in real time, reducing its dependency on SMEs
How we built it
Live data connection
Connected the existing system to the cloud through Azure IoT Hub, so data streams in real time during fracturing operations.
Data collection and analysis
Gathered data from fracturing jobs across different sites and analyzed it to understand the available features and their scales.
Expert-labeled training data
Worked with SMEs to label specific fracturing jobs, creating a dataset for the model to learn from.
Model selection
Tested multiple machine learning pipelines, models and settings. Random Forest and LightGBM performed best.
Live predictions
Azure Durable Functions listen to the IoT Hub data stream and predict each sleeve’s state during fracturing jobs.
Nearly as accurate as human experts, with less dependence on them
Technologies
Domains
Have a Query? Let's Have a Conversation
Do you have a project in mind? We’d love to know what it is. Email us or send a message from our contact page.
Have a similar problem?
Tell us what you are trying to solve and our team will get back to you.