Rig-Site Drilling Mud Prediction With ML Feedback Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing drilling mud formulations are inefficient and unreliable, leading to significant non-productive time (NPT) and unsatisfactory safety and efficiency records due to reliance on trial and error laboratory testing and human intuition.
Innovation Solution
A machine-learning based approach that predicts rheological properties and sensor-based performance of drilling mud using empirical relationships and AI/ML algorithms, enabling real-time formulation optimization at the rig-site through a materials-mud-properties and sensor-mud-properties relationship.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If trial and error laboratory testing and human intuition are used for drilling mud formulation, then formulation development can proceed with simple methods, but significant non-productive time is lost and reliability is reduced
Solution Approach 1:
The system performs preliminary actions by pre-establishing empirical relationships between compositional data and rheological properties, and between rheological properties and sensor responses. These relationships are built beforehand through laboratory testing and machine learning training, enabling rapid predictions during actual drilling operations without requiring trial-and-error testing at the rig site.
Solution Approach 2:
The patent replaces the mechanical trial-and-error testing process with an automated machine learning system that uses empirical relationships and algorithms to predict rheological properties and optimize formulations. This substitution transforms manual, time-consuming laboratory testing into an automated computational process that runs in real-time at the rig site.
2Ease of manufacture
If trial and error laboratory testing and human intuition are used for drilling mud formulation, then formulation development can proceed with simple methods, but safety and efficiency records remain unsatisfactory
Solution Approach 1:
The system implements continuous feedback by monitoring sensor responses from the drilling mud in real-time, comparing actual performance against predicted rheological properties, and using this feedback to update and refine the machine learning models. This closed-loop feedback mechanism enables continuous optimization of drilling mud formulations, improving safety and efficiency records through data-driven adjustments rather than relying on human intuition.
Solution Approach 2:
The system enables self-service by allowing the machine learning model to automatically optimize drilling mud formulations without continuous human intervention. The system autonomously analyzes compositional data, predicts rheological properties, interprets sensor responses, and generates formulation recommendations, freeing operators from manual testing and decision-making while improving reliability through consistent, data-driven optimization.
3Loss of time
If machine-learning based real-time prediction and optimization is implemented, then non-productive time is reduced and reliability is enhanced, but system complexity increases
Solution Approach 1:
The system achieves universality by creating a multi-functional machine learning platform that handles multiple tasks: predicting rheological properties from compositional data, interpreting sensor responses, optimizing formulations, and providing real-time recommendations. This single integrated system replaces multiple separate processes including laboratory testing, manual analysis, and formulation development, reducing overall system complexity despite the advanced capabilities provided.
Solution Approach 2:
The patent uses empirical relationships as intermediaries between compositional data and rheological properties, and between rheological properties and sensor responses. These pre-established mathematical relationships act as mediators that bridge different data types and prediction targets, simplifying the overall system architecture by breaking down complex predictions into manageable sequential steps rather than requiring a single monolithic model.
4Productivity
If machine-learning based real-time prediction and optimization is implemented, then non-productive time is reduced by 13-15%, but computational requirements and data processing needs increase
Solution Approach 1:
The system performs preliminary computational work by pre-training machine learning models and establishing empirical relationships during the formulation development phase. This preliminary action transfers computational burden from real-time operations to the training phase, allowing rapid predictions during actual drilling with minimal energy consumption while maintaining high productivity benefits.
Data Source
AI summary
A computer-implemented method for machine learning based rig-site on-demand drilling mud characterization, property prediction and optimization is described. The method includes predicting rheological properties of drilling mud from compositional data based on a first empirical relationship across the compositional data. The method also includes predicting sensor based rheological performance of drilling mud based on a second relationship between the predicted rheological properties of the drilling mud and sensor responses from one or more sensors. Additionally, the method includes providing feedback to the one or more sensors, wherein the feedback comprises the sensor based rheological performance, sensor responses, and a formulation associated with the drilling mud, wherein the feedback enables updating the formulation associated with the drilling mud.