Composite Well Curve Generation Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing well construction operations face delays and inefficiencies due to deviations from well plans, which are often repeated in similar drilling conditions, highlighting the need for improved rig activity monitoring and reporting to optimize drilling processes.
Innovation Solution
A system utilizing a machine learning processor to analyze historical wellbore data, generate a composite well curve by selecting high-performing sections from multiple wellbores, and provide optimized operational parameters for future drilling, thereby creating an optimized composite well curve that can inform users of potential issues and enhance drilling efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional well construction operations follow standard well plans, then operational consistency is maintained, but repeated deviations cause rig delays and reduced productivity
Solution Approach 1:
The system performs preliminary analysis of historical wellbore data before actual drilling operations begin. The machine learning processor pre-processes performance data from multiple previous wellbores to generate optimized composite well curves, so that when drilling starts, the optimal path and parameters are already determined, preventing delays rather than reacting to them
Solution Approach 2:
The system establishes a feedback loop where performance data from completed wellbores is continuously fed back into the machine learning processor. This feedback mechanism allows the system to learn from past deviations and successes, continuously improving future well curve generation and operational planning to prevent repeated mistakes
2Productivity
If historical performance data from multiple wellbores is analyzed to create optimized paths, then drilling efficiency improves, but data processing complexity and computational requirements increase
Solution Approach 1:
The patent replaces manual or simple automated data processing methods with a machine learning processor. This substitution enables the system to handle complex multi-wellbore performance data, identify patterns, and generate optimized composite well curves automatically, managing computational complexity through specialized AI algorithms rather than traditional processing methods
3Reliability
If the system generates detailed composite well curves with advisory messages, then operational guidance and issue detection improve, but system complexity and computational resources required increase
Solution Approach 1:
The system applies local quality by providing targeted advisory messages at specific wellbore depths and locations rather than uniform guidance throughout. The machine learning processor identifies critical sections where deviations are most likely to occur and generates specific advisory messages for those locations, concentrating computational resources where they provide maximum operational value
Data Source
AI summary
Method is provided that can include operations for selecting one or more wellbores from a wellbore historical database, retrieving performance data for the one or more wellbores, determining performance indices of the one or more wellbores relative to wellbore depth based on the performance data, determining which of the one or more wellbores performed best at each depth based on the performance indices at each depth for a selected performance criteria, selecting the best performance at each depth from the one or more wellbores, and generating a composite well curve based on the best performance at each depth. A method is provided that can include operations for inputting performance data from multiple wellbores into a machine learning processor, processing the performance data based on a selected performance criteria, and generating an optimized composite well curve along with an optimized set of operational parameters for drilling a future wellbore.