Hybrid Drilling Model Optimizes Trajectory via ML Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wellbore drilling technologies face challenges in accurately planning and adjusting directional drilling trajectories, often resulting in deviations from the intended path due to overestimation of drilling tool capabilities by physical modeling, leading to increased drilling time and potential missed targets.
Innovation Solution
The integration of physics-based modeling with machine learning corrections, specifically using Gaussian process regression, to predict and optimize drilling parameters such as dogleg severity and rate of penetration, allowing for real-time trajectory adjustments and new trajectory design based on tool performance capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physics-based modeling is used to predict tool performance, then trajectory planning can be performed, but the model often overestimates drilling tool capabilities leading to trajectory deviations
Solution Approach 1:
The patent implements feedback by using machine learning models to continuously correct and refine the physics-based tool performance predictions. The ML models learn from actual drilling data and adjust their predictions to better match real-world performance, creating a closed-loop system that improves accuracy over time.
Solution Approach 2:
The patent changes the parameters of the prediction system by combining multiple modeling approaches (physics-based models with machine learning corrections). This hybrid approach allows the system to capture both the fundamental physical relationships and the complex, non-ideal behaviors observed in actual drilling operations.
2Adaptability or versatility
If directional drilling is applied in myriad applications and formations, then wellbore trajectory complexity increases, but this leads to more difficult trajectory planning and control
Solution Approach 1:
The patent segments the complex trajectory planning problem into manageable components by evaluating different trajectory sections independently. The system divides the wellbore trajectory into discrete segments and applies optimized planning to each section, making the overall complex problem more tractable.
Solution Approach 2:
The patent introduces dynamic trajectory planning that adapts to changing conditions during drilling. The system can modify trajectory parameters in real-time based on actual tool performance and formation conditions, allowing the drilling operation to respond dynamically to unexpected challenges while maintaining versatility across different applications.
3Manufacturing precision
If trajectory corrections are applied in near real-time, then target accuracy can be maintained, but this requires continuous monitoring and re-planning increasing operational complexity
Solution Approach 1:
The patent applies preliminary action by pre-calculating and preparing multiple trajectory options and correction strategies before actual drilling occurs. The system pre-evaluates potential trajectory variations and prepares correction plans in advance, so that when real-time adjustments are needed, the operator can quickly select from pre-prepared options rather than performing complex calculations during the drilling operation.
Data Source
AI summary
A method to design well trajectories includes determining dogleg severity as a function of inclination, and a corresponding rate of penetration performance of the tool by a hybrid model including physical modelling and machine learning correction. The method includes solving for optimal steering parameters to predict a dogleg severity as close as possible to a desired dogleg severity at a given inclination of the trajectory, which is repeated for dogleg severity and inclination combinations of interest. The rate of penetration for feasible points is also determined and a rate of penetration (or time-to-target) map can be produced. Potential trajectories are then evaluated relative to the map to estimate drilling time-to-target, and an optimal trajectory can be selected that has a lowest time-to-target while also being feasible for the tool and optionally avoiding risks or downhole obstacles.