ML Formation Top Prediction Using Real-Time Drilling Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining formation tops during drilling are inefficient, relying on simplistic assumptions and data from offset wells or downhole sensors, which can be unreliable due to measurement lag and inherent uncertainties, leading to inaccurate formation picks and potential well integrity issues.
Innovation Solution
A computer-implemented method using machine learning analytics to predict formation tops by correlating real-time drilling parameters such as ROP, torque, RPM, WOB, and SPP, allowing for iterative model refinement and validation, enabling accurate formation picks at the drilling bit level.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If data from offset wells or downhole sensors are used to determine formation tops, then the process is simpler, but the accuracy and reliability of formation picks deteriorate due to measurement lag and inherent uncertainties
Solution Approach 1:
The patent replaces traditional mechanical/sensor-based measurement systems with a machine learning-based predictive system. Instead of relying on physical downhole sensors that suffer from measurement lag, the system uses machine learning models trained on drilling parameters (ROP, torque, RPM, WOB, SPP) to predict formation tops in real-time, eliminating the measurement lag issue while maintaining simplicity through software-based solutions
Solution Approach 2:
The patent introduces machine learning models as an intermediary between raw drilling data and formation top determination. The ML models process and correlate multiple drilling parameters to predict formation tops, acting as a mediator that transforms complex, noisy sensor data into accurate formation picks without the limitations of direct sensor measurements
2Measurement precision
If real-time drilling parameters are analyzed using machine learning, then the accuracy of formation picks improves, but the computational complexity and data processing requirements increase
Solution Approach 1:
The patent applies preliminary action by pre-training machine learning models using historical drilling data before actual drilling operations. The models are trained offline on datasets containing drilling parameters and corresponding formation tops, so that during real-time drilling, the pre-trained models can quickly make predictions without requiring complex real-time computation, thus reducing computational complexity while maintaining high accuracy
Solution Approach 2:
The patent transforms the problem from direct sensor measurement to parameter correlation analysis. Instead of measuring formation properties directly, the system changes the approach to analyzing correlations between drilling parameters (ROP, torque, RPM, WOB, SPP) and formation characteristics, using machine learning to identify patterns that indicate formation tops, thereby managing computational complexity through parameter transformation
3Reliability
If multiple drilling parameters are correlated to predict formation tops, then the reliability of formation picks improves, but the difficulty of detecting and measuring increases
Solution Approach 1:
The patent implements feedback mechanisms where the machine learning model continuously receives drilling parameter data, makes predictions, and the predictions are validated against actual formation data when available. This feedback loop allows the system to learn and improve over time, increasing reliability while the automated nature of the feedback process manages the complexity of parameter correlation through systematic validation
Data Source
AI summary
Some implementations of the present disclosure provide a computer-implemented method that includes: accessing measurement data obtained from a drilling operation, wherein the measurement data show multiple measurements during the drilling operation when a drilling bit is located at a range of depths; based on the measurement data, using machine learning analytics to construct a model that predicts a formation top when the drilling bit reaches a depth; determining a correlation between the measurement data and the predicted formation top; and in response to determining the correlation exceeds a pre-determined threshold, applying the model to predict a formation top when the drilling bit reaches the depth.