Predicting Hydrocarbon Shows Ahead of Drilling Bit
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current drilling operations lack efficient methods to predict the presence of hydrocarbons ahead of a drilling bit, leading to uncertain decisions on whether to continue drilling exploration wells, resulting in potential wastage of resources.
Innovation Solution
A machine learning-based system that uses input data such as drill bit location, depth, weight on bit, rotations per minute, lagged lithology percentages, and real-time mud gas logs to predict hydrocarbon show indicators, specifically hydrocarbon wetness, 1000 feet ahead of the drilling bit, employing algorithms like the Haworth Wetness Formula and isolation forest for data cleaning and outlier removal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional drilling operations are used without predictive analytics, then operational simplicity is maintained, but decision-making accuracy and resource efficiency deteriorate due to uncertainty in hydrocarbon presence prediction
Solution Approach 1:
The patent replaces traditional mechanical drilling decision-making processes with a machine learning-based predictive system. The system uses algorithms trained on historical drilling data to predict hydrocarbon presence ahead of the drilling bit, substituting expert geological judgment and manual analysis with automated computational models that process multiple data streams simultaneously.
Solution Approach 2:
The system performs preliminary predictions of hydrocarbon presence before the drilling bit actually reaches the target zone. By analyzing real-time drilling data and comparing it with trained models, the system forecasts hydrocarbon shows 100-1000 feet ahead of the bit, allowing advance decision-making about well continuation or abandonment before resources are committed to drilling into non-productive zones.
2Productivity
If real-time data processing and machine learning models are implemented, then decision-making speed and accuracy improve, but computational requirements and processing time increase
Solution Approach 1:
The machine learning models are trained offline on extensive historical drilling datasets before deployment. This preliminary training phase allows the models to learn complex patterns and relationships in the data, so that during actual drilling operations, predictions can be made rapidly by applying the pre-trained models to real-time data streams without requiring extensive computational processing at the moment of decision-making.
Solution Approach 2:
The system continuously monitors real-time drilling data and compares predicted hydrocarbon shows with actual outcomes as the drilling bit progresses. This feedback loop allows the system to validate predictions and potentially retrain or adjust models based on actual results, improving future prediction accuracy while maintaining efficient real-time operation through the use of established model frameworks.
Data Source
AI summary
Systems and methods include techniques for predicting hydrocarbon show indicators classifying a presence of hydrocarbons at a pre-determined distance ahead of a drilling bit. Input data is received that identifies, for different depths of a well that is being drilled, a drill bit location, a depth, a weight on bit, rotations per minute, a rate of penetration, lagged lithology percentages, and real-time mud gas logs. Data cleaning is performed on the input data using an isolation forest algorithm to remove outliers. A sequence of attributes for the well being drilled is identified from the input data, where the sequence of attributes includes the input data measured at a sequence of depths in the well. Hydrocarbon show indicators classifying a presence of hydrocarbons at a pre-determined distance ahead of a drilling bit are predicted in real time using machine learning on the sequence of attributes received while drilling the well.