Neural Network Formation Property Prediction from DDR
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional inversion algorithms for deep directional resistivity (DDR) measurements in subsurface formation evaluation are computationally expensive and prone to getting stuck in local minima, leading to slow and unreliable formation property predictions.
Innovation Solution
A neural network-based formation property prediction model is pretrained with DDR data and tool parameters to quickly predict formation properties such as resistivity, anisotropy, and azimuth without performing inversion calculations, utilizing Monte Carlo dropout for uncertainty quantification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional inversion algorithms are used for DDR measurements, then formation property predictions are obtained, but computational cost is high and results are unreliable due to getting stuck in local minima
Solution Approach 1:
The system performs preliminary action by pre-computing a lookup table containing formation property predictions for various DDR measurement scenarios before actual drilling operations. This pre-computed table stores the relationship between DDR measurements and formation properties, eliminating the need for complex real-time inversion calculations during drilling, thus reducing computational complexity while maintaining prediction reliability
Solution Approach 2:
The system creates a simplified copy or approximation of the complex inversion process by using a pre-computed lookup table that replicates the essential relationships between DDR measurements and formation properties. This copy allows rapid prediction without executing the full computational inversion algorithm, resolving the contradiction between reliability and computational complexity
2Productivity
If traditional inversion algorithms are used for DDR measurements, then formation property predictions are obtained, but prediction speed is slow due to computational expense
Solution Approach 1:
The system performs preliminary action by pre-computing formation property predictions for a range of possible DDR measurements and storing these results in a lookup table before actual drilling operations begin. During drilling, the system simply queries this pre-computed table rather than performing energy-intensive real-time inversion calculations, dramatically increasing productivity while reducing computational energy consumption
Solution Approach 2:
The system implements dynamics by adapting the prediction approach based on the specific DDR measurement values encountered during drilling. Instead of always performing full inversion calculations, the system dynamically queries the pre-computed lookup table with the actual measurement values, optimizing the balance between prediction accuracy and computational efficiency in real-time
Data Source
AI summary
A system and method for predicting formation properties is described. For example, a computing device may receive deep directional resistivity (DDR) measurement data from one or more DDR sensors. The computing device may apply a formation property prediction model to the DDR measurement data, the formation property prediction model pretrained to identify predicted formation parameters based on input DDR data, formation properties of the input DDR data, and tool parameters. The computing device may receive the predicted formation parameters for a subsurface beyond the wellbore in response to applying the formation property prediction model to the DDR measurement data.