Machine Learning Model for Downhole Resistivity Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for correcting environmental effects in deep subsurface drilling rely on user judgment, leading to inconsistencies across wells due to differing expert interpretations, which can be time-consuming and result in inaccurate resistivity measurements.
Innovation Solution
A data processing system utilizing a supervised machine learning model to identify and correct environmental effects in logging-while-drilling data, such as resistivity phase and attenuation signatures, to generate corrected resistivity values and flag affected data, thereby providing a standardized and accurate representation of the subsurface.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional methods using user judgment are used to correct environmental effects, then flexibility in interpretation is maintained, but inconsistencies across wells occur due to differing expert interpretations
Solution Approach 1:
The patent replaces the mechanical system of human expert judgment with an automated machine learning model that processes resistivity log data. The model uses trained algorithms to identify and correct environmental effects such as shoulder bed effects, invasion effects, and anisotropy effects, eliminating variability between different experts while maintaining consistent correction methodologies across all wells.
Solution Approach 2:
The patent changes the parameter of correction methodology from subjective human judgment to objective algorithmic processing. The machine learning model analyzes multiple parameters including resistivity phase signatures, attenuation signatures, and depth information to automatically determine appropriate corrections, transforming the correction process from a qualitative to a quantitative approach.
2Reliability
If conventional manual correction methods are used, then expertise can be applied to complex cases, but the process is time-consuming
Solution Approach 1:
The patent replaces time-consuming manual correction processes with automated machine learning model execution. The model rapidly processes resistivity log data, identifying environmental effects and applying corrections without the time constraints of manual analysis, while maintaining or improving correction accuracy through consistent application of trained algorithms.
Solution Approach 2:
The patent applies preliminary action by pre-training the machine learning model on extensive datasets before deployment. The model is trained in advance on labeled data representing various environmental effects and their corrections, enabling it to quickly and accurately correct new data without requiring time-consuming manual analysis during the actual correction process.
3Productivity
If automated machine learning models are used to correct environmental effects, then consistency and speed are improved, but complexity of the system increases
Solution Approach 1:
The patent applies universality by designing a machine learning model that handles multiple environmental effects (shoulder bed effects, invasion effects, anisotropy effects, dielectric effects) within a single integrated system. The model performs multiple functions including data processing, effect identification, correction calculation, and quality control, reducing the need for separate specialized systems for each correction type.
4Productivity
If automated machine learning models are used, then manual labor is reduced, but the initial setup and training requirements increase
Solution Approach 1:
The patent applies preliminary action by performing extensive model training and validation before deployment. Labeled datasets representing various environmental effects are prepared in advance, and the machine learning model is trained and tested to ensure accurate correction performance. This preliminary work, while resource-intensive, establishes a robust system that delivers high efficiency during operational use.
Data Source
AI summary
Methods and systems for drilling a well into a subsurface formation are configured for performing a downhole measurement to generate measured logging-while-drilling (LWD) data; retrieving a machine learning model that is trained using labeled LWD data, the labeled LWD data representing one or more environmental effects each causing a respective data signature in the LWD data, each respective data signature being associated with a corresponding label identifying the environmental effect; inputting the LWD data into the machine learning model; generating, by the machine learning model based on the inputting, a classification output representing at least one environmental effect represented in the LWD data; and generating, based on the classification output, corrected LWD data that removes a change to the LWD data caused by the environmental effect. The systems and methods are configured for drilling a well into the subsurface based on the corrected LWD data.