LWD Resistivity Prediction via Feature Map for Deep Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing logging-while-drilling (LWD) resistivity (Rt) curve acquisition methods are inaccurate, leading to insufficient precision in locating areas with high hydrocarbon potential during deep oil and gas navigation, due to instrument deficiencies and severe formation interference.
Innovation Solution
A method and system for evaluating sand shale formation physical properties, involving data acquisition, correlation value calculation, outlier elimination, and LWD Rt prediction using a trained model, which constructs a two-dimensional input feature map and sliding units to predict the LWD Rt curve and locate high hydrocarbon potential areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LWD Rt logging is used to measure formation resistivity, then real-time resistivity data can be obtained during drilling, but the accuracy is insufficient due to instrument deficiency or malfunction and severe formation interference
Solution Approach 1:
The patent introduces an intermediary prediction model that uses multiple logging curves (acoustic, density, neutron, photoelectric factor) as mediators to indirectly predict the LWD Rt curve. Instead of directly measuring resistivity with faulty instruments, the system uses these intermediary parameters to calculate and reconstruct the resistivity curve, thereby overcoming the harmful effects of instrument deficiency and formation interference.
Solution Approach 2:
The patent transforms the measurement approach by changing from direct resistivity measurement to predicting resistivity through parameter relationships. It establishes correlation models between LWD Rt and other logging parameters (acoustic transit time, density, neutron porosity, photoelectric factor), then uses these parameter changes and relationships to reconstruct the resistivity curve, improving measurement accuracy despite formation interference.
2Measurement precision
If multiple logging curves are used to predict LWD Rt curve, then prediction accuracy can be improved, but the complexity of data processing increases
Solution Approach 1:
The patent segments the data processing into distinct modules: data acquisition from multiple sources, correlation value calculation for each logging curve, outlier detection and elimination using isolated forest algorithm, and final curve reconstruction. This segmentation reduces processing complexity by organizing the complex multi-parameter prediction task into manageable, sequential steps.
Solution Approach 2:
The patent performs preliminary actions by pre-calculating correlation values between each logging curve and the LWD Rt curve before actual prediction. It also pre-processes the data by eliminating outliers using the isolated forest algorithm. These preliminary actions simplify the main prediction task and reduce the complexity of real-time data processing.
3Reliability
If outlier data is eliminated to improve prediction accuracy, then the reliability of LWD Rt prediction is improved, but the processing time increases
Solution Approach 1:
The patent replaces traditional mechanical outlier detection methods with an isolated forest algorithm, which is a computational approach that efficiently identifies outliers by building isolation trees. This substitution improves reliability by more accurately distinguishing true outliers from valid data points, while the algorithmic approach is optimized to minimize processing time compared to exhaustive manual or traditional statistical methods.
Data Source
AI summary
A sand shale formation physical property evaluation method and system for precise deep oil and gas navigation aims to solve the problem that the prior art cannot acquire a real-time and accurate logging-while-drilling (LWD) Rt curve. The method includes: acquiring basic data of a target well location as well as basic data and an LWD resistivity (Rt) of an adjacent well; dividing the data into different groups; retaining data with a maximum correlation value with the LWD Rt in each group of data; eliminating outliers, and performing standardization; constructing a two-dimensional input feature map by taking the correlation value and standardized data as a weight; acquiring an LWD Rt prediction curve based on the two-dimensional input feature map; calculating a hydrocarbon parameter in a window based on the LWD Rt prediction curve; and locating an area with a high hydrocarbon potential based on the hydrocarbon parameter at each position.