Well Log Normalization Through Probability-Density Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing well log normalization methods, particularly for gamma ray data, are inefficient for large datasets and fail to account for non-stationary geology, requiring subjective well selection and manual intervention, leading to potential distortion and erroneous interpretations.
Innovation Solution
A method involving clustering of well logs based on probability density functions to normalize gamma ray and bulk density logs, using a computer processor to align individual logs with the mean response of their respective clusters, thereby removing tool-specific noise and enabling consistent normalization across multiple wells.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional 2-point normalization method is used, then manual intervention and subjective well selection are required, but this leads to inefficiency and inability to scale to large datasets
Solution Approach 1:
The system performs automated clustering and normalization without requiring manual selection of reference wells. The algorithm automatically identifies clusters based on probability density functions and computes normalization factors, enabling the system to serve itself without human intervention for routine normalization tasks.
Solution Approach 2:
The manual mechanical process of selecting reference wells and computing normalization factors is replaced with automated computational algorithms. The system uses probability density function analysis and clustering algorithms to automatically determine normalization parameters, substituting human expertise with automated computational mechanisms.
2Quantity of substance
If traditional 2-point normalization method is used, then only a few wells can be normalized manually, but this creates difficulty in scaling to large datasets with tens of thousands of wells
Solution Approach 1:
The system segments the large dataset of well logs into clusters based on similar probability density functions. This segmentation allows the normalization process to be applied independently to each cluster, enabling parallel processing and scaling to large numbers of wells while maintaining computational efficiency.
Solution Approach 2:
The clustering algorithm serves multiple functions simultaneously: it groups similar wells, identifies representative members, computes normalization factors, and prepares data for analysis. This multi-functionality eliminates the need for separate manual processes and enables efficient handling of large datasets.
3Measurement precision
If traditional normalization method is used, then subjective selection of good wells is required, but this introduces bias and requires trained professional intervention
Solution Approach 1:
The system automatically identifies appropriate reference wells and computes normalization factors without requiring subjective human judgment. The probability density function-based clustering objectively determines which wells should be used as references, eliminating bias while maintaining accuracy.
Solution Approach 2:
The system changes the approach from subjective well selection to objective parameter-based clustering. By using probability density functions and statistical parameters (mean, standard deviation, skewness, kurtosis), the system transforms the normalization process into an objective, reproducible calculation that does not depend on human judgment.
4Reliability
If traditional 2-point method is used, then assumption of similar distribution shapes is required, but this fails for non-stationary geology causing distortion
Solution Approach 1:
The system segments wells into clusters based on their probability density function characteristics. This segmentation allows each cluster to have its own normalization parameters, enabling the system to adapt to non-stationary geology where different regions have different distribution characteristics. Each cluster can be normalized independently according to its specific properties.
Solution Approach 2:
The normalization approach transitions from a static, one-size-fits-all method to a dynamic, adaptive system. The clustering algorithm automatically adjusts groupings and normalization parameters based on the actual data characteristics, allowing the system to respond to varying geologic conditions without requiring manual reconfiguration.
Data Source
AI summary
A method is described for well log normalization that receives well logs including at least gamma ray logs; clustering the well logs into a plurality of well log clusters based on an interval of interest within the well logs, wherein the clustering is done based on probability density functions; for each well log cluster, normalizing each well log within the well log cluster towards a mean response of an aggregated population of the well log cluster to generate normalized well logs; and displaying the normalized well logs.