Core-Log Mineral Assemblage Transfer Learning for Noisy Formation Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing formation evaluation methods face challenges in accurately inferring mineralogy due to limited information content and quality in downhole measurements, often grouping minerals inaccurately and introducing noise and biases.
Innovation Solution
A method utilizing core mineralogy databases to generate optimized mineral assemblages through principal component analysis (PCA) and non-negative matrix factorization (NNMF), defining assemblage endpoints, and performing petrophysical evaluations based on selected assemblages and formation fluids.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of minerals included in the evaluation is increased, then the accuracy of the evaluation is improved, but the evaluation becomes less robust and more prone to noise and biases in the logs
Solution Approach 1:
The patent segments the formation minerals into distinct assemblages (e.g., quartz-feldspar-mica assemblage, clay assemblage, carbonate assemblage) based on spectral characteristics. This segmentation allows the evaluation to handle multiple minerals systematically while maintaining robustness by grouping them into manageable assemblages that can be evaluated separately using limited log data.
Solution Approach 2:
The patent transforms the problem from directly evaluating individual minerals to evaluating mineral assemblages defined by spectral parameters. By changing the parameter space from individual mineral concentrations to assemblage-level spectral signatures, the method achieves accurate mineralogy inference with limited log information while maintaining evaluation robustness.
2Ease of operation
If downhole elemental spectroscopy tools are used to provide elemental concentrations, then mineral separation becomes easier, but the information content and quality of older generation tools are limited
Solution Approach 1:
The patent creates a virtual core mineralogy database that copies and extends the information from limited downhole logs. By building a comprehensive mineral assemblage database from available spectral data and core analyses, the system reconstructs detailed mineralogy information that exceeds the direct measurement capabilities of the downhole tools, effectively copying core-quality information to log-scale evaluations.
Solution Approach 2:
The patent introduces mineral assemblages as an intermediary between raw spectral measurements and individual mineral concentrations. This intermediary layer allows the system to bridge the information gap between limited downhole log data and comprehensive mineralogy evaluation, enabling accurate mineral separation even when direct elemental spectroscopy information is limited or of lower quality.
3Ease of manufacture
If generic shale and Quartz-Feldspar-Mica assemblages are used to simplify evaluation, then the process becomes easier, but the accuracy of mineralogy inference is reduced
Solution Approach 1:
The patent implements a dynamic assemblage system where mineral assemblage definitions are not fixed but are derived and optimized based on the specific formation being evaluated. The system dynamically adjusts assemblage characteristics to match the actual mineralogy present in each formation, allowing the evaluation to maintain simplicity while achieving high accuracy by adapting to local conditions rather than applying generic templates.
Data Source
AI summary
This invention describes a method of conducting formation evaluation and in particular inferring formation mineralogy when the formation is complex, and when the set of available downhole measurements is limited in information content or quality. One or more embodiments of the method can use a core mineralogy database to generate effective mineral assemblage that captures the natural systematic correlations between series of minerals.