Well Log Data Conditioning via Quantified Uncertainty Error Bounds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing geophysical well log data conditioning process is challenging due to borehole washouts, drilling fluid invasion, and inconsistencies in log data acquired by different service companies, leading to time-consuming and potentially misleading seismic inversion results, especially when dealing with large numbers of wells.
Innovation Solution
A computer-implemented method that determines uncertainty in measured elastic logs, generates simulated logs to account for errors, and establishes error bounds to condition well log data efficiently, flagging and ranking wells with excessive uncertainties for further investigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional manual log conditioning processes are used to handle borehole washouts and drilling fluid invasion, then data quality can be improved through careful inspection and correction, but the processing time increases significantly especially for projects with thousands of wells
Solution Approach 1:
The patent creates simulated elastic logs by adding quantified uncertainties to measured elastic logs, generating multiple simulated versions that represent possible data variations. This copying approach allows automated statistical analysis without manual intervention, maintaining data quality assessment while dramatically reducing processing time for large-scale projects
Solution Approach 2:
The patent transforms the log conditioning approach by changing from manual inspection parameters to automated statistical parameters. By introducing quantified uncertainties and computing error bounds through simulated logs, the system automatically identifies problematic intervals without human intervention, resolving the time-quality contradiction
2Adaptability or versatility
If log data from different service companies using different tools is integrated, then data coverage and project scope can be expanded, but inconsistencies and measurement variations increase making conditioning more challenging
Solution Approach 1:
The patent segments the log conditioning process by applying quantified uncertainties specific to each service company's tools and measurement methods. By treating each data source independently with its own uncertainty characteristics, the system maintains measurement precision while accommodating diverse data coverage from multiple service companies
Solution Approach 2:
The patent introduces quantified uncertainties as an intermediary layer between measured elastic logs from different service companies and the final seismic inversion results. This intermediary allows the system to harmonize inconsistent measurements by accounting for tool-specific variations, enabling seamless integration of multi-source data
3Measurement precision
If comprehensive manual inspection and correction of elastic logs is performed to ensure accuracy, then measurement precision improves, but the complexity and time consumption of the process increases
Solution Approach 1:
The patent enables the log conditioning process to serve itself by automatically identifying problematic intervals through error bound calculations. The system uses quantified uncertainties and simulated logs to self-diagnose data quality issues without manual inspection, maintaining high measurement precision while reducing process complexity
Solution Approach 2:
The patent implements feedback by computing error bounds and comparing them against threshold values to automatically flag intervals requiring further investigation. This feedback mechanism replaces complex manual inspection with a simple automated decision process, maintaining accuracy while simplifying the overall workflow
Data Source
AI summary
Example computer-implemented methods, media, and systems for well log data conditioning using quantified uncertainties in rock physics and seismic inversion results. One example computer-implemented method includes determining, using multiple measured elastic logs corresponding to multiple wells, one or more first elastic attributes of the multiple wells. Respective uncertainty is added to each of the multiple measured elastic logs. Multiple simulated elastic logs are generated based on the added respective uncertainty and the multiple measured elastic logs. One or more second elastic attributes are determined using the multiple simulated elastic logs. A respective error bound for each of the multiple measured elastic logs is determined based on the respective uncertainty added to each of the multiple measured elastic logs, the one or more first elastic attributes, and the one or more second elastic attributes. Conditioning of a measured elastic well log is performed using the determined error bounds.