NMR T1-T2 Reservoir Characterization for Fluid Producibility
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods struggle to reliably characterize and model hydraulic fractures in unconventional reservoirs like shale gas and shale oil reservoirs, leading to inaccurate water cut predictions and insufficient distinction or quantification of producible oil, which affects drilling and completion strategies.
Innovation Solution
An integrated machine learning framework using unsupervised learning on nuclear magnetic resonance (NMR) measurements to generate an interpreted NMR log, quantifying fluid producibility parameters such as fluid porosity and saturation, and providing a production characterization of the reservoir.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional static formation evaluation models are used to estimate water saturation, then the modeling process is simple, but the accuracy of water cut prediction is insufficient and does not align with dynamic production data
Solution Approach 1:
The patent transforms static formation evaluation into dynamic prediction by incorporating time-dependent production data and using machine learning models that evolve with reservoir performance. This changes the parameter state from static snapshots to dynamic trajectories, improving water saturation accuracy while managing complexity through automated algorithms.
Solution Approach 2:
The patent introduces machine learning models as intermediaries between static formation data and dynamic production outcomes. These models act as mediators that integrate multiple data sources (logs, core data, production data) and translate them into accurate water saturation predictions, resolving the contradiction between simplicity and accuracy.
2Measurement precision
If traditional formation evaluation methods are used, then the process is straightforward, but the ability to distinguish and quantify producible oil is insufficient
Solution Approach 1:
The patent segments hydrocarbons into distinct categories (producible oil, immovable hydrocarbons, water) using machine learning classification of NMR data. This segmentation enables precise quantification of producible oil by separating it from non-producible phases, while the automated nature of the segmentation process manages the inherent complexity.
Solution Approach 2:
The patent applies dynamic machine learning models that adapt to reservoir behavior over time, enabling continuous refinement of producible oil quantification. This dynamic approach improves accuracy by capturing changing reservoir conditions while the automated modeling manages complexity through iterative learning rather than manual analysis.
3Reliability
If hydraulic fracture characterization is attempted using conventional methods, then the approach is simple, but the reliability of fracture modeling is insufficient
Solution Approach 1:
The patent implements feedback loops where production data continuously informs and refines fracture characterization models. This feedback mechanism improves reliability by validating and adjusting fracture models against actual reservoir performance, while the automated feedback processing manages the complexity of iterative model refinement.
Solution Approach 2:
The patent creates composite models that integrate multiple data types (NMR logs, core data, production data, geomechanics) to characterize hydraulic fractures. This composite approach improves reliability by combining complementary information sources, while the integrated modeling framework manages complexity through unified analysis rather than separate studies.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables precise characterization of hydrocarbon and water saturation aligned with dynamic production data, facilitating optimized well placement and completion strategies, reducing costs and improving resource development efficiency.
Implementation Method 1
obtaining spin-lattice relaxation time (T1) and the spin-spin relaxation time (T2) log data
Implementation Method 2
separating the NMR T1-T2 log data into a plurality of relaxed fluid signatures, each having a different T1-T2 signature
Data Source
AI summary
Implementations described and claimed herein provide systems and methods for developing resources from a reservoir. In one implementation, obtaining nuclear magnetic resonance (NMR) log data is obtained for one or more wells of the reservoir. The NMR data is captured using one or more logging tools. An interpreted NMR log is generated by quantifying one or more fluid producibility parameters. The one or more fluid producibility parameters are quantified by processing the NMR log data using automated unsupervised machine learning. A production characterization of the reservoir is generated based on the interpreted NMR log, with the reservoir being developed based on the production characterization.