Neural Network Permeability Prediction for Lost Circulation Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining fracture sizes and selecting lost circulation materials (LCMs) in subsurface formations are limited by the availability of logging tool measurements, particularly in predicting permeability and addressing lost circulation events during drilling and cementing operations.
Innovation Solution
A method using a computer processor to obtain NMR data and acquired permeability data, and then employing a neural network to predict permeability and determine fracture sizes within a geological formation. Based on these predictions, the appropriate type of LCM is selected and a command is transmitted to the well system to trigger the use of this LCM during operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If logging tool measurements are used to determine formation properties, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent uses NMR logging data as a proxy or copy to predict permeability, avoiding the need for direct permeability measurements. The neural network model creates a computational copy of the relationship between NMR parameters and permeability, allowing indirect determination of formation permeability through readily available NMR data rather than complex dedicated permeability logging tools
Solution Approach 2:
The patent replaces physical/mechanical permeability measurement systems with a computational approach using neural networks. Instead of using mechanical logging tools that directly measure permeability, the system substitutes a data-driven computational model that processes NMR data to predict permeability values, thereby reducing device complexity while maintaining measurement capability
2Reliability
If real-time logging tool measurements are required for lost circulation prevention, then reliability is improved, but loss of time occurs due to measurement unavailability
Solution Approach 1:
The patent performs preliminary actions by training the neural network model in advance using historical NMR data and corresponding permeability measurements. This pre-trained model can then rapidly predict permeability and identify lost circulation risks during drilling operations without requiring time-consuming real-time measurements, thus maintaining reliability while reducing time loss
Solution Approach 2:
The system dynamically adapts by using the trained neural network to continuously evaluate current NMR data against learned patterns from training data. This dynamic evaluation allows the system to provide real-time or near-real-time predictions of permeability changes and lost circulation risks without the delay of traditional measurement methods
3Ease of operation
If traditional methods are used to determine fracture size and select LCM, then ease of operation is maintained, but manufacturing precision of LCM selection deteriorates
Solution Approach 1:
The patent introduces an intermediary computational layer (the neural network model) that bridges the gap between simple NMR data acquisition and precise LCM selection. This intermediary automatically processes NMR data to predict permeability and determine optimal LCM types, maintaining ease of operation while dramatically improving selection accuracy through data-driven insights that would be difficult to obtain through traditional manual methods
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
This method enables the prediction of permeability and fracture sizes without relying on real-time logging tool measurements, allowing for the effective selection and deployment of LCMs to prevent lost circulation events and minimize formation damage.
Implementation Method 1
obtaining, by a computer processor, first nuclear magnetic resonance (NMR) data
Data Source
AI summary
A method may include obtaining first nuclear magnetic resonance (NMR) data and acquired permeability data regarding a geological region of interest. The method may further include determining, using a neural network and second NMR data, predicted permeability data regarding a predetermined formation within the geological region of interest. The neural network may be trained using the first NMR data and the acquired permeability data. The method may further include determining a predetermined fracture size within the predetermined formation based on the predicted permeability data. The method may further include determining a predetermined type of lost circulation material (LCM) based on the predetermined fracture size. The method may further include transmitting a command to a well system that triggers a well operation using the predetermined type of LCM.