Wellbore Trajectory Control Using DNN Reservoir Property Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for wellbore trajectory planning and control lack accuracy and efficiency, particularly in real-time applications, due to variable formation conditions and limited reservoir modeling.
Innovation Solution
A system utilizing a deep neural network (DNN) model trained with hybrid well properties and geostatistical data to generate accurate projections for wellbore trajectories, enabling real-time optimization and directional control through a steerable drill string.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional reservoir modeling methods are used for wellbore trajectory planning, then the process can be implemented with conventional tools, but the accuracy and real-time performance of trajectory projections are insufficient
Solution Approach 1:
The patent replaces traditional mechanical reservoir modeling methods with a deep neural network system that uses seismic attributes and well data to predict reservoir properties. The DNN model processes geological data computationally rather than through conventional numerical modeling, enabling both high accuracy in trajectory projections and real-time processing performance that traditional methods cannot achieve simultaneously
Solution Approach 2:
The patent transforms the approach by changing from deterministic reservoir modeling parameters to probabilistic predictions based on seismic attribute analysis. The system uses multiple seismic attributes (impedance, velocity, density) combined with well log data to generate probability distributions for reservoir properties, allowing accurate trajectory planning with real-time updates as new data becomes available
2Manufacturing precision
If complex reservoir modeling is performed to improve trajectory accuracy, then more accurate projections are achieved, but the computational time and processing complexity increase
Solution Approach 1:
The patent performs preliminary action by training the deep neural network model offline using extensive well data and seismic attributes before actual drilling operations. The pre-trained model contains learned relationships between seismic characteristics and reservoir properties, allowing rapid real-time predictions during drilling without requiring complex computational reservoir modeling at that stage, thus achieving high precision without real-time computational delays
Solution Approach 2:
The system creates a virtual copy of the reservoir model through the DNN's internal representations learned from training data. This neural network model acts as a surrogate that replicates complex reservoir behavior patterns without requiring the actual complex physical modeling computations during drilling operations, enabling fast accurate predictions
3Productivity
If real-time trajectory optimization is implemented, then drilling efficiency is improved, but the system complexity and data processing requirements increase
Solution Approach 1:
The patent implements a universal deep neural network model that performs multiple functions: predicting reservoir properties, optimizing drilling trajectories, and providing real-time guidance. This single multi-functional system replaces what would otherwise require separate specialized systems for each function, managing complexity while enabling real-time optimization for improved drilling efficiency
Solution Approach 2:
The system achieves self-service through the DNN's ability to automatically learn and adapt from incoming well data and seismic attributes during drilling operations. The model continuously refines its predictions based on actual measurements without requiring manual intervention or complex external processing systems, simplifying the overall system architecture while maintaining real-time optimization capabilities
Data Source
AI summary
Certain aspects and features relate to a system for trajectory planning and control for new wellbores. Data can be received for multiple existing wells associated with a subterranean reservoir and used to train a deep neural network model to make accurate well property projections at any other location in the reservoir. A model of features for specific well locations based on seismic attributes of the well location can be automatically generated, and the model can be used in drilling trajectory optimization. In some examples, the system builds a deep neural network (DNN) model based on the statistical features, and trains the DNN model using Bayesian optimization to produce an optimized DNN model. The optimized model can be used to provide drilling parameters to produce an optimized trajectory for a new well.