Neural Network Seismic Inversion for Drilling Strategy Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional seismic imaging tools with high propagation rates reduce vertical resolution, leading to significant uncertainties in reservoir characterization, which is crucial for hydrocarbon estimation and drilling operations, especially in complex reservoirs like carbonates.
Innovation Solution
A method using a trained neural network to enhance seismic signal processing, incorporating real-time well data during drilling to optimize the drilling strategy by iteratively retraining the network with fresh data and improving seismic image definition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If high propagation rate is used in conventional seismic imaging, then processing speed is improved, but vertical resolution deteriorates
Solution Approach 1:
The patent applies dynamics by making the wavelet propagation rate variable rather than fixed. The system dynamically adjusts the propagation rate based on local subsurface properties identified during drilling, allowing faster processing in some regions and higher resolution in others, thus resolving the contradiction between processing speed and vertical resolution
Solution Approach 2:
The patent changes the propagation rate parameter adaptively during the drilling process. By modifying this key parameter based on fresh well data and seismic feedback, the system can optimize both processing efficiency and image quality for different geological conditions without being constrained by a fixed high propagation rate
2Speed
If high propagation rate is used in carbonates, then processing speed is improved, but calculation uncertainty increases significantly
Solution Approach 1:
The patent implements feedback by continuously incorporating fresh well data obtained during drilling into the seismic processing system. This feedback loop allows the system to adjust propagation rates and processing parameters in real-time, reducing calculation uncertainties in carbonate reservoirs while maintaining processing speed through adaptive optimization
Solution Approach 2:
The patent applies preliminary action by pre-identifying carbonate formations and other complex reservoir types using initial seismic data and well logs before drilling begins. This allows the system to prepare appropriate processing parameters and propagation rates in advance, preventing excessive calculation uncertainties before they occur
3Device complexity
If conventional seismic processing is used, then device complexity is reduced, but seismic image definition deteriorates
Solution Approach 1:
The patent applies universality by creating a multi-functional processing system that combines conventional seismic processing with machine learning algorithms, real-time well data integration, and adaptive propagation rate adjustment. This unified system performs multiple functions (processing, analysis, optimization) that would otherwise require separate systems, improving image definition without proportionally increasing complexity
Data Source
Figure 1
Figure 2a~2b
Figure 3a~3b
AI summary
Disclosed is a method for optimizing a drilling strategy during drilling of a well within a subsurface volume comprising a reservoir zone. The method comprises obtaining a trained neural network, having been trained on training data relating to the subsurface volume to infer well data from seismic data and fresh well data from a portion of said well which has been drilled. The trained neural network is further trained using training data comprising said fresh well data and associated seismic data from the subsurface volume and then used to determine inversion well data relating to a region of the reservoir comprising an intended well path for the well. The drilling strategy for at least a next portion to be drilled is optimized based on said inversion well data.