Three-Dimensional Electromagnetic Inversion Using Fast Neural Operators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computing technologies face inefficiencies and delays in processing large datasets from wellbore operations, leading to suboptimal decision-making during drilling and other wellbore activities, which can result in equipment damage, increased costs, and reduced operational efficiency.
Innovation Solution
Utilizing neural operators, such as Fourier Neural Operators (FNO) and Physics-Informed Neural Operators (PINO), to perform fast data inversion and evaluation of wellbore conditions, replacing computationally intensive iterative differential equations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional iterative differential equations are used to process wellbore data, then measurement precision is improved, but processing time increases significantly
Solution Approach 1:
The patent creates a surrogate model that copies the essential behavior of the complex forward model but executes much faster. The surrogate model is trained to replicate the responses of the complex electromagnetic forward model, allowing rapid inversion without repeatedly solving the complex differential equations. This copying approach maintains measurement precision while dramatically reducing processing time.
Solution Approach 2:
The patent replaces the mechanical/mathematical system of iterative differential equation solving with a machine learning-based surrogate model. Instead of using traditional numerical methods to solve Maxwell's equations iteratively, the system uses a pre-trained neural network that has learned the relationships between inputs and outputs, substituting complex computational mechanics with data-driven prediction.
2Measurement precision
If computationally intensive iterative differential equations are used, then measurement precision is improved, but productivity decreases
Solution Approach 1:
The patent performs preliminary action by pre-training the surrogate model offline using a comprehensive dataset generated from the complex forward model. This training phase captures the essential relationships between electromagnetic measurements and formation properties. Once trained, the surrogate model can be applied rapidly during actual wellbore operations, enabling real-time decision-making without the computational burden of iterative solving.
Solution Approach 2:
The surrogate model creates a computational copy that replicates the inverse problem-solving capability of the complex iterative system but executes orders of magnitude faster. This copying allows the system to maintain high productivity by processing formation property inversions in seconds rather than minutes or hours, directly improving wellbore operation efficiency.
3Productivity
If real-time data processing is implemented, then operational efficiency is improved, but device complexity increases
Solution Approach 1:
The surrogate model provides a simplified computational copy that maintains real-time processing capability while reducing algorithmic complexity. Instead of implementing complex iterative differential equation solvers in real-time, the system uses a pre-trained neural network that performs inference through simple forward propagation, significantly reducing the computational complexity of real-time decision-making systems.
Solution Approach 2:
The patent shifts complexity to the preliminary training phase rather than the execution phase. By pre-computing and storing the surrogate model during an offline training period, the system eliminates the need for complex real-time computations during actual operations. This separation allows real-time decision-making to be achieved with simpler, faster inference code.
Data Source
AI summary
Described herein are systems and techniques for monitoring for monitoring and evaluating conditions associated with a wellbore and wellbore operations that use neural operators instead of computationally intensive iterative differential equations. Such systems and techniques allow for determinations to be made as operations associated with a wellbore are performed. Instead of having to wait for computationally intensive tasks to be performed or take risks of proceeding with a wellbore operation without real-time evaluations being performed, these wellbore operations may be continued while determinations are timely made, thus improving operation of computing systems that perform evaluations and that make decisions regarding safely and efficiently performing wellbore operations such as drilling a wellbore, cementing wellbore casings in place, or injecting fluids into formations of the Earth.