Multi-Layer DTBB Inversion for Geosteering Model Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional distance to bed boundary (DTBB) inversion techniques in hydrocarbon exploration are prone to inaccuracies due to incorrect initial formation models, leading to local minima and increased uncertainty, especially when dealing with multiple formation layers, as they rely on deterministic approaches and pre-selected layer numbers, which may not accurately represent the actual formation structure.
Innovation Solution
The implementation of multi-layer DTBB inversion using multiple initial guesses based on randomly sampled formation parameters, where only qualified models that fit actual measurements within a given error tolerance are selected for inversion, allowing for global convergence and improved formation evaluation and geosteering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional deterministic inversion techniques are used with pre-selected layer numbers, then the inversion process is computationally efficient, but the accuracy and reliability of formation modeling deteriorates due to local minima and incorrect initial assumptions
Solution Approach 1:
The patent applies preliminary action by generating multiple initial formation models with different layer configurations before performing inversion. This pre-preparation of diverse initial guesses allows the inversion process to start from multiple points in the solution space, increasing the likelihood of finding the global minimum rather than getting trapped in local minima, thereby improving formation model accuracy while maintaining computational efficiency through parallel processing of these preliminary models
Solution Approach 2:
The patent implements parameter changes by varying the number of formation layers and initial model parameters across multiple initial guesses. Instead of using a fixed pre-selected layer number, the system explores different layer configurations (e.g., 3-layer, 5-layer, 7-layer models) and selects the best-fitting model based on inversion results, thus resolving the contradiction between computational efficiency and modeling accuracy
2Adaptability or versatility
If the number of formation layers is increased to represent complex formations, then the model's ability to represent actual formation structure improves, but the inversion becomes more prone to local minima and ill-conditioning
Solution Approach 1:
The patent applies segmentation by dividing the inversion process into multiple independent inversion operations, each starting from a different initial model with a specific layer configuration. Instead of performing a single inversion with a complex high-layer model that is prone to instability, the system segments the problem into several inversions with different layer numbers (e.g., 3-layer, 5-layer, 7-layer) and selects the most reliable result, thus maintaining both adaptability to complex formations and inversion stability
Solution Approach 2:
The patent implements partial action by performing inversion with multiple layer configurations rather than attempting to use a single comprehensive high-layer model. This approach uses partial models (with fewer layers) as initial guesses that are easier to invert reliably, then refines the solution by comparing results across different layer numbers, achieving reliable inversion while still capturing complex formation structures
3Measurement precision
If gradient-free stochastic inversion techniques are used to optimize layer number, then the accuracy of layer number selection improves, but the computational time and resources required increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-generating a set of initial formation models with different layer configurations before the inversion process. This preliminary preparation allows the system to evaluate multiple layer numbers deterministically in parallel, avoiding the need for time-consuming gradient-free optimization during actual downhole operations, thus achieving accurate layer number selection with reduced computational time
Solution Approach 2:
The patent implements dynamics by making the layer number selection process adaptive and flexible. Instead of using a fixed deterministic approach or computationally expensive stochastic optimization, the system dynamically evaluates multiple initial models with different layer numbers and selects the best-fitting model based on inversion results, achieving a balance between accuracy and computational efficiency that adapts to the specific formation being analyzed
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 approach enables more efficient and accurate geosteering by refining formation property approximations across multiple layers, reducing noise and improving operational performance by selecting only qualified models for inversion, leading to better path adjustments during downhole operations.
Implementation Method 1
The transmitting antenna is used to generate electromagnetic fields in the surrounding formation. In turn, the electromagnetic fields in the formation induce a voltage in each receiving antenna.
Data Source
AI summary
System and methods for geosteering inversion are provided. A downhole tool's response along a path of a wellbore to be drilled through a formation is predicted over different stages of a downhole operation, based on each of a plurality of initial models of the formation. Each initial model represents a different number of formation layers over a specified range. The tool's actual response with respect to one or more formation parameters is determined, based on measurements obtained during a current stage of the operation. The actual response is compared with that predicted from each of the initial models. At least one of the models is selected as an inversion model, based on the comparison and a selection criterion. Inversion is performed for subsequent stages of the operation along the wellbore path, based on the selected model. The wellbore path is adjusted for the subsequent stages, based on the inversion results.