Acoustic Velocity Estimation via Multi-Well Regression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating pressure and shear velocities in rocks beneath the Earth's surface, particularly in hydrocarbon-bearing formations, face inaccuracies and incompleteness in well log data due to the cost and complexity of coring operations, leading to data 'gaps' along the well length, which hinder the generation of accurate acoustic velocity models for seismic inversion.
Innovation Solution
A method utilizing well logging data from multiple wellbores, including Nuclear Magnetic Resonance and element composition scanning data, to establish a regression model, such as a neural network, for estimating primary and secondary acoustic velocities, which can then be applied to generate accurate velocity data sets for a target wellbore, enabling the creation of a three-dimensional subsurface image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If core samples are obtained along the entire length of the well to directly obtain rock samples for acoustic velocity determination, then measurement precision is improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent uses well log data from multiple wells as copies or proxies to infer acoustic velocity properties for the target well. Instead of directly measuring core samples from every location, the system creates a velocity model by copying information from nearby wells and using regression analysis to transfer knowledge across wellbores, thereby avoiding the need for extensive physical coring operations.
Solution Approach 2:
The patent introduces well log data (such as gamma ray, resistivity, and acoustic impedance logs) as intermediary variables that correlate with acoustic velocity. These intermediary measurements are easier to obtain than direct core samples, and they serve as proxies to infer the desired velocity properties through established relationships and regression models.
2Device complexity
If well log data is used to estimate acoustic velocities, then device complexity is reduced, but measurement precision deteriorates due to data gaps and one-dimensional limitations
Solution Approach 1:
The patent merges well log data from multiple different wells into a unified regression model. By combining datasets from several wellbores, the system compensates for the one-dimensional limitations and data gaps in individual wells, creating a more robust and accurate velocity estimation for the target well through multi-well integration.
Solution Approach 2:
The patent transitions from one-dimensional well log data (depth-only) to a three-dimensional velocity model by incorporating spatial relationships between multiple wells. The regression model uses coordinates and relative positions of multiple wellbores to extrapolate properties from measured locations to surrounding regions, adding spatial dimensions to the originally one-dimensional well log measurements.
3Productivity
If seismic inversion is performed with incomplete velocity models, then productivity is maintained, but measurement precision deteriorates leading to inaccurate hydrocarbon detection
Solution Approach 1:
The patent performs preliminary velocity model construction using well log data and regression analysis before conducting the main seismic inversion process. This preliminary action creates an improved initial velocity model that incorporates multi-well information, which then serves as a better starting point for seismic inversion, ultimately improving hydrocarbon detection accuracy without significantly delaying the overall workflow.
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 allows for the generation of accurate primary and secondary acoustic velocity data sets, even in complex formations like carbonate rocks, enhancing the resolution and quality of subsurface mapping and hydrocarbon detection by filling data gaps and improving the accuracy of seismic inversion results.
Implementation Method 1
The logging tool 1 may include a Nuclear Magnetic Resonance (NMR) logging tool
Implementation Method 2
A sonic tool sends an acoustic wave into the formation at one point (source) and the wave travels through the rocks until reaching one or more receiver on the sonic tool
Data Source
AI summary
A method of estimating the primary and secondary acoustic velocities, Vp and Vs, of formation surrounding a first wellbore includes obtaining well logging data for a multiplicity of other wellbores to collect, for each other wellbore a plurality of input data sets including at least a Nuclear Magnetic Resonance logging data set, and an element composition scanning data set. The operation further collects, for each other wellbore, at least one output data set including a primary and secondary velocity data set. The method incudes training or establishing at least one regression model using said input and output data sets, obtaining well logging data for said first wellbore to obtain a corresponding plurality of input data sets, and applying the obtained corresponding plurality of input data sets to the trained or established regression model to generate as an output of the regression model, an output data set for the first wellbore including a primary and secondary velocity data set.