Geological Attribute Estimation Using Similarity Embeddings

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for estimating geological attributes of prospects in the energy industry are labor-intensive and inaccurate, particularly when dealing with sparse data, and require extensive physical explorations.

Innovation Solution

Implementing a machine learning model, such as an autoencoder, to reduce dimensionality and compare embeddings of field and prospect datasets, enabling accurate estimation of attributes based on similar fields.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional methods are used to estimate geological attributes of prospects, then physical explorations can be conducted, but the process is labor-intensive and inaccurate particularly when dealing with sparse data

Engineering Contradiction:

Improveaccuracy of estimating geological attributesVSAvoidlabor-intensive process requiring extensive physical explorations

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent creates digital copies (embeddings) of geological datasets from mature fields and uses these copies to estimate attributes of prospects. Instead of conducting extensive physical explorations, the system copies relevant information from similar, well-studied fields through machine learning models, thereby reducing time and labor while maintaining accuracy even with sparse data.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent replaces manual, labor-intensive physical exploration methods with an automated machine learning system. The mechanical process of conducting field surveys and manual data analysis is substituted by an electronic system that uses trained models to automatically estimate geological attributes, significantly reducing time loss while improving measurement precision.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Productivity

If dimensionality reduction is applied to datasets representing physical characteristics of volumes, then comparison efficiency is improved, but data processing complexity increases

Engineering Contradiction:

Improvedata processing efficiency and accuracyVSAvoidcomplexity of machine learning model processing

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent applies dimensionality reduction techniques in advance to transform complex geological datasets into simplified embeddings before comparison. By pre-processing the data into a lower-dimensional representation, the system reduces the complexity of subsequent comparisons while maintaining the essential characteristics needed for accurate attribute estimation, thereby improving overall productivity.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12572851B2Automated similarity measurement and property estimation

Publication Date: 2026.03.10 SCHLUMBERGER TECH CORP

AI summary

A method includes receiving an input dataset representing one or more physical characteristics of a volume, generating an embedding by reducing a dimensionality associated with the input dataset using a trained machine learning model, comparing the embedding with a plurality of other embeddings generated by reducing a dimensionality of other datasets representing one or more physical characteristics of other volumes, selecting one or more of the other embeddings of the one or more other datasets based at least in part on comparing, and estimating one or more attributes of the volume based at least in part on the one or more other datasets corresponding to the selected one or more of the other embeddings.