Formation Evaluation Synthesis From Surface Drilling Data

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

LWD tools for formation evaluation are expensive and can significantly increase the cost of drilling operations, making them economically unviable for some projects, particularly those with numerous wells in the same geographical region.

Innovation Solution

Utilize a trained machine learning model, such as a generative adversarial network (GAN), to synthesize formation evaluation data based on drilling mechanics data collected from surface sensors, reducing the need for downhole measurements during drilling.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If LWD tools are used to collect formation evaluation data during drilling, then the accuracy and reliability of formation evaluation data is improved, but the cost of drilling operations increases significantly

Engineering Contradiction:

Improveformation evaluation data accuracyVSAvoiddrilling cost

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent creates a virtual copy of LWD formation evaluation data by training a machine learning model on historical LWD data from offset wells. The trained model then generates synthetic formation evaluation data for the current well based on surface drilling mechanics data, eliminating the need for expensive physical LWD tools while maintaining data accuracy comparable to real LWD measurements.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent replaces expensive, complex LWD downhole tools with inexpensive surface sensors that collect drilling mechanics data. The machine learning model processes this readily available surface data to generate formation evaluation information, substituting costly specialized equipment with cheap, easily deployable surface-based alternatives.

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

2Measurement precision

If LWD tools are deployed to obtain formation evaluation logs, then the quality of formation evaluation data is improved, but the device complexity and operational complexity increase

Engineering Contradiction:

Improveformation evaluation data qualityVSAvoiddrilling system complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts the formation evaluation data generation function from the complex downhole LWD tool system and relocates it to a surface-based machine learning model. By separating the data collection (simple surface sensors) from the data generation (AI model processing), the system eliminates the need for complex downhole electronics, power systems, and measurement instruments while maintaining measurement precision through intelligent data synthesis.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent replaces the mechanical and electronic LWD measurement system with an information-processing system. Instead of using physical sensors downhole to directly measure formation properties, the system uses surface sensors to capture drilling mechanics data, then employs a machine learning model to computationally synthesize formation evaluation logs, substituting physical measurement infrastructure with intelligent algorithms.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If real-time formation evaluation data is collected using LWD tools, then the ability to control drilling trajectory and make real-time decisions is improved, but the cost and time required for drilling operations increases

Engineering Contradiction:

Improvereal-time drilling control capabilityVSAvoiddrilling cost

Core Design Contradiction:

ProductivityVSQuantity of substance

Solution Approach 1:

The patent performs preliminary training of the machine learning model using historical LWD data from offset wells before the actual drilling operation. This pre-training phase allows the model to learn the relationships between surface drilling mechanics data and formation evaluation data. During actual drilling, the pre-trained model can immediately generate formation evaluation data in real-time without requiring expensive LWD tools, enabling rapid real-time drilling control decisions at lower cost.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12585040B2Machine learning synthesis of formation evaluation data

Publication Date: 2026.03.24 SCHLUMBERGER TECH CORP

AI summary

An apparatus and method utilize a trained machine learning model to synthesize formation evaluation data such as formation tops and LWD logs. In some instances, the synthesis of formation evaluation data may further be based upon drilling mechanics data collected during drilling, thus effectively enabling formation evaluation data to be synthesized primarily based upon surface measurements collected in real time, and in many cases without the need for collecting downhole measurements during drilling. In addition, in some instances, a machine learning model implemented as a generative adversarial network (GAN) may be used to synthesize formation evaluation data, with drilling mechanics data collected during drilling also used in some instances.