Real-time Permeability Estimation from Mud Gas Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating formation hydrocarbon mobility during well drilling are time-consuming and costly, as they require stopping drilling for sampling and measurement, resulting in non-productive time and delayed reservoir quality evaluation.
Innovation Solution
A computer-implemented method using machine learning to generate real-time estimates of formation hydrocarbon mobility from mud gas data, leveraging historical data to establish nonlinear relationships between gas measurements and permeability, enabling real-time permeability logging and decision-making.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional sampling and measurement procedures are used to determine formation permeability and fluid viscosity, then measurement precision is improved, but productivity deteriorates due to drilling stoppage and non-productive time
Solution Approach 1:
The patent replaces traditional mechanical sampling and laboratory measurement systems with a machine learning-based computational system. The ML model processes real-time mud gas data (compositional and volumetric measurements) to estimate formation permeability and hydrocarbon mobility continuously during drilling, eliminating the need to stop drilling for physical sampling and laboratory analysis.
Solution Approach 2:
The patent introduces machine learning models as an intermediary between mud gas measurements and formation property estimation. The ML models are trained on historical data containing both mud gas measurements and core-derived permeability values, then deployed to predict formation properties in real-time from ongoing mud gas data, serving as a bridge between available measurements and desired formation characteristics.
2Measurement precision
If core sampling and laboratory measurement are performed to obtain formation permeability, then measurement precision is improved, but loss of time increases due to post-drilling processing
Solution Approach 1:
The patent performs preliminary training of machine learning models using historical mud gas data and corresponding core-derived permeability values before actual drilling operations. This pre-training phase establishes the relationship between mud gas measurements and formation properties, enabling real-time predictions during drilling without requiring subsequent laboratory processing or waiting for core analysis results.
Solution Approach 2:
The patent substitutes the time-consuming mechanical and laboratory-based core sampling, transportation, and analysis process with a computational machine learning system that processes digital mud gas data in real-time, providing immediate permeability estimates without physical sample handling or laboratory processing delays.
3Productivity
If machine learning models are trained on historical mud gas data to estimate formation properties, then productivity is improved through real-time decision-making, but device complexity increases
Solution Approach 1:
The patent develops machine learning models with multi-functionality that can estimate multiple formation properties (permeability, hydrocarbon mobility, reservoir quality) from a single set of mud gas measurements. The same trained model serves multiple evaluation purposes, reducing the need for separate specialized systems for each property estimation and simplifying the overall operational workflow.
4Measurement precision
If traditional methods are used to evaluate reservoir quality, then measurement precision is improved through direct sampling, but loss of information increases due to delayed data availability
Solution Approach 1:
The patent enables continuous estimation of formation properties and reservoir quality throughout the drilling process by processing ongoing mud gas measurements through the trained machine learning model. This continuous evaluation provides real-time feedback on reservoir characteristics, maintaining an up-to-date understanding of formation properties without interruption or delay, unlike discrete sampling methods that provide only periodic snapshots.
Data Source
AI summary
Systems and methods include a method for generating a real-time permeability log. Historical mud gas-permeability data is received from previously-drilled and logged wells. The historical mud gas-permeability data identifies relationships between gas measurements obtained during drilling and permeability determined after drilling. A formation hydrocarbon mobility model is trained using machine learning and the historical mud gas data. Real-time gas measurements are obtained during drilling of a new well. A real-time permeability log is generated for the new well using the formation hydrocarbon mobility model and real-time gas measurements.