Artificial Neural Network for Downhole Fluid Property Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Downhole fluid analysis tools are limited by extreme conditions, restricting the measurement of fluid properties to a small subset of those obtainable in conventional laboratory analysis, and struggle to accurately predict properties like PVT characteristics in real-time.
Innovation Solution
The use of an artificial neural network (ANN) to predict and estimate fluid properties, such as gas/oil ratio and formation volume factor, by learning correlations between input data from downhole fluid analysis tools, including concentrations of methane and ethane, and generating uncertainty values for accuracy assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If downhole fluid analysis tools are used to measure fluid properties in real-time, then productivity and decision-making speed are improved, but measurement precision and reliability of fluid property data deteriorate due to extreme downhole conditions
Solution Approach 1:
The patent introduces an artificial neural network as an intermediary computational model that bridges the gap between limited downhole measurements and comprehensive fluid property prediction. The ANN processes available downhole data (temperature, pressure, basic composition) and generates accurate predictions for properties that cannot be directly measured in the downhole environment, thereby maintaining measurement precision while enabling real-time analysis capability
Solution Approach 2:
The patent replaces complex mechanical measurement systems with computational modeling. Instead of using sophisticated physical instruments that would be too large and power-intensive for downhole deployment, the system uses software-based neural networks to predict fluid properties, substituting physical measurement mechanisms with information processing mechanisms
2Measurement precision
If the sophistication of DFA measurement tools is increased to measure more fluid properties, then measurement precision is improved, but device complexity and power consumption increase, making them unsuitable for extreme downhole conditions
Solution Approach 1:
The patent segments the fluid analysis function into two parts: a simple downhole measurement component that collects basic data (temperature, pressure, composition), and a separate computational component (neural network) that performs the complex analysis. This segmentation allows the downhole tool to remain simple and suitable for extreme conditions, while still achieving comprehensive fluid property measurement through the computational model
Solution Approach 2:
The patent creates a computational copy of the complex fluid property relationships that would exist in a surface laboratory. The neural network is trained on extensive laboratory data and then deployed downhole to replicate laboratory-quality analysis using only simple downhole inputs, effectively copying the analytical capability without requiring the physical complexity of laboratory equipment
Data Source
AI summary
Apparatus and methods to perform downhole fluid analysis using an artificial neural network are disclosed. A disclosed example method involves obtaining a first formation fluid property value of a formation fluid sample from a downhole fluid analysis process. The first formation fluid property value is provided to an artificial neural network, and a second formation fluid property value of the formation fluid sample is generated by means of the artificial neural network.