Water-Cut Laser Sensing With Self-Calibrating Unsupervised Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing laser sensors require calibration for accurate water-cut determination in two-phase flows and cannot generalize to three-phase flows due to variable path lengths, and supervised machine learning methods are costly and time-consuming.
Innovation Solution
An unsupervised machine learning model, specifically an autoencoder system, self-calibrates to determine water-cut and gas fraction in a three-phase flow by using a training data set to estimate these values without labeled data, adapting to sensor drift and variable path lengths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If supervised machine learning methods are used to improve water-cut estimation accuracy, then measurement precision is improved, but the cost and time required for data collection and model training increase significantly
Solution Approach 1:
The system performs self-calibration by automatically training the machine learning model on field data without requiring manual intervention or pre-labeled calibration data. The model trains autonomously using unsupervised learning, eliminating the need for time-consuming manual data collection and labeling processes while maintaining accurate water-cut estimation
Solution Approach 2:
The system performs preliminary calibration actions automatically during the training phase, where the model learns from field data before being deployed for production use. This preliminary self-calibration eliminates the need for subsequent manual calibration operations and ensures the model is ready for accurate measurements from the outset
2Measurement precision
If existing laser sensors are used for two-phase flows, then measurement precision is improved, but the system cannot generalize to three-phase flows with variable path lengths
Solution Approach 1:
The system adapts to variable path lengths by dynamically adjusting the laser wavelength and modifying the machine learning model's input parameters based on the detected three-phase flow conditions. This allows the sensor to maintain measurement precision across different path length conditions while extending capability to three-phase flows
Solution Approach 2:
The machine learning model is designed to handle multiple flow types (two-phase and three-phase) and various flow conditions through a single unified model structure. The model processes different spectral signatures from water, oil, and gas phases simultaneously, providing universal applicability across different flow configurations without requiring separate calibration for each scenario
3Reliability
If manual calibration is performed to prevent sensor drift, then reliability is improved, but the ease of operation deteriorates due to required manual intervention
Solution Approach 1:
The system performs automatic self-calibration by continuously training the machine learning model on new field data without requiring manual intervention. The model autonomously updates its parameters to compensate for sensor drift, maintaining reliability while eliminating the need for operational manual calibration procedures
Solution Approach 2:
The system implements continuous feedback loops where the model processes new spectral data from the laser sensor, compares it against learned patterns, and automatically adjusts its parameters to correct for drift. This closed-loop feedback mechanism maintains sensor accuracy over time while requiring no manual operational input
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
The model provides accurate water-cut and gas fraction estimation in three-phase flows, continuously adapting to maintain sensor accuracy and prevent drift without manual calibration, thus improving oilfield management decisions.
Implementation Method 1
A water-cut laser sensor may be used to determine the water-cut in a fluid mixture
Data Source
AI summary
Systems and methods for a self-calibrating three-phase flow water-cut laser sensing using an unsupervised machine learning model are disclosed. The methods include creating a training data set, wherein the training data set comprises training mixture spectra; training, using the training data set, an unsupervised machine learning model to estimate an estimated water-cut and an estimated path-length fraction value, wherein, via the training, the unsupervised machine learning model calibrates itself to determine the estimated water-cut and the estimated path-length fraction value; obtaining an observed mixture spectrum from a water-cut laser sensor; estimating, using the trained unsupervised machine learning model, the estimated water-cut and the estimated path-length fraction value from the observed mixture spectrum; determining, from the estimated path-length fraction value, an estimated gas fraction value; and determining a composition of fluids in a separator using the estimated water-cut and the estimated gas fraction value.