Chemical Formulation Testing With ML-Based Pre-Test Recommendation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The selection and development of specialty chemicals for oil and gas production, such as demulsifiers, corrosion inhibitors, and defoamers, is typically an empirical process due to complicated formulation and application scenarios, lacking a systematic and efficient approach.
Innovation Solution
A computing device with a tester interface and pre-test recommendation module that utilizes machine learning algorithms to analyze historical test results, cluster candidate chemical formulations, and train predictors to generate optimized formulations, reducing the empirical nature of the process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If empirical process is used for specialty chemical development, then flexibility in handling complex formulation scenarios is maintained, but development efficiency and time consumption deteriorate
Solution Approach 1:
The system performs preliminary actions by pre-processing historical test data, building knowledge graphs, and training machine learning models in advance. This allows the system to quickly retrieve and analyze relevant information during the chemical development process, significantly improving development efficiency without sacrificing the ability to handle complex formulation scenarios.
Solution Approach 2:
The patent introduces an AI assistant as an intermediary between the complex formulation scenarios and the development process. The AI assistant uses knowledge graphs and machine learning models to interpret complex scenarios, retrieve relevant historical data, and provide development recommendations, thereby improving efficiency while maintaining flexibility.
2Measurement precision
If traditional testing methods are used, then comprehensive performance evaluation is achieved, but time consumption and resource usage increase
Solution Approach 1:
The system applies partial action by using machine learning models to predict performance indicators for only the most promising chemical formulations identified through knowledge graph analysis. This selective approach maintains measurement precision for key candidates while significantly reducing the overall time and resources required compared to testing all possible formulations.
Solution Approach 2:
The system implements feedback mechanisms where test results from actual experiments are continuously fed back into the knowledge graph and used to retrain machine learning models. This improves prediction accuracy over time, allowing for more precise performance evaluation with reduced testing requirements.
Data Source
AI summary
Technologies for specialty chemical development and testing include devices and methods for receiving a test description. The test description is indicative of test parameters for a test of a chemical formulation, which may be an oil field specialty chemical. The devices and methods may include searching a database of historical test results based on similarity to the test parameters to generate multiple candidate chemical formulations. The devices and methods may cluster the candidate chemical formulations with an unsupervised machine learning algorithm to select a representative chemical formulation for each cluster. The devices and methods may include training a predictor based on test results using a supervised machine learning algorithm. Multiple virtual formulations may be generated and performances of each virtual formulation may be predicted with the predictor. Other embodiments are described and claimed.