Wax Risk Prediction Using Machine Learning and Pipe Flow Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting wax deposition in hydrocarbons are time-consuming and unreliable, leading to inefficiencies and high costs in hydrocarbon production, transportation, and refining due to the use of a trial-and-error approach for selecting wax inhibitors, which often fail to perform effectively with real fluids.
Innovation Solution
A method using advanced data analytics and pipe flow modeling to rapidly evaluate wax risks by analyzing hydrocarbon samples with machine learning algorithms and chemical additive predictive models, enabling precise prediction of wax deposition and selection of optimal inhibitors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional trial-and-error methods are used to select wax inhibitors, then chemical selection can be performed, but the process is time-consuming and often fails to perform effectively with real fluids
Solution Approach 1:
The patent applies preliminary action by performing comprehensive chemical characterization and predictive modeling before field testing. The system analyzes crude oil composition, predicts wax inhibitor performance using machine learning models, and screens candidates in silico before bench-top validation, thereby reducing both time and improving reliability of chemical selection
Solution Approach 2:
The patent replaces the mechanical trial-and-error approach with data-driven machine learning models. The system uses computational algorithms to predict wax inhibitor effectiveness based on crude oil composition and environmental conditions, substituting physical experimentation with virtual screening to accelerate the selection process while maintaining or improving accuracy
2Measurement precision
If bench-top tests are conducted to evaluate wax inhibitor performance, then chemical effectiveness can be assessed, but the process requires significant time and multiple tests
Solution Approach 1:
The patent performs preliminary computational screening and crude oil characterization before conducting bench-top tests. Machine learning models predict which inhibitors are most likely to succeed, allowing the experimental program to focus only on the most promising candidates, thereby maintaining measurement precision while significantly reducing overall testing time
Solution Approach 2:
The patent implements a tiered testing approach where not all possible inhibitor-c Crude combinations are tested experimentally. Instead, the system performs partial testing on selected candidates based on predictive model rankings, achieving sufficient evaluation accuracy without the excessive time required for exhaustive testing of all possibilities
3Ease of manufacture
If synthetic waxy fluids are used for comparative evaluations, then tests can be performed more easily, but the results do not accurately reflect performance with real crude oil
Solution Approach 1:
The patent introduces machine learning predictive models as an intermediary between synthetic fluid tests and real crude oil performance. The system uses these models to translate and adjust results from simplified synthetic tests to predict actual performance in real crude oil, thereby maintaining the ease of conducting synthetic tests while recovering predictive accuracy through computational correction
Solution Approach 2:
The patent applies parameter changes by adjusting the interpretation of test results based on crude oil composition parameters. The machine learning system modifies predicted performance metrics according to specific crude oil characteristics such as composition, temperature, and pressure conditions, thereby adapting results from standardized tests to reflect real-world performance in diverse crude oil systems
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Described herein are systems and methods for evaluating and mitigating the wax risks of a given hydrocarbon composition such as crude oil. The disclosed systems and methods enable rapid and ready prediction of wax risks using algorithms based on a small sample of the hydrocarbon composition. The wax risks are predicted using predictive models developed from machine learning. The disclosed systems and methods include mitigation strategies for wax risks that can include chemical additives, operation changes, and/or hydrocarbon blend.