Neural Network Impurity Removal Prediction for Parallel Reactors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The RDS process faces inefficiencies in measuring and removing impurities like sulfur, nitrogen, and CCR from high-sulfur heavy oil, leading to catalyst aging and increased costs due to the difficulty in calculating catalyst inactivity and impurity removal amounts.
Innovation Solution
A neural network-based prediction method is developed to learn from chemical process data, using two neural network models for parallel reactors to predict impurity removal amounts, incorporating an aging factor and weight clipping to ensure physicochemical consistency, and update daily using a receding horizon method.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional measurement and calculation methods are used to determine impurity removal amounts, then direct measurement of catalyst inactivity and removal amounts is attempted, but it is difficult to measure and utilize all the many variables and parameters required, leading to inefficiency
Solution Approach 1:
The patent replaces direct physical measurement of catalyst inactivity and impurity removal amounts with a neural network-based prediction system. The neural network model learns from operational data to predict removal amounts, eliminating the need for complex direct measurement systems while maintaining accuracy.
Solution Approach 2:
The patent introduces a neural network model as an intermediary between operational parameters and impurity removal amounts. This intermediary learns the complex relationships from data, simplifying the measurement process by predicting outcomes based on readily available operational variables rather than requiring direct measurement of all parameters.
2Reliability
If catalyst is used for extended periods to reduce replacement costs, then catalyst life is extended, but catalyst aging occurs due to deposition of metal elements, leading to reduced efficiency in impurity removal
Solution Approach 1:
The patent implements a feedback mechanism where the neural network continuously predicts impurity removal amounts based on current catalyst state and operational conditions. This feedback allows operators to monitor catalyst performance degradation in real-time and make informed decisions about optimal catalyst replacement timing, balancing extended use with maintained efficiency.
Solution Approach 2:
The patent uses the neural network to predict future catalyst performance based on current aging trends. This preliminary assessment allows operators to plan catalyst replacement proactively before complete deactivation occurs, optimizing the balance between extending catalyst life and maintaining removal efficiency.
3Measurement precision
If more variables and parameters are measured to improve prediction accuracy, then prediction precision is improved, but the complexity of data collection and processing increases
Solution Approach 1:
The neural network model serves multiple functions: it processes various operational parameters, predicts impurity removal amounts, and adapts to different catalyst aging stages. This multi-functionality allows the system to maintain high prediction accuracy using a unified approach that handles multiple variables without requiring separate complex measurement systems for each parameter.
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
This approach enables reliable prediction of impurity removal amounts, optimizing process conditions for efficient impurity removal, reducing catalyst replacement costs, and maintaining catalyst activity, with prediction errors averaging around 1-2.5% and ensuring adherence to physiochemical principles.
Implementation Method 1
making neural network models learn using data from a chemical process and predicting removal amounts of impurities removed in the chemical process using the learned neural network models
Implementation Method 2
The impurities are included in the feed supplied to reactors of the chemical process and are removed by reacting with catalysts disposed in the reactors
Implementation Method 3
The neural network models may be updated daily by applying a receding horizon method after learning for a certain period of time
Data Source
AI summary
An impurity removal amount prediction method in chemical process is provided. The impurity removal amount prediction method in chemical process according to the embodiments of the present invention comprises making neural network models learn using data from a chemical process and predicting removal amounts of impurities removed in the chemical process using the learned neural network models. The impurities are included in the feed supplied to reactors of the chemical process and are removed by reacting with catalysts disposed in the reactors. The reactors include a first reactor and a second reactor arranged in parallel. The neural network models include a first neural network model applied to the first reactor and a second neural network model applied to the second reactor. The neural network models learn in the direction to minimize the difference between the output data of the first neural network model and the second neural network model.