FCC Reactor Severity Control for Volume Gain Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing control systems struggle to accurately predict and optimize volume gain in fluid catalytic cracking units due to highly dynamic and non-linear relationships, making it difficult to maximize product volume while accounting for reactor severity and environmental constraints.
Innovation Solution
A neural network-based control system that uses a processing circuit to calculate volume gain by comparing input and output oil volumes, generating a target severity to optimize temperature modulation in the reactor, and incorporates a neural network model trained with historical data to predict and adjust reactor conditions for optimal performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional control systems are used to manage fluid catalytic cracking units, then the system structure is simple and easy to operate, but the system cannot accurately predict and optimize volume gain due to highly dynamic and non-linear relationships
Solution Approach 1:
The patent replaces traditional mechanical control systems with a neural network-based predictive control system. The neural network model processes historical and real-time data to predict volume gain with high accuracy, substituting the simple but inaccurate traditional control approach with a complex computational intelligence system that can handle non-linear relationships.
Solution Approach 2:
The patent introduces a neural network model as an intermediary between the physical cracking process and the control system. This intermediary processes complex non-linear relationships and provides accurate predictions of volume gain, enabling optimized control decisions without requiring direct complex modeling of the physical system.
2Productivity
If a neural network-based control system is implemented to optimize volume gain, then prediction accuracy improves, but the device complexity and training requirements increase
Solution Approach 1:
The patent performs preliminary training of the neural network model using historical data before deployment. This offline training phase prepares the model to make accurate predictions during online operation, allowing the system to achieve high productivity once trained without requiring complex real-time computations during actual cracking operations.
Solution Approach 2:
The patent implements a feedback mechanism where the neural network model continuously receives real-time operational data and adjusts its predictions accordingly. This feedback loop enables the system to adapt to changing conditions and maintain high productivity while managing the complexity through iterative learning rather than complex real-time control logic.
3Quantity of substance
If reactor severity is increased to maximize product volume, then volume gain increases, but environmental constraints and system stability may be compromised
Solution Approach 1:
The patent uses the neural network model to predict volume gain as a function of reactor severity parameters, enabling the system to identify the optimal parameter setting that maximizes product volume while staying within environmental constraints. The model processes multiple parameters simultaneously to find the best operating point.
Solution Approach 2:
The patent applies partial optimization by focusing the neural network model specifically on predicting volume gain rather than optimizing all possible parameters. This selective approach allows the system to increase product volume through targeted severity adjustments without unnecessarily pushing environmental constraints to their limits.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A control system for automatic operation of a fluid catalytic cracking unit includes a reactor severity control device operable to modulate a temperature affecting volume gain within the fluid catalytic cracking unit and a controller. The controller includes a processing circuit configured to calculate the volume gain within the fluid catalytic cracking unit by comparing a volume based on one or more input oil feeds to the fluid catalytic cracking unit to a volume of one or more output oil products of the fluid catalytic cracking unit. The processing circuit is further configured to use a neural network model to generate a target severity predicted to optimize the volume gain within the fluid catalytic cracking unit. The processing circuit is further configured to operate the reactor severity control device using the target severity to modulate the temperature affecting the volume gain within the fluid catalytic cracking unit.