Distillation Column MPC Using Transformed Temperatures and Feedforward
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Model predictive control systems face challenges in controlling distillation columns with multiple steady-state temperature profiles and require complex tuning due to the interdependence of manipulated variables on multiple controlled variables, leading to difficulties in maintaining product purity.
Innovation Solution
A method is introduced where a model predictive controller uses transformed temperatures and feed forward variables to enhance control aggressiveness and simplify tuning by decoupling the effects of manipulated variables on controlled variables, allowing for more precise control of reflux and feed flow rates in distillation columns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If model predictive control is used to control distillation columns with multiple steady-state temperature profiles, then product purity can be maintained, but the controller requires complex tuning due to interdependence of manipulated variables on multiple controlled variables
Solution Approach 1:
The patent segments the control problem by introducing selective measurement of temperature changes at specific locations in the distillation column. Instead of treating the entire column temperature profile as a single controlled variable, the system divides it into multiple measurement points, each providing independent information about different sections of the column. This segmentation reduces the interdependence between manipulated variables and controlled variables, simplifying controller tuning while maintaining product purity.
Solution Approach 2:
The patent introduces an intermediary measurement system that indirectly monitors the effect of manipulated variables on product purity. By measuring temperature changes at intermediate points in the column rather than directly measuring product composition, the system creates a simplified control loop. This intermediary measurement approach reduces the complexity of tuning while still ensuring product purity through the relationship between temperature profiles and separation efficiency.
2Reliability
If traditional model predictive control is used, then controlled variables can be predicted based on manipulated variable changes, but the response is too conservative and cannot aggressively maintain product purity in columns with multiple steady-state profiles
Solution Approach 1:
The patent applies local quality by implementing different control strategies for different sections of the distillation column. By selectively measuring temperature changes at specific locations rather than treating the column uniformly, the controller can apply more aggressive control actions in critical sections while maintaining stability in other areas. This localized approach enables aggressive product purity maintenance without requiring overly conservative overall control.
Solution Approach 2:
The patent implements preliminary action by continuously monitoring temperature changes at multiple points in the column to detect early deviations from the desired temperature profile. This early detection allows the controller to take preventive control actions before product purity is compromised, enabling more aggressive control strategies that maintain purity without waiting for significant deviations to occur.
3Adaptability or versatility
If multiple manipulated variables are used to control the distillation column, then control flexibility is improved, but the interdependence of these variables on multiple controlled variables makes tuning difficult
Solution Approach 1:
The patent segments the control variables by assigning specific temperature measurement points to control specific manipulated variables. This segmentation creates a more structured control architecture where the relationship between manipulated variables and controlled variables is simplified, reducing tuning difficulty while preserving the flexibility to adjust multiple parameters for optimal performance.
Data Source
AI summary
A method of controlling a distillation column having control valves to control both reflux and the vapor rate within the column. In accordance with the present invention, a temperature sensed in a top section of the column is magnified and utilized within the model predictive controller so that control is more aggressive as temperatures increase beyond a threshold temperature. Additionally, in the distillation column, or in fact in any other system in which two or more manipulated variables control two or more common controlled variables, special modeling techniques are utilized to make controller tuning easier to accomplish. In such modeling techniques, each manipulated variable is assumed to be able to have an effect on a controlled variable by a single step response model and other step response models are utilized so that the other manipulated variable(s) that also would have an effect on the same controlled variable are taken into account by the controller as feed forward variables.