Cascaded MPC for Plant-Wide Continuous and Batch Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial process control and automation systems face challenges in optimizing both continuous and batch operations, particularly in integrating intermediate components produced under different time frames, leading to inefficiencies and manual adjustments in production planning.
Innovation Solution
A cascaded Model Predictive Control (MPC) system is implemented, where a master MPC controller uses a planning model to perform plant-wide optimization, honoring constraints from slave MPC controllers, and receives proxy limit values to adjust process variables within safe operating limits, enabling real-time closed-loop optimization of both continuous and batch processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a master MPC controller performs plant-wide optimization including both continuous and batch operations, then productivity and profitability are improved, but device complexity and control system complexity increase
Solution Approach 1:
The control system is segmented into a master MPC controller for plant-wide optimization and multiple slave MPC controllers for individual unit optimization. Each slave controller manages specific continuous or batch operations, while the master controller coordinates them collectively, dividing the complex control task into manageable segments that can operate semi-independently.
Solution Approach 2:
The master MPC controller acts as an intermediary that receives proxy limits from slave controllers and translates plant-wide optimization goals into coordinated control actions. It mediates between the optimization needs of continuous operations and batch operations, balancing their competing requirements for resources and time.
2Manufacturing precision
If the master MPC controller integrates constraints from multiple slave MPC controllers, then manufacturing precision and constraint satisfaction improve, but measurement precision and information processing requirements increase
Solution Approach 1:
The slave MPC controllers provide proxy limits to the master controller, which are simplified representations or copies of their full constraint sets. These proxy limits capture the essential boundaries and capabilities of each slave controller without requiring the master controller to process complete constraint specifications, reducing information processing complexity while maintaining constraint satisfaction.
3Productivity
If real-time closed-loop optimization is implemented for both continuous and batch processes, then productivity increases, but loss of time for computation and coordination increases
Solution Approach 1:
Slave MPC controllers pre-compute and provide proxy limits to the master controller before the master optimization cycle begins. This preliminary action allows the master controller to receive constrained information in advance, reducing the computational burden during real-time optimization and minimizing computation time while maintaining coordinated control.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Constraints are received on initial components and intermediate components. Information is received on the products to be produced including a quantity of each of the products to be produced and a specification that specifies how the intermediate components are to be combined to form each of the products. An optimization is performed that includes the continuous conversion of initial components into the intermediate components as well as subsequent production of the products, subject to the constraints on each of the initial components, the constraints on each of the intermediate components, and the quantity of each of the products to be produced.