Perfusion Bioprocess Control With Predictive Chromatography Coupling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing bioprocess control methods, particularly in perfusion bioreactors, struggle with cyclic behavior and inability to adapt to metabolic shifts, leading to unstable process performance and inefficiencies in cell culture and downstream product purification.
Innovation Solution
A model predictive control approach that integrates a dynamic model of the bioprocess with a predictive model of downstream chromatographic processes to optimize manipulated variables, considering operational constraints and economic costs, using a state observer and machine learning for metabolic parameter estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If simple PID control loops are used to control feed flow rate, bleed flow rate, and harvest flow rate, then the control implementation is straightforward and easy to operate, but the process exhibits cyclic behavior and cannot adapt to metabolic shifts, resulting in unstable process performance
Solution Approach 1:
The patent transitions from fixed-parameter PID control to dynamic parameter control through model predictive control. The system continuously adjusts control parameters based on real-time state estimates and predicted process behavior, enabling adaptation to metabolic shifts while maintaining stability. The objective function and constraints are dynamically updated based on chromaticography predictions, allowing the system to respond to changing process conditions without cyclic behavior.
2Ease of operation
If simple PID control loops are used to control feed flow rate, bleed flow rate, and harvest flow rate, then the control implementation is straightforward and easy to operate, but the process exhibits cyclic behavior and cannot adapt to metabolic shifts, resulting in unstable process performance
Solution Approach 1:
The patent implements a closed-loop feedback system where state observers continuously estimate process variables, and model predictive control uses these estimates to predict future behavior and adjust control actions. The system incorporates feedback from chromaticography predictions to adapt to metabolic shifts, creating a responsive control mechanism that maintains adaptability while managing complexity through structured algorithms.
3Reliability
If model predictive control is used to determine adjustments of manipulated variables, then process stability is improved and cyclic behavior is eliminated, but the system complexity increases due to integration of dynamic models and optimization algorithms
Solution Approach 1:
The patent introduces state observers as intermediary components that bridge the gap between simple measurements and complex control requirements. These observers estimate unmeasured process variables and provide refined state information to the model predictive control system, reducing the direct complexity burden. The chromaticography predictions serve as another intermediary layer that translates process states into actionable constraints and objectives, managing system complexity through structured intermediate processing steps.
4Ease of manufacture
If upstream cell culture process is operated without coordination with downstream product purification, then the upstream process can be optimized independently for cell growth and productivity, but the overall system efficiency is reduced due to mismatches in operational parameters
Solution Approach 1:
The patent merges upstream cell culture control with downstream chromaticography optimization into a unified model predictive control system. The chromaticography predictions provide feedback to the upstream process control, creating coordinated operation where both processes are optimized simultaneously. The objective function integrates both upstream productivity goals and downstream operational constraints, ensuring that feed flow rate, harvest flow rate, and other manipulated variables are adjusted to maximize overall system efficiency rather than just upstream performance.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
The present disclosure provides methods, systems and products for controlling a bioprocess comprising a cell culture in a bioreactor, wherein the bioprocess is operated as a perfusion bioprocess. the methods use a model predictive control to identify values of manipulated variables that optimise an objective function, the objective function including one or more constraints and/or one or more terms that are associated with predictions obtained from a method of simulating a chromatographic process that is coupled with the bioprocess for processing of a harvest stream of the bioprocess.