Conexiant’s news site is now MDSpire News. Learn more
Advertisement
What Will It Take to Control Continuous Biomanufacturing?
Review examines how PAT, model-based control, reinforcement learning, and digital twins could maintain product quality from perfusion culture through downstream purification
The review characterizes continuous biomanufacturing as a process-control and system-integration challenge requiring coordination among sensors, models, and control systems.
2
PID control suits stable, well-measured variables, whereas model predictive control can manage interacting variables, delayed measurements, and operating constraints.
3
Process analytical technology supports control, but sensor cost and calibration drift can limit in-line and online measurements; at-line sampling may suit slower processes.
4
Soft sensors can estimate unmeasured attributes but remain vulnerable to sensor noise and model error, which could prompt inappropriate control actions.
5
The authors propose introducing digital twins gradually, with advanced control justified by measurable gains in product recovery, consistency, or risk management.
