A Model Way to Control Fermentation

Pressure Control: The pressure of the vessel is regulated by a PID loop. Constant pressure will be maintained normally, but the pressure can be activated for DO₂ control if the agitator and/or air flow are at maximum limits.

Waksman’s engineers encountered some problems with the data highway link between the computer and the external CPU. Data sampling time and heavy-duty tasking led to freeze-ups in the CPU. Eventually, a more distributed system was designed and implemented. The problem continued, albeit with less frequency. With a redesign in the external CPU software, the freeze-up problem should be resolved.

During the first six months of operation with the new control system we experimented with PID values in particular. PID control loops and an advanced model-predictive coordinated controller (CC) form the heart of the control system logic. The best solution we found was using the new system’s CC block to keep DO₂ content to desired levels.

Model-Predictive Coordinated Control

The CC block controls a single process variable by manipulating three different controller outputs. The goal of the CC block is three-fold: reject any disturbances to the process, respond quickly to setpoint changes, and optimize the available controller outputs during steady-state control. The CC block meets these goals by using two independent lists named "Active" and "Target." The Active List tells the CC block which controller output device is the strongest and most effective.

Since agitator speed, air flow and pressure affect the DO₂, the CC is used as the main controller to control DO₂. A typical control strategy requires the CC to use the agitator as primary control to maintain DO₂, followed by air flow and, finally, pressure. Each successive controller is activated when maximum setpoint operating values are obtained in the preceding controller. RPM, air flow and pressure can also be activated as singular, individual DO₂ controllers.

The Target List is used by the CC block to control the long-term steady-state value of each controller output. Simultaneously with the process variable growing towards its desired setpoint, the CC block drives all available controller outputs to their target values without upsetting the process. This means that target outputs can be set to the desired optimal and/or effective control conditions. If faulty conditions threaten to spoil a batch, strategies can be modified at any point during the fermenting process.

Ultimately, using CC has allowed Waksman the flexibility needed for faster and tighter control of DO₂ levels in relation to process requirements, and helped the facility better accommodate a larger number of clients and a more varied client base.

Ken Callanan is Supervisor, Cell & Cell Products Fermentation at Waksman Institute of Microbiology, Rutgers University.