TABLE OF CONTENTS Adding feedback measurement is essential for important process outputs because of the uncertainties and variability both in the process inputs and within the process. This is particularly true for bioreactors because of the unknowns in the metabolic pathways and kinetics of the cells. Using measurement in a feedback control loop offers automatic compensation. Since the proportional-integral-derivative (PID) controller is the predominant controller used in industry for basic feedback control, this chapter focuses on how to set up, tune, and optimize the PID.
Chapter 2 discussed how a control loop's ultimate performance depends on a process model but that the actual performance is determined by the PID controller's tuning settings. This chapter details the procedure for computing the PID tuning settings from the parameters of a process model. Though a dynamic model is implied in the tuning, as long as the PID controller is stable, it corrects for unknowns, loads, and disturbances.
In general, the user backs off from the "hottest" tuning settings for the "tightest" control in order to reduce the potential for oscillations in the process variable or manipulated variables. A tradeoff must always be made between performance and robustness. High controller gains transfer more variability from the process variable to the manipulated variable. Fortunately, bioreactor loops have fewer interactions than do many other unit operations, so variability in the manipulated variable is less disruptive. However, fluctuations in the controller output as a result of process or measurement noise that exceeds the final element's resolution can disturb...
