Enhancing pH control in a bioreactor through experimental system identification and dynamic analysis

The acidic by-products produced during fermentation can cause a drop in pH, which in turn affects the microorganisms' growth and the product's formation. In order to keep pH at the desired level, process control becomes necessary. The aim of this study is to develop a predictive model for pH behavior during the fermentation of Clostridium acetobutylicum through dynamic analysis and system identification. The First Order Plus Dead Time (FOPDT) model and the second-order Autoregressive Moving Average with Exogenous (ARMAX) model were the two approaches that were compared. While the FOPDT model was used to derive the PID controller parameters through transient analysis, the Smith and linear regression methods, the ARMAX model—identified with the Recursive Least Squares (RLS) method—was chosen for its better accuracy in capturing input-output dynamics. PID tuning was done with the Cohen-Coon method. The simulation results showed that setpoint tracking was successfully done, and the ARMAX model provided a more accurate representation of the system. The optimized PID controller recorded the minimum Integral of Squared Error (ISE) value of 50.82. This study points out effective modeling and control strategies for the production of stable pH during fermentation, thus providing very useful knowledge for other bioprocesses that require precise control.