A novel internal cascade control structure is proposed for the improved control of non-self regulating (integrating) processes. Like traditional cascade architectures, this structure consists of outer-loop and inner-loop controllers. In this instance, however, both loops compute the set point error using the same measured process variable. This structure significantly improves the set point tracking response of integrating processes as compared to the traditional single-loop architecture. It also improves rejection of Type I and Type II disturbances to a lesser degree.

This novel architecture uses an inner controller to stabilize the non-self regulating process, causing it to behave as a self regulating process. This is beneficial because the outer controller can then be tuned using any of the proven methods for self regulating processes. An inner loop stabilized as a self regulating process also enables the use of model-based architectures including Smith predictor, feed forward with feedback trim, and model predictive control (MPC).

The proposed internal cascade structure facilitates a sequential tuning procedure using the design method for tuning traditional cascade systems, beginning with the inner loop and followed by the outer loop. This contrasts with the internal feedback structure proposed by Sung and Lee (1996) that requires the use of tuning correlations specifically derived for that structure.

While other tuning correlations are readily applicable, this work demonstrates the use of the internal model control (IMC) correlation to tune the outer controller. The stability and robustness of the proposed architecture is demonstrated through mathematical analysis.

This novel architecture is demonstrated using a bench-scale pumped tank experimental system, a classical example of an integrating process, and a number of simulated industrial systems. The performance of the proposed internal cascade structure is compared with single-loop and internal feedback structures. Variation in controller tunings demonstrates the adaptability of the proposed architecture to real-world applications where the definition of optimal control is highly dependent upon the particular process under control.

Reference: Sung, S.W. and Lee, I-B. (1996). Limitations and Countermeasures of PID Controllers, Ind. Eng. Chem. Res., 35, 2596-2610.