Modeling Identification and IMC based PID Control of Multivariable Nonlinear Processes

Abstract

Design of efficient control systems is vital for any process industry for maintaining the product quality, meeting the safety needs, improving the energy efficiency and reducing the environmental pollution. The conventional Proportional Integral Derivative (PID) controllers are commonly used in majority (over 95%) of the process industries due to their simple configuration and wide range of applications. Tuning of PID controller is, however, a challenging task since it involves an in-depth understanding of both dynamic and static behaviours of process. Model based controller design techniques like the Direct Synthesis (DS) method and the Internal Model Control (IMC) method have come up as superior alternatives to the conventional PID controllers since they can be implemented within the PID controller framework without any additional hardware requirements. Moreover, the DS and Internal Model Control (IMC) based PID controllers have the added advantage of possessing only one tuning parameter as compared to three in the PID controller. A process transfer function, derived from an appropriate mathematical model is an inherent necessity for the design of model based control systems. The process modeling activity is broadly classified into two categories: (a) Theoretical modeling and (b) Process Identification. Theoretical models are based on first principles and rigorous in nature. An in depth understanding of the physical and chemical nature of the process is the primary requirement for the development of theoretical models. Process identification, on the other hand, involves development of empirical and black (purely box data driven) models, based on extensive experimental/plant data. From the controller design perspective, processes are categorized as Single Input Single Output (SISO) or Multiple Input Multiple Output (MIMO) processes. The Single Input Single Output (SISO) processes are simper to design since they have only one control loop involving one controlled variable (CV) and one manipulated variable (MV).