(264a) Control of Networks: The Two Port Approach

I will describe a new approach for modeling complex process networks which is based on the conjugate two-port representation. One class of ports represents flow of physical variables and allows the connection of process systems together to form complex networks of physical devices (flowsheets). The second class of ports represents how the process system is connected with the information system and it allows for characterization of the information flow, signal processing, process control and optimization. The fact that the underlying algebraic structures for process and signal flow variables are quite different is largely ignored in the current literature on process control.

The main advantage of a two-port, rather than the current one-part approach, is that information and process flow are modeled as different mathematical constructs. In thermodynamics this can be viewed as the difference between extensive and intensive variables, whereas in electrical circuits it can be viewed as the difference between the current ant the voltage. Current actually flows, whereas voltage represents a signal. Process flows obey conservation laws and do not have a-priori defined directionality. Information flow does not obey conservation principles but have directionality. These differences lead to a set of important fundamental differences which has consequences for how model the process flows and signal flows. The theory also shows why it is difficult to model chemical processes in a natural way using a block-diagram oriented language like Simlink whereas it is much easier to model process systems using the port-based system Modelica.

In the presentation I will review a number of different approaches to modeling physical and non-physical systems including the Network theory of Gilles, the thermodynamic theory of Ydstie and Alonso, the bond-graph based methods, methods based on block-diagram algebra (like Simulink) and the port Hamiltonian approaches. I believe it is important for the chemical engineering process control community to come to grips with all these different approaches as we seek a unified and universal framework to model the integration of process control, optimization and physical systems in order to achieve safe and well managed chemical process systems.