A Thermodynamic Approach to Measurement Selection for Process Networks

A thermodynamic approach is proposed to design decentralized controllers to guarantee the structural stability of chemical plants. Chemical plants are presented as inventory networks using mass and energy balances. A hierarchical decentralized inventory controller is developed to ensure that all the states of a chemical plant converge to their desired stationary points. We elaborate the impact of the process measurement and controlled variables selection on inventory controllers. We show the existence of a strongly concave entropy function for a single phase system. The strong concavity condition guarantees a one to one correspondence between the process state and its conjugate variable. The bijection provides an upper and a lower bound for the Lyapunov function. The Lyapunov function is thus a proper storage function.