Closed-loop interaction and performance considerations for decentralized control of two-by-two multivariable systems

This paper analyzes process and control interaction in multivariable two-input two-output processes. It is well known that the off-diagonal terms in the matrix transfer functions introduce interaction effects that are to be taken into account at the controller design stage. When decentralized control is to be used, it is of primary importance to have knowledge of the constraints imposed by local designs into the behavior of the overall system. Special attention has to be paid to the potential instability caused by interaction effects generated by the control action on the other loop. This paper provides an analysis by means of an interaction measure defined in terms of the joint effect of process and control interaction. This interaction measure appears in a determinant way into the stability constraints as well as providing the limits for the desired closed-loop dynamics for each independent control loop.