Feedforward for stabilization in the presence of constraints

It is well known that feedback is required for stabilization of open loop unstable systems. Feedforward from disturbances is normally used to improve control performance at high frequencies, beyond the achievable bandwidth for stable feedback control. This paper studies how disturbance measurements instead can contribute to avoiding input constraints, and thereby avoid the loss of stabilizing feedback caused by input constraints becoming active. The results are demonstrated both for simple linear systems and in a simulation study of an exothermic continuous stirred tank reactor.