High performance processes require design that goes close to design boundaries and specifications while still guaranty robust performances. In order to go tighter to the boundaries of process performances much attention was been devoted to integrated design and control in which decision about the dynamics as well the design are taken by optimal fashion. Rigorous methods solving design and control lead to very challenging mathematical formulation which often becomes intractable. Our group presented a novel approach which embeds control decisions in search for optimal design performances under uncertainty. It does so by working in a reduced work space in which key design variables from the system are determined while embedded control decisions are taken in a close to optimal fashion. The two stage optimization therefore reduce drastically the number of decision variables such as pairing off manipulated and control also reduces the state space and make some problems of industrial interest tractable for computationally analysis that previously were not been able to solve.

In this presentation we will overcome limitations of early approach based on diagonal identification and expand the adaptive identification to a full state representation. This state representation will still show to be computably very efficient using recursive techniques. Case studies will demonstrate the increase performances of the adaptive full state identification and illuminate problem with highly correlated multi variable systems which were previously very difficult to address. We will also present methods to illuminate the offsets and by that improve the quality of previous approach to integrated design and control.