Robust tuning of multivariable systems with decentralized control and decouplers

In this report, the objective is to study how to tune PI controllers in a decentralized control structure for multivariable systems. Robustness with respect to (i) parameter changes, (ii) multivariable interactions including opening and closing other loops and (iii) input saturation are tuning criteria. The SIMC tuning rules are extended to the proposed ‘robust’ SIMC settings which takes expected worst case operation into account. Use of decouplers to maintain the robustness criteria are also discussed as well as tuning of the PI controllers when the decouplers are added to the control scheme. The proposed techniques are first tested on a simple multivariable system, thereafter an analytic model of a high pressure vessel is derived and the controllers are designed first for SISO cases and then for the complete multivariable vessel. Based on the different simulations, the proposed robust settings manage to control the systems in a satisfactory manner, both in the nominal case and in the worst case. Combined with a one-way decoupler all of the robustness criteras are met.