Fault-tolerant spatial control of a large pressurised heavy water reactor by fast output sampling technique

A method is presented to design a spatial control system of a large pressurised heavy water reactor (PHWR) with the constraint that the closed loop system be stable even if one sensor or actuator has failed. Linear state models are obtained corresponding to the normal operating condition and different failure modes of the reactor. Each model is found to possess the two time scale property. Using similarity transformation each two time scale model is converted into a block diagonal form by which two subsystems, namely a fast subsystem and a slow subsystem, are easily obtained. State feedback is designed separately for the slow and the fast subsystems for each model. Then a composite state feedback gain is obtained from the state feedback gains computed for the slow and fast subsystems separately for each model. The composite state feedback so designed assigns the poles at arbitrary locations for the respective models. These composite state feedback gains are realised simultaneously by fast output sampling gains. Thus the states of the system are not needed for feedback purposes. An LMI formulation is used to overcome the undesired effects of poor error dynamics and noise sensitivity which are encountered if state feedback is realised exactly.