An optimization based output power regulation in small modular reactors

Abstract

Efficient controller tuning for novel designs of nuclear reactors is an ongoing research issue. An offline optimal PI regulator tuning has been accordingly addressed herein for the power control of a sodium-cooled small modular reactor. A dynamic model has been developed employing a 1-Fuel/2-Coolant (1F/2C) layout for the core and a permanent magnet dc motor as the control rod governor. A state space representation is characterized and the pertaining stability region is delineated in the (kP,kI) plane for the plant linearized at the nominal operating point. Upon selection of an appropriate search scope within this zone, Differential Evolution (DE) and Firefly optimization algorithms are conducted in parallel to minimize certain error-based performance measures for the closed loop system. The step response for the closed loop nonlinear plant is resorted in this regard which would yield optimal PI pairs for each performance index. The two optimization schemes are thus compared in terms of convergence behavior and it can be figured out that the DE approach outperforms the Firefly scheme in this work. A less rippled convergence regime steadily stabilized at the optimal point is inspected in the former strategy. The latter method though quickly situates at the proximity of the optimal location, experiences persistent fluctuations therein which calls for a deeper search space comprising more samples.