Autonomous reactor control using model based predictive control for space propulsion applications

Reliable reactor control is important to reactor safety, both in terrestrial and space systems. For a space system, where the time for communication to Earth is significant, autonomous control is imperative. Based on feedback from reactor diagnostics, a controller must be able to automatically adjust to changes in reactor temperature and power level to maintain nominal operation without user intervention. Model-based predictive control (MBPC) is investigated as a potential control methodology for reactor start-up and transient operation in the presence of either a constant or a time varying external source. Bragg-Sitton and Holloway [Bragg-Sitton, S.M., Holloway, J.P., 2004. Reactor start-up and control methodologies. In: El-Genk, M. (Ed.), Proceedings of the Space Technology and Applications International Forum (STAIF-2004), AIP Conference Proceedings 699, pp. 614-622.] assessed the applicability of MBPC to reactor start-up from a cold, zero-power condition in the presence of a time-varying external radiation source, where large fluctuations in the external radiation source can significantly impact a reactor during start-up operations. Here the MBPC algorithm is applied using the point kinetics model to describe the reactor dynamics, with a single group of delayed neutrons and a fast neutron lifetime of 10-7 s. Controller stability is assessed by carefully considering the dependencies of each component in the defined cost (objective) function and its subsequent effect on the selected 'optimal' control maneuvers. Additional analysis demonstrates the effectiveness of the controller when a lower fidelity reactor kinetics model is adopted for the model system versus using a full six-group delayed neutron representation in the point kinetics equations to represent the 'real' system operation