Time-Optimal Control of Reactor Power

Abstract

Control laws that permit adjustments in reactor power to be made in minimum time and without overshoot have been formulated and demonstrated. These control laws, which are derived from the standard and alternate dynamic period equations, are closed-form expressions of general applicability. These laws were deduced by noting that if a system is subject to one or more operating constraints, then the time-optimal response is to move the system along those constraints. Given that nuclear reactors are subject to limitations on the allowed reactor period, a time-optimal control law would step the period from infinity to the minimum allowed value, hold the period at that value for the duration of the transient, and then step the period back to infinity. The change in reactor power would therefore be accomplished in minimum time. This particular response can be achieved by exploiting the balance between the prompt and delayed neutron populations. Specifically, the reactor period can be made to change essentially as a step by initiating the transient with a very high rate of change of reactivity (prompt effect) and then cutting back on that rate as the delayed effects build in. The resulting control laws are superior to other forms of time-optimal control because they are general-purpose, closed-form expressions that are both mathematically tractable and readily implemented. Moreover, these laws include provisions for the use of feedback. The results of simulation studies and actual experiments on the 5 MWt MIT Research Reactor in which these time-optimal control laws were used successfully to adjust the reactor power are presented.