Dynamic characteristics and real-time control of a particle-to-sCO2 moving bed heat exchanger assisted by BP neural network

Abstract

The transient response of a particle-to-sCO2 moving bed heat exchanger (MBHE) is investigated by a continuum model when perturbations are applied to the particle inlet temperature, sCO2 inlet temperature, and the sCO2 mass flow rate. Upon disturbance, the outlet temperatures of the sCO2 and particle flows deviate from the designed values and gradually approach a new steady state. The transient process can be described by exponential functions when the disturbance is applied to one of the input parameters while the transition becomes unpredictable when multiple input parameters are simultaneously varied. Two real-time control strategies are then adopted to maintain the sCO2 outlet temperature at its designed value. In strategy 1, the particle flow rate is adjusted; In strategy 2, the sCO2 flow is divided into two branches: one branch flows through the heat exchanger for heating, and the other enters a bypass. The values of the adjusting parameters are instantly determined based on the BP neural network. By setting 50 random disturbance cases, we demonstrate that the control strategy based on sCO2 bypass and BP neural network can effectively make the outlet temperatures of particle and sCO2 flows reach their designed values with minimal deviations.