Control System Design for a Solar Receiver-Reactor

Abstract

Solar-thermal receiver-reactor (RR) units are being developed to harness the technology's potential for use in both electrical power generation and materials processing industries. However, there is still substantial work needed to commercialise the technology and a key facet to this is the design of suitable control systems. The objective of this thesis was to describe a control system design methodology for a solar-thermal RR. A prototype RR currently being developed for application within a novel concentrated solar power (CSP) system is taken as the case study. The prototype RR has been manufactured and is in the process of being commissioned for testing within the Australian National University (ANU) high-flux solar simulator (HFSS) facility. A system-level description of the novel concentrated solar power (CSP) station was used to highlight the general control objectives, which were found to be: (1) the control of exhaust temperature and; (2) the maximisation of solar-to-thermochemical power efficiency. Consideration was given to the operating environment being the ANU HFSS facility and therefore the performance of the HFSS was experimentally validated. Conventional and state-feedback controllers were generated and assessed using a low-order dynamic model of the RR. More specifically, four controllers were designed and assessed: (1) a PI controller for exhaust temperature control; (2) a two-input two-output (TITO) decentralised PI controller; (3) a linear-quadratic-integral (LQI) controller developed using a linearised model of the RR at the nominal operating point, and; (4) a LQI controller developed using system identification techniques. The LQI controllers both required state estimation and thus state observer theory was used to develop estimators. All controllers showed ability to automate the operation of the RR, via manipulation of the input flows, and provided good disturbance rejection performance. The multiple-input and multiple-output (MIMO) controllers were all able to maintain exhaust temperature and maximise power efficiency. The LQI controllers showed greater robustness during simulated performance testing, with potential for further improvement. However, tuning of the state feedback controllers was found to be more difficult as compared to the conventional controllers. The practical implementation of the control system will require a range of sensors, actuators, input/output (I/O) modules, computers and networking equipment with associated software. Within the laboratory environment, a three level control hierarchy is recommended whereby sensor signals are passed to the supervisory computer running the RR controller algorithm which determines the set points for embedded controllers operating the actuators. Additionally, the implementation of the two-phase (gas-solid) generation system will be a critical step in the practical implementation and will require the customised fabrication of a solid particle feeder with precision control of the driver motor. Thus, the design methodology of a controller for electric motor speed control was studied and an experiment conducted. Future work should involve experimental validation and correction of the dynamic model and controller algorithms. Experimental refinement of the controller algorithms will also be required with consideration to practical parameters such as sampling frequency, computational delays and network-induced delays. Additionally, the lab based environment will be ideal for the controlled testing of more advanced non-linear and predictive control strategies. It is suggested the addition of feed-forward control to all controllers would be advantageous due to the nature of the direct normal irradiation (DNI) disturbance variable in the field. Show more