Optimal Real-Time Predictive Control for Maximising the Power-Take-Off Efficiency of the WaveRAM Wave Energy Converter

Abstract

The aim of this thesis is to develop a control strategy for optimising the power-take-off (PTO) efficiency of the newly developed WaveRAM (WRAM) wave energy converter. The three main research streams established to satisfy these requirements have been: 1) the development of a wave-to-wire numerical model of the WRAM and its PTO system; 2) the implementation of a real-time model predictive control (RT MPC) algorithm; and 3) the construction of a laboratory test rig, capable of emulating the behaviour of a sea-going device. Given that the WRAM is intended for deployment at sea in the near future, the control algorithm must be capable of operating in real-time. This has been the primary motivation of the second research stream, where portions of the model created in the first stream were used as the controlled plant. In order to verify the algorithm's performance prior to sea-deployment, the laboratory test facility from the third stream ensures that the algorithm can operate successfully within practical constraints.

With regards to the first research stream, many parts of the numerical model for WRAM were derived based on well-accepted fundamental physics in the literature. For the subsystems where the fundamental physics were not well-known, or parameters were captured more accurately with physical equipment, new models were derived. The well-accepted components include the WRAM's hydrodynamics, air chamber pneumatics and the turbine's characteristics. The unknown components were within the turbo-generator system, where not all parameter values were available at design. As such, system identification was adopted to identify the unknown values from laboratory experiments. Results of the system identification process showed a closeness-of-fit of 96% between the experimental results and theoretical model. The WRAM's wave-to-wire model was then completed, with the turbine-to-wire portion being controlled by the RT MPC algorithm.

The real-time control algorithm, adopted in the second research stream, follows an approach in the literature that implements fast constrained MPC using Laguerre networks. This has been applied to the WRAM's PTO, with focus placed on maximising the turbine's operating efficiency. Motivation for this algorithm was based on its ability to perform optimal control with practical constraints and fast execution speeds. Results from this research have shown that speeds in the order of tens of μs are possible for this application, much faster than what has been reported in the literature for wave energy applications. The high speed of the algorithm makes it possible for real-time implementation on standard industrial control systems. Another component of the research was to investigate the performance of alternative cost functions in the optimal control problem. Results showed that care is needed in choosing an appropriate cost function, depending on the type of wave energy converter being controlled. Further questions were raised as to whether the closed-form method of RT MPC is appropriate for both cost functions used.

Lastly, in the third stream, an emulation platform has been produced, allowing for advanced control algorithms to be tested on realistic hardware, prior to implementation at sea. The platform has been integrated into an existing real-time hybrid test (RTHT) facility, such that aspects of the WRAM model, not implemented on the emulator platform, can be simulated within a real-time feedback loop. The RTHT approach helps bridge the gap between academic and industrial partners, where advanced theoretical algorithms can be tested in a safe laboratory environment. Industrial partners can then draw confidence in the technology before bearing the financial risk of failure at sea. The emulation platform developed in this research is the first of its kind, as it is capable of emulating the bidirectional, irregular flow conditions typical of oscillating water columns. Furthermore, it can do so within a RTHT framework. Other similar test facilities offer only a subset of these features.

The main contributions of the research include: a complete wave-to-wire model of the WRAM that incorporates existing well-accepted fundamental principles with newly formulated models of standard industry equipment; the demonstration of execution speeds for RT MPC in the order of μs; and a novel WRAM emulation platform, capable of generating air flow conditions typical for oscillating water columns. Several new research opportunities have surfaced, including: implementation of the control approach within the RTHT framework; further investigation into alternative cost functions of the optimal control problem; exploration into alternative methods for solving quadratic programming problems in real-time; further investigation of the control tuning parameters; consideration of alternative electrical generators; and examination of electrical power quality for grids.