Gradient-ascent latching controllers for ocean wave energy converters

The potential of ocean wave energy as a renewable energy source was recognised long ago, however, despite decades of research and development effort wave energy has yet to be commercialised. The primary challenge with wave energy harvesting is in the conversion of the highly variable broadband wave energy resource to grid frequency electricity. Feedback controlled oscillating systems have been considered extensively over the years with a significant number of control methods developed. Whilst feedback controlled oscillating wave energy converters (WECs) have demonstrated high efficacy in numerical simulations and idealised laboratory experiments, these results are yet to be achieved for a WEC in real ocean conditions. The work herein will attempt to contribute towards real world wave energy conversion by investigating simple, practical WEC controllers which are effective in real seas. This work investigates the efficacy of gradient-ascent control algorithms for sea state based latching control of an oscillating water column (OWC) WEC. The gradientascent controllers, which are model independent and make decisions solely from real-time data, will attempt to optimise the latching duration of an OWC based on sea state, considering a passive power take-off (PTO) with fixed geometry (i.e. a static turbine in series with a latching valve). Two gradient-ascent controllers are considered in this work: a standard maximum power point tracking (MPPT) controller, as frequently used in solar and wind applications, and a novel maximum capture width tracking (MCWT) controller, being an MPPT controller modified to account for incoming wave conditions as well as WEC power output. Essentially, two research questions will be addressed: (i) is the imposition of a fixed latching duration based on sea state, as opposed to the incident wave time-series, a suitable control objective for an OWC; and (ii) can gradient-ascent control effectively optimise this latching duration in a random wave environment? These research questions will be addressed through both numerical simulations and small scale experiments in a wave flume. A wave-to-turbine model of an OWC is presented which combines linear hydrodynamic wave theory, thermodynamics of an ideal gas, and published experimental turbine data. Small scale experiments are conducted using an in-house designed experimental rig. The performance of a latching controlled OWC with PTO of fixed geometry is investigated in irregular waves. It will be shown through both statistical analysis and Monte-Carlo simulations that mean turbine power and capture width are random variables which, in stochastic seas, can be considered Gaussian. Consequently, it is demonstrated that optimal OWC performance is not a singular datum, but instead a region defined by the amount of power/capture width time-series information available. In stochastic seas this region can be quantified with 95% confidence intervals for optimal performance. The efficacy of the gradient-ascent latching controllers are tested in seas with both stationary and non-stationary (i.e. time-varying) spectral characteristics. The conditions for which gradient-ascent control is effective in irregular waves are presented, which are subsequently used to explain controller behaviour. In stationary seas it will be shown that both the MPPT and MCWT controllers can optimise performance to within the confidence bounds that optimal performance can be known, whereas in non-stationary seas the novel MCWT controller is shown to be superior. The performance achievable by a sea state based latching control strategy is investigated for various levels of turbine damping. Plant performance is analysed in both the time and frequency domains. It is revealed that sea state based latching acts as a passive non-linear damper and only improves performance for significantly underdamped OWCs. The performance produced by the optimised sea state based latching strategy is similar to that achievable by a well tuned passively damped PTO, hence sea state based latching is unlikely to be useful for a real OWC. Finally, small scale experiments are performed in a wave flume to validate the results from simulations, including the Gaussian nature of performance metrics, the behaviour of MPPT/MCWT latching controllers in stationary and non-stationary seas, and the efficacy of sea state based latching. Experimental results are all in agreement with simulation results.Thesis (Ph.D.) -- University of Adelaide, School of Electrical and Mechanical Engineering, 202