Peter Kracht, Boris Fischer, Sebastian Perez-Becker, and Jean-Baptiste Richard – MARINET Technical Report, 2013
Many advanced control schemes have been proposed for wave energy converters (WEC), which offer the chance of significantly increasing the energy yield. Assessing the available literature so far most of these control schemes have only been investigated by simulations. These simulations are often based on assumptions such as that linear wave theory is applicable, that perfect models for the WEC/PTOs etc. are available or that an ideal prediction of the wave excitation force some distance in the future is available. Comparably little information can be found on the testing of these control schemes in a real life environment. In this project two advanced control schemes were applied on a 1/20-scale PTO-model of a point absorber type WEC. The idea of the first control scheme is to estimate the wave excitation force by an observer and control the velocity of the WEC so that it is in phase with the wave excitation force, which is a necessary condition for maximized energy yield. The second control scheme also uses an estimated wave excitation force. The estimated wave excitation force is fed to a short-term predictor, which estimates the future wave excitation force for a certain time horizon. This estimated future wave excitation force is fed to a so-called model predictive controller (MPC), which calculates an optimal control force for the WEC. Both kinds of controllers make use of quantities which are likely not to be available with finite precision. In case of the first controller the results depend on the accuracy of the wave excitation force estimation, which in turn depends on factors like the accuracy of the WEC-model used in the observer and also the accuracy of the measurements (signal processing delay, noise etc.). Since in the MPC the estimated wave excitation force is also used, the controller will suffer from the same inaccuracies. Additionally, the estimated wave excitation force is predicted, which is likely to introduce additional uncertainties in the system. Testing of the two control schemes under ideal circumstances (simulation) showed that both can significantly increase the energy yield (up to 50% depending on the sea state).
The aim of the project is the testing of the two controller schemes under circumstances closer to reality, so to investigate if the uncertainties mentioned above will compromise their effectiveness. The two controllers were run on the 1/20-scale model in irregular sea states. First tests with the observer revealed, that the offline optimized observer showed quite a high phase lag in the real environment, thus compromising the idea to operate the WEC in phase with the wave excitation force. A modified observer comprising a different signal model, significantly reduced the phase lag in the estimated wave excitation force. Further investigations showed that there was comparably high friction (both dynamic and static friction) present in the system. Unfortunately this issue couldn’t be thoroughly dealt with during the test period due to the limited test time. Using the modified observer it was possible to increase the energy yield by around 10-15% compared to an optimized PI-controller (the quasi standard controller for WEC of the point absorber type).Also the MPC worked stable and produced reasonable results. Due to the limited time for commissioning and testing the MPC (<1 day), sound analysis, tuning and optimisation of this controller were not possible, and roughly 10% less energy than with the PI-controller were achieved in a single test with an irregular sea state. In conclusion it can be said, that on the one hand the results obtained in the wave tank were less good as
compared to the results obtained from simulations (based on ideal assumptions). On the other hand, both controllers worked stable and showed reasonable results. One of the controllers actually gained an increased efficiency. Two additional aspects make it reasonable to assume that further improvements are possible. Firstly design optimizations can possibly lower the relative friction in a full-scale WEC or the friction can be accounted of in the model used in the control strategy. Secondly the portion of the phase lag in the estimated wave excitation force, which is caused by signal processing delays, will be reduced in a full-scale WEC, where the wave period times are
In sum, the tank test showed that it is possible to increase the energy yield of WECs by using advanced control schemes. Further investigations are currently carried out based on the data from the tests to improve future testing both on wave tank models and full-scale devices.