Comparing Controllers for Dynamic Positioning of Ships in Extreme Seas
MetadataVis full innførsel
The primary purpose of this study is to develop two different dynamic position-ing (DP) controllers for a model-scale supply vessel Cybership III, and determinewhich control strategy provides a safer, smarter and greener control when the ves-sel is exposed to extreme seas. DP has since the 1960s contributed to the safetyand efficiency of oil-related and other operations for marine vessels. With thiscomputer-controlled system, vessels such as research vessels, supply vessels andcruise ships can automatically maintain position and heading by using propellerswhen the vessel is exposed to wind, waves and current. A nonlinear passive observer (NPO), proportional-integral-derivative with acceler-ation feedback (PID-AFB) controller, sliding-mode control (SMC) and a referencemodel for Cybership III were derived. Test cases were first performed with theMarine Cybernetics Simulator (MCSim) in calm, harsh and extreme seas. Thissimulator consists of a high fidelity process plant model of Cybership III in Mat-lab/Simulink together with models for generation of environmental forces. Sec-ondly, tests in calm and harsh seas were performed on the model vessel in theMarine Cybernetics Laboratory (MCLab) at MARINTEK. The performance fortests in MCSim was evaluated by integral squared error (ISE), integral absoluteerror (IAE) and integral time-weighted absolute error (ITAE). The measures ofperformance for tests in the MCLab were performed by using ISE, IAE and ITAEin combination with the energy consumption of the thrusters on Cybership III inorder to generate a cost function. This cost function is a new method used to eval-uate the performance of the controllers with the aim of achieving a safer operationmeaning good accuracy, a greener operation by minimizing energy consumption anda smarter operation by achieving the two latter simultaneously. Other measuresof performance used to evaluate controllers on Cybership III and similar vesselsin MCLab have as far as the knowledge of the author, only been performed withrespect to the error of the positions and heading. This thesis uses a new measure ofperformance not only evaluating the error, but also the energy consumption duringtests, thus evaluating the controller with more realistic performance measures ofperformance. The PID-AFB-controller provided the best performance when the vessel was testedin extreme seas in MCSim and was evaluated only by ISE, IAE and ITAE, whilethe SMC achieved the best performance when tested in calm and harsh seas bothin the MCLab and MCSim. The tests performed in MCSim were not comparablewith the tests performed in the MCLab and the reasons for this may be inaccuratemodel parameters, the lack of a proper thruster model, 3D effects of the wavesin the basin and reflection on the tank wall. In retrospect of the experiments,Cybership III was found to weight 14% more than modeled in MCSim and hence the test cases in MCLab was considered to be more credible.