dc.contributor.advisor Imsland, Lars nb_NO dc.contributor.advisor Kjerstad, Oivind Kåre nb_NO dc.contributor.advisor Skjetne, Roger nb_NO dc.contributor.author Wold, Henrik Emil nb_NO dc.date.accessioned 2014-12-19T14:07:37Z dc.date.available 2014-12-19T14:07:37Z dc.date.created 2013-09-19 nb_NO dc.date.issued 2013 nb_NO dc.identifier 649597 nb_NO dc.identifier ntnudaim:8758 nb_NO dc.identifier.uri http://hdl.handle.net/11250/261003 dc.description.abstract The commercial industry has initiated work on how to make it feasible to enter the Arctic seas. Ice loads affects most aspects of the Arctic operation, and the marine crafts must be able to handle them all. The DP control system, and thus the thrust allocation, is not designed to handle ice loads and will not work properly \cite{Moran}. The main purpose of this master thesis is to enhance the thrust allocation for handling ice loads. This is done by including thruster dynamics and adding thruster ice clearance by thruster wake. When the ice loads are too high for the DP system to handle, a prioritization of the degrees of freedom is included to achieve predictable degradation of performance. To predict possible drift-offs, energy analysis will be used to investigate if the control forces integrated over time contain enough energy to withstand the ice loads.The thrust allocation is based on numerical optimization and implemented in $Matlab$. To make the thrust allocation more realistic, thruster dynamics are added. The first method is to low-pass filter the control forces, and the second is to add restrictions on the change of control forces. To clear the ice away from the hull, thruster ice clearance is implemented. The first solution is to let the algorithm calculate the azimuth angles within predefined sectors, and secondly to force the azimuth thrusters to follow predefined references in control forces and azimuth angles.A case study is done to investigate the performance of the thrust allocation algorithm, where towing tank measurement data from CIV Arctic is used as input. To measure the performance of the thrust allocation, the magnitude of the slack term, $\bm{s}^\top \bm{Q} \bm{s}$, gives a first impression. For further investigation, the error between the forces and moments from the ice loads and the achieved forces and moments from the thrust allocation is used. The results from the case study indicated that when the ice loads were high, the prioritization of degrees of freedom was followed. Both with and without thruster dynamics the error in produced thrust was less than 8 [\%] for small ice conditions, but increased rapidly for 1.2 [m] of ice. The thruster dynamics did not increase the error significantly, except an increase in yaw error for light ice conditions when the low-pass filter was applied. By adding thruster ice clearance, the error in produced thrust increased. Corresponding results were found for the energy considerations. The chosen thrust allocation algorithm gave satisfactory results. By decreasing the ice concentrations, for instance by using ice management, the performance was improved. Adding restrictions on the change of control forces was found to be the best way of including thruster dynamics, because then the restrictions were implicit in the thrust allocation algorithm. Two solutions were also proposed for implementing thruster ice clearance. By letting the thrust allocation find the azimuth angles, the performance of the algorithm was better than by forcing the control forces and azimuth angles to follow predefined references. In spite of this, the second solution was found to be the best in practice because the vessel operator has more control over the thrusters. Some recommendations for future work are to include all the components of the DP control system, do a more advanced implementation of the thruster dynamics and a more detailed energy analysis. nb_NO dc.language eng nb_NO dc.publisher Institutt for teknisk kybernetikk nb_NO dc.title Thrust allocation for DP in ice nb_NO dc.type Master thesis nb_NO dc.source.pagenumber 110 nb_NO dc.contributor.department Norges teknisk-naturvitenskapelige universitet, Fakultet for informasjonsteknologi, matematikk og elektroteknikk, Institutt for teknisk kybernetikk nb_NO
﻿