Sensitivities in fatigue analysis of offshore wind turbine support structures
MetadataVis full innførsel
A fatigue analysis gives an answer to the question of whether a structure will resist fatigue throughout its lifetime or not. The hot spot stress approach is one of the methods used to find the solution, by looking at points around the circumference of a weld, and finding the stress values at these points. This master thesis investigates fatigue analysis using the hot spot stress approach, through a sensitivity analysis By changing what parameters and methods the fatigue analysis uses, several fatigue damage values are calculated for each joint. The design chosen for the study is a single support structure for an offshore wind turbine (OWT) the UpWind reference jacket from Phase I of the Offshore Code Comparison Collaboration Continuation (OC4) project.  The jacket structure supports the NREL 5-MW reference wind turbine.  It consist of K-joints, X-joints and Y-joints. With fatigue damage being the design driving factor, the method for estimating fatigue damage may necessitate different design in the structure.  That means that different fatigue damage estimations may result in different costs and a more reliable structure. As anything can be safe and reliable if you just overspend and over-design it, an ideal fatigue analysis can give reliable results, without being too conservative. During the master work, an application was developed in the .NET Framework to work in conjuncture with FEDEM WindPower, a simulation program used for wind turbine systems. The developed application performs all extractions and calculations needed for estimating the fatigue damage using the hot spot stress method. The HSSs calculated are linearly extrapolated structural hot spot stress, with the nominal stress in the chord included for the HSSs on the chord side of the weld. The application performs the fatigue damage estimations using several procedures of the hot spot stress approach. This leads to several fatigue damage values for each joint, giving the basis for the sensitivity analysis concerning the hot spot stress approach. Common for all variations in the options, are the load simulations and the usage of hot spot stress approach, rainflow counting, S-N curve and Palmgren-Miner rule. This is all explained in chapter 2. They also have in common general fixity conditions, chord-end fixity parameter equal to 0.7, and for the joints with two braces, loads occur in both. The variations consist of the L parameter used in the SCF equations, the order in which maximum fatigue damage concerning one hot spot was kept while the others were excluded, the total number of hot spots around the circumference of the weld, and the choice regarding which equations to use for the SCFs at the crown for axial load. For the choice of equations to use for the SCFs, one can choose between solely geometrical SCFs, and SCFs including nominal stress. Here, both options are used. Only Y-joints are affected by changes in the L parameter after a certain limit, but until this limit is reached it affects all joints. This is because Y-joints are the only ones, regarding the structure analysed and the choices made in this thesis, that have SCFs containing L. K-joints are those joints that are most affected by which order one chooses to keep the maximum value amongst the hot spots. This indicates that the spot most prone to fatigue damage varies more within the K-joints than in the other two joint types. X-joints and K-joints are equally affected by the number of hot spot one uses. Increasing the number of hot spots from 8 to 16 can for several joints give an increase ratio equal to 16 %, the ratio being the fatigue damage estimated using 16 hot spots divided by the fatigue damage estimated using 8 hot spots. Increasing the number of hot spots either gives the same value or a higher value, because increasing the number of hot spots indicates adding hot spots in between the existing ones. The optional SCFs at the crown for axial load are only evaluated for Y-joints. The equations containing nominal stress shows, for most of the joints, much lower fatigue damage estimations than the solely geometrical ones, for L values at 7 and higher. The limited selection of parameters, due to limited amount of time, results in a sensitivity analysis that is not sufficient to come up with a final, clear answer regarding the ultimate fatigue analysis. This thesis is therefore intended to be used, with the developed application, as the basis for a larger sensitivity analysis that could provide further knowledge on the preferable procedure concerning fatigue analysis using the hot spot stress approach. This master thesis has been performed in cooperation with Fedem Technology AS in Trondheim. The app development was assisted by Fedem Technology AS representatives Ole-Ivar Holte, Runar Heggelien Refsnæs and Daniel Zwick. Inge Lotsberg at DNV GL AS has contributed concerning the fatigue analysis, as has Sebastian Schafhirt, who has also helped in relation to the research method. Michael Muskulus has been the main supervisor and contributed regarding the research question.