| dc.description.abstract | The master project was ready to start in mid-February, so some time was lost in the beginning. The company, Cambi prepared a workstation for me in their office in Asker. The work started with reading about the problem and preparing for an FEA-analysis. Their plant, EGE near Vormsund, was visited during early stages of the project. A better understanding about the vibrations was achieved when they were felt firsthand. There were some problems installing the necessary software, but with some help it was sorted. The modelling of the area of interest was a straight forward job and the model was soon ready to be imported into a FEA- program. The analysis was done in Inventor, but the program did not have the proper tools to get any results. Abaqus was the program used at NTNU which I could get help from supervisors to do the analysis. Unfortunately my main supervisor got ill and was hospitalized for a period of about 2 months. This made supervision difficult in this period and the FEA analysis work was abandoned. This resulted in the change of the focus of this thesis, into doing material testing of the failed material. The actual nozzle which failed was located in UK and was also repaired on site. This made the material test on the actual nozzle impossible. After consulting with Institute for Construction Technic, it was decided to simplify the specimens into a flat dogbone shaped specimen cut out from a plate with a weld across the center. It was welded from one side with no heat treating to resemble the actual case. The delivery of the welded plate was outsourced to a local company, which struggled to deliver on time. Because it was delivered so late, the material testing was done by the people at Institute for Construction Technic, with the main supervisor present. The testing of the specimens was done in one machine and the results were received at the end of this work period. The stress level was initially set slightly high in order to get a lower life and faster testing, but this turned out to be wrong. The stress level needed to be raised as the results were not consistent. The three last specimens had grooves along the weld toe of the weld which changed the stress concentration and the load on the specimen. The results of the tested specimens where plotted to the DNV curve for this weld type. Based on the comparison, some recommendations can be done when redesigning the equipment. The results have been beneficial for Cambi when it comes to understanding the design flaws of a weld and how much the weld influences the design life. | en |
