Analysis and prediction of welding deformations of ship panels in prefabrication process

Abstract

In the shipbuilding industry problems related to the welding process are very often identified, especially because it must fulfil several requirements as for design as for production, moreover it needs to be performed under strict supervision and also by very skilled workers. Intending to minimize the effects caused by the welding in the shipyards, over the year many works have been developed showing methods to predict and mitigate these effects, mainly through Finite Element Analyses where is possible to verify points of improvement along the welding process, for example in terms of process sequence or welding parameter. In this current work is proposed a practical way to predict the welding-induced deformation of 5-mm-thick steel ship panels prefabricated in a traditional Polish shipyard through a number of experiments and data collection performed on the butt welding line using Submerged Arc Welding (SAW). For these experiments a Design of Experiment (DoE) has been organized so as to get different condition of welding energy as well as the values of the pressure applied by the devices used during the welding and then different trials and data collection were carefully carried out in the shipyard to realize what are the most important welding parameter which are contributing for the problem (welding deformation). The data available have been analyzed by the statistical software Design-Expert® and considerations about data arrangement have also been done aiming to provide the best condition to describe the welding deformation. As a result this work provides an approach to predict the welding-induced deformation within the workshop by way of numerical equation where the main welding parameters were taken into account and dully weighted by the software. Some partial evaluations of the accuracy of the equations were achieved and their results considered satisfactory regarding to the comparison between the actual values versus predicted values. Another important issue reached is the data arrangement used to compute the welding deformation and finally this thesis is completed with recommendations and suggestions for the future investigation.