Robust multi-criteria optimisation of welded joints

Civilisation has depended on welded structures to facilitate production and improve the quality of life. Welds are used to create infrastructure upon which we rely, such as transportation, oil and gas piping, shipbuilding, bridges and buildings, and to produce the equipment that makes all of this happen. In short, the joining of two metals through welding has immensely contributed to our society.

A critical factor in the strength of welded joints is the geometry of the joints, and for this reason a robust optimisation of geometrical parameters of welded joints has been conducted in order to establish the optimum and most robust design in the presence of variation amongst geometrical parameters.

A parametric finite element analysis, using Python script, has been performed with the objective to investigate the effect of the welded joint parameters on the stress concentration factors under tensile and bending load. The results indicate that the parametric model, which is generated by Python script, can be used in a wide range of welded geometry, and has the capacity to reduce the time of computation. Additionally, an experimental study, including the geometrical identification of the welded joints, tensile test, hardness test and fatigue, has also been performed.

In order to select the best optimisation algorithms, different optimisation algorithms and performance metrics with various types of problem were examined in this study. The results from this part show the accuracy of Circumscription Metric (CM) in comparison to Pair wise Metric (PW) - which is used widely in optimisation studies. Furthermore, the results show that the Fast Multi-objective Optimisation Algorithm (FMOGA-II) outperformed other optimisation algorithms used during this study.

In this study, a new methodology for selecting the most robust designs from within the Pareto set has been developed. Finally, a traditional and robust optimisation of a butt welded joint has been performed by establishing a link between an optimisation software package and parametric finite element, the results of which show the ability of this approach to extract the robust optimal designs from the Pareto front.