Sensitivity Analysis for Prediction of Bead Geometry using Plasma Arc Welding in Bellows Segment

Recently automated welding systems have received much attention because they are highly suitable not only to increase production rate and quality, but also to decrease cost and time to manufacture for a given product. To get the desired quality welds, it is essential to have complete control over the relevant process parameters in order to obtain the required bead geometry and which is also based on weldability. However, new algorithms need to be developed to make effective use of automated arc welding process.  

Previous works on relationship between the process parameters and bead geometry in arc welding process can be grouped into two distinct areas: empirical methods based on studies of actual welding situations [1]-[2] and theoretical studies based on heat flow theory [3,4]. Despite the large number of attempts to analysis arc welding process, there is still the lack of algorithms that can predict bead geometry over a wide range of welding conditions. Sensitivity analysis, a method to identify critical parameters and rank them by their order of importance, is paramount in model validation where attempts are made to compare the calculated output to the measured data. This type of analysis can study which parameters must be most accurately measured, thus determining the input parameters exerting the most influence upon model outputs. It differs considerably from the usual approach of perturbing a process parameter of a known amount and evaluating the new results. Various statistical techniques such as regression analysis, Response Surface Methodology (RSM) and Taguchi method have been applied to modeling and optimization of bead geometry in Gas Metal Arc(GMA) welding[5]-[8]. Palani and Murugan[9] developed mathematical model for prediction of bead geometry in Flux Core Arc(FCA) welding using a RSM method. Taguchi method was also used to analyze the effect of each process parameter on the bead geometry [10]-[11]. Effect of pulse current on bead profiles of Gas Tungsten Arc(GTA) welded aluminum alloy joints has been studied [12]. Because the solution of a mathematical model to predict bead geometry is complex and the parameters involved are highly coupled, some researchers have resorted to Artificial Intelligence(AI) such as Artificial Neural Network(ANN) and Genetic Algorithm(GA) techniques based on large experimental databases [13]-[14]. Also, Son and Kwak [15] established a sensitivity formulation for eigen values, including repeated eigenvalues, with respect to the change of boundary conditions. The tangential design velocity component was employed to present the change of boundary conditions. The practical PAW process is widely appreciated in the aerospace, chemical, naval, nuclear industries, etc[16]. Prasada et al. investigated the weld quality characteristics of pulsed current micro-plasma arc welded austenitic stainless steels and considered weld pool geometry, microstructure, grain size, hardness and tensile  properties  as  weld  quality  characteristics.