Abstract Weld bead geometry is a critical factor for determining the quality of welded joints, for this the welding process input parameters play a key role. In this study, the relationships between welding process variables and the size of the weld bead produced by pulsed GMAW process were investigated by a neural network trained with Bayesian-Regulation Back Propagation algorithm and a second degree regression models. A series of experiments were carried out by applying a Box-Behnken design of experiment. The results showed that both models can predict well the bead geometry. However, the neural network model had a slightly better performance than the second-order regression model. Both models can be used for further analyses and using them may surmount or reduce the need of experimental procedures especially in thermal analysis validations of welding finite element modelling.

Resumo: O principal objetivo do presente trabalho é analisar a influência da vazão de água e da geometria na entrada de um calorímetro de fluxo contínuo sobre a eficiência térmica do arco. O procedimento experimental consiste em testar diferentes vazões de água e três configurações geométricas na região de entrada de água no calorímetro: com rolha reta, com difusor cônico e com obstáculo. Os experimentos foram planejados e os resultados avaliados com base na análise de variância estatística de um único fator, no caso, a vazão de água na entrada do calorímetro. Os cordões de solda foram depositados através do processo Gas Metal Arc Welding – GMAW, sobre tubos de aço baixo carbono, usando os mesmos parâmetros e tempo de soldagem. A maior eficiência térmica média de 80,5% foi obtida para a vazão de 4 l/min, com baixo erro estatístico, utilizando rolha de entrada com geometria de difusor cônico, sendo que para vazões menores ou maiores ocorreu diminuição nos valores da eficiência medida. O modelo em que o fluxo entra diretamente no tubo apresentou todos os valores de eficiência térmica do arco com pequeno decréscimo numérico se comparados com o difusor cônico, enquanto a rolha com obstáculo apresentou elevado erro estatístico.

Abstract: The main aim of the present work is to analyze the influence of water flow rate and inlet geometry on the arc thermal efficiency of a continuous water flow calorimeter. The experimental procedure consists of varying water flow rate and testing three different calorimeter inlet seal geometries: straight seal, conical diffuser seal and seal with water flux obstacle. The experiments were designed and the results evaluated based in a one-factor statistical analysis of variance, in this case the inlet calorimeter water flow. The welding beads were deposited on low carbon steel pipes by Gas Metal Arc Welding – GMAW process, using the same parameters and welding time. The highest average thermal efficiency is 80.5% to water flow of 4 l/min, with a low statistical error, using the conical diffuser seal inlet geometry, whereas for smaller or higher flow rates the measured efficiency values were lower. The inlet with straight seal model shown all the arc thermal efficiency values with slightly lower numerical values compared with conical diffuser, while the seal with flux obstacle exhibited high statistical error.