The study of thermal stress in the friction stir welding (FSW) process is of particular importance due to its direct effect on the quality of manufactured part. The present research work is concerned with implementation of novel method to find the best solution area in friction stir welding process. The numerical simulation of FSW has been done using LS-Dyna software. The investigated variables include the rotational speed, transverse speed and shoulder diameter. Due to the necessity of reducing the tensile-thermal stress in FSW operation, the aim of current research work is to minimize this factor. Each variable has five levels and according to Taguchi method, the number of numerical tests was reduced to 25. After performing the simulation of FSW operation, the hierarchical clustering on inputs and output variables has been performed and based on similarity level (more than 80 %), the number of clusters was computed to ten. The extraction rules acquired from the employing of clusters were considered as input of the fuzzy inference system using MATLAB software. According to the surface and heat map plots obtained from the fuzzy inference system, the solution area was determined. Based on the response sensitivity analysis in the range of process variables, the most dependency of the tensile-thermal stress changes was on the rotational speed. In the appropriate feasible solutions set, the rotational speed was gained about 710 rpm and the ranges of the transverse speed and shoulder diameter were computed to, in turn, [70–110] (mm/min) and [19.5–23] (mm).
