Optimization of tube drawing process through FE analysis, intelligent computation, and experimental verification

In present study, an attempt based on a hybrid approach was made to find optimal tube drawing parameters combination (i.e., die semi-angle, friction coefficient, tube entrance velocity, deformation length, and primary tube diameter) achieving minimum forming force, minimum dimensional error along with maximum thickness in longitudinal and circumferential directions. In order to form a design matrix, Taguchi L27 design was used and required data were derived from finite element model of the process. Adaptive neuro-fuzzy inference system (ANFIS) was then used to correlate process inputs to and responses aforementioned response outputs. Prediction accuracy of created ANFIS models was further checked by FE simulation and experimental testing data. Finally, the models were integrated to form objective function that was optimized by teaching-learning-based optimization (TLBO) algorithm. To check the adequacy of optimization, a finite element run and an experiment were carried out by obtained optimal parameter setting and the results were compared. The findings indicated that that the proposed methodology, that is, FEA-ANFIS-TLBO has superiority in identifying the optimal results.