Experimental Study and Artificial Neural Networks Prediction of Effective Parameters in Continuous Dieless Wire Drawing

Abstract

The dieless drawing process is an innovative methodemanated and appeared in coincidence with development of theconcept of metal superplasticity. It is utilized from the localheating of a wire or tube to a specified temperature and followedby a local cooling, so an additional deformation is inhibited. Inthis study, a special dieless drawing machine was designed tocarry out an experimental program on SUS304-stainless steel wirehaving diameter of (1.6-2) mm to investigate the main processparameters such as speeds, heat quantity, heating coil width andheating-cooling separation distance. Also, a numerical modelbased on thermo-mechanical analysis was developed and validatedwith experimental program. Furthermore, an artificial neuralnetwork ANN model based on current experimental data wasprepared to predict the dieless drawing behavior. A maximum areareduction of 40.7% was obtained in single pass. A 3.12mm/sfeeding velocity and 4.97mm/s drawing velocity were realizedthrough the experimental tests. The results showed that bothdrawing force and wire profile were effected by increasing offeeding speed, heating coil width and separation distance. Also, itis confirmed that strain rate was reduced by increasing the heatingcoil width and the reduction ratio was promoted. A maximumerror of 21% was recorded between ANN model and experimentalresults. The results showed a good agreement amongexperimental, numerical and ANN models.