An Experimental Approach towards an Equation for the Prediction of Piercing Mandrels Wear

Abstract

Analysis of the relationship between variables involved in the wear of mandrels used for seamless tube making was carried out. Laboratory-scale mandrels were pierced in sequences with different process parameters. Process data were acquired during piercing events and digital photographs of the mandrels’ silhouette were taken to study their profile evolution. Direct variables were registered and others were created. Data were first analysed in ranges of furnace temperature and cooling water flow. A self-organizing map (a kind of artificial neural network) was trained with experimental data for further multi-variable analysis. Results were analysed from the data clusters found and conclusions were extracted. Furnace temperature and cooling water flow were important for their influence on wear but a definite relationship could not be established. Results showed that the mandrel shape and wear condition could be inferred through the study of the axial force curve. A first equation for mandrel wear prediction is introduced, taking into account the relations found between the most important variables involved in the process. Nevertheless, additional research must be done considering the billet material stress fields and friction contact conditions.