The present paper deals with the use of the Artificial Intelligence for the quality control of a seamless tube production mill. In particular, the production quality of seamless tubes with the Mannesmann process is strongly affected by the position of the plug as a function of worked material and cage geometry which can result in both internal oxidation of the tube and accelerated plug wear. In the practice, the mill operator modifies the plug position on the basis of his long-term operational experience. In this context, the availability of a tool able to support the operator in the decision taking process is fundamental for improving the internal surface and dimensional quality of the pipes and the plug life. The information necessary to build the Artificial Intelligence system derives from a research project carried out by our research group and the Tenaris Dalmine SpA company. In the study the rolling process was simulated by rigid-plastic FEM code DEFORM ™ and it was simplified in a two dimensional model. The effectiveness of a 2D model was validated by a 3D simulation of the process where it was found that the axial stresses in the bar are constant along the bar axis and only 10% of the transversal stresses and, as a consequence, they were neglected. All the process data come from the industrial practice. The simulative parameter used to study the hole formation in the bar was the damage calculated according to the Cockroft-Latham criterion. To validate the simulation results and to identify the critical damage value, the position of the starting of the crack (Mannesmann cone) was measured in the simulations and in TENARIS-Dalmine experimental tests. The model was tested under different billet materials and mill geometries obtaining always a good agreement being the error less than 10%. The successful results of the simulative tests will be used in the present research to develop and to teach a neural network system which aims to check and improve the quality of TENARIS-Dalmine products and to realize an on-line process optimisation. Since the ability of the neural network in predicting the optimal response greatly depends on its layout, several ANN configurations will be tested in order to identify the best one able to furnish the most reliable indications. The final goal is to implement this ANN into the supervision system of the plant in order to help the mill operator in setting the optimal plug position.

A neural network application to improve the quality of seamless tubes

GIARDINI, Claudio;
2006-01-01

Abstract

The present paper deals with the use of the Artificial Intelligence for the quality control of a seamless tube production mill. In particular, the production quality of seamless tubes with the Mannesmann process is strongly affected by the position of the plug as a function of worked material and cage geometry which can result in both internal oxidation of the tube and accelerated plug wear. In the practice, the mill operator modifies the plug position on the basis of his long-term operational experience. In this context, the availability of a tool able to support the operator in the decision taking process is fundamental for improving the internal surface and dimensional quality of the pipes and the plug life. The information necessary to build the Artificial Intelligence system derives from a research project carried out by our research group and the Tenaris Dalmine SpA company. In the study the rolling process was simulated by rigid-plastic FEM code DEFORM ™ and it was simplified in a two dimensional model. The effectiveness of a 2D model was validated by a 3D simulation of the process where it was found that the axial stresses in the bar are constant along the bar axis and only 10% of the transversal stresses and, as a consequence, they were neglected. All the process data come from the industrial practice. The simulative parameter used to study the hole formation in the bar was the damage calculated according to the Cockroft-Latham criterion. To validate the simulation results and to identify the critical damage value, the position of the starting of the crack (Mannesmann cone) was measured in the simulations and in TENARIS-Dalmine experimental tests. The model was tested under different billet materials and mill geometries obtaining always a good agreement being the error less than 10%. The successful results of the simulative tests will be used in the present research to develop and to teach a neural network system which aims to check and improve the quality of TENARIS-Dalmine products and to realize an on-line process optimisation. Since the ability of the neural network in predicting the optimal response greatly depends on its layout, several ANN configurations will be tested in order to identify the best one able to furnish the most reliable indications. The final goal is to implement this ANN into the supervision system of the plant in order to help the mill operator in setting the optimal plug position.
book chapter - capitolo di libro
2006
Ceretti, Elisabetta; Giardini, Claudio; Attanasio, Aldo
File allegato/i alla scheda:
Non ci sono file allegati a questa scheda.
Pubblicazioni consigliate

Aisberg ©2008 Servizi bibliotecari, Università degli studi di Bergamo | Terms of use/Condizioni di utilizzo

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10446/19959
Citazioni
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact