Analysis of Process Control of Roller Leveling by Utilization of a FE Model with Integrated Closed-loop Control Structures

• Applicant: Professor Dr.-Ing. Gerhard Hirt
• Subject Area: Primary Shaping and Reshaping Technology, Additive Manufacturing
• Term: from 2013 to 2016
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 238040321

Roller leveling is essential to remove flatness defects and residual stresses of e.g. flat products. In contrast to other manufacturing technologies the setting of the leveler is mostly determined by experience. However, stricter tolerances and new materials are showing more and more the limits of this simple approach. Therefore it is vital to gain an improved understanding and to develop a better control of the process.A closed-loop control of the setting of leveling rolls would be a sensible solution. But all existing approaches have three major disadvantages:1) A measurement of the resulting flatness leads to a significant dead time in the control system. As a consequence the beginning of the strip will not be leveled properly.2) Variations in the properties of the work piece (yield stress, strip thickness and initial curvature) will not be detected before they have passed the whole machine.3) No practical measurement techniques are available to measure a residual curvature influenced by gravity in a tension free outlet of the leveler.Especially the last aspect makes experimental research of the influence between target properties (curvature, residual stresses) and real measurement data (e.g. forces and moments at the rolls, deflection of the strip) difficult to conduct. Therefore, a FE model shall be used to create an idealized environment (e.g. without influence of gravity on the curvature) to analyze these influences. Additionally implementing closed-loop control systems in the FE simulation creates a tool to test different control systems virtually. At a later stage of the project the closed-loop control system is tested by practical experiments.Already conducted research has shown that process parameters (e.g. forces and moments) can be utilized before and during the process to identify deviations from the expected work piece properties. Model values created beforehand can then provide the right settings for the machine. Thus, the long term target is to develop a new control concept that utilizes process data starting already during the filling of the machine. This can lead to reduced investments as well as maintenance costs, less wasted material and a higher availability of the subsequent processing line.

• DFG Programme: Research Grants