Thermal detection of cracks on glowing wire during the process of rolling with regard to reduction of rejections
During the rolling process laps can occur, which are a sort of surface crack. If the depth of the crack exceeds a prescribed limit, the whole rolled wire coil is rejected. This means very high production loss of materials and energy. Laps result from wrong settings and can be avoided by the correction of the roll passes.
In order to eliminate the cause of this failure as early as possible, the crack has to be found already at the glowing wire during its production in the rolling mills. This requires a non-destructive, fully automated, inline testing at temperature values of 800-1100°C and at a speed of 30-40m/s. Within the scope of this project a thermo-graphical test method has been investigated, by applying first either an additional heating or a short cooling to the rolled wire. Around a surface crack an inhomogeneous temperature distribution occurs, which can be recorded by an infrared camera. The heating is done inductively, where the wire is moved through an induction coil. In the first part of the project measurements in the laboratory were accomplished.
For these measurements the wire was heated first to a temperature value of 800-1100°C and an additional short heating or cooling was applied. Different materials (steel grades, titan alloys, and so on) have been tested and compared for their heating and cooling behavior. Model calculations and finite element simulations were carried out, in order to evaluate the temperature distribution for different material parameters, crack forms and for different crack depth values. Additionally, the behavior of the wire moving with high speed has been also simulated.
After successful completion of the laboratory measurements and the simulations, measurements were executed directly at the rolling mill of one of the project partners. In the case of a glowing metallic surface the intensity of the emitted energy is sufficient to use cameras for the visible range. Such cameras are much cheaper than infrared cameras.
In the scope of the project the possibility to create a temperature calibration for a camera in the visible range has been investigated, which is reproducible and accurate enough to use for such an application. The image processing is an important part of an automated optical inspection. The computer analyzes the recorded images, identifies the eventual defects and makes a fully automatic decision. An additional challenge is to make this analysis in real-time. Because of the high speed of the material a very high recording frequency and very quick image processing is required. In the scope of the project the usage of a graphic board with CUDA capability has been tested for the acceleration of the image processing.
The results showed that detection of the laps and surface cracks under labor circumstances at 1000°C is well possible, but the industrial realization of this technique because of the very high material speed and because of the oxide scale at the surface of the rolled wire is difficult.
Dr. Beata Oswald-Tranta
Lehrstuhl für Automation, Montanuniversität Leoben
Prof. Paul O’Leary, Carmen Brebric, DI Gernot Wally, Mario Sorger, Andre Pura
Project or cooperation partner
- Dr. Wolfgang Schützenhöfer, Dr. Walter Zleppnig
Boehler Edelstahl GmbH & Co KG
- Dr. Ronald Ofner
Hot Vision Research GmbH
- Dr. Johann Reisinger, Dr. Mark Tratnig