Internal stress and flatness after tension levelling

For this project a large number of technical exchanges allowed all partners to increase their knowledge on the principles of tension levelling and on the study of internal stresses after tension levelling. The aim of the work of BRITISH STEEL was to determine if significant quantified relationships could be established between internal stress differentials through the strip thickness (of between 0.4 to 1.8mm) and observed strip curvature. If suitable relationships could be established then these could be used to assess the influence of tension levelling on internal stress distributions/strip curvature. For the measurement of internal stress the Stresscan 500C (based on Barkhausen noise principle) was deemed to be the most suitable on the market. In addition, previous work using this system had shown significant relationships between stress data and strip curvature on thinner (<0.2mm) tinplate material. The Stresscan system was calibrated so that actual stress data could be obtained. Attempts were made to draw comparisons with the x-ray system and the 3MA (also based on magnetic properties), as used by BFI in their part of the project. This comparison proved inconclusive and was believed to be due to the differing depths of measurement of the magnetic systems. Also, the systems were measuring in areas close to the surface where it was shown that there were rapid changes in internal stress for temper rolled/tension levelled strip. The examination of stress data and strip shape showed that it varied from no correlation to very strong but the overall conclusion was that there was no reliable relationship between the two parameters. For strip curvature, in this case longbow or strip curvature in the rolling direction, there was a positive relationship found between the difference in stress (top to bottom surfaces) and longbow. This was for those samples which exhibited longbow and which were flattened when the stress measurements were made. This simulated on-line conditions, where observations showed that in continuous form the strip did not exhibit any sign of longbow. There was no relationship found between stress and curvature for samples measured in the non flattened condition. Although the relationship found was a strong one it was considered not to be sensitive enough for the Stresscan system to be used as an accurate measure of stress. The reason for the scatter of results could be due to the fact that magnetic systems are also affected by the grain size. There was some evidence that grain size variation in the surface layer had a large influence on stress. The effect of this needs to be determined and, if this can be accounted for, then the relationship found between stress and longbow should be considered as a basis for a control system. This system would be based on maintaining the top and bottom readings within certain limits.