Analysis And Verification Of The Vertical Rolling Force And Shapes Based On Cosine Function Dog-Bone Model In Rough Rolling Process

With the development of the hot rolling processing, the quality requirements of the hot rolling production are valued by scholars both domestic and overseas. Vertical rolling plays an important role in control of the strip width during the hot rolling processing. As it is known, when width-to-thickness ratio is very large, the dog bone shapes will become in the edge of the strip after vertical rolling. To predict the force and the dog bone shapes of the vertical rolling is necessary for the steel production. Unfortunately, the formulas which are used to calculate the rolling force in vertical rolling are all based on the horizontal rolling processing. When they are used, they will be added by some increment. But the horizontal rolling is different from vertical rolling. Using the formulas of horizontal rolling processing to predict the value of vertical rolling processing should be considered incorrect. After these formulas are used in the field, a lot of time is needed to the self-learning and self-adaption. When it is under the real working conditions, plenty of empirical models are used. However, it is obviously that these models aren’t built on theories. So it is necessary that there should be a formulation based on analytical solution.In this paper, with researching the reference about the vertical rolling process, two assumptions are proposed. One of them is plane deformation assumption, the other one is the rigid-plastic assumption. The two assumptions are verified by both the reference and experiment. Then a velocity field is established. Through the cosine function dog-bone model the rolling deformation power is obtained. Minimum the power the rolling force is gotten, as well as the real shape parameters.Using the Finite element simulation, the rolling force can be compared. The accuracy of the prediction also can be obtained. Comparing the traditional models, FEM and cosine function model in this paper, the errors among them are less than10%. Also the model is compared with the real data in the field, the accuracy is proved.