Reliability Analysis of Rolled Strip Exit Profile Considering Random Nature of Critical Rolling Variables

摘要

According to the results provided in the previous section, some observations can be made about the reliability of the strip crown after rolling, and what factors can be addressed to improve the quality and yield of the rolled steel. In this work, several assumptions were made regarding the choice and nature of the random variables. Based on experience, it was decided to consider as random the entry strip crown, the work roll crown, the strip compressive yield stress, and the elastic modulus of the work rolls. Additional assumptions included relational independence between the random variables and probability density functions with normal (Gaussian) distributions. Although the statistical information used was based on best estimates only, this work may motivate a new direction in application of reliability analysis technology to the steel industry. Despite the estimations made, it may be concluded that the analysis methods presented here illustrate one approach to identify the likely failure modes for the physical flatness quality of rolled strip. Equipped with the manufacturing resources to allow a detailed study, engineers in the rolling industry may gain new insight into how to improve the quality of rolled steel by addressing the randomness of troublesome. Finally, evaluation of the reliability of strip exit crowns may lead to more refined mill design and control capabilities. Mill designers may be able to quantitatively account for the randomness of specific parameters at a customer's manufacturing plant and tailor mill designs accordingly. By performing reliability-based roll profile optimization with grinding performance .data from the customer, shorter mill commissioning times may be seen New real-time control philosophies, such as reliability-based pass schedule optimization and reliability-based flatness control with statistical target flatness validation may improve productivity and provide quantifiable measures on the likelihood of meeting performance guarantees.