5. plate mill application Examples
5.2 Nanjing Iron & Steel Co. (NISCO)
Primary purpose for the Level 2 model improvement is for the optimal draft schedule in order to achieve better plate shape and steel properties. NISCO also uses Tippins Level 2 model.
The improvements performed for NISCO include all those done in the EOS project. In the following section, work beyond EOS scope is introduced . Even for the same improvement, the NISCO project overwhelmingly uses automatic programming in the data design, different from EOS case, so that the similar project in the future can be performed much more efficiently.
Level 2 grades were designed based on several hundred existing steel grades. A very comprehensive model grade specification was established based on NISCO’s current and future needs, in which every Level 2 grade may consist of up to 180 model grades. This is under the practice that most steel grades are currently rolled with the controlled rolling. In addition, grade family category is also redesigned, based on chemical composition. Before this project, NISCO classified its grade and grade family based on the purpose of steel application and mechanical property, so many different steel grades had quite similar chemical compositions. From technical aspect, they should belong to the same grade though they are for different applications. To some extent, the steel grade consolidation and the grade family design have stimulated the technical upgrade in NISCO, and the model grade design has paved the way for several metallurgical modeling projects in the future.
Automatic Design of
Flow Stress Coefficients
With the application of the Guided Two-Parameter Learning, large number of flow stress coefficients for the model grades in each of the three temperature regions were designed.
Due to the highly diversified product mix and high requirements for controlled rolling, there were about 6500 model grades and accordingly about 20,000 sets of flow stress coefficients C1, C2, C3, and C4. In this case, manual or semi-manual design, as in the past projects, cannot meet the requirement. Automatic design, by applying a long list of existing and newly created models and data, was performed.
Other Physical/Mechanical Data and File Generation
Because the Level 2 grades and grade families are all newly designed based on totally different criteria, physical and mechanical data were also newly created.
Both the family files (each for one grade family) and grade files (each for a model grade) were regenerated, and all the data in each of those files were adjusted or recalculated. With the large number of grade files, they had to be created programmatically. Creation of grade files was done by coping format from old files and replacing the old data with new ones. The newly designed data were stored in the database table, for the program to access.
Resume Pass Model Improvement
During the hold, there are metallurgical processes incurred inside steel, so in the pass right after the hold, the so-called resume pass, there is significant force error.
Model modification was made for the resume pass.
Improvements on Draft Scheduling Logics and others
With the progress of the improvements, several more weaknesses were identified. One of them is the draft scheduling logics. NISCO did some improvements in user interface, to allow operators to handle many issues; but as to the logic improvement such as that on draft scheduling and shape control, still no sufficient work was done. Typical weakness exists in, that when force prediction is getting more accurate, the number of passes tends to increase, and the draft in the second last and/or third last pass sometimes is too high. Existing draft schedule logics prefer the case with lower predicted force, while the new force model is very accurate, with errors only about one half of the error incurred in the old model, and overwhelmingly below 5%.
In the current Tippins package, the logics for draft scheduling based on predicted force, consists of three steps, taking finishing pass logics as example:
- initial design of the draft schedule;
- established equation system to examine whether a list of constrains are met; and
- if one or more constrains cannot be satisfied, creating a schedule based on lowered requirement.
Recent diagnoses indicate that modifications to all the three steps are necessary. Some existing logics have been modified, such as in the step (1) and (3) above in certain situation, the initially assumed equal draft rule has been eliminated. Several further changes are
made, so the draft schedule are in high quality.
Further improvements included the continued shape control, temperature improvement connected with the Frits, and product head end and tail end shapes, as well as the continued application of metallurgical models in the Level 2 model.
Shape control through Level 2 is of very high economic benefit. Certain improvement would lead to an annual return of a million US dollar.