• Static and dynamic recrystallization and grain growth, as well as precipitation. Learning procedure could be designed and optimized. In this aspect the semi-empirical model may also be used.

• Retained strain, as mentioned earlier. Theoretical and semi-empirical approaches with microstructure simulation would be particularly carried out. Empirical models collected in the past could be evaluated and combined with the microstructure simulation.

• Microstructure-affected flow stress. Studies would also cover the initial grain size effect and the prediction of the flow stress based on the chemical composition, etc.

• The hold and the resume pass. Phenomenon involved in the hold and resume pass would be further studied. Microstructure simulation and flow stress modeling would be carried out for the hold period.

• Microstructure evolution and finish steel properties based on the given draft-schedule, to verify that various requirements (shape, properties, etc.) are satisfied. Option to use various algorithms (both linear and nonlinear ones) could be provided.

The newly developed Level 2 model may include following modules

• Rolling Process Models, to perform model calculations for, such as force, temperature, roll flattening, roll deflection, thermal crown, roll wear, steel deformation. It should call the metallurgical modules to determine microstructure, retained strain and flow stress, etc.

• Metallurgical Models, to determine retained strain, grain size, rolled steel properties, etc., by integrating with intelligent learning such as neural network, fuzzy logic and expert system.

• Expert System, which consists of logics, data and influence factors on the mechanical, thermal and metallurgical parameters depending on rolling and thermal processes. Past mill-experiences could be programmed as a portion of the expert system.

• System Learning, based on rolling process models, in which the neural network provides the correcting factors for the model coefficients, and fuzzy logic rules and expert system provide guidelines (upper and lower boundaries, etc.) for the learning.

• Draft Scheduling, which is based on various requirements (shape, properties, etc.) and various algorithms, with special attention paid to nonlinear algorithms. Microstructure and finish properties will be predicted for every newly generated pass schedule.