Development Of Model And Algorithm For Hot Strip Mill Scheduling

During past few decades, with the global competition the steel manufacturing enterprises are faced with the challenges to reducing operational costs and making the business more flexible and dynamic under an uncertain worldwide marketplace. It has been recognized that the advanced planning and scheduling (APS) plays an important role in making a seamless integration of business and manufacturing to optimize business performance with low costs and superior customer services. The"make-to-order"business model has been popularly used in many modern steel mills. It should be noted that the hot strip mill (HSM) is one of the most important production lines to produce hot rolled strip products in either an integrated mill or a mini-mill. From system point of view, the HSM does not only produce its finished hot rolled products, but also build up the work-in-process (WIP) inventory for variety of cold mills. Consequently, producing right products at right time with lowest costs becomes the general goal of a HSM production unit. Being a constrained combinatorial optimization the HSM production scheduling problem is a typical NP-hard problem.After investigation of HSM scheduling problem, a multiple round scheduling model is proposed to describe both objective criteria and constraints commonly concerned in production scale HSM scheduling. To solve the relevant optimal scheduling problem, a hybrid genetic algorithm (GA) is proposed to build up a HSM campaign scheduling scenario with multiple scheduling rounds.In addition to the campaign building, this thesis also proposes a dynamic scheduling (or rescheduling) solution with uncertain changes in HSM business and/or manufacturing environments, for example, the changes in manufacturing orders and unexpected machine breakdown, etc.A simulation software prototype coded with VC++ was developed based on scheduling model and hybrid GA. The simulation results with benchmark data show that the proposed solutions may provide reasonable solutions for both multi-round campaign building and dynamic rescheduling.