Study On Modeling And Optimization Algorithm For Slab Design Problem In Iron And Steel Industry

In recent years, with the increasing competition in the steel industry, every iron and steel enterprise is improving core competence and reducing cost. The production requirements are becoming more and more low-volume and high-variety, which contradicts the mass production in the steel industry. Production management optimization is the principle method to solve this problem. As one of the important parts in production management, slab design problem refers to that planner organizes multiple orders with the same or similar specifications producing on one slab, of which the objectives are minimizing surplus and maximizing furnace efficiency. Therefore, studying slab design problem is of great significance for iron and steel enterprises to save energy and reduce cost.According to the source of slab, there are two types of slab design problem:First, iron and steel enterprises use their own made slabs, of which the length is adjustable, to production and processing. Second, enterprises buy slabs, of which the specifications are limited, to production and processing. The second type is investigated in this paper. This problem is originated from real production, and it is widely spread in the steel industry. It is a multi-objective combinatorial optimization problem, and there is no algorithm that is able to solve this problem within a short time. This paper studies slab design problem, which aims to meet the need on site.With the characters of slab design problem, this paper builds a multi-objective nonlinear mix-integer programming model, of which the objects are minimizing surplus and maximizing furnace efficiency. For the sake of solving this problem, the programming model is transformed into a multi-objective linear mix-integer programming one, of which the constraints are that the total order length and unit length are both interval values. Due to the large scale and real-time, a hybrid method combined with heuristic and MILP is proposed in this paper. There are three phases in this method:The first phase is to build integer programming to find the best combinations of every single order or between every two orders, of which the objective is only minimizing surplus. The second phase is to change the combinations mentioned, of which the objective is to maximize the furnace efficiency. The third phase is to build a mix-integer linear programming model based on the combinations. This method makes use of the flexibility and rapidity of heuristic. It decomposes the problems into three small ones, and optimizes the two objectives in different phases. The method also makes use of the solution accuracy from mix-integer linear programming. Computational experiments with practical data demonstrate that this hybrid method is able to find a near optimal solution within an acceptable time, which not only meets the need of real-time production, but also guarantees the quality of solution.Based on the hybrid method, slab design optimization system is developed, which is under testing in a enterprise. The result shows that this system is able to effectively reduce the surplus and improve the furnace efficiency. Moreover, this system also brings down the labor intensity and largely reduces the cost.