Research On Group Design And Optimum Utilization Of Steel Production Material

In order to gain advantage in a competitive market environment, iron and steel enterprises need adapt to the diverse needs of the market. The diversity of the steel products in both steel-grade and specification will lead to the difficulty in batching the steel process, which will increase difficulty in organizing production, and thus will inevitably lead to low utilization rate of production equipment and low lumber recovery of materials. On the other hand, rising prices of iron ore and other raw material exacerbates the burden of production costs of the enterprises. Therefore, in order to play a role in mass production of large-scale production equipment and reduce material waste in steel production, iron and steel enterprises, on the one hand, need to make a production design for grouping orders with similar specification requirements, and on the other hand, need to make a reasonable optimum reutilization for surplus materials.In this dissertation, the steel production started from the steelmaking area to the hot rolling area is taken as research background. Two typical production material group design problems (i.e. master-plate design problem, steel-grade consolidation and assortment problem) are addressed. These two problems are respectively formulated as mathematical programming models, and solved by developing the corresponding hybrid metaheuristics. Besides, the lower bounds to the optimal solutions of the corresponding problems are constructed to test the performance of the developed algorithms. One typical production material optimum utilization problem (i.e. inventory slab allocation problem) is addressed. This problem is respectively formulated as a static (and dynamic) mathematical programming model. For the dynamic problem, a column-generation based branch-and-price algorithm is developed to solve the medium-size instances. For the static problem, a hybrid metaheuristic is developed to approximately solve the large-size instances, and a lower bound to the problem’s optimum is developed to evaluate the performance of the developed algorithm. A decision support system that incorporates the mathematical programming model and algorithm is developed and applied to the practical production in an iron and steel enterprise in China. The content is summarized as follows. (1) The task of the problem is to pack customer rectangle order-plates into master-plates under consideration of satisfying guillotine cuts, no rotation and no overlap constraints. The aim of the problem is to minimize the material waste and maximize the utilization rate of designing master-plate. This problem is formulated as a nonlinear mathematical programming model. A hybrid metaheuristic, which combines scatter search and tabu search, is developed to obtain the near-optimal solution. Besides, a bounding method based on split-recompose policy is proposed in order to verify the performance of the developed algorithm. Computational results show that the developed hybrid metaheuristic can effectively solve the problem.(2) The static steel-grade consolidation and assortment problem is to determine the steel-grade number and category in order to satisfy all production orders according to order quality reqirements and steel-grade specifications. The aim of the problem is to minimize the total production costs which are related to the quality specification distinctness between order and steel-grade, the use of steel-grade, and the penalty for unsatisfied orders. The problem is formulated as an integer programming model. A hybrid metaheuristic that combines scatter search and variable neighborhood search is developed for solving this problem. And a lower bound to the problem optimality is obtained by developing Lagrangian relaxation in order to evaluate the performance of the developed algorithm. Computational results show that he developed hybrid metaheuristic can quickly obtain the near-optimum solution (the errors between the near-optimum solution and lower bounds obtained in all the instances are less than5%).(3) The difference between the dynamic problem and the static problem is that the dynamic problem is the steel-grade consolidation and assortment problem with multi-periond consideration, while only one period is considered in the static problem. This problem is formulated as an integer programming model in which the objectives include minimizing the cost for assigning steel-grades to orders, the setup cost for utilizing new steel-grades, and the other related penalty cost. A hybrid metaheuristic that combines scatter search and filter-and-fan is developed for solving this problem. Computational results obtained by comparing the CPLEX solver with the developed algorithm show that the developed hybrid metaheuristic can obtain the satisfactory near-optimum solution, and consume less computational time than CPLEX.(4) The static inventory slab allocation problem is to allocate inventory slabs to the orders by considering the customer requirements for slabs and the properties of the inventory slabs in storage. The aim of the problem is to maximize material utilization rate, materila weight, and customer satisfaction. The problem is formulated as a mix-integer programming model, and a hybrid metaheuristic which combines scatter search that performs as a diversification mechanism and variable depth search that provides intensification for further exploration is developed for solving this problem. Moreover, two improvement strategies are used to enhance the solution quality and one speed-up policy is used to improve the solution efficiency. A column-generation based lower bound is provided to evaluate the performance of the hybrid metaheuristic. Computational experiments on real-world data collected from a large steel company show that the hybrid metaheuristic can approximately solve large-size instances, and the solution is better than the one obtained by human planner.(5) The difference between the dynamic problem and the static problem is that the dynamic problem is the inventory slab allocation problem with multi-periond consideration, while only one period is considered in the static problem. The aim of the dynamic inventory slab allocation problem is to plan the production for inventory slabs so that the total cost of allocation and inventory holding cost are simultaneity minimized. This problem is formulated as a0-1integer problem, and a branch-and-price algorithm is first designed for solving the problem. In this algorithm, a linear relaxation based heuristic is developed to generate feasible solution to this problem. Computational results show that the branch-and-price algorithm is capable of generating the optimum solution to the medium-sized problem.(6) In order to apply the obtained research results, an inventory slab allocation decision support system that combines the established integer programming model, a developed hybrid metaheuristic, and human-machine interaction technique is developed with taking inventory slab yard in one lagre iron and steel enterprise as background. This system with graphical human-machine interaction interface is provided, and planner can set the choose condition according to the practical production environment. The system enable automatic transferring and allocating inventory slabs to orders, and upload the plann automatically. The system also achieves the planning result evaluation, process parameter settings, and maintenance functions. This system is now available with a domestic iron and steel enterprise production planning management system to achieve complete docking, and has been online running in the enterprise.