Magnitogorsk Iron and Steel Works (MMK) saved 500.9 million rubles in 2018 through the use of a mathematical model to optimize the delivery and consumption of coal. These savings were achieved by improving the structure of coal charge and the quality of coke.
This model is part of MMK’s large-scale digitalization strategy. The program is based on the creation of a comprehensive multi-level model of sinter production, designed to optimize the production processes of the first processing stage.
The optimization model for the delivery and consumption of coal raw materials at MMK has been in operation since February 2017. MMK developed the model themselves, based on the study of coke and blast furnace production at the site since 2011. Implementation and development of the model is carried out at MMK’s science and technology center, by a group of mathematical modeling and system-analytical researchers. The model optimizes the process both for the technologists and for the delivery specialists. The goal is to obtain the required quality of coke at the minimum cost.
To date, the model is used for purchasing coal concentrates and integrating the knowledge and requirements of all MMK services involved in this process: this includes category management of delivery (owner model), the science and technology center, coke production and economy management. Monthly calculations are made as part of the process of purchasing coal concentrates:
– a forecast quality of coke when given the structure and volume of raw coal;
– optimization of the structure of the delivery and consumption of coal charge;
– analysis of the sensitivity of the optimization calculation to prices and calculation of the recommended effective prices for the purchase of coal concentrates.
All calculations are performed in the multi-user web interface – AIS “Model of optimization of delivery and consumption of coal raw materials”.
This is not the first successful mathematical modeling project to optimize production processes of the first stage of conversion (the contribution of the first stage to the cost of production is about 50%). The optimization mathematical model for delivery and consumption of iron ore raw materials is being developed and the automated information system “optimum cast iron” is being created. Big data technologies and artificial neural networks are used in development, but traditional physical and chemical models are also used, in particular, to take into account the dynamic characteristics of the domain process. (MMK/Ukrainian metal)