Research of joint work of titanium sponge plots by a model

The necessity of increasing the efficiency of production management systems of titanium sponge by the method of Krol is justified. A system analysis of the joint operation of magnesium electrolysis stage and the reduction of titanium sponge stage is performed. The task to eliminate violations in the provision of the reduction stage with bottom buckets, leading to a decrease in production efficiency, is given. To study the possibility of solving the problem, a software library of simulation and multi-agent modeling is developed. A model of sponge production is created with the library. The model is equipped with measurements on location, the statistical distributions is assumed to be uniform due to insufficient data. A study of the influence of a number of controllable factors on the performance of the plots, the waiting time of bottom buckets in the queue for magnesium chloride discharge and the percentage of violations in the provision of the restoration site with bottom buckets is made with the software model. According to the results of the first part of the research, most of the factors were discarded as insignificant regarding to their influence on the result. Further, the joint influence of such factors as the number of bottom buckets, the discharge time of magnesium chloride from the titanium sponge recovery apparatus, and the transport time of the bottom bucket to the magnesium electrolysis section is investigated. It is found out that a joint impact of factors produces a synergistic effect which significantly exceeds the model’s response to independent changes. In the first approximation, limits of the factors to provide process violations exception are determined. It is shown that the refinement of statistical distributions can significantly affect to the simulation results. As a conclusion the qualitative results of the study can be considered adequate, but to clarify the quantitative results, more measurements and more accurate statistical distributions of random variables are needed.