Acceleration of simulation models for raw materials thermal treatment

Abstract

Parallel data processing belongs in the present time to the basic approaches. Its realization is possible by using of multi-core processors or we can use new trend with graphic accelerators on the new type of graphic cards. However this process is not straightforward and it requires an adequate model structure and application program parallelization. Adequate model structure includes not only parallelization of computing operations but also creates convenient application surroundings. Application of exploited approaches of developed mathematical model for granular materials thermal treatment includes model generation for the setting of all parameters and parallelization mode according to the selected thermal apparatus. According to the model complexity, adequate parallelization platform has to be selected. Presented application is on the rotary furnace. Original mathematical model was modified by using the graphical accelerators. Calculation parts of the model were divided into the particular computing kernels of the accelerator. At executed experiments computing time was reduced from almost three hours to less than five minutes. This approach enables us to use the simulation models in the real time.

Simulation Analysis of Preheater Charge to the Rotary Furnace

“…The model used the replacement partial models, where resolving power is given by size of elementary particles and by the option of partial processes models. For this reason the thermal apparatus is decomposed into zones and elements [6,7]. A high quantity of simulation alternatives can be created by arranging the zones and the parameters, or the media flow directions.…”