USINGTILIZING ARTIFICIAL NEURAL NETWORK MODELINGFORTO OPTIMIZEING HPGR OPERATION IN A CLOSED CIRCUIT FOR PREPARING IRON ORE PELLETIZING PLANT FEED
Leila Hosseini*, Mohsen Zare, Hamed Abedini, AND Rasoul Hejazi
Fakoor Sanat Tehran Company
(*Corresponding author: )
The high pressure grinding roll (HPGR) is nowadays presentlyalso being used for iron pellet feed preparation as the regrinding stage of the iron concentrate in order to increase the Blainevalue of the iron concentrate. In line 5 of Gol-E-Gohar Iron Concentrate Plant, final concentrate dewatered by vacuum belt filters has aBlaine value of 1,150 cm2/gr. In order to increase the blianeBlaine value of the concentrate to 1,850 cm2/g—r, suitable for pelletizing process, —this concentrate is reground in an HPGR.In order to achieve this final blaine value, cConcentrate ground in the center area of the HPGR rolls (approx. 40% of the HPGR whole product) is considered as the final product and the part ground in the edges of the HPGR (approx. 60% of the HPGR whole product) is recycled back to the HPGRSThe studies have revealed that two group of factors influence the HPGR`s performance:; operating parameters of the HPGR itself and qualitative and quantitative specifications of the feed. In this study, samples were taken from the HPGR feed and product at fixed time intervals and operating parameters of the HPGR at the sampling times were recorded. Results showed that level of the feed hopper, power of the HPGR, rolls pressure and rollsrRpm are the most influential operating parameters of the HPGR, whereasileblaineBlainenumber value and moisture of the dewatered concentrate and amount of the recycled productare the most effective qualitative factors. It has also been revealed that the feed qualitative and quantitative specifications affect the operating parameters of the HPGR as well. In order to optimize these parameters and minimize the fluctuations in the feed specifications, the system has been modeled by using an artificial neural network. The results indicated that the proposed model can accurately estimate the effect of operating and qualitative parameters on the final blaineBlainenumbervalue.
KEYWORDS
HPGR, Pelletizing, Pellet Feed Regrinding, Blaine Value, Artificial Neural Network