The ore quality at mining faces in the KGHM underground copper ore mines can be determined based on channel samples and block models built in the Datamine system. Unfortunately, even very accurate information regarding the quality of deposits at mining faces does not translate into the possibility of predicting the composition of feed to enrichment plants. The mixed ore from mining faces is cyclically loaded by trucks onto belt conveyors, which in turn convey it to shafts. When transported on the conveyors, mixed ore portions from many loading points form a divisional stream, which then goes to the main haulage conveyors where ore streams from various divisions are combined. The way of filling the bunkers, as well as ore flow temporary stoppages, changes the sequence of ore mixing and its arrival, which hinders the ability to track its quality. In the current study, radio-frequency identification (RFID) was proposed for tracking ore composition. A complementary method of ore quality prediction comes from simulating the tagged ore flow within the FlexSim software package. The discrete element method (DEM) of simulations, verified by experiments with RFID tags, can determine the ore flow through bunkers. Forecasts of ore feed composition for the next shift can be prepared with actual plans of mining division operations, the filling level of bunkers and the work plan of the transport system.