Modeling of iron ore pelletization using 3 ** (k–p) factorial design of experiments and polynomial surface regression methodology

Abstract

The process of size enlargement or agglomeration finds a variety of applications in material processing and utilization. Pelletization is one such process which uses water as medium and revolving units (Disc/Drums) to form spherical pellets from fine particulates. Green or wet pelletization is the first step in pelletization process and is of critical importance since the effectiveness of the subsequent stages of drying and in duration depends on the quality and quantity of green pellets. This research presents the work carried out to develop model equations to predict the size distribution of pellets at any given level of intervals. Modeling of pelletization was oriented towards predicting the size distribution of pellets at any given level of variables .The prediction of pellet size distribution involves quantification of the self preserving curve and correlation between D 50 and the variables. A new model has been developed to predict the size distribution of the pellets using advanced statistical software “STATISTICA”. The equation Y = − 0.3757X 2 + 1.6256X-0.74 where “Y” is the cumulative wt.% passing and “X” is D/D 50 was used to predict the pellet size distribution. Correlation between D 50 and variables was given by “D 50” = 4.226 + (3.106*M) − (0.544*M 2) + (2.044*I) − (0.644*I 2) + (0.2444*T) − (0.028*T 2) − (0.058*M*I) + (0.0917*I*T) using quadratic response surface methodology. The mean pellet diameter “D 50” observed versus predicted was compared. A polynomial regression equation was used to quantify the characteristic curve of iron ore slimes agglomeration process. This can be utilized to predict the complete agglomerate size distribution irrespective of the operating conditions and the size of the pelletizer if a relationship such as agglomerate median product size D 50, as a function of the operating conditions is made available.