Automatic image processing based size characterization of green pellets

Abstract

The size of pellets has a significant effect on the performance of blast furnaces. Based on experience, the universally accepted size of pellets for efficient blast furnace operations is between 9 millimeters to 16 millimeters. But presence of smaller size pellets or fines lowers the blast furnace stack permeability, increases dust losses and may lower the maximum permissible blast temperature for smooth operation of the furnace. Pellets too big are also undesirable, particularly if its reducibility is low and is poor in strength, thus undergoing physical degradation during furnace operation. This paper relates to a non-contact method of measuring the size of green pellets being fed into the Blast furnace. A method is developed for the automatic size characterization of green pellets in the conveyor belts before the pellets are being indurated in a travelling grate furnace to confer the required physical and metallurgical properties. The method employs imaging and illumination devices installed in the conveyor belt in the pellet plant of Tata Steel and then, processing those images using imaging algorithms to obtain the size distribution. Apart from the size distribution of the green pellets, the system also identifies the pelletizing disc from which the green balls originated so as to effectively control the pelletizing process.