Blast furnace operation relies heavily on coke quality and therefore the accurate characterisation of coke can lead to the stable operation of a very expensive asset. The use of petrographic analysis to identify morphologies in metallurgical coke can indicate several key parameters related to its performance in blast furnaces. These include reactivity, which is an important factor for determining the coke’s residence time in the furnace and the formation of the hearth layer; and strength, which is indicative of the coke’s ability to withstand the weight of the ferrous burden as it reduces in the furnace.

Standard reactivity and strength testing (such as CSR, CRI, Irsid or Micum tests) can take up to 6 hours to perform, while petrographic analysis can be conducted in under an hour. Furthermore, coke petrography can highlight key operational elements of coke-making. Automating the analysis of petrographic analysis can help further speed up characterisation and the communication of these results to lay audiences which, in turn, can help translate these results into operational indicators far quicker. Additionally, a novel technique, Visual Reactivity Analysis (VRA), can be used to model the thermal properties of coke along with more conventional ThermoGravimetric Analysis (TGA) techniques. VRA allows the observation of coke combustion in real time. The change in shape of the sample can be quantified using image analysis techniques and used to infer its reactivity at any given time, thereby granting access to a wealth of information on how coke combusts over the complete range of temperatures and atmospheres during the experiment. A combination of these three techniques can thus be deployed to gain invaluable information on the properties of coke.

Keywords: Petrography, Image Processing, Image Classification, Metallurgical Coke

Acknowledgement: This research is funded and supported by Engineering and Physical Sciences Research Council (EPSRC), Tata Steel Strip Products UK (TSSPUK) and the University of Nottingham. The project is an Engineering Doctorate at the Centre for Doctoral Training in Carbon Capture and Storage and Cleaner Fossil Energy at the University of Nottingham, grant number EP/L016362/1. The authors would like to thank EPSRC, Tata Steel Strip Products UK (TSSPUK) and the Centre for Doctoral Training in Carbon Capture and Storage and Cleaner Fossil Energy for financial support and guidance.