Technical Market Support » Metallurgical Coal
Understanding of coke strength and its dependence on original coal blend parameters is one of the major technical marketing support tasks required to improve prediction of the performance of coals and blends during coking.
This project applied novel structural identification and characterisation techniques to high resolution optical photomicrographs of coke with the aim of understanding the dependence between coke strength, its structural characteristics and the parent coal blend. Automated segmentation of Inert Maceral Derived Components (IMDC) and Reactive Maceral Derived Components (RMDC) allowed measurement of a broad range of parameters characterising IMDC. This characterisation included IMDC boundaries and parameters characterising connections between IMDC and RMDC. It also included measurements of IMDC porosity which relates to IMDC fusibility.
Two new approaches using image analysis software have been investigated. In the first case, separation of large porosity objects into individual pores has been performed (Separated Pores), while in the second approach, weak walls have been removed and hence potential Weak Areas have been identified and measured.
After the reactive part of the coke is fluidised, part of it coalesces into structures relatively thicker than walls, which can be described as "nodes". When the standard wall thickness is measured, all parts of the matrix, including not only walls but also nodes, are taken into account. In this work, nodes and walls were segmented separately from each other, which allowed estimation of the effect of nodes and walls on coke strength as distinct variables. This segmentation also allowed measurement of Modified Wall Thickness when only actual walls were considered. Finally, the narrowest sections of each wall, ie, the "necks", have been identified, allowing estimation of average neck thickness together with Specific Wall Neck Thickness, ie, the total thickness of all necks per unit area. This approach emphasised the fact that, for coke strength, not only the thickness of the walls is important, but also the overall amount of walls (and necks) per unit area.
Altogether more than 250 parameters characterising coke structure were calculated and studied. Calculations showed that all new concepts and approaches provide useful information and give a better understanding of the connection between coke structure and coke strength indices.
The relationships between coal and coke parameters (strength indices, etc. measured on each coke and supplied by individual companies), as well as between coke parameters, were also studied. Some of the coke parameters showed good correlation with coal parameters and with each other. At the same time, some of the coke parameters were found to have quite a poor correlation with each other, even when both parameters were representing coke strength.
A huge set of data was produced during this project and, while the authors put most of their effort into developing the new processing methodologies, performing the actual data processing for the cokes, and providing conclusions based on this processing within the project scope, more information about the relationships between coke parameters and coke structure can certainly be extracted from the images and measured data with further effort. The project produced a significant amount of new information, which can be attributed to several novel approaches used during this project, including averaging of traditional descriptors by object area rather than number, automated segmentation of IMDC and RMDC, identification and measurement of IMDC boundary areas, separating pores and identifying Weak Areas, subdivision of the coke matrix into Nodes and Walls, and calculation of Neck Thickness and Specific Neck Thickness.
High correlation coefficients between coke strength indexes, coal characteristics and calculated parameters characterising coke texture, as well as high R-square values obtained in the modelling of coke strength indices, confirm the correctness of the chosen methodology and usefulness of the novel procedures for description of coke structure.
It is strongly recommended that more cokes should be incorporated in the models in order to explain the behaviour of outliers (Cokes 10 and 11), extend the range of parameters modelled, and develop more robust modelling and calculations (possibly non-linear).